PM Tech: Document Title

π PM Tech PMA212816C
Document Title
4Gb (256M x 16) DDR4 SDRAM C Die Datasheet
This document is a general product description and subject to change without notice.
4GBIT DDR4 DRAM
Key Features
 VDD=VDDQ=1.2V +/- 0.06V (1.14V~1.26V)  Programmable data strobe preambles
 VPP = 2.5V, -0.125V/+0.25V  Data strobe preamble training
 On-die, internal, adjustable VREFDQ generation  Driver strength selected by MRS
 1.2V pseudo open-drain I/O  Dynamic On Die Termination
 16 internal banks (x4, x8): 4 groups of 4 banks  Two Termination States such as RTT_PARK and
each RTT_NOM switchable by ODT pin
 8 internal banks (x16): 2 groups of 4 banks each  Asynchronous RESET pin
 8n-bit prefetch architecture  Internal (self) calibration: Internal self calibration
 Fully differential clock inputs (CK, CK) operation through ZQ pin (RZQ: 240 ohm ± 1%)
 Differential Data Strobe (DQS, DQS)  TDQS (Termination Data Strobe) supported (x8
 On chip DLL align DQ, DQS and DQS transition only)
with CK transition  Write and read leveling
 DM masks write data-in at the both rising and  Databus write cyclic redundancy check (CRC)
falling edges of the data strobe  Maximum power saving mode
 All addresses and control inputs except data, data  Temperature controlled refresh (TCR)
strobes and data masks latched on the rising  Low-power auto self refresh (LPASR)
edges of the clock  Fine granularity refresh
 Programmable CAS latency 9, 10, 11, 12, 13, 14,  Per-DRAM addressability
15, 16, 17, 18, 19 and 20 supported  Geardown mode
 Programmable additive latency 0, CL-1, and CL-2  Self refresh abort
supported (x4/x8 only)  Command/Address (CA) parity
 Programmable CAS Write latency (CWL) = 9, 10,  Data bus inversion (DBI) for data bus
11, 12, 14, 16, 18  Connectivity Test Mode (x16)
 Programmable burst length 4/8 with both nibble  PPR is supported
sequential and interleave mode  JEDEC standard package
 BL switch on the fly - 78ball FBGA(x4/8)
 Average Refresh Cycle (Tcase of 0 oC~ 95oC) - 96ball FBGA(x16)
- 7.8 μs at 0oC ~ 85°C  Lead free & RoHS compliant
- 3.9 μs at 85oC ~ 95°C  JEDEC JESD-79-4 compliant
1600 1866 2133 2400 2666 2933 3200

Speed Unit
11-11-11 13-13-13 15-15-15 17-17-17 19-19-19 20-20-20 22-22-22
tCK(min) 1.25 1.071 0.937 0.833 0.75 0.682 0.625 ns
CAS Latency 11 13 15 17 19 20 22 nCK
tRCD(min) 13.75 13.92 14.06 14.16 14.25 13.64 13.75 ns
tRP(min) 13.75 13.92 14.06 14.16 14.25 13.64 13.75 ns
tRAS(min) 35 34 33 32 32 32 32 ns
tRC(min) 48.75 47.92 47.06 46.16 46.25 45.64 45.75 ns
Rev 1.1 1 September 2020

π PM Tech
Addressing
PMA212816C
Organization 1024Mbx4 512Mb x 8 256Mb x 16
Bank group 4 4 2
Bank group address BG[1:0] BG[1:0] BG0
Bank per group 4 4 4
Bank address BA[1:0] BA[1:0] BA[1:0]
Row address 64K (A[15:0]) 32K (A[14:0]) 32K (A[14:0])
Column address 1K (A[9:0]) 1K (A[9:0]) 1K (A[9:0])
Page Size 512B 1KB 2KB
tREFI Tc≦ 85°C:7.8μs, Tc>85°C:3.9μs
tRFC 260ns
Package
4Gb Dimension Ball pitch

(Org. / Package) (mm) (mm)
256Mbx16 96-ball FBGA 7.5 x 13 0.80

Ordering Information
Speed
Organization Part No. Package
Clock (MHz) Data Rate (Mb/s) CL-TRCD-TRP
Commercial Grade
96-ball
256Mx16 PMA212816CBR-SEDN 1333 DDR4-2666 19-19-19
BGA
Industrial Grade
96-ball
256Mx16 PMA212816CBR-SEIN 1333 DDR4-2666 19-19-19
BGA

Package Ballout and Addressing

x4 Package Ballout (Top view) : 78ball FBGA Package
1 2 3 4 5 6 7 8 9
DM_n
A VDD VSSQ TDQS_c DBI_n VSSQ VSS A
TDQS_t
B VPP VDDQ DQS_c DQ1 VDDQ ZQ B
C VDDQ DQ0 DQS_t VDD VSS VDDQ C
D VSSQ NC DQ2 DQ3 NC VSSQ D
E VSS VDDQ NC NC VDDQ VSS E
F VDD NC ODT CK_t CK_c VDD F
G VSS NC CKE CS_n NC TEN G

WE_n CAS_n RAS_n
H VDD ACT_n VSS H
A14 A15 A16
A10 A12
J VREFCA BG0 BG1 VDD J
AP BC_n
K VSS BA0 A4 A3 BA1 VSS K
L RESET_n A6 A0 A1 A5 ALERT_n L
M VDD A8 A2 A9 A7 VPP M
N VSS A11 PAR NC A13 VDD N

π PM Tech
PMA212816C
1 2 3 4 5 6 7 8 9
DM_n
A VDD VSSQ TDQS_c DBI_n VSSQ VSS A
TDQS_t
B VPP VDDQ DQS_c DQ1 VDDQ ZQ B
C VDDQ DQ0 DQS_t VDD VSS VDDQ C
D VSSQ DQ4 DQ2 DQ3 DQ5 VSSQ D
E VSS VDDQ DQ6 DQ7 VDDQ VSS E
F VDD NC ODT CK_t CK_c VDD F
G VSS NC CKE CS_n NC TEN G

WE_n CAS_n RAS_n
H VDD ACT_n VSS H
A14 A15 A16
A10 A12
J VREFCA BG0 BG1 VDD J
AP BC_n
K VSS BA0 A4 A3 BA1 VSS K
L RESET_n A6 A0 A1 A5 ALERT_n L
M VDD A8 A2 A9 A7 VPP M
N VSS A11 PAR NC A13 VDD N

π PM Tech
PMA212816C
1 2 3 4 5 6 7 8 9
A VDDQ VSSQ DQU0 DQSU_c VSSQ VDDQ A
B VPP VSS VDD DQSU_t DQ9 VDD B
C VDDQ DQ12 DQ10 DQ11 DQ13 VSSQ C
D VDD VSSQ DQU6 DQ15 VSSQ VDDQ D

UDM_n/ LDM_n/
E VSS VSSQ VSSQ VSS E
UDBI_n LDBI_n
F VSSQ VDDQ DQSL_c DQ1 VDDQ ZQ F
G VDDQ DQ0 DQSL_t VDD VSS VDDQ G
H VSSQ DQ4 DQ2 DQ3 DQ5 VSSQ H

VDD VDDQ DQ6 DQ7 VDDQ VDD J
J
K VSS CKE ODT CK_t CK_c VSS K
WE_n RAS_n
L VDD ACT_n CS_n VDD L
A14 A16
A12 CAS_n
M VREFCA BG0 A10/AP VSS M
BC_n A15
N VSS BA0 A4 A3 BA1 TEN N
P RESET_n A6 A0 A1 A5 ALERT_n P
R VDD A8 A2 A9 A7 VPP R
T VSS A11 PAR NC A13 VDD T

Package Dimensions
x16 96ball Package Outline Drawing

π PM Tech
Input/Output Functional Description
PMA212816C
Symbol Type Function
Clock: CK_t and CK_c are differential clock inputs. All address and control input
CK_t, CK_c Input signals are sampled on the crossing of the positive edge of CK_t and negative
edge of CK_c.
Clock Enable: CKE HIGH activates, and CKE Low deactivates, internal clock
signals and device input buffers and output drivers. Taking CKE Low provides
Precharge Power-Down and Self-Refresh operation (all banks idle), or Active
Power-Down (row Active in any bank). CKE is asynchronous for Self-Refresh exit.
CKE Input After VREFCA and VREFDQ have become stable during the power on and
initialization sequence, they must be maintained during all operations (including
Self-Refresh). CKE must be maintained high throughout read and write accesses.
Input buffers, excluding CK, CK_c, ODT and CKE, are disabled during power-
down. Input buffers, excluding CKE, are disabled during Self-Refresh.
Chip Select: All commands are masked when CS_n is registered HIGH. CS_n
CS_n Input provides for external Rank selection on systems with multiple Ranks. CS_n is
considered part of the command code.
Chip ID: These inputs are used only when devices are stacked; that is, they are
C0,C1,C2 used in 2H, 4H, and 8H stacks for x4 and x8 configurations (these pins are not
(CKE1, CS1_n, Input used in the x16 configuration). DDR4 will support a traditional DDP package,
ODT1) which uses these three signals for control of the second die (CS1_n, CKE1,
ODT1).
On Die Termination: ODT (registered HIGH) enables RTT_NOM termination

resistance internal to the DDR4 SDRAM. When enabled, ODT is only applied to
each DQ, DQS_t, DQS_c and DM_n/DBI_n/TDQS_t, NU/TDQS_c (When TDQS is
ODT Input enabled via Mode Register A11=1 in MR1) signal for x8 conurations. For x16
conuration ODT is applied to each DQ, DQSU_t, DQSU_c, DQSL_t, DQSL_c,
DMU_n, and DML_n signal. The ODT pin will be ignored if MR1 is programmed to
disable RTT_NOM.
Activation Command Input: ACT_n defines the Activation command being entered
ACT_n Input along with CS_n. The input into RAS_n/A16, CAS_n/A15 and WE_n/A14 will be
considered as Row Address A16, A15 and A14.
Command Inputs: RAS_n/A16, CAS_n/A15 and WE_n/A14 (along with CS_n)

RAS_n/A16, define the command being entered. Those pins have multi function. For example,
CAS_n/A15, Input for activation with ACT_n Low, those are Addressing like A16, A15 and A14 but for
WE_n/A14 non-activation command with ACT_n High, those are Command pins for Read,
Write and other command defined in command truth table.
Input Data Mask and Data Bus Inversion: DM_n is an input mask signal for write
data. Input data is masked when DM_n is sampled LOW coincident with that input
data during a Write access. DM_n is sampled on both edges of DQS. DM is
muxed with DBI function by Mode Register A10, A11, A12 setting in MR5. For x8
DM_n/DBI_n/ Input/
device, the function of DM or TDQS is enabled by Mode Register A11 setting in
TDQS_t Output
MR1. DBI_n is an input/output identifying wherther to store/output the true or
inverted data. If DBI_n is LOW, the data will be stored/output after inversion inside
the DDR4 SDRAM and not inverted if DBI_n is HIGH.
TDQS is only supported in x8.
Bank Group Inputs: BG0 - BG1 define to which bank group an Active, Read, Write
or Precharge command is being applied. BG0 also detemines which mode register
BG0B- G1 Input
is to be accessed during a MRS cycle. x4/x8 have BG0 and BG1 but x16 has only
BG0.
Bank Address Inputs: BA0 - BA1 define to which bank an Active, Read, Write or
BA0B- A1 Input Precharge command is being applied. Bank address also determines if the mode
register or extended mode register is to be accessed during a MRS cycle.

Address Inputs: Provied the row address for ACTIVATE Commands and the
column address for Read/Write commands th select one location out of the
memory array in the respective bank. (A10/AP, A12/BC_n, RAS_n/A16,
A0-A17 Input
CAS_n/A15 and WE_n/A14 have additional functions, see other rows. The
address inputs also provide the op-code during Mode Register Set commands.
A17 is only defined for the x4 configration.
Auto-precharge: A10 is sampled during Read/Write commands to determine

whether Autoprecharge should be performed to the accessed bank after the
Read/Write opera-tion. (HIGH: Autoprecharge; LOW: no Autoprecharge).A10 is
A10 / AP Input
sampled during a Pre-charge command to determine whether the Precharge
applies to one bank (A10 LOW) or all banks (A10 HIGH). If only one bank is to be
precharged, the bank is selected by bank addresses.
Burst Chop: A12 / BC_n is sampled during Read and Write commands to
A12 / BC_n Input determine if burst chop (on-the-fly) will be performed. (HIGH, no burst chop; LOW:
burst chopped). See command truth table for details.
Active Low Asynchronous Reset: Reset is active whenRESET_n is LOW, and
inactive when RESET_n is HIGH. RESET_n must be HIGH during normal
RESET_n Input
operation. RESET_n is a CMOS rail to rail signal with DC high and low at 80% and
20% of VDD.
Data Input/ Output: Bi-directional data bus. If CRC is enabled via Mode register
then CRC code is added at the end of Data Burst. Any DQ from DQ0~DQ3 may
DQ Input/Output indicate the internal VREF level during test via Mode Register Setting MR4
A4=High. During this mode, RTT value should be set to Hi-Z. Refer to vendor
specific datasheets to determine which DQ is used.
Data Strobe: output with read data, input with write data. Edge-aligned with read
data, centered in write data. For x16, DQSL corresponds to the data on DQL0-
DQS_t, DQS_c, DQL7; DQSU corresponds to the data on DQU0-DQU7. The data strobe DQS_t,
DQSU_t, DQSU_c, Input/Output DQSL_t, and DQSU_t are paired with differential signals DQS_c, DQSL_c, and
DQSL_t, DQSL_c DQSU_c, respectively, to provide differential pair signaling to the system during
reads and writes. DDR4 SDRAM supports differential data strobe only and does
not support single-ended.
Termination Data Strobe: TDQS_t/TDQS_c is applicable for x8 DRAMs only.

When enabled via Mode Register A11 = 1 in MR1, the DRAM will enable the same
termination resistance function on TDQS_t/TDQS_c that is applied to
TDQS_t, TDQS_c Ouput DQS_t/DQS_c. When disabled via mode register A11 = 0 in MR1, DM/DBI/TDQS
will provide the data mask function or Data Bus Inversion depending on MR5; A11,
12, 10 and TDQS_c is not used. x4/x16 DRAMs must disable the TDQS function
via mode register A11 = 0 in MR1.
Command and Address Parity Input : DDR4 Supports Even Parity check in DRAM
with MR setting. Once it’s enabled via Register in MR5, then DRAM calculates
PAR Input Parity with ACT_n, RAS_n/A16, CAS_n/A15, WE_n/A14, BG0-BG1, BA0-BA1,
A17-A,0 and C0- C2(3DS devices). Input parity should maintain at the rising edge
of the clock and at the same time with command & address with CS_n LOW.
Alert: It has multi functions such as CRC error flag, Command and Address Parity
error flag as Output signal. If there is error in CRC, then Alert_n goes LOW for the
period time interval and goes back HIGH. If there is error in Command Address
ALERT_n Output Parity Check, then Alert_n goes LOW for relatively long period until on going
DRAM internal recovery trans-action to complete. During Connectivity Test mode,
this pin works as input. Using this signal or not is dependent on system. In case of
not connected as Signal, ALERT_n Pin must be bounded to VDD on board.
Connectivity Test Mode Enable: Required on x16 devices and optional input on
x4/x8 with densities equal to or greater than 8Gb. HIGH in this pin will enable
Connectivity Test Mode operation along with other pins. It is a CMOS rail to rail
TEN Input
signal with AC high and low at 80% and 20% of VDD. Using this signal or not is
dependent on System. This pin may be DRAM internally pulled low through a weak
pull-down resistor to VSS.

NC No Connect: No internal electrical connection is present.
VDDQ Supply DQ Power Supply: 1.2 V +/- 0.06 V
VSSQ Supply DQ Ground
VDD Supply Power Supply: 1.2 V +/- 0.06 V
VSS Supply Ground
VPP Supply DRAM Activation Power Supply: 2.5V (2.375V min , 2.75 max)
VREFCA Supply Reference voltage for CA
ZQ Supply Reference Pin for ZQ calibration
NOTE : Input only pins (BG0-BG1,BA0-BA1, A0-A17, ACT_n, RAS_n/A16, CAS_n/A15, WE_n/A14, CS_n, CKE, ODT, and
RESET_n) do not supply termination.

Simplified State Diagram

Abbr. Function Abbr. Function Abbr. Function
ACT Activate Read RD,RDS4, RDS8 PDE Enter Power-down
PRE Precharge Read A RDA, RDAS4, RDAS8 PDX Exit Power-down
PREA PRECHARGE All Write WR, WRS4, WRS8 with/without CRC SRE Self-Refresh entry
MRS Mode Register Set Write A WRA,WRAS4, WRAS8 with/without CRC SRX Self-Refresh exit
REF Refresh, Fine granularity Refresh RESET Start RESET procedure MPR Multi Purpose Register
TEN Boundary Scan Mode Enable

Basic Functionality
The DDR4 SDRAM is a high-speed dynamic random-access memory internally configured as sixteen-banks,
4 bank group with 4 banks for each bank group for x4/x8 and eight-banks, 2 bank group with 4 banks for
each bank group for x16 DRAM. The DDR4 SDRAM uses a 8n prefetch architecture to achieve high-speed
operation. The 8n prefetch architecture is combined with an interface designed to transfer two data words per
clock cycle at the I/O pins. A single read or write operation for the DDR4 SDRAM consists of a single 8n-bit
wide, four clock data transfer at the internal DRAM core and eight corresponding n-bit wide, one-half clock
cycle data transfers at the I/O pins.
Read and write operation to the DDR4 SDRAM are burst oriented, start at a selected location, and continue
for a burst length of eight or a ‘chopped’ burst of four in a programmed sequence. Operation begins with the
registration of an ACTIVATE Command, which is then followed by a Read or Write command. The address
bits registered coincident with the ACTIVATE Command are used to select the bank and row to be activated
(BG0-BG1 in x4/x8 and BG0 in x16 select the bankgroup; BA0-BA1 select the bank; A0-A17 select the row;
refer to “DDR4 SDRAM Addressing” on Section 2.7 for specific requirements). The address bits registered
coincident with the Read or Write command are used to select the starting column location for the burst
operation, determine if the auto precharge command is to be issued (via A10), and select BC4 or BL8 mode
‘on the fly’ (via A12) if enabled in the mode register.
Prior to normal operation, the DDR4 SDRAM must be powered up and initialized in a predefined manner.
The following sections provide detailed information covering device reset and initialization, register definition,
command descriptions, and device operation.
RESET and Initialization Procedure

For power-up and reset initialization, in order to prevent DRAM from functioning improperly default values for
the following MR settings need to be defined.
 Gear down mode (MR3 A[3]) : 0 = 1/2 Rate
 Per DRAM Addressability (MR3 A[4]) : 0 = Disable
 Max Power Saving Mode (MR4 A[1]) : 0 = Disable
 CS to Command/Address Latency (MR4 A[8:6]) : 000 = Disable
 CA Parity Latency Mode (MR5 A[2:0]) : 000 = Disable
 Hard post package repair mode (MR4 A[13]): 0 = disable
 Soft post package repair mode (MR4 A[5]): 0 = disable
Power-up Initialization Sequence

The following sequence is required for POWER UP and Initializatio.
1. Apply power (RESET_n is recommended to be maintained below 0.2 x VDD; all other inputs may be
undefined). RESET_n needs to be maintained for minimum 200us with stable power. CKE is pulled “ Low”
anytime before RESET_n being de-asserted (min. time 10ns) . The power voltage ramp time between
300mV to VDD min must be no greater than 200ms; and during the ramp, VDD ≥ VDDQ and (VDD-VDDQ)
< 0.3V. VPP must ramp at the same time or earlier than VDD and VPP must be equal to or higher than
VDD at all times.
During power-up, the supply slew rate is governed by the limits stated in the table below, and either of the
following conditions must be met:

π PM Tech
Supply Power-up Slew Rate
PMA212816C
Symbol Min Max Unit Note

VDD_SL,
VDDQ_SL, 0.004 600 V/ms Measured between 300mV and 80% of supply minimum
VPP_SL
VDD_ona,
N/A 200 ms VDD(Q) maximum ramp time from 300mV to VDD(Q) minimum
VDDQ_ona
Note:20 MHz band-limited measurement
 Condition A :
 VDD and VDDQ are driven from a single power converter output, AND
 The voltage levels on all pins other than VDD,VDDQ,VSS,VSSQ must be less than or equal to VDDQ
and VDD on one side and must be larger than or equal to VSSQ and VSS on the other side.
 VTT is limited to 0.76V max once power ramp is finished, AND
 VREFCA tracks VDD/2.
Or
 Condition B :
 Apply VDD without any slope reversal before or at the same time as VDDQ
 Apply VDDQ without any slope reversal before or at the same time as VTT & VREFCA.
 Apply VPP without any slope reversal before or at the same time as VDD.
 The voltage levels on all pins other than VDD,VDDQ,VSS,VSSQ must be less than or equal to VDDQ
and VDD on one side and must be larger than or equal to VSSQ and VSS on the other side.
2. After RESET_n is de-asserted, wait for another 500us until CKE becomes active. During this time, the
DRAM will start internal initialization; this will be done independently of external clocks.
3. Clocks (CK_t,CK_c) need to be started and stabilized for at least 10ns or 5tCK (which is larger) before
CKE goes active. Since CKE is a synchronous signal, the corresponding setup time to clock (tIS) must be
met. Also a Deselect command must be registered (with tIS set up time to clock) at clock edge Td. Once
the CKE registered “High” after Reset, CKE needs to be continuously registered “High” until the
initialization sequence is finished, including expiration of tDLLK and tZQinit
4. The DDR4 SDRAM keeps its on-die termination in high-impedance state as long as RESET_n is asserted.
Further, the SDRAM keeps its on-die termination in high impedance state after RESET_n deassertion until
CKE is registered HIGH. The ODT input signal may be in undefined state until tIS before CKE is
registered HIGH. When CKE is registered HIGH, the ODT input signal may be statically held at either
LOW or HIGH. If RTT_NOM is to be enabled in MR1 the ODT input signal must be statically held LOW. In
all cases, the ODT input signal remains static until the power up initialization sequence is finished,
including the expiration of tDLLK and tZQinit.
5. After CKE is being registered high, wait minimum of Reset CKE Exit time, tXPR, before issuing the first
MRS command to load mode register. (tXPR=Max(tXS, 5nCK)]
6. Issue MRS Command to to load MR3 with all application settings, wait tMRD.
7. Issue MRS command to load MR6 with all application settings, wait tMRD.
13. Issue ZQCL command to starting ZQ calibration

π PM Tech
14. Wait for both tDLLK and tZQ init completed
PMA212816C
15. The DDR4 SDRAM is now ready for read/Write training (include VREF training and Write leveling).
RESET and Initialization Sequence at Power-On Ramping
NOTE 1 From the time point Td until Tk, a DES command must be applied between MRS and ZQCL commands.
NOTE 2 MRS commands must be issued to all mode registers that have defined settings.

π PM Tech
RESET Initialization with Stable Power Sequence
PMA212816C
The following sequence is required for RESET at no power interruption initialization:

1. Assert RESET_n below 0.2 × VDD any time reset is needed (all other inputs may be undefined). RESET
must be maintained for a minimum of 100ns. CKE is pulled LOW before RESET_n is de-asserted
(minimum time 10ns).
2. Follow Steps 2 to 10 in the Reset and Initialization Sequence at Power-on Ramping procedure.
3. The Reset sequence is now completed, DDR4 SDRAM is ready for Read/Write training (include VREF
training and Write leveling)
RESET Procedure at Power Stable Condition

NOTE 1 From the time point Td until Tk, a DES command must be applied between MRS and ZQCL commands.
NOTE 2 MRS commands must be issued to all mode registers that have defined settings.

π PM Tech
Programming Mode Registers
PMA212816C
For application flexibility, various functions, features, and modes are programmable in seven Mode Registers
(MRn), provided by the DDR4 SDRAM, as user defined variables and they must be programmed via a Mode
Register Set (MRS) command. The mode registers are divided into various fields depending on the
functionality and/or modes. As not all the Mode Registers (MR#) have default values defined, contents of
Mode Registers must be initialized and/or re-initialized, i. e. written, after power up and/or reset for proper
operation. Also the contents of the Mode Registers can be altered by re-executing the MRS command during
normal operation. When programming the mode registers, even if the user chooses to modify only a sub-set
of the MRS fields, all address fields within the accessed mode register must be redefined when the MRS
command is issued. MRS and DLL RESET commands do not affect array contents, which means these
commands can be executed any time after power-up without affecting the array contents. MRS Commands
can be issued only when DRAM is at idle state. The mode register set command cycle time, tMRD is
required to complete the write operation to the mode register and is the minimum time required between two
MRS commands shown in the tMRD Timing figure.
Some of the mode register settings affect address/command/control input functionality. In these cases, the
next MRS command can be allowed when the function being updated by the current MRS command is
completed. These MRS commands don’t apply tMRD timing to the next MRS command; however, the input
cases have unique MR setting procedures, so refer to individual function descriptions.
tMRD Timing
NOTE 1 This timing diagram depicts CA Parity Mode "Disabled" case.

NOTE 2 tMRD applies to all MRS commands with the following exceptions:
 Geardown Mode
 CA Parity Lantency Mode
 CS to Command/Address Lantency Mode
 Per DRAM Addressability Mode
 VREFDQ training value, VREFDQ training mode, and VREFDQ Training Range

π PM Tech
The MRS command to non-MRS command delay, tMOD, is required for the DRAM to update features,
PMA212816C
except DLL RESET, and is the minimum time required from an MRS command to a non-MRS command,
excluding DES, as shown in the tMOD Timing figure.
tMOD Timing
NOTE 1 This timing diagram depicts CA Parity Mode "Disabled" case.

NOTE 2 List of MRS commands exception that do not apply to tMOD.
• DLL Enable, DLL Reset
• VREFDQ training value, internal VREF monitor, VREFDQ training mode, and VREFDQ Training Range
• Geardown Mode
• Per DRAM Addressability Mode
• Maximum Power Saving Mode
• CA Parity Mode
The mode register contents can be changed using the same command and timing requirements during
normal operation as long as the device is in idle state; that is, all banks are in the precharged state with tRP
satisfied, all data bursts are completed, and CKE is HIGH prior to writing into the mode register. If the
RTT_NOM feature is enabled in the mode register prior to and/or after an MRS command, the ODT signal
must continuously be registered LOW, ensuring RTT is in an off state prior to the MRS command. The ODT
signal may be registered HIGH after tMOD has expired. If the RTT_NOM feature is disabled in the mode
register prior to and after an MRS command, the ODT signal can be registered either LOW or HIGH before,
during, and after the MRS command. The mode registers are divided into various fields depending on
functionality and modes.
In some mode register setting cases, function updating takes longer than tMOD. This type of MRS does not
apply tMOD timing to the next valid command, excluding DES. These MRS command input cases have
unique MR setting procedures, so refer to individual function descriptions.

π PM Tech
Mode Register
PMA212816C
Mode Register 0 (MR0)

Address Operating Mode Description
BG1 RFU 0 = must be programmed to 0 during MRS
000 = MR0
001 = MR1
010 = MR2
011 = MR3
BG0, BA1:BA0 MR Select
100 = MR4
101 = MR5
110 = MR6
111 = RCW1
A17 RFU 0 = must be programmed to 0 during MRS
A13,A11:A9 WR and RTP 2, 3 See Table : Write Recovery and Read to Precharge
0 = NO
A8 DLL Reset
1 = Yes
0 = Normal
A7 TM
1 = Test
A6:A4,A2 CAS Latency4 See Table: CAS Lantency
0 = Sequential
A3 Read Burst Type
1 = Interleave
00 = 8 (Fixed)
01 = BC4 or 8 (on the fly)
A1:A0 Burst Length
10 = BC4 (Fixed)
11 = Reserved
NOTE
1. Reserved for Register control word setting. DRAM ignores MR command with BG0, BA[1:0]=111 and doesn’t respond.
When RFU MR code setting is inputted, DRAM operation is not defined.
2. WR (write recovery for autoprecharge) min in clock cycles is calculated by following rounding algorithm. The WR value in
the mode register must be programmed to be equal or larger than WRmin. The programmed WR value is used with tRP to
determine tDAL.
3. The table shows the encodings for Write Recovery and internal Read command to Precharge command delay. For actual
Write recovery timing, please refer to AC timing table.
4. The table only shows the encodings for a given CAS Latency. For actual supported CAS Latency, please refer to speed bin
tables for each frequency. CAS Latency controlled by A12 is optional for 4Gb device.
5. A13 for WR and RTP setting is optional for 4Gb device.
Write Recovery and Read to Precharge (cycles)

A13 A11 A10 A9 WR RTP WR_CRC_DM
0 0 0 0 10 5 TBD
0 0 0 1 12 6 TBD
0 0 1 0 14 7 TBD
0 0 1 1 16 8 TBD
0 1 0 0 18 9 TBD
0 1 0 1 20 10 TBD
0 1 1 0 24 12 TBD
0 1 1 1 22 11 Reserved
1 0 0 0 26 13

π PM Tech
CAS Latency
PMA212816C
A12 A6 A5 A4 A2 CAS Latency Note
0 0 0 0 0 Reserved 1
0 0 0 0 1 Reserved
0 0 0 1 0 11 1
0 0 0 1 1 12
0 0 1 0 0 13 1
0 0 1 0 1 14
0 0 1 1 0 15 1
0 0 1 1 1 Reserved
0 1 0 0 0 18
0 1 0 0 1 20
0 1 0 1 0 22
0 1 0 1 1 24
0 1 1 0 1 17 1
0 1 1 1 0 19 1
1 0 0 0 0 Reserved
1 0 0 0 1 Reserved
1 0 0 1 0 Reserved
1 0 0 1 1 Reserved
1 0 1 0 0 Reserved
1 0 1 0 1 Reserved
1 0 1 1 0 Reserved
1 0 1 1 1 Reserved
1 1 0 0 0 Reserved
Note 1: This CL setting is related to read DBI usage only and please check “Speed bin” section and have a proper
corresponding option to use.

π PM Tech
PMA212816C
000 = MR0
001 = MR1
010 = MR2
011 = MR3
100 = MR4
101 = MR5
110 = MR6
111 = RCW 3
0 = Output buffer enabled

A12 Qoff 1
1 = Output buffer disabled
0 = Disable
A11 TDQS enable
1 = Enable
A10, A9, A8 RTT_NOM See Table: RTT_NOM
0 = Disable
A7 Write Leveling Enable
1 = Enable
A6, A5 RFU 0 = must be programmed to 0 during MRS
00 = 0 (AL disabled)
01 = CL-1
A4, A3 Additive Latency
10 = CL-2
11 = Resrved
A2, A1 Output Driver Impedance Control See Table: Output Driver Impedance Control
0 = Disable 2
A0 DLL Enable
1 = Enable
NOTE
1. Outputs disabled -DQs, DQS_ts, DQS_cs.
2. States reversed to “0 as Disable” with respect to DDR4.
3. Reserved for Register control word setting.DRAM ignores MR command with BG0,BA[1:0]=111 and doesn’t respond. When
RFU MR code setting is inputted, DRAM operation is not defined.
RTT_NOM
A10 A9 A8 RTT_NOM
0 0 0 Disabled
0 0 1 RZQ/4
0 1 0 RZQ/2
0 1 1 RZQ/6
1 0 0 RZQ/1
1 0 1 RZQ/5
1 1 0 RZQ/3
1 1 1 RZQ/7

π PM Tech
Output Driver Impedance Control
PMA212816C
A2 A1 Output Driver Impedance Control
0 0 RZQ/7 (34 ohm)
0 1 RZQ/5 (48 ohm)
1 0 Reserved
1 1 Reserved

000 = MR0
001 = MR1
010 = MR2
011 = MR3
100 = MR4
101 = MR5
110 = MR6
111 = RCW1

0 = Disable
A12 Write CRC
1 = Enable
A11:A9 RTT_WR See Table: RTT_WR
00 = Manual Mode (Normal Operaing Temperature Range)

Low Power Array Self 01 = Manual Mode (Reduced Operating Temperature Range)
A7:A6
Refresh (LP ASR) 10 = Manual Mode (Extended Operating Temperature Range)
11 = ASR Mode (Auto Self Refresh)
A5:A3 CAS Write Latency(CWL) See Table: CWL (CAS Write Latency)
A2:A0 RFU 0 = must be programmed to 0 during MRS
NOTE 1. Reserved for Register control word setting. DRAM ignores MR command with BG0, BA[1:0]=111 and doesn’t respond.

π PM Tech
RTT_WR
PMA212816C
A11 A10 A9 RTT_WR
0 0 0 Dynamic ODT Off
0 0 1 RZQ/2 (120 ohm)
0 1 0 RZQ/1 (240 ohm)
0 1 1 Hi-Z
1 0 0 RZQ/3 (80 ohm)
1 0 1 Reserved
1 1 0 Reserved
1 1 1 Reserved
CAS Write Latency (CWL)
Speed bin in MT/s

A5 A4 A3 CWL
1tCK WRITE 2tCK WRITE
preamble preamble1
0 0 0 9 1600
0 0 1 10 1866
0 1 0 11 2133/1600
0 1 1 12 2400/1866
1 0 0 14 2666/2133 2400
1 0 1 16 2933, 3200/2400 2666/2400
1 1 0 18 2666 2933, 3200/2666
1 1 1 20 2933, 3200 2933, 3200
NOTE 1. The 2Tck WRITE preamble is valid for DDR4-2400/2666/2933 Speed Grade. For the 2nd Set of 2Tck
WRITE preamble, no additional CWL is needed.

π PM Tech
PMA212816C
000 = MR0
001 = MR1
010 = MR2
BG0, 011 = MR3
MR Select
BA1:BA0 100 = MR4
101 = MR5
110 = MR6
111 = RCW1
00 = Serial
01 = Parallel
A12:A11 MPR Read Format
10 = Staggered
11 = Reserved
A10:A9 Write CMD Latency when CRC See Table:

and DM are enabled Write Command Latency when CRC and DM are both enabled
A8:A6 Fine Granularity Refresh Mode See Table: Fine Granularity Refresh Mode
0 : Disable
A5 Temperature sensor readout
1: Enable
0 = Disable
A4 Per DRAM Addressability
1 = Enable
0 = 1/2 Rate
A3 Geardown Mode
1 = 1/4 Rate
0 = Normal
A2 MPR Operation
1 = Dataflow from/to MPR
00 = Page0
01 = Page1
A1:A0 MPR page Selection 10 = Page2
11 = Page3
See Table: MPR Data Format
NOTE 1. Reserved for Register control word setting.DRAM ignores MR command with BG0,BA[1:0]=111 and doesn’t respond.
Fine Granularity Refresh Mode
A8 A7 A6 Fine Granularity Refresh
0 0 0 Normal (Fixed 1x)
0 0 1 Fixed 2x
0 1 0 Fixed 4x
0 1 1 Reserved
1 0 0 Reserved
1 0 1 Enable on-the-fly 1x/2x
1 1 0 Enable on-the-fly 1x/4x
1 1 1 Reserved

π PM Tech
Write Command Latency when CRC and DM are both enabled
PMA212816C
A10 A9 CRC+DM Write Command Latency Speed Bin
0 0 4nCK 1600
0 1 5nCK 1866, 2133, 2400, 2666
1 0 6nCK 2933, 3200
1 1 Reserved Reserved
NOTE
1. The WRITE command latency (WCL) must be set when both write CRC and DM are enabled for write CRC persistent
mode
2. At less than or equal to 1600 then 4nCK; neither 5nCK nor 6nCK
3. At greater than 1600 and lessthan or equal to 2666 then 5nCK; neither 4nCK nor 6nCK
4. At greater than 2666 and less than or equal to 3200 then 6nCK; neither 4nCK nor 5nCK
MPR Data Format

MPR page0 (Training Pattern)
MPR
Address [7] [6] [5] [4] [3] [2] [1] [0] Note
Location
00 = MPR0 0 1 0 1 0 1 0 1
01 = MPR1 0 0 1 1 0 0 1 1 Read/ Write
BA1:BA0 (default value)
10 = MPR2 0 0 0 0 1 1 1 1
11 = MPR3 0 0 0 0 0 0 0 0
MPR page1 (CA Parity Error Log)

MPR
Address [7] [6] [5] [4] [3] [2] [1] [0] Note
Location
00 = MPR0 A[7] A[6] A[5] A[4] A[3] A[2] A[1] A[0]
CAS_n/ WE_n/
01 = MPR1 A[13] A[12] A[11] A[10] A[9] A[8]
A15 A14
RAS_n/
BA1:BA0 10 = MPR2 PAR ACT_n BG[1] BG[0] BA[1] BA[0] A[17] 3 Read only
A16
CA Parity CA Parity Latency 4

CRC Error
11 = MPR3 Error
Status
Status MR5.A[2] MR5.A[1] MR5.A[0]
NOTE
1. MPR used for C/A parity error log readout is enabled by setting A[2] in MR3
2. For higher density of DRAM, where A[17] is not used, MPR2[1] should be treated as don’t care.
3. MPR page 1 used for CA parity error log readout is enabled by setting A[2] in MR3.
4. MPR3 bit 0~2 (CA parity latency) reflects the latest programmed CA parity latency values.

π PM Tech
MPR page2 (MRS Readout)
PMA212816C
Address MPR Location [7] [6] [5] [4] [3] [2] [1] [0] Note
Temperature CRC Write
hPPR sPPR RTT_WR RTT_WR
Sensor Status Enable
00 = MPR0
- - MR2 - - MR2 MR2
- - A11 - - A12 A10 A9
VREF DQ
Gear down
Traing VREF DQ training Value
Enable
range
01= MPR1
MR6 MR6
BA1:BA0 Read only
A6 A5 A4 A3 A2 A1 A0 A3
CAS Latency CAS Write Latency
10 = MPR2 MR0 MR2

A6 A5 A4 A2 A12 A5 A4 A3
RTT_NOM RTT_PARK Driver Impedance
11 = MPR3 MR1 MR5 MR1

A10 A9 A6 A8 A7 A6 A2 A1
MR bit for Temperature
MR3 bit A5=1: DRAM updates the temperature sensor status to MPR Page 2 (MPR0 bits A4:A3). Temperature data is
guaranteed by the DRAM to be no more than 32ms old at the time of MPR Read of the Temperature Sensor Status bits.
MR3 bit A5=0: DRAM disables updates to the temperature sensor status in MPR Page 2(MPR0-bit A4:A3)
MPR0 bit A4 MPR0 bit A3 Refresh Rate Range

0 0 Sub 1X refresh ( > tREFI)
0 1 1X refresh rate(= tREFI)
1 0 2X refresh rate(1/2 x tREFI)
1 1 Reserved
MPR page3 (Vendor use only)1
Address MPR Location [7] [6] [5] [4] [3] [2] [1] [0] Note
00 = MPR0 don’t care
BA1:BA0 Read only
NOTE 1 MPR page3 is specifically assigned to DRAM. Actual encoding method is vendor specific.

π PM Tech
PMA212816C

000 = MR0
001 = MR1
010 = MR2
011 = MR3
100 = MR4
101 = MR5
110 = MR6
111 = RCW1
0 = Disable
A13 hPPR
1 = Enable
0 = 1 nCK
A12 Write Preamble
1 = 2 nCK
0 = 1 nCK
A11 Read Preamble
1 = 2 nCK
Read Preamble Training 0 = Disable
A10
Mode 1 = Enable
0 = Disable
A9 Self Refresh Abort
1 = Enable
000 = Disable
001 = 3
010 = 4
011 = 5
A8:A6 CS to CMD/ADDR Latency Mode (cycles) 100 = 6
101 = 8
110 = Reserved
111 = Reserved
(See Table: CS to CMD/ADDR Latency Mode Setting)
0 = Disable
A5 sPPR
1 = Enable
0 = Disable
A4 Internal VREF Monitor
1 = Enable
0 = Disable
A3 Temperature Controlled Refresh Mode 1 = Enable
0 = Normal
A2 Temperature Controlled Refresh Range 1 = Extended
0 = Disable
A1 Maximum Power Down Mode 1 = Enable
NOTE 1 Reserved for Register control word setting .DRAM ignores MR command with BG0,BA1;BA0=111 and doesn’t
respond. When RFU MR code setting is inputted, DRAM operation is not defined.

π PM Tech
CS to CMD / ADDR Latency Mode Setting
PMA212816C
A8 A7 A6 CAL
0 0 0 Disable
0 0 1 3
0 1 0 4
0 1 1 5
1 0 0 6
1 0 1 8
1 1 0 Reserved
1 1 1 Reserved

000 = MR0
001 = MR1
010 = MR2
011 = MR3
100 = MR4
101 = MR5
110 = MR6
111 = RCW1
0 = Disable
A12 Read DBI
1 = Enable
0 = Disable
A11 Write DBI
1 = Enable
0 = Disable
A10 Data Mask
1 = Enable
0 = Disable
A9 CA parity Persistent Error
1 = Enable
A8:A6 RTT_PARK See Table: RTT_PARK
0 = ODT input buffer is activated
A5 ODT Input Buffer during Power Down mode
1 = ODT input buffer is deactivated
0 = Clear
A4 CA Parity Error Status
1 = Error
0 = Clear
A3 CRC Error Clear
1 = Error
A2:A0 CA Parity Latency Mode See Table: CA Parity Latency Mode
NOTE
1. Reserved for Register control word setting.DRAM ignores MR command with BG0,BA[1:0]=111 and doesn’t respond.
2. When RTT_NOM Disable is set in MR1, A5 of MR5 will be ignored.

π PM Tech
RTT_PARK
PMA212816C
A8 A7 A6 RTT_PARK
0 0 0 RTT_PARK Disable
0 0 1 RZQ/4
0 1 0 RZQ/2
0 1 1 RZQ/6
1 0 0 RZQ/1
1 0 1 RZQ/5
1 1 0 RZQ/3
1 1 1 RZQ/7
CA Parity Latency Mode
A2 A1 A0 PL Speed Bin
0 0 0 Disable
0 0 1 4 1600/1866/2133
0 1 0 5 2400/2666
0 1 1 6 2933/3200
1 0 0 8 RFU
1 0 1 Reserved
1 1 0 Reserved
1 1 1 Reserved
NOTE 1 Parity latency must be programmeaccording to timing parameters by speed grade table.

π PM Tech
PMA212816C

000 = MR0
001 = MR1
010 = MR2
011 = MR3
100 = MR4
101 = MR5
110 = MR6
111 = RCW1
A12:A10 tCCD_L See Table: tCCD_L & tDLLK
A9, A8 RFU 0 = must be programmed to 0 during MRS
0 = Disable(Normal operation Mode)
A7 VREFDQ Training Enable
1 = Enable(Training Mode)
0 = Range 1
A6 VREFDQ Training Range
1 = Range 2
A5:A0 VREFDQ Training Value See Table: VREFDQ Training Values
NOTE 1 Reserved for Register control word setting.DRAM ignores MR command with BG0,BA[1:0]=111 and doesn’t respond.
tCCD_L & tDLLK
A12 A11 A10 tCCD_L.min (nCK)1 tDLLKmin (nCK)1 Note

0 0 0 Reserved 1333Mbps
597
0 0 1 5 >1333Mbps and 1866Mbps (1600/1866Mbps)
0 1 0 6 768 >1866Mbps and 2400Mbps (2133/2400Mbps)
0 1 1 7 >2400Mbps and 2666Mbps (2666Mbps)

1024
1 0 0 8 >2666Mbps and 3200Mbps (2933/3200Mbps)
1 0 1
1 1 0 Reserved
1 1 1
NOTE 1 tCCD_L/tDLLK should be programmed according to the value defined in AC parameter table per operating frequency.

π PM Tech
VREFDQ Training Values
PMA212816C
A5:A0 Range1 Range2 A5:A0 Range1 Range2

00 0000 60.00% 45.00% 01 1010 76.90% 61.90%
00 0001 60.65% 45.65% 01 1011 77.55% 62.55%
00 0010 61.30% 46.30% 01 1100 78.20% 63.20%
00 0011 61.95% 46.95% 01 1101 78.85% 63.85%
00 0100 62.60% 47.60% 01 1110 79.50% 64.50%
00 0101 63.25% 48.25% 01 1111 80.15% 65.15%
00 0110 63.90% 48.90% 10 0000 80.80% 65.80%
00 0111 64.55% 49.55% 10 0001 81.45% 66.45%
00 1000 65.20% 50.20% 10 0010 82.10% 67.10%
00 1001 65.85% 50.85% 10 0011 82.75% 67.75%
00 1010 66.50% 51.50% 10 0100 83.40% 68.40%
00 1011 67.15% 52.15% 10 0101 84.05% 69.05%
00 1100 67.80% 52.80% 10 0110 84.70% 69.70%
00 1101 68.45% 53.45% 10 0111 85.35% 70.35%
00 1110 69.10% 54.10% 10 1000 86.00% 71.00%
00 1111 69.75% 54.75% 10 1001 86.65% 71.65%
01 0000 70.40% 55.40% 10 1010 87.30% 72.30%
01 0001 71.05% 56.05% 10 1011 87.95% 72.95%
01 0010 71.70% 56.70% 10 1100 88.60% 73.60%
01 0011 72.35% 57.35% 10 1101 89.25% 74.25%
01 0100 73.00% 58.00% 10 1110 89.90% 74.90%
01 0101 73.65% 58.65% 10 1111 90.55% 75.55%
01 0110 74.30% 59.30% 11 0000 91.20% 76.20%
01 0111 74.95% 59.95% 11 0001 91.85% 76.85%
01 1000 -75.60% 60.60% 11 0010 92.50% 77.50%
01 1001 76.25% 61.25% 11 0011 to 111111 Reserved Reserved

DRAM MR7 Ignore
The DDR4 SDRAM shall ignore any access to MR7 for all DDR4 SDRAM.Any bit setting within MR7 may not
take any effect in the DDR4 SDRAM.

π PM Tech
Command Description and Operation
PMA212816C
Command Truth Table

Note 1,2,3 and 4 apply to the entire Command truth table
Note 5 applies to all Read/Write commands.
[BG=Bank Group Address, BA=Bank Address, RA=Row Address, CA=Column Address, BC_n=Burst Chop,
X=Don’t Care, V=Valid H or L].
CKE A17,
RAS_n CAS_n WE_n/ BG0- BA0- C2- A12/ A10/ A0-
Function Symbol CS_n ACT_n A13, NOTE
Prev. Pres. /A16 /A15 A14 BG1 BA1 C0 BC_n AP A9
A11
Mode Register Set MRS H H L H L L L BG BA V OP Code 12
Refresh REF H H L H L L H V V V V V V V
Self Refresh Entry SRE H L L H L L H V V V V V V V 7,9
H X X X X X X X X X X X 7,8,9,
Self Refresh Exit SRX L H
10
L H H H H V V V V V V V
Single Bank Precharge PRE H H L H L H L BG BA V V V L V
Precharge all Banks PREA H H L H L H L V V V V V H V
RFU RFU H H L H L H H RFU
Bank Activate ACT H H L L Row Address(RA) BG BA V Row Address (RA)
Write (Fixed BL8 or BC4) WR H H L H H L L BG BA V V V L CA
Write (BC4, on the Fly) WRS4 H H L H H L L BG BA V L V L CA
Write (BL8, on the Fly) WRS8 H H L H H L L BG BA V H V L CA

Write with Auto Precharge
WRA H H L H H L L BG BA V V V H CA
(Fixed BL8 or BC4)
WRAS4 H H L H H L L BG BA V L V H CA
(BC4, on the Fly)
WRAS8 H H L H H L L BG BA V H V H CA
(BL8, on the Fly)
Read (Fixed BL8 or BC4) RD H H L H H L H BG BA V V V L CA
Read (BC4, on the Fly) RDS4 H H L H H L H BG BA V L V L CA
Read (BL8, on the Fly) RDS8 H H L H H L H BG BA V H V L CA

Read with Auto Precharge
RDA H H L H H L H BG BA V V V H CA
(Fixed BL8 or BC4)
RDAS4 H H L H H L H BG BA V L V H CA
(BC4, on the Fly)
RDAS8 H H L H H L H BG BA V H V H CA
(BL8, on the Fly)
No Operation NOP H H L H H H H V V V V V V V 10
Device Deselected DES H H H X X X X X X X X X X X
Power Down Entry PDE H L H X X X X X X X X X X X 6
Power Down Exit PDX L H H X X X X X X X X X X X 6
ZQ calibration Long ZQCL H H L H H H L V V V V V H V
ZQ calibration Short ZQCS H H L H H H L V V V V V L V
NOTE
1. All DDR4 SDRAM commands are defined by states of CS_n, ACT_n, RAS_n/A16, CAS_n/A15, WE_n/A14 and CKE at the
rising edge of the clock. The MSB of BG, BA, RA and CA are device density and configuration dependant. When ACT_n =
H; pins RAS_n/A16, CAS_n/A15, and WE_n/A14 are used as command pins RAS_n, CAS_n, and WE_n respectively.
When ACT_n = L; pins RAS_n/A16, CAS_n/A15, and WE_n/A14 are used as address pins A16, A15, and A14
respectively.
2. RESET_n is Low enable command which will be used only for asynchronous reset so must be maintained HIGH during
any function.
3. Bank Group addresses (BG) and Bank addresses (BA) determine which bank within a bank group to be operated upon.
For MRS commands the BG and BA selects the specific Mode Register location.
4. “V” means “H or L (but a defined logic level)” and “X” means either “defined or undefined (like floating) logic level”.

π
5.
PM Tech
Burst reads or writes cannot be terminated or interrupted and Fixed/on-the-Fly BL will be defined by MRS.
PMA212816C
6. The Power Down Mode does not perform any refresh operation.
7. The state of ODT does not affect the states described in this table. The ODT function is not available during Self Refresh.
8. Controller guarantees self refresh exit to be synchronous.
9. VPP and VREF(VREFCA) must be maintained during Self Refresh operation.
10. Mode Exit
11. Refer to the CKE Truth Table for more detail with CKE transition.
12. During a MRS command A17 is Reserved for Future Use and is device density and configuration dependent.
CKE Truth Table

CKE
Command (N)3
Current State2 Previous Cycle1 Present Cycle1 Action (N)3 NOTE
RAS_n, CAS_n, WE_n, CS_n
(N-1) (N)
L L X Maintain Power-Down 14, 15
Power Down
L H DESELECT Power Down Exit 11, 14
L L X Maintain Self Refresh 15, 16

Self Refresh
L H DESELECT Self Refresh Exit 8, 12, 16
Bank(s) Active H L DESELECT Active Power Down Entry 11, 13, 14
Reading H L DESELECT Power Down Entry 11, 13, 14, 17
Writing H L DESELECT Power Down Entry 11, 13, 14, 17
Precharging H L DESELECT Power Down Entry 11, 13, 14, 17
Refreshing H L DESELECT Precharge Power Down Entry 11
H L DESELECT Precharge Power Down Entry 11,13, 14, 18

All Banks Idle
H L REFRESH Self Refresh Entry 9, 13, 18
For more details with all signals See “Command Truth Table”. 10
NOTE
1. CKE (N) is the logic state of CKE at clock edge N; CKE (N-1) was the state of CKE at the previous clock edge.
2. Current state is defined as the state of the DDR4 SDRAM immediately prior to clock edge N.
3. COMMAND (N) is the command registered at clock edge N, and ACTION (N) is a result of COMMAND (N),ODT is not
included here.
4. All states and sequences not shown are illegal or reserved unless explicitly described elsewhere in this document.
5. The state of ODT does not affect the states described in this table. The ODT function is not available during Self-
Refresh.
6. During any CKE transition (registration of CKE H->L or CKE L->H), the CKE level must be maintained until 1nCK prior to
tCKEmin being satisfied (at which time CKE may transition again).
7. DESELECT and NOP are defined in the Command Truth Table.
8. On Self-Refresh Exit DESELECT commands must be issued on every clock edge occurring during the tXS period. Read
or ODT commands may be issued only after tXSDLL is satisfied.
9. Self-Refresh mode can only be entered from the All Banks Idle state.
10. Must be a legal command as defined in the Command Truth Table.
11. Valid commands for Power-Down Entry and Exit are DESELECT only.
12. Valid commands for Self-Refresh Exit are DESELECT only except for Gear Down mode and Max Power Saving exit.
NOP is allowed for these 2 modes.
13. Self-Refresh can not be entered during Read or Write operations. For a detailed list of restrictions See “Self-Refresh
Operation” and See “Power-Down Modes”.
14. The Power-Down does not perform any refresh operations.
15. “X” means “don’t care“ (including floating around VREF) in Self-Refresh and Power-Down. It also applies to Address
pins.
16. VPP and VREF(VREFCA) must be maintained during Self-Refresh operation.
17. If all banks are closed at the conclusion of the read, write or precharge command, then Precharge Power-Down is
entered, otherwise Active Power-Down is entered.
18. ‘Idle state’ is defined as all banks are closed (tRP, tDAL, etc. satisfied), no data bursts are in progress, CKE is high, and
all timings from previous operations are satisfied (tMRD, tMOD, tRFC, tZQinit, tZQoper, tZQCS, etc.) as well as all Self-
Refresh exit and Power-Down Exit parameters are satisfied (tXS, tXP,etc).


π PM Tech
Burst Length, Type, and Order
PMA212816C
Accesses within a given burst may be programmed to sequential or interleaved order. The burst type is
selected via bit A3 of Mode Register MR0. The ordering of accesses within a burst is determined by the burst
length, burst type, and the starting column address as shown in Table below. The burst length is defined by
bits A0-A1 of Mode Register MR0. Burst length options include fixed BC4, fixed BL8, and ‘on-the-fly’ which
allows BC4 or BL8 to be selected coincident with the registration of a Read or Write command via A12/BC_n.
Starting
Burst Column burst type = Sequential burst type = Interleaved
Read/ Write NOTE
Length Address (decimal) A3=0 (decimal) A3=1
(A2,A1,A0)
000 0,1,2,3,T,T,T,T 0,1,2,3,T,T,T,T 1,2,3
001 1,2,3,0,T,T,T,T 1,0,3,2,T,T,T,T 1,2,3
010 2,3,0,1,T,T,T,T 2,3,0,1,T,T,T,T 1,2,3
011 3,0,1,2,T,T,T,T 3,2,1,0,T,T,T,T 1,2,3
READ
100 4,5,6,7,T,T,T,T 4,5,6,7,T,T,T,T 1,2,3
4 Chop
101 5,6,7,4,T,T,T,T 5,4,7,6,T,T,T,T 1,2,3
110 6,7,4,5,T,T,T,T 6,7,4,5,T,T,T,T 1,2,3
111 7,4,5,6,T,T,T,T 7,6,5,4,T,T,T,T 1,2,3
0, V, V 0,1,2,3,X,X,X,X 0,1,2,3,X,X,X,X 1,2,4,5
WRITE
1, V, V 4,5,6,7,X,X,X,X 4,5,6,7,X,X,X,X 1,2,4,5
000 0,1,2,3,4,5,6,7 0,1,2,3,4,5,6,7 2
001 1,2,3,0,5,6,7,4 1,0,3,2,5,4,7,6 2
010 2,3,0,1,6,7,4,5 2,3,0,1,6,7,4,5 2
011 3,0,1,2,7,4,5,6 3,2,1,0,7,6,5,4 2
READ
8 100 4,5,6,7,0,1,2,3 4,5,6,7,0,1,2,3 2
101 5,6,7,4,1,2,3,0 5,4,7,6,1,0,3,2 2
110 6,7,4,5,2,3,0,1 6,7,4,5,2,3,0,1 2
111 7,4,5,6,3,0,1,2 7,6,5,4,3,2,1,0 2
WRITE V, V, V 0,1,2,3,4,5,6,7 0,1,2,3,4,5,6,7 2,4
NOTE
1. In case of burst length being fixed to 4 by MR0 setting, the internal write operation starts two clock cycles earlier than for
the BL8 mode. This means that the starting point for tWR and tWTR will be pulled in by two clocks. In case of burst length
being selected on-the-fly via A12/BC_n, the internal write operation starts at the same point in time like a burst of 8 write
operation. This means that during on-the-fly control, the starting point for tWR and tWTR will not be pulled in by two clocks.
2. 0...7 bit number is value of CA[2:0] that causes this bit to be the first read during a burst.
3. Output driver for data and strobes are in high impedance.
4. V : A valid logic level (0 or 1), but respective buffer input ignores level on input pins.
5. X : Don’t Care.
BL8 Burst order with CRC Enabled

DDR4 SDRAM supports fixed write burst ordering [A2:A1:A0=0:0:0] when write CRC is enabled in BL8
(fixed).

π PM Tech
DLL-off Mode
PMA212816C
DDR4 SDRAM DLL-off mode is entered by setting MR1 bit A0 to “0”; this will disable the DLL for subsequent
operations until A0 bit is set back to “1”. The MR1 A0 bit for DLL control can be switched either during
initialization or later. Refer to “Input clock frequency change”.
The DLL-off Mode operations listed below are an optional feature for DDR4 SDRAM. The maximum clock
frequency for DLL-off Mode is specified by the parameter tCKDLL_OFF. There is no minimum frequency limit
besides the need to satisfy the refresh interval, tREFI.
Due to latency counter and timing restrictions, only one value of CAS Latency (CL) in MR0 and CAS Write
Latency (CWL) in MR2 are supported. The DLL-off mode is only required to support setting of both CL=10
and CWL=9. When DLL-off Mode is enabled, use of CA Parity Mode is not allowed.
DLL-off mode will affect the Read data Clock to Data Strobe relationship (tDQSCK), but not the Data Strobe
to Data relationship (tDQSQ, tQH). Special attention is needed to line up Read data to controller time domain.
Compared with DLL-on mode, where tDQSCK starts from the rising clock edge (AL+CL) cycles after the
Read command, the DLL-off mode tDQSCK starts (AL+CL - 1) cycles after the read command.
Another difference is that tDQSCK may not be small compared to tCK (it might even be larger than tCK) and
the difference between tDQSCKmin and tDQSCKmax is significantly larger than in DLL-on mode. The
tDQSCK(DLL_off) values are vendor specific.
The timing relations on DLL-off mode READ operation are shown in the following Timing Diagram (CL=10,
BL=8, PL=0):
DLL-Off Mode Read Timing Operation

π PM Tech
DLL On/Off Switching Procedure
PMA212816C
DDR4 DLL-off mode is entered by setting MR1 bit A0 to “0”; this will disable the DLL for subsequent
operations until the A0 bit is set back to “1”.
DLL On to DLL Off Procedure

To switch from DLL “on” to DLL “off” requires the frequency to be changed during self refresh, as outlined in
the following procedure:
1. Starting from Idle state (All banks pre-charged, all timings fulfilled, and DRAMs On-die Termination
resistors, RTT_NOM, must be in high impedance state before MRS to MR1 to disable the DLL.)
2. Set MR1 bit A0 to “0” to disable the DLL.
3. Wait tMOD.
4. Enter Self Refresh Mode; wait until (tCKSRE) is satisfied.
5. Change frequency, in guidance with “Input clock frequency change” on Section 4.6.
6. Wait until a stable clock is available for at least (tCKSRX) at DRAM inputs.
7. Starting with the Self Refresh Exit command, CKE must continuously be registered HIGH until all tMOD
timings from any MRS command are satisfied. In addition, if any ODT features were enabled in the mode
registers when Self Refresh mode was entered, the ODT signal must continuously be registered LOW until
all tMOD timings from any MRS command are satisfied. If RTT_NOM features were disabled in the mode
registers when Self Refresh mode was entered, ODT signal is Don’t Care.
8. Wait tXS_Fast or tXS_Abort or tXS, then set Mode Registers with appropriate values (especially an
update of CL, CWL and WR may be necessary. A ZQCL command may also be issued after tXS_Fast).
- tXS - ACT, PRE, PREA, REF, SRE, PDE, WR, WRS4, WRS8, WRA, WRAS4, WRAS8, RD, RDS4,
RDS8, RDA, RDAS4, RDAS8
- tXS_Fast - ZQCL, ZQCS, MRS commands. For MRS command, only DRAM CL and WR/RTP register in
MR0, CWL register in
MR2 and geardown mode in MR3 are allowed to be accessed provided DRAM is not in per DRAM
addressibility mode. Access to other DRAM mode registers must satisfy tXS timing.
- tXS_Abort - If the MR4 bit A9 is enabled then the DRAM aborts any ongoing refresh and does not
increment the refresh counter.
The controller can issue a valid command after a delay of tXS_abort. Upon exit from Self-Refresh, the
DDR4 SDRAM requires a minimum of one extra refresh command before it is put back into Self-Refresh
Mode. This requirement remains the same irrespective of the setting of the MRS bit for self refresh abort.
9. Wait for tMOD, and then the DRAM is ready for the next command.

π PM Tech
DLL Switch Sequence from DLL On to DLL Off
PMA212816C
COMMAND
NOTE
1. Starting with idle state. RTT in stable.
2. Disable DLL by setting MR1 bit A0 to “0”.
3. Enter SR.
4. Change frequency.
5. Clock must be stable in tCKSRX.
6. Exit SR.
7. Update mode registers allowed with DLL_off parameters setting.

π PM Tech
DLL Off to DLL On Procedure
PMA212816C
To switch from DLL off to DLL on (with required frequency change) during self refresh:
1. Starting from the idle state (all banks pre-charged, all timings fulfilled, and DRAM on-die termination
resistors (RTT_NOM) must be in the high impedance state before self refresh mode is entered.)
2. Enter Self Refresh mode; wait until tCKSRE satisfied.
3. Change frequency, following the guidelines in the “Input Clock Frequency Change” section.
4. Wait until a stable clock is available for at least (tCKSRX) at DRAM inputs.
5. Starting with the Self Refresh Exit command, CKE must continuously be registered HIGH until tDLLK
timing from the subsequent DLL RESET command is satisfied. In addition, if any ODT features were
enabled in the mode registers when self refresh mode was entered, the ODT signal must continuously be
registered LOW until tDLLK timings from the subsequent DLL RESET command is satisfied. If RTT_NOM
was disabled in the mode registers when self refresh mode was entered, the ODT signal is "Don't Care."
6. Wait tXS or tXS_ABORT, depending on bit A9 in MR4, then set MR1 bit A0 to “1” to enable the DLL.
7. Wait tMRD, then set MR0 bit A8 to “1” to start DLL Reset.
8. Wait tMRD, then set mode registers with appropriate values (especially an update of CL, CWL, and WR
may be necessary. After tMOD is satisfied from any proceeding MRS command, a ZQCL command may
also be issued during or after tDLLK.)
9. Wait for tMOD, then DRAM is ready for the next command. (Remember to wait tDLLK after DLL RESET
before applying any command requiring a locked DLL.) In addition, wait for tZQoper in case a ZQCL
command was issued.
DLL Switch Sequence from DLL Off to DLL On
NOTE
1. Starting with Idle State
2. Enter SR
3. Change Frequency
4. Clock must be stable tCKSRX
5. Exit SR
6. Set DLL-on by MR1 A0=’1’
7. Start DLLReset
8. Update rest MR register values after tDLLK (not shown in the diagram)
9. Ready for valid command after tDLLK (not shown in the diagram)

π PM Tech
Input Clock Frequency Change
PMA212816C
After the DDR4 SDRAM is initialized, the DDR4 SDRAM requires the clock to be “stable” during almost all
states of normal operation. This means that, once the clock frequency has been set and is to be in the
“stable state”, the clock period is not allowed to deviate except for what is allowed for by the clock jitter and
SSC (spread spectrum clocking) specifications. The input clock frequency can be changed from one stable
clock rate to another stable clock rate only when in self refresh mode. Outside of self refresh mode, it is
illegal to change the clock frequency.
After the DDR4 SDRAM has been successfully placed in to Self-Refresh mode and tCKSRE has been
satisfied, the state of the clock becomes a don’t care. Once a don’t care, changing the clock frequency is
permissible, provided the new clock frequency is stable prior to tCKSRX. When entering and exiting Self-
Refresh mode for the sole purpose of changing the clock frequency, the Self-Refresh entry and exit
specifications must still be met as outlined in “Self-Refresh Operation”.
For the new clock frequency, additional MRS commands to MR0, MR2, MR3, MR4, MR5, and MR6 may
need to be issued to program appropriate CL, CWL, Gear-down mode, Read & Write Preamble, Command
Address Latency (CAL Mode), Command Address Parity (CA Parity Mode), and tCCD_L/tDLLK value.
In particular, the Command Address Parity Latency (PL) must be disabled when the clock rate changes, ie.
while in Self Refresh Mode. For example, if changing the clock rate from DDR4-2133 to DDR4-2933 with CA
Parity Mode enabled, MR5[2:0] must first change from PL = 4 to PL = disable prior to PL = 6. A correct
procedure would be to (1) change PL = 4 to disable via MR5 [2:0], (2) enter Self Refresh Mode, (3) change
clock rate from DDR4-2133 to DDR4-2933, (4) exit Self Refresh Mode, (5) Enable CA Parity Mode setting PL
= 6 via MR5 [2:0].
If the MR settings that require additional clocks are updated after the clock rate has been increased, i.e. after
exiting self refresh mode, the required MR settings must be updated prior to removing the DRAM from the
IDLE state, unless the DRAM is RESET. If the DRAM leaves the idle state to enter self refresh mode or ZQ
Calibration, the updating of the required MR settings may be deferred to after the next time the DRAM enters
the IDLE state.
If MR6 is issued prior to Self Refresh Entry for new tDLLK value, then DLL will relock automatically at Self
Refresh Exit. However, if MR6 is issued after Self Refresh Entry, then MR0 must be issued to reset the DLL.
The DDR4 SDRAM input clock frequency is allowed to change only within the minimum and maximum
operating frequency specified for the particular speed grade. Any frequency change below the minimum
operating frequency would require the use of DLL_on mode to DLL_off mode transition sequence (see DLL
On/Off Switching Procedure).

π PM Tech
Write Leveling
PMA212816C
For better signal integrity, DDR4 memory modules use fly-by topology for the commands, addresses, control
signals, and clocks. Fly-by topology has benefits from the reduced number of stubs and their length, but it
also causes flight-time skew between clock and strobe at every DRAM on the DIMM. This makes it difficult
for the controller to maintain tDQSS, tDSS, and tDSH specifications. Therefore, the device supports a write
leveling feature to allow the controller to compensate for skew. This feature may not be required under some
system conditions, provided the host can maintain the tDQSS, tDSS, and tDSH specifications.
The memory controller can use the write leveling feature and feedback from the device to adjust the DQS
(DQS_t, DQS_c) to CK (CK_t, CK_c) relationship. The memory controller involved in the leveling must have
an adjustable delay setting on DQS to align the rising edge of DQS with that of the clock at the DRAM pin.
The DRAM asynchronously feeds back CK, sampled with the rising edge of DQS, through the DQ bus. The
controller repeatedly delays DQS until a transition from 0 to 1 is detected. The DQS delay established though
this exercise would ensure the tDQSS specification. Besides tDQSS, tDSS and tDSH specifications also
need to be fulfilled. One way to achieve this is to combine the actual tDQSS in the application with an
appropriate duty cycle and jitter on the DQS signals. Depending on the actual tDQSS in the application, the
actual values for tDQSL and tDQSH may have to be better than the absolute limits provided in the AC Timing
Parameters section in order to satisfy tDSS and tDSH specifications. A conceptual timing of this scheme is
shown below.
Write Leveling Concept
DQS_t - DQS_c driven by the controller during leveling mode must be terminated by the DRAM based on
ranks populated. Similarly, the DQ bus driven by the DRAM must also be terminated at the controller.
All data bits should carry the leveling feedback to the controller across the DRAM configurations X4, X8, and
X16. On a X16 device, both byte lanes should be leveled independently. Therefore, a separate feedback
mechanism should be available for each byte lane. The upper data bits should provide the feedback of the
upper diff_DQS(diff_UDQS) to clock relationship whereas the lower data bits would indicate the lower
diff_DQS(diff_LDQS) to clock relationship.

π PM Tech
DRAM Setting for Write Leveling and DRAM Termination Function in that Mode
PMA212816C
DRAM enters into Write leveling mode if A7 in MR1 set ’High’ and after finishing leveling, DRAM exits from
write leveling mode if A7 in MR1 set ’Low’. Note that in write leveling mode, only DQS_t/DQS_c terminations
are activated and deactivated via ODT pin, unlike normal operation.
MR Settings for Leveling Procedures
Function MR1 Enable Disable

Write leveling enable A7 1 0
Output buffer mode (Qoff) A12 0 1
DRAM Termination Function in Leveling Mode
ODT pin at DRAM DQS_t/DQS_c termination DQs termination
RTT_NOM with ODT HIGH On Off
RTT_PARK with ODT LOW On Off
NOTE
1. In write-leveling mode with its output buffer disabled (MR1[bitA7] = 1 with MR1[bitA12]=1) all RTT_NOM and RTT_PARK
settings are allowed; in write-leveling mode with itsoutput buffer enabled (MR1[bitA7] = 1 with MR1[bitA12] = 0) all
RTT_NOM and RTT_PARK settingsare allowed.
2. Dynamic ODT function is not available in Write Leveling Mode. DRAM MR2 bits A[11:9] must be ‘000’ prior to entering
Write Leveling Mode.

π PM Tech
Procedure Description
PMA212816C
The Memory controller initiates Leveling mode of all DRAMs by setting bit A7 of MR1 to 1. When entering
write leveling mode, the DQ pins are in undefined driving mode. During write leveling mode, only DESELECT
commands are allowed, as well as an MRS command to change Qoff bit (MR1 [A12]) and an MRS command
to exit write leveling (MR1 [A7]). Upon exiting write leveling mode, the MRS command performing the exit
(MR1 [A7]=0) may also change MR1 bits of A12-A8 ,A2-A1. Since the controller levels one rank at a time,
the output of other ranks must be disabled by setting MR1 bit A12 to 1. The Controller may assert ODT after
tMOD, at which time the DRAM is ready to accept the ODT signal.
The Controller may drive DQS_t low and DQS_c high after a delay of tWLDQSEN, at which time the DRAM
has applied on-die termination on these signals. After tDQSL and tWLMRD, the controller provides a single
DQS_t, DQS_c edge which is used by the DRAM to sample CK_t - CK_c driven from controller.
tWLMRD(max) timing is controller dependent.
DRAM samples CK_t - CK_c status with rising edge of DQS_t - DQS_c and provides feedback on all the DQ
bits asynchronously after tWLO timing. There is a DQ output uncertainty of tWLOE defined to allow
mismatch on DQ bits. The tWLOE period is defined from the transition of the earliest DQ bit to the
corresponding transition of the latest DQ bit. There are no read strobes (DQS_t/DQS_c) needed for these
DQs. Controller samples incoming DQs and decides to increment or decrement DQS_t - DQS_c delay
setting and launches the next DQS_t/DQS_c pulse after some time, which is controller dependent. Once a 0
to 1 transition is detected, the controller locks DQS_t - DQS_c delay setting and write leveling is achieved for
the device.
Write Leveling Sequence (DQS Capturing CK LOW at T1 and CK HIGH at T2)
NOTE
1. The device drives leveling feedback on all DQs.
2. MRS: Load MR1 to enter write leveling mode.
3. diff_DQS is the differential data strobe. Timing reference points are the zero crossings.
4. DQS_t is shown with a solid line; DQS_c is shown with a dotted line.
5. CK_t is shown with a solid dark line; CK_c is shown with a dotted line.
6. DQS needs to fulfill minimum pulse width requirements, tDQSH (MIN) and tDQSL (MIN),
7. as defined for regular WRITEs; the maximum pulse width is system dependent.
8. tWLDQSEN must be satisfied following equation when using ODT:
• DLL = Enable, then tWLDQSEN > tMOD (MIN) + DODTLon + tADC
• DLL = Disable, then tWLDQSEN > tMOD (MIN) + tAONAS

π PM Tech
Write-Leveling Mode Exit
PMA212816C
Write leveling mode should be exited as follows:

1. After the last rising strobe edge (see ~T0), stop driving the strobe signals (see ~Tc0). Note that from this
point on, DQ pins are in undefined driving mode and will remain undefined, until tMOD after the respective
MR command (Te1).
2. Drive ODT pin LOW (tIS must be satisfied) and continue registering LOW (see Tb0).
3. After RTT is switched off, disable write leveling mode via the MRS command (see Tc2).
4. After tMOD is satisfied (Te1), any valid command can be registered. (MR commands can be issued after
tMRD [Td1]).
Write Leveling Exit

π PM Tech
Temperature-Controlled Refresh Mode
PMA212816C
Normal tREFI Refresh (TCR Disabled)

During normal operation, TCR mode disabled, the DRAM must have a Refresh command issued once every
tREFI, except for what is allowed by posting (see Refresh Command section). This means a refresh
command must be issued once every 3.9μs if TC greater than or equal to 85°C, and once every 7.8μs if TC
less than 85°C as shown in table below.
Normal Temperature Extended Temperature

Temperature
External Refresh Internal Refresh External Refresh Internal Refresh
Period Period Period Period
Tc < +45°C
7.8μs 7.8μs
+45°C ≦ Tc < +85°C 3.9μs 1 3.9μs 1
+85°C ≦ Tc < +95°C NA

NOTE
1. If TC is less than 85°C, the external refresh period can be 7.8μs instead of 3.9μs.
This mode is enabled and disabled by setting bit A3 in MR4. Two modes are supported that are selected by
bit A2 setting in MR4.
When TCR mode is enabled, the device will register the externally supplied REFRESH command and adjust
the internal refresh period to be longer than tREFI of the normal temperature range, when allowed, by
skipping REFRESH commands with the proper gear ratio. TCR mode has two ranges to select between the
normal temperature range and the extended temperature range; the correct range must be selected so the
internal control operates correctly. The DRAM must have the correct refresh rate applied externally; the
internal refresh rate is determined by the DRAM based upon the temperature.
TCR Mode - Normal Temperature Range

REFRESH commands are to be issued to DRAM with the refresh period equal to or shorter than tREFI of
normal temperature range (0°C to +85°C). In this mode, the system guarantees that the DRAM temperature
does not exceed 85°C. The DRAM may adjust the internal refresh period to be longer than tREFI of the
normal temperature range by skipping external REFRESH commands with the proper gear ratio when TC is
below 45°C. Not more than three fourths of external refresh commands are skipped at any temperature in
this mode. The internal refresh period is automatically adjusted inside the DRAM and the DRAM controller
does not need to provide any additional control.
TCR Mode - Extended Temperature Range
REFRESH commands should be issued to DRAM with the refresh period which is tREFI of extended
temperature range (+85°C to +95°C). In this mode, the system guarantees that the DRAM temperature does
not exceed 95°C. Even though the external Refresh supports the extended temperature range, the DRAM
will adjust its internal refresh period to tREFI of the normal temperature range by skipping external
REFRESH commands with proper gear ratio when operating in the normal temperature range (0°C to +85°C).
The DRAM may furthur adjust the internal refresh period to be longer than tREFI of the normal temperature
range by skipping external REFRESH commands with the proper gear ratio when TC is below 45°C. The
internal refresh period is automatically adjusted inside the DRAM and the DRAM controller does not need to
provide any additional control.

π PM Tech
Normal tREFI Refresh (TCR Enabled)
PMA212816C
Normal Temperature Extended Temperature

Temperature
External Refresh Internal Refresh External Refresh Internal Refresh
Period Period Period Period
Tc < +45°C > 7.8μs >> 3.9μs
7.8μs
+45°C ≦ Tc < +85°C 7.8μs 3.9μs 1 >3.9μs
+85°C ≦ Tc < +95°C NA 3.9μs

NOTE
1. If the external refresh period is slower than 3.9μs, the device will refresh internally at too slow of a refresh rate and will
violate refresh specifications.
Normal temperature mode

Once this mode is enabled by setting bit A3=1 and A2=0 in MR4, Refresh commands should be issued to
DDR4 SDRAM with the refresh period equal to or shorter than tREFI of normal temperature range (0°C -
85°C). In this mode, the system guarantees that the DRAM temperature does not exceed 85°C.
Below 45°C, DDR4 SDRAM may adjust internal refresh period to be longer than tREFI of the normal
temperature range by skipping external refresh commands with proper gear ratio. The internal refresh period
adjustment is automatically done inside the DRAM and user does not need to provide any additional control.
Extended temperature mode

Once this mode is enabled by setting bit A3=1 and A2=1 in MR4, Refresh commands should be issued to
DDR4 SDRAM with the refresh period equal to or shorter than tREFI of extended temperature range (85°C -
95°C).
In the normal temperature range (0°C - 85°C), DDR4 SDRAM adjusts its internal refresh period to tREFI of
the normal temperature range by skipping external refresh commands with proper gear ratio. Below 45°C,
DDR4 SDRAM may further adjust internal refresh period to be longer than tREFI of the normal temperature
range. The internal refresh period adjustment is automatically done inside the DRAM and user does not need
to provide any additional control.

π PM Tech
Fine Granularity Refresh Mode
PMA212816C
Mode Register and Command Truth Table

The Refresh cycle time (tRFC) and the average Refresh interval (tREFI) of DDR4 SDRAM can be
programmed by MRS command. The appropriate setting in the mode register will set a single set of Refresh
cycle time and average Refresh interval for the DDR4 SDRAM device (fixed mode), or allow the dynamic
selection of one of two sets of Refresh cycle time and average Refresh interval for the DDR4 SDRAM
device(on-the-fly mode). The on-the-fly mode must be enabled by MRS before any on-the-fly-Refresh
command can be issued.
MR3 definition for Fine Granularity Refresh Mode
A8 A7 A6 Fine Granularity Refresh
0 0 0 Normal mode (Fixed 1x)
0 0 1 Fixed 2x
0 1 0 Fixed 4x
0 1 1 Reserved
1 0 0 Reserved
1 0 1 On-the-fly 1x/2x
1 1 0 On-the-fly 1x/4x
1 1 1 Reserved
There are two types of OTF modes (1x/2x and 1x/4x modes) that are selectable by programming the
appropriate values into the mode register. When either of the two OTF modes is selected (‘A8=1’), DDR4
SDRAM evaluates the BG0 bit when a REFRESH command is issued, and depending on the status of BG0,
it dynamically switches its internal refresh configuration between 1x/2x or 1x/4x modes, then executes the
corresponding REFRESH operation.
REFRESH Command Truth Table
A0-9,
RAS_n CAS_n WE_n A10/
Function CS_n ACT_n BG1 BG0 BA0-1 A11-12, MR3[8:6]
/A15 /A14 /A13 AP
A16-20
Refresh
(Fixed rate) L H L L H V V V V V A8 = ‘0’
Refresh
(on-the-fly 1x) L H L L H V L V V V A8 = ‘1’
Refresh A8:A7:A6=‘1
(on-the-fly 2x) 01’
L H L L H V H V V V
Refresh A8:A7:A6=‘1
(on-the-fly 4x) 10’

π PM Tech
tREFI and tRFC Parameters
PMA212816C
The default Refresh rate mode is fixed 1x mode where Refresh commands should be issued with the normal
rate, i.e., tREFI1 =tREFI(base) (for Tcase<=85°C), and the duration of each refresh command is the normal
refresh cycle time (tRFC1). In 2x mode (either fixed 2x or on-the-fly 2x mode), Refresh commands should be
issued to the DRAM at the double frequency (tREFI2 = tREFI(base)/2) of the normal Refresh rate. In 4x
mode, Refresh command rate should be quadrupled (tREFI4 = tREFI(base)/4). Per each mode and
command type, tRFC parameter has different values.
The refresh command that should be issued at the normal refresh rate and has the normal refresh cycle
duration may be referred to as a REF1x command. The refresh command that should be issued at the
double frequency (tREFI2 = tREFI(base)/2) may be referred to as a REF2x command. Finally, the refresh
command that should be issued at the quadruple rate (tREFI4 = tREFI(base)/4) may be referred to as a
REF4x command.
In the Fixed 1x Refresh rate mode, only REF1x commands are permitted. In the Fixed 2x Refresh rate mode,
only REF2x commands are permitted. In the Fixed 4x Refresh rate mode, only REF4x commands are
permitted. When the on-the-fly 1x/2x Refresh rate mode is enabled, both REF1x and REF2x commands are
permitted. When the on-the-fly 1x/4x Refresh rate mode is enabled, both REF1x and REF4x commands are
permitted.
tREFI and tRFC Parameters

Refresh
Parameter 2Gb 4Gb 8Gb 16Gb Unit
Mode
tREFI(base) 7.8 7.8 7.8 7.8 μs
0°C <= TCASE <= 85°C tREFI(base) tREFI(base) tREFI(base) tREFI(base) μs

tREFI1
1X mode 85°C < TCASE <= 95°C tREFI(base)/2 tREFI(base)/2 tREFI(base)/2 tREFI(base)/2 μs
tRFC1(min) 160 260 350 550 ns
0°C <= TCASE <= 85°C tREFI(base)/2 tREFI(base)/2 tREFI(base)/2 tREFI(base)/2 μs

tREFI2
tRFC2(min) 110 160 260 350 ns
0°C <= TCASE <= 85°C tREFI(base)/4 tREFI(base)/4 tREFI(base)/4 tREFI(base)/4 μs

tREFI4
tRFC4(min) 90 110 160 260 ns

π PM Tech
Changing Refresh Rate
PMA212816C
If Refresh rate is changed by either MRS or on-the-fly, new tREFI and tRFC parameters would be applied
from the moment of the rate change. When the REF1x command is issued to the DRAM, then tREF1 and
tRFC1 are applied from the time that the command was issued. And then, when REF2x command is issued,
then tREF2 and tRFC2 should be satisfied.
On-the-fly Refresh Command Timing
The following conditions must be satisfied before the Refresh rate can be changed. Otherwise, data retention
of DDR4 SDRAM cannot be guaranteed.
1. In the fixed 2x Refresh rate mode or the on-the-fly 1x/2x Refresh mode, an even number of REF2x
commands must be issued to the DDR4 SDRAM since the last change of the Refresh rate mode with an
MRS command before the Refresh rate can be changed by another MRS command.
2. In the on-the-fly 1x/2x Refresh rate mode, an even number of REF2x commands must be issued between
any two REF1x commands.
3. In the fixed 4x Refresh rate mode or the on-the-fly 1x/4x Refresh mode, a multiple of-four number of
REF4x commands must be issued to the DDR4 SDRAM since the last change of the Refresh rate with an
MRS command before the Refresh rate can be changed by another MRS command.
4. In the on-the-fly 1x/4x Refresh rate mode, a multiple-of-four number of REF4x commands must be issued
between any two REF1x commands.
There are no special restrictions for the fixed 1x Refresh rate mode. Switching between fixed and on-the-fly
modes keeping the same rate is not regarded as a Refresh rate change.

π PM Tech
Usage with Temperature Controlled Refresh Mode
PMA212816C
If the temperature controlled refresh mode is enabled, then only the normal mode (fixed 1x mode, MR3 A[8:6]
= 000) is allowed. If any other refresh mode than the normal mode is selected, then the temperature
controlled refresh mode must be disabled.
Self Refresh Entry and Exit

DDR4 SDRAM can enter Self Refresh mode anytime in 1x, 2x and 4x mode without any restriction on the
number of Refresh commands that has been issued during the mode before the Self Refresh entry. However,
upon Self Refresh exit, extra Refresh command(s) may be required depending on the condition of the Self
Refresh entry. The conditions and requirements for the extra Refresh command(s) are defined as follows
1. There are no special restrictions on the fixed 1x Refresh rate mode.
2. In the fixed 2x Refresh rate mode or the enable-on-the-fly 1x/2x Refresh rate mode, it is recommended
that there should be an even number of REF2x commands before entry into Self Refresh since the last
Self Refresh exit or REF1x command or MRS command that set the refresh mode. If this condition is met,
no additional refresh commands are required upon Self Refresh exit.
3. In the case that this condition is not met, either one extra REF1x command or two extra REF2x
commands are required to be issued to the DDR4 SDRAM upon Self Refresh exit. These extra Refresh
commands are not counted toward the computation of the average refresh interval (tREFI).
4. In the fixed 4x Refresh rate mode or the enable-on-the-fly 1x/4x Refresh rate mode, it is recommended
that there should be a multiple-of-four number of REF4x commands before entry into Self Refresh since
the last Self Refresh exit or REF1x command or MRS command that set the refresh mode. If this
condition is met, no additional refresh commands are required upon Self Refresh exit. In the case that this
condition is not met, either one extra REF1x command or four extra REF4x commands are required to be
issued to the DDR4 SDRAM upon Self Refresh exit. These extra Refresh commands are not counted
toward the computation of the average refresh interval (tREFI).

π PM Tech
Multi Purpose Register (MPR)
PMA212816C
The multipurpose register (MPR) function, MPR Access Mode, is used to write/read specialized data to/from
the DRAM. The MPR consists of four logical Pages, MPR Page 0 through MPR Page 3, with each Page
having four 8-bit registers, MPR0 through MPR3.
MPR mode enable and Page selection is done with MRS commands. Data Bus Inversion (DBI) is not allowed
during MPR Read operation. Prior to issuing the MRS command, all banks must be in the idle state (all
banks precharged and tRP met). After MPR is enabled, any subsequent RD or RDA commands will be
redirected to a specific mode register.
Once the MPR Access Mode is enabled (MR3 A[2] = 1), only the following commands are allowed: MRS, RD,
RDA WR, WRA, DES, REF and Reset; RDA/WRA have the same functionality as RD/WR which means the
auto precharge part of RDA/WRA is ignored. The mode register location is specified with the READ
command using address bits. The MR is split into upper and lower halves to align with a burst length
limitation of 8. Power Down mode and Self-Refresh command are not allowed during MPR enable Mode.
No other command can be issued within tRFC after a REF command has been issued. 1x Refresh is only
allowed when MPR mode is Enable. While in MPR Access Mode, MPR read or write sequences must be
completed prior to a refresh command.
MR3 Setting for the MPR Access Mode
Address Operation Mode Description

00 = Serial ...........
01 = Parallel
A[12:11] MPR data read format 10 = Staggered ....
11 = Reserved
0 = Standard operation (MPR not enabled)
A2 MPR access
1 = MPR data flow enabled
00 = Page 0 ....
01 = Page 1
A[1:0] MPR page selection 10 = Page 2 ....
11 = Page 3
DRAM Address to MPR UI Translation
MPR Location [7] [6] [5] [4] [3] [2] [1] [0]
DRAM address – Ax A7 A6 A5 A4 A3 A2 A1 A0
MPR UI – UIx UI0 UI1 UI2 UI3 UI4 UI5 UI6 UI7

π PM Tech
MPR Page and MPRx Definitions
PMA212816C
MPR Page 0 – Read or Write (Data Patterns)

Address MPR Location [7] [6] [5] [4] [3] [2] [1] [0] note
00 = MPR0 0 1 0 1 0 1 0 1
01= MPR1 0 0 1 1 0 0 1 1 Read/Write
BA1:BA0
10 = MPR2 0 0 0 0 1 1 1 1 (default value)
11 = MPR3 0 0 0 0 0 0 0 0
MPR Page 1 – Read-only (CA Parity Error Log)
00 = MPR0 A[7] A[6] A[5] A[4] A[3] A[2] A[1] A[0]
CAS_n/ WE_n/
01 = MPR1 A[13] A[12] A[11] A[10] A[9] A[8]
A15 A14
Read-
BA1:BA0 RAS_n/
10 = MPR2 PAR ACT_n BG[1] BG[0] BA[1] BA[0] A[17] only
A16
CA Parity CA Parity Latency
CRC Error
11 = MPR3 Error C[2] C[1] C[0]
Status
Status MR5.A[2] MR5.A[1] MR5.A[0]
MPR Page 2 – Read-only (MRS Readout)
CRC
Temperature
RFU RFU RFU Write RTT_WR
Sensor Status
Enable
00 = MPR0
- - - - - MR2 MR2
- - - - - A12 A10 A9
VREFDQ Gear-
Training VREFDQ Training Value down
Range Enable
01= MPR1
MR6 MR6
BA1:BA0 Read-only
A6 A5 A4 A3 A2 A1 A0 A3
CAS Latency RFU CAS Write Latency
10 = MPR2 MR0 - MR2
A6 A5 A4 A2 - A5 A4 A3
RTT_WR RTT_Park Driver Impedance
11 = MPR3 MR1 MR5 MR2
A10 A9 A6 A8 A7 A6 A2 A1

π PM Tech
MPR Page 3 – Read-only (Vendor Purpose Only)
PMA212816C
01= MPR1 don’t care

BA1:BA0 Read-only
MPR Reads
MPR reads are supported using BL8 and BC4 modes. Burst length on-the-fly is not supported for MPR reads.
Data bus inversion (DBI) is not allowed during MPR READ operation; the device will ignore the Read DBI
enable setting in MR5 [12] when in MPR mode. READ commands for BC4 are supported with a starting
column address of A[2:0] = 000 or 100. After power-up, the content of MPR Page 0 has the default values.
MPR page 0 can be rewritten via an MPR WRITE command. The device maintains the default values unless
it is rewritten by the DRAM controller. If the DRAM controller does overwrite the default values (Page 0 only),
the device will maintain the new values unless re-initialized or there is power loss.
Timing in MPR mode:

 Reads (back-to-back) from Page 0 may use tCCD_S or tCCD_L timing between READ commands
 Reads (back-to-back) from Pages 1, 2, or 3 may not use tCCD_S timing between READ commands;
tCCD_L must be used for timing between READ commands
The following steps are required to use the MPR to read out the contents of a mode register (MPR Page x,
MPRy).
1. The DLL must be locked if enabled.
2. Precharge all; wait until tRP is satisfied.
3. MRS command to MR3[2] = 1 (Enable MPR data flow), MR3[12:11] = MPR read format, and MR3[1:0]
MPR page.
a. MR3[12:11] MPR read format:
1. 00 = Serial read format
2. 01 = Parallel read format
3. 10 = staggered read format
4. 11 = RFU
b. MR3[1:0] MPR page:
1. 00 = MPR Page 0
2. 01 = MPR Page 1
3. 10 = MPR Page 2
4. 11 = MPR Page 3
4. tMRD and tMOD must be satisfied.
5. Redirect all subsequent READ commands to specific MPRx location.
6. Issue RD or RDA command.
a. BA1 and BA0 indicate MPRx location:
1. 00 = MPR0
2. 01 = MPR1
3. 10 = MPR2

π PM Tech
4. 11 = MPR3
PMA212816C
b. A12/BC = 0 or 1; BL8 or BC4 fixed-only, BC4 OTF not supported.

1. If BL = 8 and MR0 A[1:0] = 01, A12/BC must be set to 1 during MPR READ commands.
c. A2 = burst-type dependant:
1. BL8: A2 = 0 with burst order fixed at 0, 1, 2, 3, 4, 5, 6, 7
2. BL8: A2 = 1 not allowed
3. BC4: A2 = 0 with burst order fixed at 0, 1, 2, 3, T, T, T, T
4. BC4: A2 = 1 with burst order fixed at 4, 5, 6, 7, T, T, T, T
d. A[1:0] = 00, data burst is fixed nibble start at 00.
e. Remaining address inputs, including A10, and BG1 and BG0 are "Don’t Care."
7. After RL = AL + CL, DRAM bursts data from MPRx location; MPR readout format determined by
MR3[A12,11,1,0].
8. Steps 5 through 7 may be repeated to read additional MPRx locations.
9. After the last MPRx READ burst, tMPRR must be satisfied prior to exiting.
10. Issue MRS command to exit MPR mode; MR3[2] = 0.
11. After the tMOD sequence is completed, the DRAM is ready for normal operation from the core (such as
ACT).
MPR Readout Format

The MPR read data format can be set to three different settings: serial, parallel, and staggered.
MPR Readout Serial Format

The serial format is required when enabling the MPR function to read out the contents of an MRx,
temperature sensor status, and the command address parity error frame. However, data bus calibration
locations (four 8-bit registers) can be programmed to read out any of the three formats. The DRAM is
required to drive associated strobes with the read data similar to normal operation (such as using MRS
preamble settings).
Serial format implies that the same pattern is returned on all DQ lanes, as shown the table below, which uses
values programmed into the MPR via [7:0] as 0111 1111.
x4 Device
Serial UI0 UI1 UI2 UI3 UI4 UI5 UI6 UI7

DQ0 0 1 1 1 1 1 1 1
DQ1 0 1 1 1 1 1 1 1
DQ2 0 1 1 1 1 1 1 1
DQ3 0 1 1 1 1 1 1 1

π PM Tech
x8 Device
PMA212816C

DQ0 0 1 1 1 1 1 1 1
DQ1 0 1 1 1 1 1 1 1
DQ2 0 1 1 1 1 1 1 1
DQ3 0 1 1 1 1 1 1 1
DQ4 0 1 1 1 1 1 1 1
DQ5 0 1 1 1 1 1 1 1
DQ6 0 1 1 1 1 1 1 1
DQ7 0 1 1 1 1 1 1 1
x16 Device

DQ0 0 1 1 1 1 1 1 1
DQ1 0 1 1 1 1 1 1 1
DQ2 0 1 1 1 1 1 1 1
DQ3 0 1 1 1 1 1 1 1
DQ4 0 1 1 1 1 1 1 1
DQ5 0 1 1 1 1 1 1 1
DQ6 0 1 1 1 1 1 1 1
DQ7 0 1 1 1 1 1 1 1
DQ8 0 1 1 1 1 1 1 1
DQ9 0 1 1 1 1 1 1 1
DQ10 0 1 1 1 1 1 1 1
DQ11 0 1 1 1 1 1 1 1
DQ12 0 1 1 1 1 1 1 1
DQ13 0 1 1 1 1 1 1 1
DQ14 0 1 1 1 1 1 1 1
DQ15 0 1 1 1 1 1 1 1
MPR Readout Parallel Format

Parallel format implies that the MPR data is returned in the first data UI and then repeated in the remaining
UIs of the burst, as shown in the table below. Data pattern location 0 is the only location used for the parallel
format. RD/RDA from data pattern locations 1, 2, and 3 are not allowed with parallel data return mode. In this
example, the pattern programmed in the data pattern location 0 is 0111 1111. The x4 configuration only
outputs the first four bits (0111 in this example). For the x16 configuration, the same pattern is repeated on
both the upper and lower bytes.

π PM Tech
x4 Device
PMA212816C

DQ0 0 0 0 0 0 0 0 0
DQ1 1 1 1 1 1 1 1 1
DQ2 1 1 1 1 1 1 1 1
DQ3 1 1 1 1 1 1 1 1
x8 Device

DQ0 0 0 0 0 0 0 0 0
DQ1 1 1 1 1 1 1 1 1
DQ2 1 1 1 1 1 1 1 1
DQ3 1 1 1 1 1 1 1 1
DQ4 1 1 1 1 1 1 1 1
DQ5 1 1 1 1 1 1 1 1
DQ6 1 1 1 1 1 1 1 1
DQ7 1 1 1 1 1 1 1 1
x16 Device

DQ0 0 0 0 0 0 0 0 0
DQ1 1 1 1 1 1 1 1 1
DQ2 1 1 1 1 1 1 1 1
DQ3 1 1 1 1 1 1 1 1
DQ4 1 1 1 1 1 1 1 1
DQ5 1 1 1 1 1 1 1 1
DQ6 1 1 1 1 1 1 1 1
DQ7 1 1 1 1 1 1 1 1
DQ8 0 0 0 0 0 0 0 0
DQ9 1 1 1 1 1 1 1 1
DQ10 1 1 1 1 1 1 1 1
DQ11 1 1 1 1 1 1 1 1
DQ12 1 1 1 1 1 1 1 1
DQ13 1 1 1 1 1 1 1 1
DQ14 1 1 1 1 1 1 1 1
DQ15 1 1 1 1 1 1 1 1

π PM Tech
MPR Readout Staggered Format
PMA212816C
Staggered format of data return is defined as the staggering of the MPR data across the lanes. In this mode,
an RD/RDA command is issued to a specific data pattern location and then the data is returned on the DQ
from each of the different data pattern locations.
For the x4 configuration, an RD/RDA to data pattern location 0 will result in data from location 0 being driven
on DQ0, data from location 1 being driven on DQ1, data from location 2 being driven on DQ2, and so on.
Similarly, an RD/RDA command to data pattern location 1 will result in data from location 1 being driven on
DQ0, data from location 2 being driven on DQ1, data from location 3 being driven on DQ2, and so on.
MPR Readout Staggered Format, x4
READ MPR0 Command READ MPR1 Command READ MPR2 Command READ MPR3 Command
Stagger UI[7:0] Stagger UI[7:0] Stagger UI[7:0] Stagger UI[7:0]
DQ0 MPR0 DQ0 MPR1 DQ0 MPR2 DQ0 MPR3
It is expected that the DRAM can respond to back-to-back RD/RDA commands to the MPR for all DDR4
frequencies so that a sequence (such as the one that follows) can be created on the data bus with no
bubbles or clocks between read data. In this case, the system memory controller issues a sequence of
RD(MPR0), RD(MPR1), RD(MPR2),RD(MPR3), RD(MPR0), RD(MPR1), RD(MPR2), and RD(MPR3).
MPR Readout Staggered Format, x4 – Consecutive READs
Stagger UI[7:0] UI[15:8] UI[23:16] UI[31:24] UI[39:32] UI[47:40] UI[55:48] UI[63:56]

DQ0 MPR0 MPR1 MPR2 MPR3 MPR0 MPR1 MPR2 MPR3
For the x8 configuration, the same pattern is repeated on the lower nibble as on the upper nibble. READs to
other MPR data pattern locations follow the same format as the x4 case.
MPR Readout Staggered Format, x8 and x16
x8 READ MPR0 Command x16 READ MPR0 Command x16 READ MPR0 Command
Stagger UI[7:0] Stagger UI[7:0] Stagger UI[7:0]
DQ0 MPR0 DQ0 MPR0 DQ8 MPR0

π PM Tech
MPR READ Timing
PMA212816C
NOTE
1. Multipurpose registers Read/Write Enable (MR3 A2 = 1). Redirect all subsequent readsand writes to MPR locations.
2. Address setting:
- A[1:0] = “00”(data burst order is fixed starting at nibble, always 00b here)
- A[2]= 0(For BL = 8, burst order is fixed at 0, 1, 2, 3, 4, 5, 6, 7)
- BA1 and BA0 indicate the MPR location
- A10 and other address pins are "Don’t Care" including BG1 and BG0.
- A12 is don’t carewhen MR0 A[1:0] = 00 or 10, and must be “1”when MR0 A[1:0] = 01
3. Multipurpose registers Read/Write Disable (MR3 A2 = 0).
4. Continue with regular DRAM command.
5. Parity latency (PL) is added to data output delay when C/A parity latency mode is enabled.
MPR Back-to-Back READ Timing
NOTE
1. tCCD_S = 4, Read Preamble = 1tCK
2. Address setting:
- A[1:0] = 00(data burst order is fixed starting at nibble, always 00b here)
- A[2] = 0(for BL = 8, burst order is fixed at 0, 1, 2, 3, 4, 5, 6, 7; for BC = 4, burst order isfixed at 0, 1, 2, 3, T, T, T, T)

π PM Tech
PMA212816C
- A12 is "Don’tCare" when MR0 A[1:0] = 00 or 10, and must be “1”when MR0 A[1:0] = 01
MPR READ-to-WRITE Timing
NOTE
1. Address setting:
- A[1:0] = 00(data burst order is fixed startingat nibble, always 00 here)
- A[2]= 0(for BL = 8, burst order is fixed at 0, 1, 2, 3, 4, 5, 6, 7)
- A12 is"Don’tCare" when MR0 A[1:0] = 00, and must be 1b when MR0 A[1:0] = 01
2. Address setting:
- A [7:0] = data for MPRA10 and other address pins are don’t care.
MPR Writes
MPR Access Mode allows 8-bit writes to the MPR location using the address bus A7:A0(refer to table: DRAM
Address to MPR UI 8-bit Registers Translation)
The following steps are required to use the MPR to write to mode register MPR Page 0.
1. The DLL must be locked prior to MPR Writes. DLL is Enabled, MR1 A0 = 1
2. Precharge all; wait until tRP is satisfied.
3. MR3 A2 = 1 (Enable MPR data flow) and MR3 A[1:0] = 00 (MPR Page 0); 01, 10, 11 = Not allowed.
4. Redirect all subsequent Write commands to specific MPR location.
5. tMRD and tMOD must be satisfied.
6. Issue WR or WRA command:
a. BA1 and BA0 indicate MPR location:
00 = MPR0
01 = MPR1
10 = MPR2

π PM Tech
11 = MPR3
PMA212816C
b. A[7:0] = data for MPR Page 0, mapped A[7:0] to UI[0:7] .

c. Remaining address inputs, including A10, BG0 and BG1 are don’t care.
7. tWR_MPR must be satisfied to complete MPR Write.
8. Steps 5 through 7 may be repeated these calibration writes and reads until data capture at memory
controller is optimized.
9. After the last MPR location Write, tMPRR must be satisfied prior to exiting.
10. Issue MRS command to exit MPR mode; MR3 A2= 0.
11. Wait until tMRD and tMOD are satisfied; Continue with regular DRAM commands like Activate.
MPR WRITE and WRITE-to-READ Timing
NOTE
1. Multipurpose registers Read/Write Enable (MR3 A2 = 1).
2. Address setting:
- A [7:0] = data for MPR
- A10 and other address pins are "Don’t Care"

π PM Tech
MPR Back-to-Back WRITE Timing
PMA212816C
NOTE
1. Address setting:
- A [7:0] = data for MPR
- A10 and other address pins are "Don’t Care"
REFRESH Command Timing
NOTE
1. Multipurpose registers Read/Write Enable (MR3 A2 = 1). Redirect all subsequent readand writes to MPR locations.
2. 1x refresh is only allowed when MPR mode is enabled.

π PM Tech
READ-to-REFRESH Command Timing
PMA212816C
NOTE
1. Address setting:
- A[1:0] = 00(data burst order is fixedstarting at nibble, always 00 here)
- A[2]= 0 (for BL = 8, burst order is fixed at 0, 1, 2, 3, 4, 5, 6, 7)
- A12 is "Don’tCare" when MR0 A[1:0] = 00 or 10, and must be 1b when MR0 A[1:0] = 01
WRITE-to-REFRESH Timing
NOTE
1. Address setting:
- BA1 and BA0 indicate the MPR location; - A [7:0] = data for MPR; - A10 and other address pins are "Don’t Care"

π PM Tech
Data Mask(DM), Data Bus Inversion (DBI) and TDQS
PMA212816C
DDR4 SDRAM supports Data Mask (DM) function and Data Bus Inversion (DBI) functionon in x8 and x16
DRAM configuration. x4 DDR4 SDRAM does not support DM and DBI function. x8 DDR4 SDRAM supports
TDQS function. x4 and x16 DDR4 SDRAM does not support TDQS function.
DM, DBI & TDQS functions are supported with dedicated one pin labeled as DM_n/DBI_n/TDQS_t. The pin
is bi-directional pin for DRAM. The DM_n/DBI _n pin is Active Low as DDR4 supports VDDQ reference
termination. TDQS function does not drive actual level on the pin.
DM, DBI & TDQS functions are programmable through DRAM Mode Register (MR). The MR bit location is bit
A11 in MR1 and bit A12:A10 in MR5.
Write operation: Either DM or DBI function can be enabled but both functions cannot be enabled
simultanteously. When both DM and DBI functions are disabled, DRAM turns off its input receiver and does
not expect any valid logic level.
Read operation: Only DBI function applies. When DBI function is disabled, DRAM turns off its output driver
and does not drive any valid logic level.
TDQS function: When TDQS function is enabled, DM & DBI functions are not supported. When TDQS
function is disabled, DM and DBI functions are supported. When TDQS function is enabled, the same
termination resistance function is applied to the TDQS_t/TDQS_c pins that is applied to DQS_t/DQS_c pins.
DBI vs. DM vs. TDQS Function Matrix
Read DBI Write DBI Data Mask (DM) TDQS (x8 only)
Disabled Disabled Disabled
MR5[11] = 0 MR5[10] = 0 MR1[11] = 0
Enabled (or Disabled)
Enabled Disabled Disabled
MR5[12]=1 (or
MR5[11] = 1 MR5[10] = 0 MR1[11] = 0
MR5[12] = 0)
Disabled Enabled Disabled
MR5[11] = 0 MR5[10] = 1 MR1[11] = 0
Disabled Disabled Disabled Enabled
MR5[12] = 0 MR5[11] = 0 MR5[10] = 0 MR1[11] = 1
DM function during Write operation: DRAM masks the write data received on the DQ inputs if DM_n was
sampled Low on a given byte lane. If DM_n was sampled High on a given byte lane, DRAM does not mask
the write data and writes into the DRAM core.
DBI function during Write operation: DRAM inverts write data received on the DQ inputs if DBI_n was
sampled Low on a given byte lane. If DBI_n was sampled High on a given byte lane, DRAM leaves the data
received on the DQ inputs non-inverted.
DBI function during Read operation: DRAM inverts read data on its DQ outputs and drives DBI_n pin Low
when the number of ‘0’ data bits within a given byte lane is greater than 4; otherwise DRAM does not invert
the read data and drives DBI_n pin High.
x8 DRAM Write DQ Frame Format
Data transfer
Function
0 1 2 3 4 5 6 7
DQ[7:0] Byte 0 Byte 1 Byte 2 Byte 3 Byte 4 Byte 5 Byte 6 Byte 7
DM_n or DM0 or DM1 or DM2 or DM3 or DM4 or DM5 or DM6 or DM7 or

DBI_n DBI0 DBI1 DBI2 DBI3 DBI4 DBI5 DBI6 DBI7

π PM Tech
x8 DRAM Read DQ Frame Format
PMA212816C
Data transfer
Function
0 1 2 3 4 5 6 7
DQ[7:0] Byte 0 Byte 1 Byte 2 Byte 3 Byte 4 Byte 5 Byte 6 Byte 7
DBI_n DBI0 DBI1 DBI2 DBI3 DBI4 DBI5 DBI6 DBI7
x16 DRAM Write DQ Frame Format
Data transfer
Function
0 1 2 3 4 5 6 7
DQL[7:0] LByte 0 LByte 1 LByte 2 LByte 3 LByte 4 LByte 5 LByte 6 LByte 7

DML_n or DML0 or DML1 or DML2 or DML3 or DML4 or DML5 or DML6 or DML7 or
DBIL_n DBIL0 DBIL1 DBIL2 DBIL3 DBIL4 DBIL5 DBIL6 DBIL7
DQU[7:0] UByte 0 UByte 1 UByte 2 UByte 3 UByte 4 UByte 5 UByte 6 UByte 7
DMU_n or DMU0 or DMU1 or DMU2 or DMU3 or DMU4 or DMU5 or DMU6 or DMU7 or
DBIU_n DBIU0 DBIU1 DBIU2 DBIU3 DBIU4 DBIU5 DBIU6 DBIU7
x16 DRAM Read DQ Frame Format
Data transfer
Function
0 1 2 3 4 5 6 7
DQL[7:0] LByte 0 LByte 1 LByte 2 LByte 3 LByte 4 LByte 5 LByte 6 LByte 7
DBIL_n DBIL0 DBIL1 DBIL2 DBIL3 DBIL4 DBIL5 DBIL6 DBIL7
DQU[7:0] UByte 0 UByte 1 UByte 2 UByte 3 UByte 4 UByte 5 UByte 6 UByte 7
DBIU_n DBIU0 DBIU1 DBIU2 DBIU3 DBIU4 DBIU5 DBIU6 DBIU7

π PM Tech
ZQ Calibration Commands
PMA212816C
ZQ Calibration command is used to calibrate DRAM RON and ODT values. The device needs a longer time
to calibrate the output driver and on-die termination circuits at initialization and a relatively smaller time to
perform periodic calibrations.
The ZQCL command is used to perform the initial calibration during the power-up initialization sequence.
This command may be issued at any time by the controller depending on the system environment. The
ZQCL command triggers the calibration engine inside the DRAM and, once calibration is achieved, the
calibrated values are transferred from the calibration engine to DRAM IO, which is reflected as an updated
output driver and on-die termination values.
The first ZQCL command issued after reset is allowed a timing period of tZQinit to perform the full calibration
and the transfer of values. All other ZQCL commands except the first ZQCL command issued after RESET
are allowed a timing period of tZQoper.
The ZQCS command is used to perform periodic calibrations to account for voltage and temperature
variations. A shorter timing window is provided to perform the calibration and transfer of values as defined by
timing parameter tZQCS. One ZQCS command can effectively correct a minimum of 0.5 % (ZQ correction)
of RON and RTT impedance error within 128 nCK for all speed bins assuming the maximum sensitivities
specified in the Output Driver and ODT Voltage and Temperature Sensitivity tables. The appropriate interval
between ZQCS commands can be determined from these tables and other application-specific parameters.
One method for calculating the interval between ZQCS commands, given the temperature (Tdriftrate) and
voltage (Vdriftrate) drift rates that the device is subjected to in the application, is illustrated. The interval could
be defined by the following formula:
ZQ correction
(Tsense x Tdrift rate) + (Vsense x Vdrift rate)
where Tsense = max(dRTTdT, dRONdTM) and Vsense = max(dRTTdV, dRONdVM) define the temperature
and voltage sensitivities.
For example, if Tsense = 1.5% / oC, Vsense = 0.15% / mV, Tdrift_rate = 1oC / sec and Vdrift_rate = 15 mV
/sec, then the interval between ZQCS commands is calculated as:
0.5
= 0.133≈ 128ms
(1.5 x 1) + (0.15 x 15)
No other activities should be performed on the DRAM channel by the controller for the duration of tZQinit,
tZQoper, or tZQCS. The quiet time on the DRAM channel allows accurate calibration of output driver and on-
die termination values. Once DRAM calibration is achieved, the device should disable the ZQ current
consumption path to reduce power.
All banks must be precharged and tRP met before ZQCL or ZQCS commands are issued by the controller.
ZQ calibration commands can also be issued in parallel to DLL lock time when coming out of self refresh.
Upon self refresh exit, the device will not perform an IO calibration without an explicit ZQ calibration
command. The earliest possible time for a ZQ calibration command (short or long) after self refresh exit is XS,
XS_Abort/ XS_FAST depending on operation mode.
In systems that share the ZQ resistor between devices, the controller must not allow any overlap of tZQoper,
tZQinit, or tZQCS between the devices.

π PM Tech
ZQ Calibration Timing
PMA212816C
NOTE
1. CKE must be continuously registered HIGH during the calibration procedure.
2. During ZQ Calibration, ODT signal must be held LOW and DRAM continues to provide RTT_PARK.
3. All devices connected to the DQ bus should be High Z or RTT_PARK during the calibration procedure.

π PM Tech
DQ VREF Training
PMA212816C
The DRAM internal DQ VREF specification parameters are operating voltage range, stepsize, VREF step time,
VREF full step time and VREF valid level.
The voltage operating range specifies the minimum required VREF setting range for DDR4 DRAM devices.
The minimum range is defined by VREFmax and VREFmin as depicted in the following figure.
VREF operating range (VREFmin, VREFmax)
The VREF stepsize is defined as the stepsize between adjacent steps. VREF stepsize ranges from 0.5%
VDDQ to 0.8% VDDQ.However, for a given design, DRAM has one value for VREF step size that falls within
the range.
The VREF set tolerance is the variation in the VREF voltage from the ideal setting. This accounts for
accumulated error over multiplesteps. There are two ranges for VREF set tolerance uncertainity. The range of
VREF set tolerance uncertainty is a function of number of steps n.
The VREF set tolerance is measured with respect to the ideal line which is based on the two endpoints.
Where the endpoints are at the min and max VREF values for a specified range. An illustration depicting an
example of the stepsize and VREF set tolerance is below.
Example of VREF Set Tolerance and Step Size

The VREF increment/decrement step times are defined by VREF_time. VREF_time is defined from t0 to t1,
where t1 is referenced to when the VREF voltage is at the final DC level within the VREF valid tolerance
(VREF_val_tol).
The VREF valid level is defined by VREF_val tolerance to qualify the step time t1. This parameter is used to
insure an adequate RC time constant behavior of the voltage level change after any VREF
increment/decrement adjustment.
This parameter is only applicable for DRAM component level validation/charachterization.
VREF_time is the time including up to VREF,min to VREF,max or VREF,max to VREF,min change in VREF
voltage.
t0 - is referenced to MRS command clock
t1 - is referenced to the VREF_val_tol
VREF_time for short and long timing diagram
A MRS command to the mode register bits 5:0 of MR6 are used to program the vref value. VREFDQ training
mode is enabled/disabled by A7 of MR6 and training range can be selected by A6 of MR6. When VREFDQ
training mode is entered/exited, the following parameter needs to be satisfied to prevent current consumption
and also stable operation.
VREFDQ Training Mode Entry and Exit Timing Diagram
NOTE
1. The MR command used to enter VrefDQ Calibration Mode treats MR6 a[5:0] as don’ t care while thenext subsequent MR
command sets VrefDQ values in MR6 A[5:0]
2. Depending on the step size of the latest programmed VREF value, Vref_timemust be satisfied before disabling VrefDQ
training mode.

π PM Tech
AC Parameters of DDR4 VREFDQ Training
PMA212816C
DDR4-1600,1866,2133,2400,2666,
Speed
2933,3200 Unit NOTE
Parameter Symbol MIN MAX
VREFDQ training
Enter VrefDQ training mode to the first write
tVREFDQE 150 - ns
or VREFDQ MRS command delay
Exit VrefDQ training mode to the first write
tVREFDQX 150 - ns
command delay
VREF step single stepsize increment case
VREF step single stepsize decrement case

π PM Tech
VREF full step from VREFmin to VREFmax case
PMA212816C
VREF full step from VREFmax to VREFmin case

π PM Tech
VREFDQ Supply and Calibration Ranges
PMA212816C
The DDR4 SDRAM internally generates its own VREFDQ. DRAM internal VREFDQ specification parameters:
voltage range, step-size, VREF step time, VREF full step time, and VREF valid level. The voltage operating
range specifies the minimum required VREF setting range for DDR4 DRAM devices. The minimum range is
defined by VREFDQ,min and VREFDQ,max. A calibration sequence should be performed by the DRAM
controller to adjust VREFDQ and optimize the timing and voltage margin of the DRAM data input receivers.
VREFDQ Specification
Prarmeter Symbol Min Typ Max Unit NOTE
VREF operating point

VREFDQ R1 60% - 92% VDDQ 1, 11
Range1
VREF operating point
VREFDQ R2 45% - 77% VDDQ 1, 11
Range2
VREF Stepsize VREF,step 0.50% 0.65% 0.80% VDDQ 2
-1.625% 0.00% 1.625% VDDQ 3,4,6

VREF Set Tolerance VREF,set_tol
-0.15% 0.00% 0.15% VDDQ 3,5,7
VREF Step Time VREF,time - - 150 ns 8,9,12
VREF Valid tolerance VREF_val_tol -0.15% 0.00% 0.15% VDDQ 10
NOTE
1. VREF (DC) voltage is referenced to VDDQ(DC). VDDQ(DC) is 1.2V.
2. VREF step size increment/decrement range. VREF at DC level.
3. VREF_new = VREF_old ± n × VREF_step; n = number of steps. If increment, use “+;” if decrement, use “-”.
4. The minimum value of VREF setting tolerance = VREF_new -1.625% × VDDQ. The maximum value of VREF setting
tolerance = VREF_new + 1.625% × VDDQ for n>4.
5. The minimum value of VREF setting tolerance = VREF_new -0.15% × VDDQ. The maximum value of VREF setting tolerance
= VREF_new + 0.15% × VDDQ for n>4.
6. Measured by recording the MIN and MAX values of the VREF output over the range, drawing a straight line between those
points, and comparing all other VREF output settings to that line.
7. Measured by recording the MIN and MAX values of the VREF output across four consecutive steps (n = 4), drawing a
straight line between those points, and comparing al VREF output settings to that line.
8. Time from MRS command to increment or decrement one step size for VREF.
9. Time from MRS command to increment or decrement one step size up to the full range of VREF.
10. Only applicable for DRAM component-level test/characterization purposes. Not applicable for normal mode of operation.
VREF valid qualifies the step times, which will be characterized at the component level.
11. DRAM range 1 or range 2 is set by the MRS bit, MR6 A6.
12. If the VREF monitor is enabled, VREF_time must be derated by: +10ns if DQ bus load is 0pF and anadditional +15ns/pF of
DQ bus loading.

π PM Tech
Per-DRAM Addressability (PDA Mode)
PMA212816C
DDR4 allows programmability of a given device on a rank. As an example, this feature can be used to
program different ODT or Vref values on DRAM devices on a given rank.
1. Before entering ‘per DRAM addressability (PDA)’ mode, the write leveling is required.
2. Before entering ‘per DRAM addressability (PDA)’ mode, the following Mode Register setting is possible.
- RTT_PARK MR5 {A8:A6} = Enable
- RTT_NOM MR1 {A10:A9:A8} = Enable
3. Enable ‘per DRAM addressability (PDA)’ mode using MR3 bit “A4=1”.
4. In the ‘per DRAM addressability’ mode, all MRS command is qualified with DQ0. DRAM captures DQ0 by
using DQS_c and DQS_t signals. If the value on DQ0 is 0 then the DRAM executes the MRS command.
If the value on DQ0 is 1, then the DRAM ignores the MRS command. The controller can choose to drive
all the DQ bits.
5. Program the desired devices and mode registers using MRS command and DQ0.
6. In the ‘per DRAM addressability’ mode, only MRS commands are allowed.
7. The mode register set command cycle time at PDA mode, AL + CWL + 3.5nCK + tMRD_PDA is required
to complete the write operation to the mode register and is the minimum time required between two MRS
commands.
8. Remove the DRAM from ‘per DRAM addressability’ mode by setting MR3 bit “A4=0”. (This command will
require DQ0=0 for x4, 8, x16).
Note: Removing a DRAM from per DRAM addressability mode will require programming the entire MR3
when the MRS command is issued. This may impact some per DRAM values programmed within a rank as
the exit command is sent to the rank. In order to avoid such a case, the PDA Enable/Disable Control bit is
located in a mode register that does not have any ‘per DRAM addressability’ mode controls.
In per DRAM addressability mode, DRAM captures DQ0 using DQS_t and DQS_c like normal write operation.
However, Dynamic ODT is not supported. Extra care is required for the ODT setting. If RTT_NOM MR1 [10:8]
= Enable, DDR4 SDRAM data termination need to be controlled by ODT pin and apply the same timing
parameters as defined in Direct ODT function.
Applied ODT Timing Parameter to PDA Mode
Symbol Parameter
DODTLon Direct ODT turn on latency
DODTLoff Direct ODT turn off latency
tADC RTT change timing skew
tAONAS Asynchronous RTT_NOM turn-on delay
tAOFAS Asynchronous RTT_NOM turn-off delay

π PM Tech
MRS w/ per DRAM addressability (PDA) issuing before MRS
PMA212816C
NOTE
RTT_PARK = Enable, RTT_NOM = Enable, Write Preamble Set = 2tCK and DLL = ON
MRS w/ per DRAM addressability (PDA) Exit
NOTE
RTT_PARK = Enable, RTT_NOM = Enable, Write Preamble Set = 2tCK and DLL = ON

π PM Tech
PDA using Burst Chop 4
PMA212816C
tPDA_S = tDS and tPDA_H = tDH for all DDR4 speed bins.
Since PDA mode may be used to program optimal VREF for the DRAM, the DRAM may incorrectly read DQ
level at the first DQS edge and the last falling DQS edge. It is recommended that DRAM samples DQ0 on
either the first falling or second rising DQS edges.
This will enable a common implementation between BC4 and BL8 modes on the DRAM. Controller is
required to drive DQ0 to a ‘Stable Low or High’ during the length of the data transfer for BC4 and BL8 cases.

π PM Tech
CAL Mode (CS_n to Command Address Latency)
PMA212816C
DDR4 supports Command Address Latency, CAL, function as a power savings feature. CAL is the delay in
clock cycles between CS_n and CMD/ADDR defined by MR4[A8:A6].
CAL gives the DRAM time to enable the CMD/ADDR receivers before a command is issued. Once the
command and the address are latched,the receivers can be disabled. For consecutive commands, the
DRAM will keep the receivers enabled for the duration of the command sequence.
Definition of CAL
CAL operational timing for consecutive command issues
The following tables show the timing requirements for tCAL and MRS settings at defferent data rates.
CS to Command Address Latency
Parameter Symbol DDR4-1600 DDR4-1866 DDR4-2133 DDR4-2400 Unit
CS to Command Address Latency tCAL max(3 nCK, 3.748 ns) nCK
NOTE Geardown mode is not supported for speed bins below DDR4-2666.
Parameter Symbol DDR4-2666 DDR4-2933 DDR4-3200 Unit
CS to Command Address Latency tCAL max(3 nCK, 3.748 ns) nCK
NOTE
In geardown mode, odd nCK values for tCAL are not supported, and nCK values must be rounded up to the next higher even
integer.

π PM Tech
MRS settings for CAL
PMA212816C
A8:A6 @ MR4 CAL(tCK cycles)

000 default(disable)
001 3
010 4
011 5
100 6
101 8
110 Reserve
111 Reserve
MRS Timings with Command/Address Latency enabled

When Command/Address latency mode is enabled, users must allow more time for MRS commands to take
effect. When CAL mode is enabled, or being enabled by an MRS command, the earliest the next valid
command can be issued is tMOD_CAL, where tMOD_CAL=tMOD+tCAL.
CAL enable timing - tMOD_CAL
NOTE
1. MRS command at Ta1 enables CAL mode
2. tMOD_CAL=tMOD+tCAL

π PM Tech
tMOD_CAL, MRS to valid command timing with CAL enabled
PMA212816C
NOTE
1. MRS at Ta1 may or may not modify CAL, tMOD_CAL is computed based on new tCAL setting.
2. tMOD_CAL=tMOD+tCAL.
When Command/Address latency is enabled or being entered, users must wait tMRD_CAL until the next
MRS command can be issued. tMRD_CAL=tMOD+tCAL.
CAL enabling MRS to next MRS command, tMRD_CAL
NOTE
1. MRS command at Ta1 enables CAL mode.
2. tMRD_CAL=tMOD+tCAL.
tMRD_CAL, mode register cycle time with CAL enabled
NOTE
1. MRS at Ta1 may or may not modify CAL, tMRD_CAL is computed based on new tCAL setting.
2. tMRD_CAL=tMOD+tCAL.

π
CRC
PM Tech PMA212816C
CRC Polynomial and logic equation

DDR4 supports CRC for write operation, and doesn’t support CRC for read operation.
The CRC polynomial used by DDR4 is the ATM-8 HEC, X^8+X^2+X^1+1
A combinatorial logic block implementation of this 8-bit CRC for 72-bits of data contains 272 two-input XOR
gates contained in eight 6 XOR gate deep trees.
The CRC polynomial and combinatorial logic used by DDR4 is the same as used on GDDR5.
CRC Error Detection Details
ERROR TYPE DETECTION CAPABILITY

Random Single Bit Error 100%
Random Double Bit Error 100%
Random Odd Count Error 100%
Random one Multi-bit UI vertical column error detection excluding DBI bits 100%
CRC Combinatorial Logic Equations

module CRC8_D72;
// polynomial: (0 1 2 8)
// data width: 72
// convention: the first serial data bit is D[71]
//initial condition all 0 implied
// "^" = XOR
function [7:0]
nextCRC8_D72;
input [71:0] Data;
input [71:0] D;
reg [7:0] CRC;
begin
D = Data;
NewCRC[0] = D[69] ^ D[68] ^ D[67] ^ D[66] ^ D[64] ^ D[63] ^ D[60] ^

D[56] ^ D[54] ^ D[53] ^ D[52] ^ D[50] ^ D[49] ^ D[48] ^
D[45] ^ D[43] ^ D[40] ^ D[39] ^ D[35] ^ D[34] ^ D[31] ^
D[30] ^ D[28] ^ D[23] ^ D[21] ^ D[19] ^ D[18] ^ D[16] ^
D[14] ^ D[12] ^ D[8] ^ D[7] ^ D[6] ^ D[0] ;
NewCRC[1] = D[70] ^ D[66] ^ D[65] ^ D[63] ^ D[61] ^ D[60] ^ D[57] ^

D[56] ^ D[55] ^ D[52] ^ D[51] ^ D[48] ^ D[46] ^ D[45] ^
D[44] ^ D[43] ^ D[41] ^ D[39] ^ D[36] ^ D[34] ^ D[32] ^
D[30] ^ D[29] ^ D[28] ^ D[24] ^ D[23] ^ D[22] ^ D[21] ^
D[20] ^ D[18] ^ D[17] ^ D[16] ^ D[15] ^ D[14] ^ D[13] ^
D[12] ^ D[9] ^ D[6] ^ D[1] ^ D[0];
NewCRC[2] = D[71] ^ D[69] ^ D[68] ^ D[63] ^ D[62] ^ D[61] ^ D[60] ^

D[58] ^ D[57] ^ D[54] ^ D[50] ^ D[48] ^ D[47] ^ D[46] ^
D[44] ^ D[43] ^ D[42] ^ D[39] ^ D[37] ^ D[34] ^ D[33] ^
D[29] ^ D[28] ^ D[25] ^ D[24] ^ D[22] ^ D[17] ^ D[15] ^
D[13] ^ D[12] ^ D[10] ^ D[8] ^ D[6] ^ D[2] ^ D[1] ^ D[0];

NewCRC[3] = D[70] ^ D[69] ^ D[64] ^ D[63] ^ D[62] ^ D[61] ^ D[59] ^

D[58] ^ D[55] ^ D[51] ^ D[49] ^ D[48] ^ D[47] ^ D[45] ^
D[44] ^ D[43] ^ D[40] ^ D[38] ^ D[35] ^ D[34] ^ D[30] ^
D[29] ^ D[26] ^ D[25] ^ D[23] ^ D[18] ^ D[16] ^ D[14] ^
D[13] ^ D[11] ^ D[9] ^ D[7] ^ D[3] ^ D[2] ^ D[1];
NewCRC[4] = D[71] ^ D[70] ^ D[65] ^ D[64] ^ D[63] ^ D[62] ^ D[60] ^

D[59] ^ D[56] ^ D[52] ^ D[50] ^ D[49] ^ D[48] ^ D[46] ^
D[45] ^ D[44] ^ D[41] ^ D[39] ^ D[36] ^ D[35] ^ D[31] ^
D[30] ^ D[27] ^ D[26] ^ D[24] ^ D[19] ^ D[17] ^ D[15] ^
D[14] ^ D[12] ^ D[10] ^ D[8] ^ D[4] ^ D[3] ^ D[2];
NewCRC[5] = D[71] ^ D[66] ^ D[65] ^ D[64] ^ D[63] ^ D[61] ^ D[60] ^

D[57] ^ D[53] ^ D[51] ^ D[50] ^ D[49] ^ D[47] ^ D[46] ^
D[45] ^ D[42] ^ D[40] ^ D[37] ^ D[36] ^ D[32] ^ D[31] ^
D[28] ^ D[27] ^ D[25] ^ D[20] ^ D[18] ^ D[16] ^ D[15] ^
D[13] ^ D[11] ^ D[9] ^ D[5] ^ D[4] ^ D[3];
NewCRC[6] = D[67] ^ D[66] ^ D[65] ^ D[64] ^ D[62] ^ D[61] ^ D[58] ^

D[54] ^ D[52] ^ D[51] ^ D[50] ^ D[48] ^ D[47] ^ D[46] ^
D[43] ^ D[41] ^ D[38] ^ D[37] ^ D[33] ^ D[32] ^ D[29] ^
D[28] ^ D[26] ^ D[21] ^ D[19] ^ D[17] ^ D[16] ^ D[14] ^
D[12] ^ D[10] ^ D[6] ^ D[5] ^ D[4];
NewCRC[7] = D[68] ^ D[67] ^ D[66] ^ D[65] ^ D[63] ^ D[62] ^ D[59] ^

D[55] ^ D[53] ^ D[52] ^ D[51] ^ D[49] ^ D[48] ^ D[47] ^
D[44] ^ D[42] ^ D[39] ^ D[38] ^ D[34] ^ D[33] ^ D[30] ^
D[29] ^ D[27] ^ D[22] ^ D[20] ^ D[18] ^ D[17] ^ D[15] ^
D[13] ^ D[11] ^ D[7] ^ D[6] ^ D[5];
nextCRC8_D72 = NewCRC;
CRC Data Bit Mapping

The Controller generates the CRC checksum and forms the write data frames as below tables.The DRAM
checks for an error in a received code word D[71:0] by comparing the received checksum against the
computed checksum and reports errors using the signal if there is a mis-match.DRAM can write data to the
DRAM core without waiting for CRC check for full writes. If bad data is written to the DRAM core, then
controller will retry the transaction and overwrite the bad data. Controller is responsible for data coherency.
CRC Data Mapping for x4 Devices, BL8

A x4 device has a CRC tree with 32 input bits. The input for the upper 40 bits D[71:32] are ‘1’s.
Transfer Burst Bit

0 1 2 3 4 5 6 7 8 9
DQ0 D0 D1 D2 D3 D4 D5 D6 D7 CRC0 CRC4

π PM Tech
PMA212816C
For a x8 DRAM the controller must send 1’s in the transfer 9 if CRC is enabled and must send 1’s in transfer
8 and transfer 9 of thelane if DBI function is enabled. A x8 device has a CRC tree with 72 input bits. The 8
bits D[71:64] are used if either Write DBI or DM is enabled. Note that Write DBI and DM function cannot be
enabled simultaneously. If both Write DBI and DM is disabled then the inputs of the 8 bits D[71:64] are ‘1’s.
Transfer Burst Bit

0 1 2 3 4 5 6 7 8 9
DQ0 D0 D1 D2 D3 D4 D5 D6 D7 CRC0 1
DQ1 D8 D9 D10 D11 D12 D13 D14 D15 CRC1 1
DQ2 D16 D17 D18 D19 D20 D21 D22 D23 CRC2 1
DQ3 D24 D25 D26 D27 D28 D29 D30 D31 CRC3 1
DQ4 D32 D33 D34 D35 D36 D37 D38 D39 CRC4 1
DQ5 D40 D41 D42 D43 D44 D45 D46 D47 CRC5 1
DQ6 D48 D49 D50 D51 D52 D53 D54 D55 CRC6 1
DQ7 D56 D57 D58 D59 D60 D61 D62 D63 CRC7 1
DM_n
D64 D65 D66 D67 D68 D69 D70 D71 1 1
DBI_n

For a x16 DRAM the controller must send 1’s in the transfer 9 if CRC is enabled and must send 1’s in
transfer 8 and transfer 9 of the DBIL_n and DBIU_n lanes if DBI function is enabled. A x16 device has two
identical CRC trees with 72 input bits each. The upper 8 bits are used if either Write DBI or DM is enabled.
Note that Write DBI and DM function cannot be enabled simultaneously. If both Write DBI and DM is
disabled then the inputs of the upper 8 bits D[143:136] and D[71:64] are ‘1’s.
A x16 device is treated as two x8 devices; a x16 device will have two identical CRC trees implemented.
CRC[7:0] covers data bits D[71:0], and CRC[15:8] covers data bits D[143:72].
Transfer Burst Bit
0 1 2 3 4 5 6 7 8 9
DQ0 D0 D1 D2 D3 D4 D5 D6 D7 CRC0 1
DQ1 D8 D9 D10 D11 D12 D13 D14 D15 CRC1 1
DQ2 D16 D17 D18 D19 D20 D21 D22 D23 CRC2 1
DQ3 D24 D25 D26 D27 D28 D29 D30 D31 CRC3 1
DQ4 D32 D33 D34 D35 D36 D37 D38 D39 CRC4 1
DQ5 D40 D41 D42 D43 D44 D45 D46 D47 CRC5 1
DQ6 D48 D49 D50 D51 D52 D53 D54 D55 CRC6 1
DQ7 D56 D57 D58 D59 D60 D61 D62 D63 CRC7 1
LDM_n
D64 D65 D66 D67 D68 D69 D70 D71 1 1
LDBI_n
DQ8 D72 D73 D74 D75 D76 D77 D78 D79 CRC8 1
DQ9 D80 D81 D82 D83 D84 D85 D86 D87 CRC9 1
DQ10 D88 D89 D90 D91 D92 D93 D94 D95 CRC10 1
DQ11 D96 D97 D98 D99 D100 D101 D102 D103 CRC11 1
DQ12 D104 D105 D106 D107 D108 D109 D110 D111 CRC12 1
DQ13 D112 D113 D114 D115 D116 D117 D118 D119 CRC13 1
DQ14 D120 D121 D122 D123 D124 D125 D126 D127 CRC14 1
DQ15 D128 D129 D130 D131 D132 D133 D134 D135 CRC15 1
UDM_n
D136 D137 D138 D139 D140 D141 D142 D143 1 1
UDBI_n

π PM Tech
CRC Error Handling
PMA212816C
CRC Error mechanism shares the same Alert_n signal for reporting errors on writes to DRAM. The controller
has no way to distinguish between CRC errors and Command/Address/Parity errors other than to read the
DRAM mode registers. This is a very time consuming process in a multi-rank configuration.
To speed up recovery for CRC errors, CRC errors are only sent back as a pulse. The minimum pulse-width is
2 clocks. The latency to Alert_n signal is defined as tCRC_ALERT in the figure below.
DRAM will set CRC Error Clear bit in A4 of MR5 to '1' and CRC Error Status bit in MPR3 of page1 to '1' upon
detecting a CRC error. The CRC Error Clear bit remains set at '1' until the host clears it explicitly using an
MRS command.
The controller upon seeing an error as a pulse width will retry the write transactions. The controller
understands the worst case delay for Alert_n (during init) and can backup the transactions accordingly or the
controller can be made more intelligent and try to correlate the write CRC error to a specific rank or a
transaction. The controller is also responsible for opening any pages and ensuring that retrying of writes is
done in a coherent fashion.
The pulse width may be seen longer than two clocks at the controller if there are multiple CRC errors as the
Alert_n is a daisy chain bus.
CRC Error Reporting
NOTE
1. CRC ALERT_PW is specified from the point where the DRAM starts to drive the signal low to the point where the DRAM
driver releases and the controller starts to pull the signal up.
2. Timing diagram applies to x4, x8, and x16 devices.
CRC Error Timing Parmeters
DDR4-1600 DDR4-1866 DDR4-2133 DDR4-2400

Parameter Symbol Unit
min max min max min max min max
CRC error to ALERT_n latency tCRC_ALERT - 13 - 13 - 13 - 13 ns
CRC ALERT_n pulse width CRC ALERT_PW 6 10 6 10 6 10 6 10 nCK
DDR4-2666 DDR4-2933 DDR4-3200

Parameter Symbol Unit
min max min max min max
CRC error to ALERT_n latency tCRC_ALERT - 13 - 13 - 13 ns
CRC ALERT_n pulse width CRC ALERT_PW 6 10 6 10 6 10 nCK

π PM Tech
CRC Frame Format with BC4
PMA212816C
DDR4 SDRAM supports CRC function for Write operation for Burst Chop 4 (BC4). The CRC function is
programmable using DRAM mode register and can be enabled for writes.
When CRC is enabled the data frame length is fixed at 10UI for both BL8 and BC4 operations. DDR4
SDRAM also supports burst length on the fly with CRC enabled. This is enabled using mode register.
CRC Data Bit Mapping for x4 Devices (BC4)

For a x4 device, the CRC tree inputs are 16 data bits; and the inputs for the remaining bits are 1.
When A2 = 1, data bits D[7:4] are used as inputs for D[3:0], D[15:12] are used as inputs to D[11:8] and so
forth for the CRC tree.
Transfer Burst Bit
0 1 2 3 4 5 6 7 8 9
A2=0
DQ0 D0 D1 D2 D3 1 1 1 1 CRC0 CRC4
DQ1 D8 D9 D10 D11 1 1 1 1 CRC1 CRC5
DQ2 D16 D17 D18 D19 1 1 1 1 CRC2 CRC6
DQ3 D24 D25 D26 D27 1 1 1 1 CRC3 CRC7
A2=1
DQ0 D4 D5 D6 D7 1 1 1 1 CRC0 CRC4
DQ1 D12 D13 D14 D15 1 1 1 1 CRC1 CRC5
DQ2 D20 D21 D22 D23 1 1 1 1 CRC2 CRC6
DQ3 D28 D29 D30 D31 1 1 1 1 CRC3 CRC7

For a x8 device, the CRC tree inputs are 36 data bits in transfer’s four through seven as 1’s.
When A2 = 0, the input bits D[67:64]) are used if DBI_n or DM_n functions are enabled; if DBI_n and DM_n
are disabled, then D[67:64]) are 1.
When A2 = 1, data bits D[7:4] are used as inputs for D[3:0], D[15:12] are used as inputs to D[11:8], and so
forth, for the CRC tree. The input bits D[71:68]) are used if DBI_n or DM_n functions are enabled; if DBI_n
and DM_n are disabled, then D[71:68]) are 1.
Transfer Burst Bit
0 1 2 3 4 5 6 7 8 9
A2=0
DQ0 D0 D1 D2 D3 1 1 1 1 CRC0 1
DQ1 D8 D9 D10 D11 1 1 1 1 CRC1 1
DQ2 D16 D17 D18 D19 1 1 1 1 CRC2 1
DQ3 D24 D25 D26 D27 1 1 1 1 CRC3 1
DQ4 D32 D33 D34 D35 1 1 1 1 CRC4 1
DQ5 D40 D41 D42 D43 1 1 1 1 CRC5 1
DQ6 D48 D49 D50 D51 1 1 1 1 CRC6 1
DQ7 D56 D57 D58 D59 1 1 1 1 CRC7 1
DM_n
D64 D65 D66 D67 1 1 1 1 1 1
DBI_n

π PM Tech
Transfer Burst Bit
PMA212816C
0 1 2 3 4 5 6 7 8 9
A2=1
DQ0 D4 D5 D6 D7 1 1 1 1 CRC0 1
DQ1 D12 D13 D14 D15 1 1 1 1 CRC1 1
DQ2 D20 D21 D22 D23 1 1 1 1 CRC2 1
DQ3 D28 D29 D30 D31 1 1 1 1 CRC3 1
DQ4 D36 D37 D38 D39 1 1 1 1 CRC4 1
DQ5 D44 D45 D46 D47 1 1 1 1 CRC5 1
DQ6 D52 D53 D54 D55 1 1 1 1 CRC6 1
DQ7 D60 D61 D62 D63 1 1 1 1 CRC7 1
DM_n
D68 D69 D70 D71 1 1 1 1 1 1
DBI_n

There are two identical CRC trees for x16 devices, each have CRC tree inputs of 36 bits.
When A2 = 0, input bits D[67:64] are used if DBI_n or DM_n functions are enabled; if DBI_n and DM_n are
disabled, then D[67:64] are 1s. The input bits D[139:136] are used if DBI_n or DM_n functions are enabled; if
DBI_n and DM_n are disabled, then D[139:136] are 1.
When A2 = 1, data bits D[7:4] are used as inputs for D[3:0], D[15:12] are used as inputs for D[11:8], and so
forth, for the CRC tree. Input bits D[71:68] are used if DBI_n or DM_n functions are enabled; if DBI_n and
DM_n are disabled, then D[71:68] are 1s. The input bits D[143:140] are used if DBI_n or DM_n functions are
enabled; if DBI_n and DM_n are disabled, then D[143:140] are 1.
Transfer Burst Bit
0 1 2 3 4 5 6 7 8 9
A2=0
DQ0 D0 D1 D2 D3 1 1 1 1 CRC0 1
DQ1 D8 D9 D10 D11 1 1 1 1 CRC1 1
DQ2 D16 D17 D18 D19 1 1 1 1 CRC2 1
DQ3 D24 D25 D26 D27 1 1 1 1 CRC3 1
DQ4 D32 D33 D34 D35 1 1 1 1 CRC4 1
DQ5 D40 D41 D42 D43 1 1 1 1 CRC5 1
DQ6 D48 D49 D50 D51 1 1 1 1 CRC6 1
DQ7 D56 D57 D58 D59 1 1 1 1 CRC7 1
LDM_n
D64 D65 D66 D67 1 1 1 1 1 1
LDBI_n
DQ8 D72 D73 D74 D75 1 1 1 1 CRC8 1
DQ9 D80 D81 D82 D83 1 1 1 1 CRC9 1
DQ10 D88 D89 D90 D91 1 1 1 1 CRC10 1
DQ11 D96 D97 D98 D99 1 1 1 1 CRC11 1
DQ12 D104 D105 D106 D107 1 1 1 1 CRC12 1
DQ13 D112 D113 D114 D115 1 1 1 1 CRC13 1
DQ14 D120 D121 D122 D123 1 1 1 1 CRC14 1
DQ15 D128 D129 D130 D131 1 1 1 1 CRC15 1
UDM_n
D136 D137 D138 D139 1 1 1 1 1 1
UDBI_n

π PM Tech
Transfer Burst Bit
PMA212816C
0 1 2 3 4 5 6 7 8 9
A2=1
DQ0 D4 D5 D6 D7 1 1 1 1 CRC0 1
DQ1 D12 D13 D14 D15 1 1 1 1 CRC1 1
DQ2 D20 D21 D22 D23 1 1 1 1 CRC2 1
DQ3 D28 D29 D30 D31 1 1 1 1 CRC3 1
DQ4 D36 D37 D38 D39 1 1 1 1 CRC4 1
DQ5 D44 D45 D46 D47 1 1 1 1 CRC5 1
DQ6 D52 D53 D54 D55 1 1 1 1 CRC6 1
DQ7 D60 D61 D62 D63 1 1 1 1 CRC7 1
LDM_n
D68 D69 D70 D71 1 1 1 1 1 1
LDBI_n
DQ8 D76 D77 D78 D79 1 1 1 1 CRC8 1
DQ9 D84 D85 D86 D87 1 1 1 1 CRC9 1
DQ10 D92 D93 D94 D95 1 1 1 1 CRC10 1
DQ11 D100 D101 D102 D103 1 1 1 1 CRC11 1
DQ12 D108 D109 D110 D111 1 1 1 1 CRC12 1
DQ13 D116 D117 D118 D119 1 1 1 1 CRC13 1
DQ14 D124 D125 D126 D127 1 1 1 1 CRC14 1
DQ15 D132 D133 D134 D135 1 1 1 1 CRC15 1
UDM_n
D140 D141 D142 D143 1 1 1 1 1 1
UDBI_n
CRC equations for x8 device in BC4 mode with A2=0 are as follows:
CRC[0] = D[69]=1 ^ D[68]=1 ^ D[67] ^ D[66] ^ D[64] ^ D[63]=1 ^ D[60]=1 ^ D[56] ^ D[54]=1 ^ D[53]=1 ^
D[52]=1 ^ D[50] ^ D[49] ^ D[48] ^ D[45]=1 ^ D[43] ^ D[40] ^ D[39]=1 ^ D[35] ^ D[34] ^ D[31]=1^ D[30]=1 ^
D[28]=1 ^ D[23]=1 ^ D[21]=1 ^ D[19] ^ D[18] ^ D[16] ^ D[14]=1 ^ D[12]=1 ^ D[8] ^ D[7]=1 ^ D[6] =1 ^ D[0] ;
CRC[1] = D[70]=1 ^ D[66] ^ D[65] ^ D[63]=1 ^ D[61]=1 ^ D[60]=1 ^ D[57] ^D[56] ^ D[55]=1 ^ D[52]=1 ^ D[51]
^ D[48] ^ D[46]=1 ^ D[45]=1 ^ D[44]=1 ^ D[43] ^ D[41] ^ D[39]=1 ^ D[36]=1 ^ D[34] ^ D[32] ^ D[30]=1 ^
D[29]=1 ^ D[28]=1 ^ D[24] ^ D[23]=1 ^ D[22]=1 ^ D[21]=1 ^ D[20]=1 ^ D[18] ^ D[17] ^ D[16] ^ D[15]=1 ^
D[14]=1 ^ D[13]=1 ^ D[12]=1 ^ D[9] ^ D[6]=1 ^ D[1] ^ D[0];
CRC[2] = D[71]=1 ^ D[69]=1 ^ D[68]=1 ^ D[63]=1 ^ D[62]=1 ^ D[61]=1 ^ D[60]=1 ^ D[58] ^ D[57] ^ D[54]=1 ^
D[50] ^ D[48] ^ D[47]=1 ^ D[46]=1 ^ D[44]=1 ^ D[43] ^ D[42] ^ D[39]=1 ^ D[37]=1 ^ D[34] ^ D[33] ^ D[29]=1 ^
D[28]=1 ^ D[25] ^ D[24] ^ D[22]=1 ^ D[17] ^ D[15]=1 ^ D[13]=1 ^ D[12]=1 ^ D[10] ^ D[8] ^ D[6]=1 ^ D[2] ^ D[1]
^ D[0];
CRC[3] = D[70]=1 ^ D[69]=1 ^ D[64] ^ D[63]=1 ^ D[62]=1 ^ D[61]=1 ^ D[59] ^ D[58] ^ D[55]=1 ^ D[51] ^ D[49]
^ D[48] ^ D[47]=1 ^ D[45]=1 ^ D[44]=1 ^ D[43] ^ D[40] ^ D[38]=1 ^ D[35] ^ D[34] ^ D[30]=1 ^ D[29]=1 ^ D[26]
^ D[25] ^ D[23]=1 ^ D[18] ^ D[16] ^ D[14]=1 ^ D[13]=1 ^ D[11] ^ D[9] ^ D[7]=1 ^ D[3] ^ D[2] ^ D[1];
CRC[4] = D[71]=1 ^ D[70]=1 ^ D[65] ^ D[64] ^ D[63]=1 ^ D[62]=1 ^ D[60]=1 ^ D[59] ^ D[56] ^ D[52]=1 ^ D[50]
^ D[49] ^ D[48] ^ D[46]=1 ^ D[45]=1 ^ D[44]=1 ^ D[41] ^ D[39]=1 ^ D[36]=1 ^ D[35] ^ D[31]=1 ^ D[30]=1 ^
D[27] ^ D[26] ^ D[24] ^ D[19] ^ D[17] ^ D[15]=1 ^ D[14]=1 ^ D[12]=1 ^ D[10] ^ D[8] ^ D[4]=1 ^ D[3] ^ D[2];
CRC[5] = D[71]=1 ^ D[66] ^ D[65] ^ D[64] ^ D[63]=1 ^ D[61]=1 ^ D[60]=1 ^ D[57] ^ D[53]=1 ^ D[51] ^ D[50] ^
D[49] ^ D[47]=1 ^ D[46]=1 ^ D[45]=1 ^ D[42] ^ D[40] ^ D[37]=1 ^ D[36]=1 ^ D[32] ^ D[31]=1 ^ D[28]=1 ^ D[27]
^ D[25] ^ D[20]=1 ^ D[18] ^ D[16] ^ D[15]=1 ^ D[13]=1 ^ D[11] ^ D[9] ^ D[5]=1 ^ D[4]=1 ^ D[3];

CRC[6] = D[67] ^ D[66] ^ D[65] ^ D[64] ^ D[62]=1 ^ D[61]=1 ^ D[58] ^ D[54]=1 ^ D[52]=1 ^ D[51] ^ D[50] ^
D[48] ^ D[47]=1 ^ D[46]=1 ^ D[43] ^ D[41] ^ D[38]=1 ^ D[37]=1 ^ D[33] ^ D[32] ^ D[29]=1 ^ D[28]=1 ^ D[26] ^
D[21]=1 ^ D[19] ^ D[17] ^ D[16] ^ D[14]=1 ^ D[12]=1 ^ D[10] ^ D[6]=1 ^ D[5]=1 ^ D[4]=1;
CRC[7] = D[68]=1 ^ D[67] ^ D[66] ^ D[65] ^ D[63]=1 ^ D[62]=1 ^ D[59] ^ D[55]=1 ^ D[53]=1 ^ D[52]=1 ^ D[51]
^ D[49] ^ D[48] ^ D[47]=1 ^ D[44]=1 ^ D[42] ^ D[39]=1 ^ D[38]=1 ^ D[34] ^ D[33] ^ D[30]=1 ^ D[29]=1 ^ D[27]
^ D[22]=1 ^ D[20]=1 ^ D[18] ^ D[17] ^ D[15] =1^ D[13]=1 ^ D[11] ^ D[7]=1 ^ D[6]=1 ^ D[5]=1;
CRC equations for x8 device in BC4 mode with A2=1 are as follows:
CRC[0] = 1 ^ 1 ^ D[71] ^ D[70] ^ D[68] ^ 1 ^ 1 ^ D[60] ^ 1 ^ 1 ^ 1 ^ D[54] ^ D[53] ^ D[52] ^ 1 ^ D[47] ^ D[44] ^
1 ^ D[39] ^ D[38] ^ 1^ 1 ^ 1 ^ 1 ^ 1 ^ D[23] ^ D[22] ^ D[20] ^ 1 ^ 1 ^ D[12] ^ 1 ^ 1 ^ D[4] ;
CRC[1] = 1 ^ D[70] ^ D[69] ^ 1 ^ 1 ^ 1 ^ D[61] ^ D[60] ^ 1 ^ 1 ^ D[55] ^ D[52] ^ 1 ^ 1 ^ 1 ^ D[47] ^ D[45] ^ 1 ^

1 ^ D[38] ^ D[36] ^ 1 ^ 1 ^ 1 ^ D[28] ^ 1 ^ 1 ^ 1 ^ 1 ^ D[22] ^ D[21] ^ D[20] ^1 ^ 1 ^1 ^ 1 ^ D[13] ^ 1 ^ D[5] ^
D[4];
CRC[2] = 1 ^ 1 ^ 1 ^1 ^1 ^ 1 ^ 1 ^ D[62] ^ D[61] ^ 1 ^ D[54] ^ D[52] ^ 1 ^ 1 ^ 1 ^ D[47] ^ D[46] ^ 1 ^ 1 ^ D[38]

^ D[37] ^ 1 ^ 1 ^ D[29] ^ D[28] ^ 1 ^ D[21] ^ 1 ^ 1 ^ 1 ^ D[14] ^ D12] ^1 ^ D[6] ^ D[5] ^ D[4];
CRC[3] = 1 ^ 1 ^ D[68] ^ 1 ^ 1 ^ 1 ^ D[63] ^ D[62] ^ 1 ^ D[55] ^ D[53] ^ D[52] ^ 1 ^ 1 ^ 1 ^ D[47] ^ D[44] ^ 1 ^

D[39] ^ D[38] ^ 1 ^ 1 ^ D[30] ^ D[29] ^ 1 ^ D[22] ^ D[20] ^ 1 ^ 1 ^ D[15] ^ D[13] ^ 1 ^ D[7] ^ D[6] ^ D[5];
CRC[4] = 1 ^1 ^ D[69] ^ D[68] ^ 1 ^ 1 ^ 1 ^ D[63] ^ D[60] ^ 1 ^ D[54] ^ D[53] ^ D[52] ^ 1 ^1 ^ 1 ^ D[45] ^ 1 ^ 1

^ D[39] ^1 ^ 1 ^ D[31] ^ D[30] ^ D[28] ^ D[23] ^ D[21] ^ 1 ^ 1 ^ 1 ^ D[14] ^ D[12] ^ 1 ^ D[7] ^ D[6];
CRC[5] = 1 ^ D[70] ^ D[69] ^ D[68] ^ 1 ^ 1 ^ 1 ^ D[61] ^ 1 ^ D[55] ^ D[54] ^ D[53] ^ 1 ^ 1 ^ 1 ^ D[46] ^ D[44] ^
1 ^ 1 ^ D[36] ^ 1 ^ 1 ^ D[31] ^ D[29] ^ 1 ^ D[22] ^ D[20] ^ 1 ^ 1 ^ D[15] ^ D[13] ^ 1 ^ 1 ^ D[7];
CRC[6] = D[71] ^ D[70] ^ D[69] ^ D[68] ^ 1 ^ 1 ^ D[62] ^ 1 ^ 1 ^ D[55] ^ D[54] ^ D[52] ^ 1 ^1 ^ D[47] ^ D[45] ^
1 ^ 1 ^ D[37] ^ D[36] ^1 ^ 1 ^ D[30] ^ 1 ^ D[23] ^ D[21] ^ D[20] ^ 1 ^ 1 ^ D[14] ^ 1 ^ 1 ^ 1;
CRC[7] = 1 ^ D[71] ^ D[70] ^ D[69] ^ 1 ^ 1 ^ D[63] ^ 1 ^ 1 ^ 1 ^ D[55] ^ D[53] ^ D[52] ^ 1 ^ 1 ^ D[46] ^ 1 ^ 1 ^

D[38] ^ D[37] ^ 1 ^ 1 ^D[31] ^ 1 ^ 1 ^ D[22] ^ D[21] ^ 1^ 1 ^ D[15] ^ 1 ^ 1 ^ 1;
Simultaneous DM and CRC Functionality

When both DM and Write CRC are enabled in the DRAM mode register, the DRAM calculates CRC before
sending the write data into the array. If there is a CRC error, the DRAM blocks the write operation and
discards the data.
CRC Simultaneous Operation Restrictions

When write CRC is enabled, neither MPR writes nor per-DRAM mode is allowed.

π PM Tech
Command Address Parity (CA Parity)
PMA212816C
[A2:A0] of MR5 are defined to enable or disable C/A Parity in the DRAM. The default state of the C/A Parity
bits is disabled. If C/A parity is enabled by programming a non-zero value to C/A Parity Latency in the mode
register (the Parity Error bit must be set to zero when enabling C/A any Parity mode), then the DRAM has to
ensure that there is no parity error before executing the command. The additional delay for executing the
commands versus a parity disabled mode is programmed in the mode register when C/A Parity is enabled
(Parity Latency) and is applied to all commands.When C/A Parity is enabled, only DES is allowed between
valid commands to prevent DRAM from any malfunctioning. CA Parity Mode is supported when DLL-on
Mode is enabled, use of CA Parity Mode when DLL-off Mode is enabled is not allowed.
C/A Parity signal (PAR) covers ACT_n, RAS_n, CAS_n, WE_n and the address bus including bank address
and bank group bits. The control signals CKE, ODT and CS_n are not included. (e.g. for a 4 Gbit x4
monolithic device, parity is computed across BG0, BG1, BA1, BA0, A16/ RAS_n, A15/CAS_n, A14/WE_n,
A13-A0 and ACT_n). (DRAM should internally treat any unused address pins as 0’s, e.g., if a common die
has stacked pins but the device is used in a monolithic application then the address pins used for stacking
should internally be treated as 0’s)
The convention of parity is even parity i.e. valid parity is defined as an even number of ones across the
inputs used for parity computation combined with the parity signal. In other words the parity bit is chosen so
that the total number of 1’s in the transmitted signal, including the parity bit, is even.
If a DRAM detects a C/A parity error in any command as qualified by CS_n then it must perform the following
steps:
 Ignore the erroneous command. Commands in max NnCK window (tPAR_UNKNOWN) prior to the
erroneous command are not guaranteed to be executed. When a READ command in this NnCK window is
not executed, the DRAM does not activate DQS outputs.
 Log the error by storing the erroneous command and address bits in the error log. (MPR page1)
 Set the Parity Error Status bit in the mode register to ‘1’. The Parity Error Status bit must be set before the
ALERT_n signal is released by the DRAM (i.e. tPAR_ALERT_ON + tPAR_ALERT_PW(min)).
 Assert the ALERT_n signal to the host (ALERT_n is active low) within tPAR_ALERT_ON time.
 Wait for all in-progress commands to complete. These commands were received tPAR_UNKOWN before
the erroneous command. If a parity error occurs on a command issued between the tXS_Fast and tXS
window after self-refresh exit then the DRAM may delay the de-assertion of ALERT_n signal as a result of
any internal on going refresh.
 Wait for tRAS_min before closing all the open pages. The DRAM is not executing any commands during
the window defined by (tPAR_ALERT_ON + tPAR_ALERT_PW).
 After tPAR_ALERT_PW_min has been satisfied, the DRAM may de-assert ALERT_n.
 After the DRAM has returned to a known pre-charged state it may de-assert ALERT_n.
 After (tPAR_ALERT_ON + tPAR_ALERT_PW), the DRAM is ready to accept commands for normal
operation. Parity latency will be in effect, however, parity checking will not resume until the memory
controller has cleared the Parity Error Status bit by writing a ‘0’(the DRAM will execute any erroneous
commands until the bit is cleared).
 It is possible that the DRAM might have ignored a refresh command during the (tPAR_ALERT_ON +
tPAR_ALERT_PW) window or the refresh command is the first erroneous frame so it is recommended that
the controller issues extra refresh cycles as needed.
 The Parity Error Status bit may be read anytime after (tPAR_ALERT_ON + tPAR_ALERT_PW) to
determine which DRAM had the error. The DRAM maintains the Error Log for the first erroneous command
until the Parity Error Status bit is reset to ‘0’.

Mode Register for C/A Parity Error is defined as follows. C/A Parity Latency bits are write only, Parity Error
Status bit is read/write and error logs are read only bits. The controller can only program the Parity Error
Status bit to ‘0’. If the controller illegally attempts to write a ‘1’ to the Parity Error Status bit the DRAM does
not guarantee that parity will be checked. The DRAM may opt to block the controller from writing a ‘1’ to the
Parity Error Status bit.
DDR4 SDRAM supports MR bit for ‘Persistent Parity Error Mode’. This mode is enabled by setting MR5
A9=High and when it is enabled, DRAM resumes checking CA Parity after the alert_n is deasserted, even if
Parity Error Status bit is set as High. If multiple errors occur before the Error Status bit is cleared the Error
log in MPR page 1 should be treated as ‘Don’t Care’. In ‘Persistent Parity Error Mode’ the Alert_n pulse will
be asserted and deasserted by the DRAM as defined with the min. and max. value for tPAR_ALERT_PW.
The controller must issue DESELECT com-mands once it detects the Alert_n signal, this response time is
defined as tPAR_ALERT_RSP.
Mode Register Setting for CA Parity

C/A Parity C/A Parity Error Status Parity Persistent Mode
Latency Speed bins Errant C/A Frame
MR5[4] MR5[9]
1
000 = Disabled -
0=clear 0 = Disabled
001= 4 Clocks 1600,1866,2133
C2-C0, ACT_n, BG1, BG0, BA0,
010= 5 Clocks 2400 BA1, PAR, A17, A16/RAS_n,
A15/CAS_n, A14/WE_n, A13:A0
011= 6 Clocks RFU 1=Error 1 = Enabled
100= 8 Clocks RFU
NOTE
1. Parity Latency is applied to all commands.
2. Parity Latency can be changed only from a C/A Parity disabled state, i.e. a direct change from PL=3 ・ PL=4 is not allowed.
Correct sequence is PL=3  Disabled  PL=4
3. Parity Latency is applied to write and read latency. Write Latency = AL+CWL+PL. Read Latency = AL+CL+PL.
The following figure captures the flow of events on the C/A bus and the ALERT_n signal.
Normal CA Parity Error Checking Operation
NOTE
1. DRAM is emptying queues. Precharge all and parity checking off until Parity Error Statusbit cleared.
2. Command execution is unknown; the corresponding DRAM internal state change mayor may not occur. The DRAM
controller should consider both cases and make sure thatthe command sequence meets the specifications.
3. Normal operation with parity latency (CA Parity Persistent Error Mode disabled). Paritychecking off until Parity Error Status
bit cleared.

π PM Tech
Persistent CA Parity Error Checking Operation
PMA212816C
NOTE
1. DRAM is emptying queues. Precharge all and parity check re-enable finished bytPAR_ALERT_PW.
3. Normal operation with parity latency and parity checking (CA Parity Persistent ErrorMode enabled).
CA Parity Error Checking - PDE/PDX
NOTE
1. Deselect command only allowed.
2. Error could be Precharge or Activate.
3. Normal operation with parity latency (CA Parity Persistent Error Mode disabled). Paritychecking is off until Parity Error
Status bit cleared.
5. Deselect command only allowed; CKE may go high prior to Td2 as long as DES commandsare issued.

π PM Tech
CA Parity Error Checking - SRE Attempt
PMA212816C
NOTE
1. Deselect command only allowed.
2. Self Refresh command error. DRAM masks the intended SRE command and enters PrechargePower Down.
3. Normal operation with parity latency (CA Parity Persistent Error Mode disabled). Paritychecking is off until Parity Error
Status bit cleared.
4. Controller cannot disable clock until it has been able to have detected a possible C/AParity error.
6. Deselect command only allowed; CKE may go high prior to Tc2 as long as DES commandsare issued.
CA Parity Error Checking – SRX

π
NOTE
PM Tech PMA212816C
1. Self Refresh Abort = Disable: MR4 A[9] = 0.

2. Input commands are bounded by tXSDLL, tXS, tXS_ABORT and tXS_FAST timing.
4. Normal operation with parity latency (CA Parity Persistent Error Mode disabled). Paritychecking off until Parity Error Status
bit cleared.
5. Only MRS (limited to those described in the Self-Refresh Operation section), ZQCS, orZQCL command allowed
6. Valid commands not requiring a locked DLL.
7. Valid commands requiring a locked DLL.
8. This figure shows the case from which the error occurred after tXS_FAST. An error mayalso occur after tXS_ABORT and
tXS.
Command/Address parity entry and exit timings

When in CA Parity mode, including entering and exiting CA Parity mode, users must wait tMRD_PAR before
issuing another MRS command, and wait tMOD_PAR before any other commands.
tMOD_PAR = tMOD + PL
tMRD_PAR = tMOD + PL
For CA parity entry, PL in the equations above is the parity latency programmed with the MRS command
entering CA parity mode.
For CA parity exit, PL in the equations above is the programmed parity latency prior to the MRS command
exiting CA parity mode.
Parity entry timing example - tMRD_PAR
NOTE tMRD_PAR = tMOD + N; where N is the programmed parity latency.

π PM Tech
Parity entry timing example - tMOD_PAR
PMA212816C
NOTE tMOD_PAR = tMOD + N; where N is the programmed parity latency.
Parity exit timing example - tMRD_PAR
NOTE tMRD_PAR = tMOD + N; where N is the programmed parity latency.
Parity exit timing example - tMOD_PAR
NOTE tMOD_PAR = tMOD + N; where N is the programmed parity latency.

π PM Tech
CA Parity Error Log Readout
PMA212816C
Address MPR Location [7] [6] [5] [4] [3] [2] [1] [0]
00=MPR0 A7 A6 A5 A4 A3 A2 A1 A0
CAS_n/
01=MPR1 WE_n/A14 A13 A12 A11 A10 A9 A8
A15
BA1:BA0 = 0:1 RAS_n/
10=MPR2 PAR ACT_n BG1 BG0 BA1 BA0 A17
A16
CRC Error CA Parity
11=MPR3 CA Parity Latency C2 C1 C0
Status Error Status
NOTE
1. MPR used for CA parity error log readout is enabled by setting A[2] in MR3
2. For higher density of DRAM, where A[17] is not used, MPR2[1] should be treated as don’t care.
3. If a device is used in monolithic application, where C[2:0] are not used, then MPR3[2:0] should be treated as don’t care.

π PM Tech
Gear Down Mode
PMA212816C
The following sequence represents for the gear down mode. The DRAM defaults in 1/2 rate(1N) clock mode
and utilizes a low frequency MRS command followed by a sync pulse to align the proper clock edge for
operating the control lines CS_n, CKE and ODT in 1/4 rate(2N) mode. For operation in 1/2 rate mode, no
MRS command for geardown or sync pulse is required. DRAM defaults is in 1/2 rate mode.
General sequence for operation in geardown during initialization
1. DRAM defaults to a 1/2 rate(1N mode) internal clock at power up/reset
2. Assertion of reset
3. Assertion of CKE enables the DRAM
4. MRS is accessed with a low frequency N*tck MRS geardown CMD Ntck static MRS command qualified
by 1N CS_n
5. DRAM controller sends a 1N sync pulse with a low frequency N*tCK NOP CMD; tSYNC_GEAR is an
even number of clocks; sync pulse on even clock boundary from MRS CMD.
6. Initialization sequence, including the expiration of tDLLK and tZQinit, starts in 2N mode after
tCMD_GEAR from 1N Sync Pulse.
Initialization sequence, including the expiration of tDLLK and tZQinit, starts in 2N mode after tCMD_GEAR
from 1N Sync Pulse.
1. MRS is set to 1, via MR3[3], with a low-frequency N × tCK gear-down MRS command.
a. The NtCK static MRS command is qualified by 1N CS_n, which meets tXS or tXS_ABORT.
b. Only a REFRESH command may be issued to the DRAM before the NtCK static MRS command.
2. The DRAM controller sends a 1N sync pulse with a low-frequency N × tCK NOP command.
a. tSYNC_GEAR is an even number of clocks.
b. The sync pulse is on even edge clock boundary from the MRS command.
3. A valid command not requiring locked DLL is available in 2N mode after tCMD_GEAR from the 1N sync
pulse.
a. A valid command requiring locked DLL is available in 2N mode after tXSDLL or tDLLK from the 1N
sync pulse.
4. If operation is in 1N mode after self refresh exit, N × tCK MRS command or sync pulse is not required
during self refresh exit. The minimum exit delay to the first valid command is tXS, or tXS_ABORT.
The DRAM may be changed from 1/4(2N) rate to 1/2(1N) rate by entering self refresh mode, which will reset
to 1N mode. Changing from 1/4(2N) to 1/2(1N) by any other means can result in loss of data and make
operation of the DRAM uncertain.
When operating in 2N gear-down mode, the following MR settings apply:

• CAS latency (MR0[6:4,2]): Even number of clocks
• Write recovery and read to precharge (MR0[11:9]): Even number of clocks
• Additive latency (MR1[4:3]): CL - 2
• CAS WRITE latency (MR2 A[5:3]): Even number of clocks
• CS to command/address latency mode (MR4[8:6]): Even number of clocks
• CA parity latency mode (MR5[2:0]): Even number of clocks

The diagram below illustrates the sequence for control operation in 2N mode during power up.
Gear down (2N) mode entry sequence during initialization
The diagram below represents the operation of geardown (1/2 rate to 1/4 rate) mode during normal operation
with CKE and Reset set high.
Gear down (2N) mode entry sequence during normal operation

π PM Tech
Gear down (2N) mode entry sequence after self refresh exit (SRX)
PMA212816C
NOTE CKE High Assert to Gear Down Enable Time (tXS, tXS_Abort) depend on MR setting. A correspondence of
tXS/tXS_Abort and MR Setting is as follows.
- MR4[A9] = 0 : tXS
- MR4[A9] = 1 : tXS_Abort
Comparison Timing Diagram Between Geardown Disable and Enable
NOTE
1. BL=8, tRCD=CL=16
2. DOUT n = data-out from column n.
3. DES commands are shown for ease of illustration; other commands may be valid at these times.
4. CA Parity = Disable, CS to CA Latency = Disable, Read DBI = Disable.

π PM Tech
Bank Access Operation
PMA212816C
DDR4 supports bank grouping: x4/x8 DRAMs have four Bank Groups (BG[1:0]) and each bank group is
comprised of four sub-banks (BA[1:0]); x16 DRAMs have two Bank Groups (BG[0]) and each bank group is
comprised of made up of four sub-banks. Bank accesses to different banks groups require less time delay
between accesses than Bank accesses to within the same banks group. Bank accesses to different bank
groups require tCCD_S short delay between commands while Bank accesses within the same bank group
require tCCD_L long delay between commands.
tCCD Timing (WRITE to WRITE Example)
NOTE
1. tCCD_S : CAS_n-to-CAS_n delay (short) : Applies to consecutive CAS_n to different Bank Group (i.e. T0 to T4)
2. tCCD_L : CAS_n-to-CAS_n delay (long) : Applies to consecutive CAS_n to the same Bank Group (i.e. T4 to T10)
tCCD Timing (READ to READ Example)
NOTE
1. tCCD_S : CAS_n-to-CAS_n delay (short) : Applies to consecutive CAS_n to different Bank Group (i.e. T0 to T4)
2. tCCD_L : CAS_n-to-CAS_n delay (long) : Applies to consecutive CAS_n to the same Bank Group (i.e. T4 to T10)

π PM Tech
tRRD Timing
PMA212816C
NOTE
1. tRRD_S : ACTIVATE to ACTIVATE Command period (short) : Applies to consecutive ACTIVATE Commands to different
Bank Group (i.e. T0 to T4)
2. tRRD_L : ACTIVATE to ACTIVATE Command period (long) : Applies to consecutive ACTIVATE Commands to the different
Banks of the same Bank Group (i.e. T4 to T10)
tFAW Timing
NOTE
tFAW : Four activate window

π PM Tech
tWTR_S Timing (WRITE to READ, Different Bank Group, CRC and DM Disabled)
PMA212816C
NOTE
tWTR_S : Delay from start of internal write transaction to internal read command to a different Bank Group
tWTR_L Timing (WRITE to READ, Same Bank Group, CRC and DM Disabled)
NOTE
tWTR_L : Delay from start of internal write transaction to internal read command to the same Bank Group

π PM Tech
Programmable Preamble
PMA212816C
Write Preamble
DDR4 supports a programmable write preamble. Write preamble modes of 1 tCK and 2 tCK are selectable
by MR4 [A12].
CWL needs to be incremented by 1nCK when 2tCK preamble is enabled.
When operating in 2tCK Write Preamble Mode, tWTR and tWR must be programmed to a value 1 clock
greater than the tWTR and tWR setting supported in the applicable speed bin.
The timing diagrams below illustrate 1 and 2 tCK preamble scenarios for consecutive write commands with
tCCD timing of 4, 5 and 6 nCK, respectively. Setting tCCD to 5nCK is not allowed in 2 tCK preamble mode.
tCCD=4 (AL=PL=0)
1tCK mode
2tCK mode

π PM Tech
tCCD=5 (AL=PL=0)
PMA212816C
1tCK mode
2tCK mode : tCCD=5 is not allowed in 2tCK mode
tCCD=6 (AL=PL=0)
1tCK mode
2tCK mode

π PM Tech
Read Preamble
PMA212816C
DDR4 supports a programmable read preamble. Read preamble modes of 1 tCK and 2 tCK are selectable by
MR4 [A11].
Read preamble modes of 1 tCK and 2 tCK are shown below.
Read Preamble Training

DDR4 supports READ preamble training via MPR reads; that is, READ preamble training is allowed only
when the DRAM is in the MPR access mode. The READ preamble training mode can be used by the DRAM
controller to train or "read level" its DQS receivers.
READ preamble training is entered via an MRS command (MR4[A10] = 1 is enabled and MR4[A10] = 0 is
disabled). After the MRS command is issued to enable READ preamble training, the DRAM DQS signals are
driven to a valid level by the time tSDO is satisfied. During this time, the data bus DQ signals are held quiet,
that is, driven HIGH.
The DQS_t signal remains driven LOW and the DQS_c signal remains driven HIGH until an MPR Page0
READ command is issued (MPR0 through MPR3 determine which pattern is used), and when CAS latency
(CL) has expired, the DQS signals will toggle normally depending on the burst length setting. To exit READ
preamble training mode, an MRS command must be issued, MR4[A10] = 0.

π PM Tech
Postamble
PMA212816C
Write Postamble
DDR4 supports a fixed Write postamble. Write postamble nominal is 0.5tck for preamble modes 1tCK, 2tCK
are shown below.
Read Postamble
DDR4 supports a fixed Read postamble. Read postamble nominal is 0.5tck for preamble modes 1tCK, 2tCK
are shown below.

π PM Tech
ACTIVATE Command
PMA212816C
The ACTIVATE command is used to open (activate) a row in a particular bank for subsequent access. The
values on the BG[1:0] inputs select the bank group, the BA[1:0] inputs select the bank within the bank group,
and the address provided on inputs A[17:0] selects the row within the bank. This row remains active (open)
for accesses until a PRECHARGE command is issued to that bank. A PRECHARGE command must be
issued before opening a different row in the same bank.
Bank-to-bank command timing for ACTIVATE commands uses two different timing parameters, depending
on whether the banks are in the same or different bank group. tRRD_S (short) is used for timing between
banks located in different bank groups. tRRD_L (long) is used for timing between banks located in the same
bank group.
Another timing restriction for consecutive ACTIVATE commands [issued at tRRD (MIN)] is tFAW (fifth
activate window). Because there is a maximum of four banks in a bank group, the tFAW parameter applies
across different bank groups (five ACTIVATE commands issued at tRRD_L (MIN) to the same bank group
would be limited by tRC).
Precharge Command
The PRECHARGE command is used to deactivate the open row in a particular bank or the open row in all
banks. The bank(s) will be available for a subsequent row activation for a specified time (tRP) after the
PRECHARGE command is issued. An exception to this is the case of concurrent auto precharge, where a
READ or WRITE command to a different bank is allowed as long as it does not interrupt the data transfer in
the current bank and does not violate any other timing parameters.
After a bank is precharged, it is in the idle state and must be activated prior to any READ or WRITE
commands being issued to that bank. A PRECHARGE command is allowed if there is no open row in that
bank (idle state) or if the previously open row is already in the process of precharging. However, the
precharge period will be determined by the last PRECHARGE command issued to the bank.
The auto precharge feature is engaged when a READ or WRITE command is issued with A10 HIGH. The
auto precharge feature uses the RAS lockout circuit to internally delay the PRECHARGE operation until the
ARRAY RESTORE operation has completed. The RAS lockout circuit feature allows the PRECHARGE
operation to be partially or completely hidden during burst READ cycles when the auto precharge feature is
engaged. The PRECHARGE operation will not begin until after the last data of the burst write sequence is
properly stored in the memory array.

Read Operation
READ Timing Definitions
Read timing shown below is applied when the DLL is enabled and locked.
Rising data strobe edge parameters:

 tDQSCK (MIN)/(MAX) describes the allowed range for a rising data strobe edge relative to CK.
 tDQSCK is the actual position of a rising strobe edge relative to CK.
 tQSH describes the DQS differential output HIGH time.
 tDQSQ describes the latest valid transition of the associated DQ pins.
 tQH describes the earliest invalid transition of the associated DQ pins.
Falling data strobe edge parameters:

 tQSL describes the DQS differential output LOW time.
tDQSQ, both rising/falling edges of DQS, no tAC defined.
READ Timing Definition

π PM Tech
READ Timing – Clock to Data Strobe Relationship
PMA212816C
Clock to Data Strobe relationship is shown below6 and is applied when the DLL is enabled and locked.
 tDQSCK min/max describes the allowed range for a rising data strobe edge relative to CK_t, CK_c.
 tDQSCK is the actual position of a rising strobe edge relative to CK_t, CK_c.
 tQSH describes the data strobe high pulse width.

 tQSL describes the data strobe low pulse width.
 tLZ(DQS), tHZ(DQS) for preamble/postamble.
Clock to Data Strobe Relationship
NOTE
1. Within a burst, the rising strobe edge will vary within tDQSCKj while at the same voltage and temperature. However,
when the device, voltage, and temperature variations are incorporated, the rising strobe edge variance window can
shift between tDQSCK (MIN) and tDQSCK (MAX).
A timing of this window's right edge (latest) from rising CK_t, CK_c is limited by a device's actual tDQSCK (MAX). A
timing of this window's left inside edge (earliest) from rising CK_t, CK_c is limited by tDQSCK (MIN).
2. Notwithstanding Note 1, a rising strobe edge with tDQSCK (MAX) at T(n) can not be immediately followed by a rising
strobe edge with tDQSCK (MIN) at T(n + 1) because other timing relationships (tQSH, tQSL) exist: if tDQSCK(n + 1) <
0: tDQSCK(n) < 1.0 tCK - (tQSH (MIN) + tQSL (MIN)) - | tDQSCK(n + 1) |.
3. The DQS_t, DQS_c differential output HIGH time is defined by tQSH, and the DQS_t, DQS_c differential output LOW
time is defined by tQSL.
4. tLZ(DQS) MIN and tHZ(DQS) MIN are not tied to tDQSCK (MIN) (early strobe case), and tLZ(DQS) MAX and
tHZ(DQS) MAX are not tied to tDQSCK (MAX) (late strobe case).
5. The minimum pulse width of READ preamble is defined by tRPRE (MIN).
6. The maximum READ postamble is bound by tDQSCK (MIN) plus tQSH (MIN) on the left side and tHZDSQ (MAX) on
the right side.
7. The minimum pulse width of READ postamble is defined by tRPST (MIN).
8. The maximum READ preamble is bound by tLZDQS (MIN) on the left side and tDQSCK (MAX) on the right side.

π PM Tech
READ Timing – Data Strobe to Data Relationship
PMA212816C
The data strobe to data relationship is shown below and is applied when the DLL is enabled and locked.

tDQSQ; both rising/falling edges of DQS, no tAC defined.
Data Strobe to Data Relationship
NOTE
1. BL = 8, RL = 11 (AL = 0, CL = 1) , Premable = 1tCK.

2. DOUTn = data-out from column n.
4. BL8 setting activated by either MR0[A1:0 = 00] or MR0[A1:0 = 01] and A12 = 1 during READ commands at T0.
5. Output timings are referenced to VDDQ, and DLL on for locking.
6. tDQSQ defines the skew between DQS to data and does not define DQS to clock.
7. Early data transitions may not always happen at the same DQ. Data transitions of a DQ can vary (either early or late) within
a burst.

π PM Tech
tLZ(DQS), tLZ(DQ), tHZ(DQS), tHZ(DQ) Calculation
PMA212816C
tHZ and tLZ transitions occur in the same time window as valid data transitions. These parameters are
referenced to a specific voltage level that specifies when the device output is no longer driving tHZ(DQS) and
tHZ(DQ), or begins driving tLZ(DQS) and tLZ(DQ). The figure below shows a method to calculate the point
when the device is no longer driving tHZ(DQS) and tHZ(DQ), or begins driving tLZ(DQS) and tLZ(DQ), by
measuring the signal at two different voltages. The actual voltage measurement points are not critical as long
as the calculation is consistent. tLZ(DQS), tLZ(DQ), tHZ(DQS), and tHZ(DQ) are defined as singled-ended
parameters.
tLZ and tHZ method for calculating transitions and begin points
NOTE
1. Vsw1 = (0.70 - 0.04) × VDDQ for both tLZ and tHZ.
2. Vsw2 = (0.70 + 0.04) × VDDQ for both tLZ and tHZ.
3. Extrapolated point (low level) = VDDQ/(50 + 34) × 34 = 0.4 × VDDQ
Driver impedance = RZQ/7 = 34ohm
VTT test load = 50ohm to VDDQ.

π PM Tech
tRPRE Calculation
PMA212816C
Method for Calculating tRPRE Transitions and Endpoints
NOTE
1. Vsw1 = (0.3 - 0.04) × VDDQ.
2. Vsw2 = (0.30 + 0.04) × VDDQ.
3. DQS_t and DQS_c low level = VDDQ/(50 + 34) × 34 = 0.4 × VDDQ

π PM Tech
tRPST Calculation
PMA212816C
tRPST Method for Calculating Transitions and Endpoints
NOTE
1. Vsw1 = (–0.3 - 0.04) × VDDQ.
2. Vsw2 = (–0.30 + 0.04) × VDDQ.
3. DQS_t and DQS_c low level = VDDQ/(50 + 34) × 34 = 0.4 × VDDQ

π PM Tech
READ Burst Operation
PMA212816C
DDR4 READ commands support bursts of BL8 (fixed), BC4 (fixed), and BL8/BC4 on-the-fly (OTF); OTF uses
address A12 to control OTF when OTF is enabled:
 A12 = 0, BC4 (BC4 = burst chop)
 A12 = 1, BL8
READ commands can issue precharge automatically with a READ with auto precharge command (RDA),
and is enabled by A10 HIGH:
 READ command with A10 = 0 (RD) performs standard read, bank remains active after READ burst.
 READ command with A10 = 1 (RDA) performs read with auto precharge, bank goes in to precharge after
READ burst.
READ Burst Operation RL = 11 (AL = 0, CL = 11, BL8)
NOTE
1. BL8, RL = 0, AL = 0, CL = 11, Preamble = 1tCK.
2. DO n = data-out from column n.
4. BL8 setting activated by either MR0[1:0] = 00 or MR0[1:0] = 01 and A12 = 1 during READ command at T0.
5. CA parity = Disable, CS to CA latency = Disable, Read DBI = Disable.

π PM Tech
READ Burst Operation RL = 21 (AL = 10, CL = 11, BL8)
PMA212816C
NOTE
1. BL8, RL = 21, AL = (CL - 1), CL = 11, Preamble = 1tCK.
2. DO n = data-out from column n.
4. BL8 setting activated by either MR0[1:0] = 00 or MR0[1:0] = 01 and A12 = 1 during READ command at T0.

π PM Tech
Consecutive READ (BL8) with 1tCK Preamble in Different Bank Group
PMA212816C
NOTE
1. BL8, AL = 0, CL = 11, Preamble = 1tCK.
2. DO n (or b) = data-out from column n (or column b).
4. BL8 setting activated by either MR0[1:0] = 00 or MR0[1:0] = 01 and A12 = 1 during READcommands at T0 and T4.
Consecutive READ (BL8) with 2tCK Preamble in Different Bank Group
NOTE
4. BL8 setting activated by either MR0[1:0] = 00 or MR0[1:0] = 01 and A12 = 1 during READ commands at T0 and T4.

π PM Tech
Nonconsecutive READ (BL8) with 1tCK Preamble in Same or Different Bank Group
PMA212816C
NOTE
1. BL8, AL = 0, CL = 11, Preamble = 1tCK, tCCD_S/L = 5.
4. BL8 setting activated by either MR0[1:0] = 00 or MR0[1:0] = 01 and A12 = 1 during READ commands at T0 and T5.
Nonconsecutive READ (BL8) with 2tCK Preamble in Same or Different Bank Group
NOTE
1. BL8, AL = 0, CL = 11, Preamble = 2tCK, tCCD_S/L = 6.
4. BL8 setting activated by either MR0[A1:0 = 00] or MR0[A1:0 = 01] and A12 = 1 during READ commands at T0 and T6.
6. tCCD_S/L = 5 isn’t allowed in 2tCK preamble mode.

π PM Tech
READ (BC4) to READ (BC4) with 1tCK Preamble in Different Bank Group
PMA212816C
NOTE
4. BC4 setting activated by either MR0[1:0] = 10 or MR0[1:0] = 01 and A12 = 0 during READ commands at T0 and T4.
READ (BC4) to READ (BC4) with 2tCK Preamble in Different Bank Group
NOTE
4. BC4 setting activated by either MR0[1:0] = 10 or MR0[1:0] = 01 and A12 = 0 during READ commands at T0 and T4.

π PM Tech
READ (BL8) to WRITE (BL8) with 1tCK Preamble in Same or Different Bank Group
PMA212816C
NOTE
1. BL = 8, RL = 11 (CL = 11, AL = 0), Read Preamble = 1tCK, WL =9 (CWL = 9, AL = 0), Write Preamble = 1tCK
2. DOUT n = data-out from column n, DIN b = data-in to column b.
4. BL8 setting activated by either MR0 A[1:0] = 00 or MR0 A[1:0] = 01and A12 = 1 during READ command at T0 and WRITE
command at T8.
5. CA Parity = Disable, CS to CA Latency = Disable, Read DBI = Disable, Write DBI = Disable, Write CRC = Disable.
READ (BL8) to WRITE (BL8) with 2tCK Preamble in Same or Different Bank Group
NOTE
1. BL = 8, RL = 11 (CL = 11, AL = 0), Read Preamble = 2tCK, WL = 10 (CWL = 9+1*5, AL = 0), Write Preamble = 2tCK
4. BL8 setting activated by either MR0[A1:A0 = 0:0] or MR0[A1:A0 = 0:1] and A12 = 1 during READ command at T0 and
WRITE command at T8.
5. When operating in 2tCK Write Preamble Mode, CWL must be programmed to a value at least 1 clock greater than the
lowest CWL setting.

π PM Tech
READ (BC4) OTF to WRITE (BC4) OTF with 1tCK Preamble in Same or Different Bank Group
PMA212816C
NOTE
1. BC = 4, RL = 11 (CL = 11, AL = 0), Read Preamble = 1tCK, WL = 9 (CWL = 9, AL = 0), Write Preamble = 1tCK
4. BC4(OTF) setting activated by MR0[A1:A0 = 0:1] and A12 = 0 during READ command at T0 and WRITE command at T6.
READ (BC4) OTF to WRITE (BC4) OTF with 2tCK Preamble in Same or Different Bank Group
NOTE
1. BC = 4, RL = 11 (CL = 11, AL = 0), Read Preamble = 2tCK, WL = 10 (CWL = 9+1*5, AL = 0), Write Preamble = 2tCK
4. BC4(OTF) setting activated by MR0[A1:A0 = 0:1] and A12 = 0 during READ command at T0 and WRITE command at T6.
lowest CWL setting.

π PM Tech
READ (BC4) Fixed to WRITE (BC4) Fixed with 1tCK Preamble in Same or Different Bank Group
PMA212816C
NOTE :
1. BC = 4, RL = 11 (CL = 11, AL = 0), Read Preamble = 1tCK, WL = 9 (CWL = 9, AL = 0), Write Preamble = 1tCK
4. BC4(Fixed) setting activated by MR0[A1:A0 = 1:0].
READ (BC4) Fixed to WRITE (BC4) Fixed with 2tCK Preamble in Same or Different Bank Group
NOTE :
4. BC4(Fixed) setting activated by MR0[A1:A0 = 1:0].
lowest CWL setting.

READ (BL8) to READ (BC4) OTF with 1tCK Preamble in Different Bank Group
NOTE :
1. BL = 8, AL =0, CL = 11 ,Preamble = 1tCK
2. DOUT n (or b) = data-out from column n (or column b).
4. BL8 setting activated by MR0[A1:A0 = 0:1] and A12 = 1 during READ command at T0
BC4 setting activated by MR0[A1:A0 = 0:1] and A12 = 0 during READ command at T4.
READ (BL8) to READ (BC4) OTF with 2tCK Preamble in Different Bank Group
NOTE :
4. BL8 setting activated by MR0[A1:A0 = 0:1] and A12 = 1 during READ command at T0.
BC4 setting activated by MR0[A1:A0 = 0:1] and A12 = 0 during READ command at T4.

π PM Tech
READ (BC4) to READ (BL8) OTF with 1tCK Preamble in Different Bank Group
PMA212816C
NOTE :
4. BC4 setting activated by MR0[A1:A0 = 0:1] and A12 = 0 during READ command at T0.
BL8 setting activated by MR0[A1:A0 = 0:1] and A12 = 1 during READ command at T4.
READ (BC4) to READ (BL8) OTF with 2tCK Preamble in Different Bank Group
NOTE :
BL8 setting activated by MR0[A1:A0 = 0:1] and A12 = 1 during READ command at T4.

π PM Tech
READ (BC4) to WRITE (BL8) OTF with 1tCK Preamble in Same or Different Bank Group
PMA212816C
NOTE :
1. BC = 4, RL = 11(CL = 11 , AL = 0 ), Read Preamble = 1tCK, WL=9(CWL=9,AL=0), Write Preamble = 1tCK
BL8 setting activated by MR0[A1:A0 = 0:1] and A12 = 1 during WRITE command at T6.
READ (BC4) to WRITE (BL8) OTF with 2tCK Preamble in Same or Different Bank Group
NOTE :
BL8 setting activated by MR0[A1:A0 = 0:1] and A12 = 1 during WRITE command at T6.
lowest CWL setting.

π PM Tech
READ (BL8) to WRITE (BC4) OTF with 1tCK Preamble in Same or Different Bank Group
PMA212816C
NOTE :
1. BL = 8, RL = 11(CL = 11 , AL = 0 ), Read Preamble = 1tCK, WL=9(CWL=9,AL=0), Write Preamble = 1tCK
BC4 setting activated by MR0[A1:A0 = 0:1] and A12 = 0 during WRITE command at T8.
READ (BL8) to WRITE (BC4) OTF with 2tCK Preamble in Same or Different Bank Group
NOTE :
1. BL = 8, RL = 11 (CL = 11, AL = 0), Read Preamble = 2tCK, WL = 10 (CWL = 9+1*5, AL = 0), Write Preamble = 2tCK
BC4 setting activated by MR0[A1:A0 = 0:1] and A12 = 0 during WRITE command at T8.
lowest CWL setting.

π PM Tech
Burst Read Operation followed by a Precharge
PMA212816C
The minimum external Read command to Precharge command spacing to the same bank is equal to AL +
tRTP with tRTP being the Internal Read Command to Precharge Command Delay. Note that the minimum
ACT to PRE timing, tRAS, must be satisfied as well. The minimum value for the Internal Read Command to
Precharge Command Delay is given bytRTP.min, A new bank active command may be issued to the same
bank if the following two conditions are satisfied simultaneously:
1. The minimum RAS precharge time (tRP.MIN) has been satisfied from the clock at which the precharge
begins.
2. The minimum RAS cycle time (tRC.MIN) from the previous bank activation has been satisfied.
Examples of Read commands followed by Precharge are show in READ to PRECHARGE with 1tCK
Preamble to READ to PRECHARGE with Additive Latency and 1tCK Preamble.
READ to PRECHARGE with 1tCK Preamble
NOTE :
1. BL = 8, RL = 11(CL = 11 , AL = 0 ), Preamble = 1tCK, tRTP = 6, tRP = 11
4. The example assumes tRAS. MIN is satisfied at Precharge command time(T7) and that tRC. MIN is satisfied at the next
Active command time(T18).

π PM Tech
READ to PRECHARGE with 2tCK Preamble
PMA212816C
NOTE :
1. BL = 8, RL = 11(CL = 11 , AL = 0 ), Preamble = 2tCK, tRTP = 6, tRP = 11
READ to PRECHARGE with Additive Latency and 1tCK Preamble
NOTE :
1. BL = 8, RL = 20 (CL = 11 , AL = CL- 2 ), Preamble = 1tCK, tRTP = 6, tRP = 11

π PM Tech
READ with Auto Precharge and 1tCK Preamble
PMA212816C
NOTE :
1. BL = 8, RL = 11 (CL = 11 , AL = 0 ), Preamble = 1tCK, tRTP = 6, tRP = 11
4. tRTP = 6 setting activated by MR0[A11:9 = 001]
5. The example assumes tRC. MIN is satisfied at the next Active command time(T18).
READ with Auto Precharge, Additive Latency and 1tCK Preamble
NOTE :
1. BL = 8, RL = 20 (CL = 11 , AL = CL- 2 ), Preamble = 1tCK, tRTP = 6, tRP = 11

4. tRTP = 6 setting activated by MR0[A11:9 = 001]
5. The example assumes tRC. MIN is satisfied at the next Active command time(T27).

π PM Tech
Burst Read Operation with Read DBI (Data Bus Inversion)
PMA212816C
Consecutive READ (BL8) with 1tCK Preamble and DBI in Different Bank Group
NOTE :
1. BL = 8, AL = 0, CL = 11, Preamble = 1tCK, tDBI = 2tCK
2. DOUT n (or b) = data-out from column n ( or column b).
3. DES commands are shown for ease of illustrat:ion; other commands may be valid at these times.
4. BL8 setting activated by either MR0[A1:A0 = 00] or MR0[A1:A0 = 0:1] and A12 = 1 during READ command at T0 and T4.
5. CA Parity = Disable, CS to CA Latency = Disable, Read DBI = Enable.
Burst Read Operation with Command/Address Parity
Consecutive READ (BL8) with 1tCK Preamble and CA Parity in Different Bank Group
NOTE :
1. BL = 8, AL = 0, CL = 11, PL = 4, (RL = CL + AL + PL = 15), Preamble = 1tCK
2. DOUT n (or b) = data-out from column n ( or column b).
4. BL8 setting activated by either MR0[A1:A0 = 0:0] or MR0[A1:A0 = 0:1] and A12 = 1 during READ command at T0 and T4.
5. CA Parity =Enable, CS to CA Latency = Disable, Read DBI = Disable.

π PM Tech
READ (BL8) to WRITE (BL8) with 1tCK Preamble and CA parity in Same or Different Bank Group
PMA212816C
NOTE :
1. BL = 8, AL = 0, CL = 11, PL = 4, (RL = CL + AL + PL = 15), Read Preamble = 1tCK, CWL=9, AL=0, PL=4,
(WL=CL+AL+PL=13), Write Preamble = 1tCK
4. BL8 setting activated by either MR0[A1:A0 = 0:0] or MR0[A1:A0 = 0:1] and A12 = 1 during READ command at T0 and Write
command at T8.
5. CA Parity = Enable, CS to CA Latency = Disable, Read DBI = Disable, Write DBI = Disable, Write CRC = Disable.

π PM Tech
Read to Write with Write CRC
PMA212816C
READ (BL8) to WRITE (BL8 or BC4:OTF) with 1tCK Preamble and Write CRC in Same or Different
Bank Group
NOTE :
1. BL = 8 ( or BC = 4 : OTF for Write), RL = 11 (CL = 11, AL = 0), Read Preamble = 1tCK, WL=9 (CWL=9, AL=0), Write
Preamble = 1tCK
2. DOUT n = data-out from column n . DIN b = data-in to column b.
4. BL8 setting activated by either MR0[A1:A0 = 0:0] or MR0[A1:A0 = 0:1] and A12 = 1 during READ command at T0 and Write
command at T8.
5. BC4 setting activated by MR0[A1:0 = 01] and A12 = 0 during Write command at T8.
6. CA Parity = Disable, CS to CA Latency = Disable, Read DBI = Disable, Write DBI = Disable, Write CRC = Enable.
READ (BC4:Fixed) to WRITE (BC4:Fixed) with 1tCK Preamble and Write CRC in Same or Different
Bank Group
NOTE :
1. BC = 4 (Fixed), RL = 11 (CL = 11, AL = 0), Read Preamble = 1tCK, WL=9 (CWL=9, AL=0), Write Preamble = 1tCK
2. DOUT n = data-out from column n . DIN b = data-in to column b.
4. BC4 setting activated by MR0[A1:A0 = 1:0].
5. CA Parity = Disable, CS to CA Latency = Disable, Read DBI = Disable, Write DBI = Disable, Write CRC = Enable.

π PM Tech
Read to Read with CS to CA Latency
PMA212816C
Consecutive READ (BL8) with CAL(3) and 1tCK Preamble in Different Bank Group
NOTE :
1. BL = 8 ,AL = 0, CL = 11, CAL = 3, Preamble = 1tCK
5. CA Parity = Disable, CS to CA Latency = Enable, Read DBI = Disable.
6. Enabling of CAL mode does not impact ODT control timings. Users should maintain the same timing relationship relative to
the command/address bus as when CAL is disabled.
Consecutive READ (BL8) with CAL(4) and 1tCK Preamble in Different Bank Group
NOTE :
1. BL = 8 ,AL = 0, CL = 11, CAL = 4, Preamble = 1tCK
5. CA Parity = Disable, CS to CA Latency = Enable, Read DBI = Disable.
6. Enabling of CAL mode does not impact ODT control timings. Users should maintain the same timing relationship relative to
the command/address bus as when CAL is disabled.

π PM Tech
Write Operation
PMA212816C
Write Timing Definitions

The write timings shown in the following figures are applicable in normal operation mode, that is, when the
DLL is enabled and locked.
NOTE
1. BL8, WL = 9 (AL = 0, CWL = 9).
2. DINn = data-in from column n.
4. BL8 setting activated by either MR0[1:0] = 00 or MR0[1:0] = 01 and A12 = 1 during WRITE command at T0.
5. tDQSS must be met at each rising clock edge.

π PM Tech
tWPRE Calculation
PMA212816C
Method for calculating tWPRE transitions and endpoints
NOTE
1. Vsw1 = (0.1) × VIH,diff,DQS.
2. Vsw2 = (0.9) × VIH,diff,DQS.

π PM Tech
tWPST Calculation
PMA212816C
Method for calculating tWPST transitions and endpoints
NOTE
1. Vsw1 =(0.9) × VIL,diff,DQS.
2. Vsw2 = (0.1) × VIL,diff,DQS.

π PM Tech
WRITE Burst Operation
PMA212816C
The following write timing diagrams are to help understanding of each write parameter'smeaning and are just
examples. The details of the definition of each parameter will be defined separately. In these write timing
diagram, CK and DQS are shown alignedand also DQS and DQ are shown center aligned for illustration
purpose.
WRITE Burst Operation WL = 9 (AL = 0, CWL = 9, BL8)
NOTE :
1. BL = 8 ,WL = 9, AL = 0, CWL = 9, Preamble = 1tCK
2. DIN n = data-in to column n.
4. BL8 setting activated by either MR0[A1:A0 = 0:0] or MR0[A1:A0 = 0:1] and A12 = 1 during WRITE command at T0.
5. CA Parity = Disable, CS to CA Latency = Disable, Write DBI = Disable.
WRITE Burst Operation WL = 19 (AL = 10, CWL = 9, BL8)
NOTE :
1. BL = 8 ,WL = 19, AL = 10 (CL-1), CWL = 9, Preamble = 1tCK

π PM Tech
Consecutive WRITE (BL8) with 1tCK Preamble in Different Bank Group
PMA212816C
NOTE :
1. BL = 8, AL = 0, CWL = 9, Preamble = 1tCK
2. DIN n (or b) = data-in to column n (or column b).
4. BL8 setting activated by either MR0[A1:A0 = 0:0] or MR0[A1:A0 = 0:1] and A12 = 1 during WRITE command at T0 and T4.
5. C/A Parity = Disable, CS to C/A Latency = Disable, Write DBI = Disable.
6. The write recovery time (tWR) and write timing parameter (tWTR) are referenced from the first rising clock edge after the
last write data shown at T17.
Consecutive WRITE (BL8) with 2tCK Preamble in Different Bank Group
NOTE :
1. BL = 8 ,AL = 0, CWL = 9 + 1 = 107, Preamble = 2tCK
2. DIN n (or b) = data-in to column n( or column b).
6. The write recovery time(tWR) and write timing parameter(tWTR) are referenced from the first rising clock edge after the last
write data shown at T18.
lowest CWL setting supported in the applicable tCK range. That means CWL = 9 is not allowed when operating in 2tCK
Write Preamble Mode.

π PM Tech
Nonconsecutive WRITE (BL8) with 1tCK Preamble in Same or Different Bank Group
PMA212816C
NOTE:
1. BL = 8 ,AL = 0, CWL = 9 , Preamble = 1tCK, tCCD_S/L = 5
Nonconsecutive WRITE (BL8) with 2tCK Preamble in Same or Different Bank Group
NOTE:
1. BL = 8 ,AL = 0, CWL = 9 + 1 = 108 , Preamble = 2tCK, tCCD_S/L = 6
6. tCCD_S/L=5 isn’t allowed in 2tCK preamble mode.
lowest CWL setting supported in the applicable tCK range.That means CWL = 9 is not allowed when operating in 2tCK

π PM Tech
WRITE (BC4) OTF to WRITE (BC4) OTF with 1tCK Preamble in Different Bank Group
PMA212816C
NOTE:
1. BC = 4, AL = 0, CWL = 9 , Preamble = 1tCK
4. BC4 setting activated by MR0[A1:A0 = 0:1] and A12 = 0 during WRITE command at T0 and T4.
WRITE (BC4) OTF to WRITE (BC4) OTF with 2tCK Preamble in Different Bank Group
NOTE:
1. BC = 4, AL = 0, CWL = 9 + 1 = 107 , Preamble = 2tCK
4. BC4 setting activated by MR0[A1:A0 = 0:1] and A12 = 0 during WRITE command at T0 and T4.
lowest CWL setting supported in the applicable tCK range.That means CWL = 9 is not allowed when operating in 2tCK

π PM Tech
WRITE (BC4) Fixed to WRITE (BC4) Fixed with 1tCK Preamble in Different Bank Group
PMA212816C
NOTE:
1. BC = 4, AL = 0, CWL = 9 , Preamble = 1tCK
WRITE (BL8) to READ (BL8) with 1tCK Preamble in Different Bank Group
NOTE:
1. BC = 4, AL = 0, CWL = 9, CL = 11, Preamble = 1tCK
2. DIN n = data-in to column n(or column b). DOUT b = data-out from column b.
4. BL8 setting activated by either MR0[A1:A0 = 0:0] or MR0[A1:A0 = 0:1] and A12 = 1 during WRITE command at T0 and
READ command at T15.
6. The write timing parameter (tWTR_S) are referenced from the first rising clock edge after the last write data shown at T13.

π PM Tech
WRITE (BL8) to READ (BL8) with 1tCK Preamble in Same Bank Group
PMA212816C
NOTE:
1. BL = 8, AL = 0, CWL = 9, CL = 11, Preamble = 1tCK
2. DIN n = data-in to column n (or column b). DOUT b = data-out from column b.
4. BL8 setting activated by either MR0[A1:A0 = 0:0] or MR0[A1:A0 = 0:1] and A12 = 1 during WRITE command at T0 and
READ command at T17.
6. The write timing parameter (tWTR_L) are referenced from the first rising clock edge after the last write data shown at T13.
WRITE (BC4)OTF to READ (BC4)OTF with 1tCK Preamble in Different Bank Group
NOTE:
4. BC4 setting activated by MR0[A1:A0 = 0:1] and A12 = 0 during WRITE command at T0 and READ command at T15.

π PM Tech
WRITE (BC4)OTF to READ (BC4)OTF with 1tCK Preamble in Same Bank Group
PMA212816C
NOTE:
4. BC4 setting activated by MR0[A1:A0 = 0:1] and A12 = 0 during WRITE command at T0 and READ command at T17.
WRITE (BC4)Fixed to READ (BC4)Fixed with 1tCK Preamble in Different Bank Group
NOTE:

π PM Tech
WRITE (BC4)Fixed to READ (BC4)Fixed with 1tCK Preamble in Same Bank Group
PMA212816C
NOTE:
WRITE (BL8) to WRITE (BC4) OTF with 1tCK Preamble in Different Bank Group
NOTE:
1. BL = 8 / BC = 4, AL = 0, CWL = 9, Preamble = 1tCK
4. BL8 setting activated by MR0[A1:A0 = 0:1] and A12 =1 during WRITE command at T0.
BC4 setting activated by MR0[A1:A0 = 0:1] and A12 =0 during WRITE command at T4.
last write data shown at T17

π PM Tech
WRITE (BC4) OTF to WRITE (BL8) with 1tCK Preamble in Different Bank Group
PMA212816C
NOTE:
4. BC4 setting activated by MR0[A1:A0 = 0:1] and A12 =0 during WRITE command at T0.
BL8 setting activated by MR0[A1:A0 = 0:1] and A12 =1 during WRITE command at T4.
last write data shown at T17
WRITE (BL8/BC4) OTF to PRECHARGE Operation with 1tCK Preamble
NOTE:
1. BL = 8 / BC = 4, AL = 0, CWL = 9, Preamble = 1tCK, tWR = 12
BL8 setting activated by MR0[A1:A0 = 0:0] or MR0[A1:0 = 01] and A12 =1 during WRITE command at T0.
6. The write recovery time (tWR) is referenced from the first rising clock edge after the last write data shown at T13.
tWR specifies the last burst write cycle until the precharge command can be issued to the same bank.

π PM Tech
WRITE (BC4) Fixed to PRECHARGE Operation with 1tCK Preamble
PMA212816C
NOTE:
1. BC = 4, AL = 0, CWL = 9, Preamble = 1tCK, tWR = 12
tWR specifies the last burst write cycle until the precharge command can be issued to the same bank.
WRITE (BL8/BC4) OTF with Auto PRECHARGE Operation and 1tCK Preamble
NOTE:
1. BL = 8 / BC = 4, AL = 0, CWL = 9, Preamble = 1tCK, WR = 12
BL8 setting activated by either MR0[A1:0 = 00] or MR0[A1:0 = 01] and A12 =1 during WRITE command at T0.
6. The write recovery time (WR) is referenced from the first rising clock edge after the last write data shown at T13.
WR specifies the last burst write cycle until the precharge command can be issued to the same bank.

π PM Tech
WRITE (BC4) Fixed with Auto PRECHARGE Operation and 1tCK Preamble
PMA212816C
NOTE:
1. BC = 4, AL = 0, CWL = 9, Preamble = 1tCK, WR = 12
WR specifies the last burst write cycle until the precharge command can be issued to the same bank.
WRITE (BL8/BC4) OTF with 1tCK Preamble and DBI
NOTE:
BL8 setting activated by either MR0[A1:A0 = 0:0] or MR0[A1:A0 = 0:1] and A12 =1 during WRITE command at T0.
5. CA Parity = Disable, CS to CA Latency = Disable, Write DBI = Enable, CRC = Disable.
6. The write recovery time (tWR_DBI) and write timing parameter (tWTR_DBI) are referenced from the first rising clock edge
after the last write data shown at T13.

π PM Tech
WRITE (BC4) Fixed with 1tCK Preamble and DBI
PMA212816C
NOTE:
1. BC = 4, AL = 0, CWL = 9, Preamble = 1tCK
5. CA Parity = Disable, CS to CA Latency = Disable, Write DBI = Enable, CRC = Disable.
6. The write recovery time (tWR_DBI) and write timing parameter (tWTR_DBI) are referenced from the first rising clock edge
after the last write data shown at T11.
Consecutive WRITE (BL8) with 1tCK Preamble and CA Parity in Different Bank Group
NOTE:
1. BL = 8, AL = 0, CWL = 9, PL = 4, Preamble = 1tCK
2. DIN n (or b) = data-in to column n(or column b).
4. BL8 setting activated by either MR0[A1:A0 = 0:0] or MR0[A1:A0 = 0:1] and A12 =1 during WRITE command at T0 and T4.
5. CA Parity = Enable, CS to CA Latency = Disable, Write DBI = Disable.

π PM Tech
Consecutive WRITE (BL8/BC4)OTF with 1tCK Preamble and Write CRC in Same or Different Bank Group
PMA212816C
NOTE:
1. BL = 8/BC = 4, AL = 0, CWL = 9, Preamble = 1tCK, tCCD_S/L = 5
4. BL8 setting activated by either MR0[A1:0 = 00] or MR0[A1:0 = 01] and A12 = 1 during WRITE command at T0 and T5.
BC4 setting activated by MR0[A1:A0 = 0:1] and A12 = 0 during WRITE command at T0 and T5.
5. C/A Parity = Disable, CS to C/A Latency = Disable, Write DBI = Disable, Write CRC = Enable.
Consecutive WRITE (BC4)Fixed with 1tCK Preamble and Write CRC in Same or Different Bank Group
NOTE:
1. BL = 8, AL = 0, CWL = 9, Preamble = 1tCK, tCCD_S/L = 5
4. BL8 setting activated by MR0[A1:A0 = 1:0] at T0 and T5.
5. CA Parity = Disable, CS to CA Latency = Disable, Write DBI = Disable, CRC = Enable.

π PM Tech
Nonconsecutive WRITE (BL8/BC4)OTF with 1tCK Preamble and Write CRC in Same or Different Bank
PMA212816C
Group
NOTE:
1. BL = 8, AL = 0, CWL = 9, Preamble = 1tCK, tCCD_S/L = 6
4. BL8 setting activated by either MR0[A1A:0 = 0:0] or MR0[A1A:0 = 0:1] and A12 =1 during WRITE command at T0 and T6.
5. BC4 setting activated by MR0[A1:A0 = 0:1] and A12 =0 during WRITE command at T0 and T6.
6. CA Parity = Disable, CS to CA Latency = Disable, Write DBI = Disable, Write CRC = Enable.

π PM Tech
Nonconsecutive WRITE (BL8/BC4)OTF with 2tCK Preamble and Write CRC in Same or Different Bank
PMA212816C
Group
NOTE:
1. BL = 8, AL = 0, CWL = 9 + 1 = 109, Preamble = 2tCK, tCCD_S/L = 7
4. BL8 setting activated by either MR0[A1:A0 = 0:0] or MR0[A1:A0 = 0:1] and A12 =1 during WRITE command at T0 and T7.
5. BC4 setting activated by MR0[A1:A0 = 0:1] and A12 =0 during WRITE command at T0 and T7.
6. CA Parity = Disable, CS to CA Latency = Disable, Write DBI = Disable, Write CRC = Enable.
7. tCCD_S/L = 6 isn’t allowed in 2tCK preamble mode.
lowest CWL setting supported in the applicable tCK range. That means CWL = 9 is not allowed when operating in 2tCK

π PM Tech
WRITE (BL8/BC4)OTF/Fixed with 1tCK Preamble and Write CRC and DM in Same or Different Bank
PMA212816C
Group
NOTE:
5. BC4 setting activated by either MR0[A1:A0 = 1:0] or MR0[A1:A0 = 0:1] and A12 = 0 during READ command at T0.
6. CA Parity = Disable, CS to CA Latency = Disable, Write DBI = Disable, Write CRC = Enable, DM = Enable.
7. The write recovery time (tWR_CRC_ DM) and write timing parameter (tWR_S_CRC_ DM/tWR_L_CRC_ DM) are
referenced from the first rising clock edge after the last write data shown at T13.

π PM Tech
Refresh Command
PMA212816C
The Refresh command (REF) is used during normal operation of the DDR4 SDRAMs. This command is non
persistent, so it must be issued each time a refresh is required. The DDR4 SDRAM requires Refresh cycles
at an average periodic interval of tREFI. When CS_n, RAS_n/A16 and CAS_n/A15 are held Low and
WE_n/A14 and ACT_n are held High at the rising edge of the clock, the chip enters a Refresh cycle. All
banks of the SDRAM must be precharged and idle for a minimum of the precharge time tRP(min) before the
Refresh Command can be applied. The refresh addressing is generated by the internal refresh controller.
This makes the address bits “Don’t Care” during a Refresh command. An internal address counter supplies
the addresses during the refresh cycle. No control of the external address bus is required once this cycle has
started. When the refresh cycle has completed, all banks of the SDRAM will be in the precharged (idle) state.
A delay between the Refresh Command and the next valid command, except DES, must be greater than or
equal to the minimum Refresh cycle time tRFC(min) as shown in Figure X. Note that the tRFC timing
parameter depends on memory density.
In general, a Refresh command needs to be issued to the DDR4 SDRAM regularly every tREFI interval. To
allow for improved efficiency in scheduling and switching between tasks, some flexibility in the absolute
refresh interval is provided for postponing and pulling-in refresh command. A maximum of 8 Refresh
commands can be postponed when DRAM is in 1X refresh mode and for 2X/4X refresh mode, 16/32 Refresh
commands can be postponed respectively during operation of the DDR4 SDRAM, meaning that at no point in
time more than a total of 8,16,32 Refresh commands are allowed to be postponed for 1X,2X,4X Refresh
mode respectively.
When 8 Refresh commands are postponed in a row, the resulting maximum interval between the surrounding
Refresh commands is limited to 9 × tREFI. In 2X and 4X Refresh mode, it’s limited to 17 x tREFI2 and 33 x
tREFI4. A maximum of 8 additional Refresh commands can be issued in advance (“pulled in”) in 1X refresh
mode and for 2X/4X refresh mode, 16/32 Refresh commands can be pulled in respectively, with each one
reducing the number of regular Refresh commands required later by one. Note that pulling in more than
8/16/32, depending on Refresh mode, Refresh commands in advance does not further reduce the number of
regular Refresh commands required later, so that the resulting maximum interval between two surrounding
Refresh commands is limited to 9 × tREFI , 17 x tRFEI2 and 33 x tREFI4 respectively. At any given time, a
maximum of 16 REF/32REF 2/64REF 4 commands can be issued within 2 x tREFI/ 4 x tREFI2/ 8 x tREFI4.
Refresh Command Timing (Example of 1x Refresh mode)
NOTE :
1. Only DES commands allowed after Refresh command registered untill tRFC(min) expires.
2. Time interval between two Refresh commands may be extended to a maximum of 9 X tREFI.

π PM Tech
Postponing Refresh Commands (Example of 1X Refresh mode)
PMA212816C
Pulling-in Refresh Commands (Example of 1X Refresh mode)

π PM Tech
Self refresh Operation
PMA212816C
The Self-Refresh command can be used to retain data in the DDR4 SDRAM, even if the rest of the system is
powered down. When in the Self-Refresh mode, the DDR4 SDRAM retains data without external
clocking.The DDR4 SDRAM device has a built-in timer to accommodate Self-Refresh operation. The Self-
Refresh-Entry (SRE) Command is defined by having CS_n, RAS_n/A16, CAS_n/A15, and CKE held low with
WE_n/A14 and ACT_n high at the rising edge of the clock.
Before issuing the Self-Refresh-Entry command, the DDR4 SDRAM must be idle with all bank precharge
state with tRP satisfied. ‘Idle state’ is defined as all banks are closed (tRP, tDAL, etc. satisfied), no data
bursts are in progress, CKE is high, and all timings from previous operations are satisfied (tMRD,
tMOD,tRFC, tZQinit, tZQoper, tZQCS, etc.). Deselect command must be registered on last positive clock
edge before issuing Self Refresh Entry command. Once the Self Refresh Entry command is registered,
Deselect command must also be registered at the next positive clock edge. Once the Self-Refresh Entry
command is registered, CKE must be held low to keep the device in Self-Refresh mode. .DRAM
automatically disables ODT termination and set Hi-Z as termination state regardless of ODT pin and
RTT_PARK set when it enters in Self-Refresh mode. Upon exiting Self-Refresh, DRAM automatically
enables ODT termination and set RTT_PARK asynchronously during tXSDLL when RTT_PARK is enabled.
During normal operation (DLL on) the DLL is automatically disabled upon entering Self-Refresh and is
automatically enabled (including a DLL-Reset) upon exiting Self-Refresh.
When the DDR4 SDRAM has entered Self-Refresh mode, all of the external control signals, except CKE and
RESET_n, are “don’t care.” For proper Self-Refresh operation, all power supply and reference pins (VDD,
VDDQ, VSS, VSSQ, VPP, and VREFCA) must be at valid levels. DRAM internal VREFDQ generator circuitry
may remain ON or turned OFF depending on DRAM design. If DRAM internal VREFDQ circuitry is turned OFF
in self refresh, when DRAM exits from self refresh state, it ensures that VREFDQ generator circuitry is
powered up and stable within tXS period. First Write operation or first Write Leveling Activity may not occur
earlier than tXS after exit from Self Refresh. The DRAM initiates a minimum of one Refresh command
internally within tCKE period once it enters Self-Refresh mode.
The clock is internally disabled during Self-Refresh Operation to save power. The minimum time that the
DDR4 SDRAM must remain in Self-Refresh mode is tCKESR. The user may change the external clock
frequency or halt the external clock tCKSRE after Self-Refresh entry is registered, however, the clock must
be restarted and stable tCKSRX before the device can exit Self-Refresh operation.
The procedure for exiting Self-Refresh requires a sequence of events. First, the clock must be stable prior to
CKE going back HIGH. Once a Self-Refresh Exit command (SRX, combination of CKE going high and
Deselect on command bus) is registered, following timing delay must be satisfied:
1. Commands that do not require locked DLL:
 tXS - ACT, PRE, PREA, REF, SRE, PDE, WR, WRS4, WRS8, WRA, WRAS4, WRAS8, tXSFast -
ZQCL, ZQCS, MRS commands.
 For MRS command, only DRAM CL and WR/RTP register in MR0, CWL register in MR2 and geardown
mode in MR3 are allowed to be accessed provided DRAM is not in per DRAM addressability mode.
Access to other DRAM mode registers must satisfy tXS timing.
Note that synchronous ODT for write commands ( WR, WRS4, WRS8, WRA, WRAS4 and WRAS8 ) and
dynamic ODT controlled by write command require locked DLL.
2. Commands that require locked DLL:
 tXSDLL - RD, RDS4, RDS8, RDA, RDAS4, RDAS8
Depending on the system environment and the amount of time spent in Self-Refresh, ZQ calibration
commands may be required to compensate for the voltage and temperature drift as described in “ZQ
Calibration Commands” on Section 4.12. To issue ZQ calibration commands, applicable timing requirements
must be satisfied.

CKE must remain HIGH for the entire Self-Refresh exit period tXSDLL for proper operation except for Self-
Refresh re-entry. Upon exit from Self-Refresh, the DDR4 SDRAM can be put back into Self-Refresh mode or
Power down mode after waiting at least tXS period and issuing one refresh command (refresh period of
tRFC). Deselect commands must be registered on each positive clock edge during the Self-Refresh exit
interval tXS. Low level of ODT pin must be registered on each positive clock edge during tXSDLL when
normal mode ( DLL-on ) is set. Under DLL-off mode, asynchronous ODT function might be allowed.
The use of Self-Refresh mode introduces the possibility that an internally timed refresh event can be missed
when CKE is raised for exit from Self-Refresh mode. Upon exit from Self-Refresh, the DDR4 SDRAM
requires a minimum of one extra refresh command before it is put back into Self-Refresh Mode.
The exit timing from self-refresh exit to first valid command not requiring a locked DLL is tXS.
The value of tXS is (tRFC+10ns). This delay is to allow for any refreshes started by the DRAM to complete.
tRFC continues to grow with higher density devices so tXS will grow as well.
A Bit A9 in MR4 is defined to enable the self refresh abort mode. If the bit is disabled then the controller uses
tXS timings.
If the bit is enabled then the DRAM aborts any ongoing refresh and does not increment the refresh counter.
The controller can issue a valid command not requiring a locked DLL after a delay of tXS_abort. Upon exit
from Self-Refresh, the DDR4 SDRAM requires a minimum of one extra refresh command before it is put
back into Self-Refresh Mode. This requirement remains the same irrespective of the setting of the MRS bit
for self refresh abort.
Self-Refresh Entry/Exit Timing
NOTE :
1. Only MRS (limited to those described in the Self-Refresh Operation section). ZQCS or ZQCL command allowed.
2. Valid commands not requiring a locked DLL.
3. Valid commands requiring a locked DLL.
4. Only DES is allowed during tXS_ABORT.

π PM Tech
Low Power Auto Self Refresh
PMA212816C
DDR4 devices support Low Power Auto Self-Refresh (LP ASR) operation at multiple temperatures ranges.
MR2 definitions for Low Power Auto Self-Refresh mode
A6 A7 Self-Refresh Operation Mode
0 0 Manual Mode – Normal operating temperature range
0 1 Manual Mode – Extended operating temperature range
1 0 Manual Mode – Lower power mode at a reduced operating temperature range
ASR Mode – automatically switching between all modes to optimize power for any of the temperature ranges
1 1
listed above
Auto Self Refresh (ASR)

DDR4 DRAM provides an Auto Self-Refresh mode (ASR) for application ease. ASR mode is enabled by
setting the above MR2 bits A6=1 and A7=1. The DRAM will manage Self Refresh entry through the
supported temperature range of the DRAM. In this mode, the DRAM will change self-refresh rate as the
DRAM operating temperature changes, lower at low temperatures and higher at high temperatures.
Manual Modes
If ASR mode is not enabled, the LP ASR Mode Register must be manually programmed to one the three
self-refresh operating modes listed above. In this mode, the user has the flexibility to select a fixed self-
refresh operating mode at the entry of the selfrefresh according to their system memory temperature
conditions. The user is responsible to maintain the required memory temperature condition for the mode
selected during the self-refresh operation. The user may change the selected mode after exiting from self
refresh and before the next self-refresh entry. If the temperature condition is exceeded for the mode selected,
there is risk to data retention resulting in loss of data.
Self Refresh Function Table
Low-Power Allowed Operating

MR2 MR2
Auto Self Self Refresh Operation temp. range for Self
A6 A7
Refresh Mode Refresh Mode
Fixed normal self-Refresh rate to maintain data retention for the

0 0 Normal normal operating temperature. User is required to ensure 85°C (0°C – 85°C)
DRAM Tcasemax is not exceeded to avoid any risk of data loss.
Fixed high self-Refresh rate to optimize data retention to support the

0 1 Extended extended temperature range (0°C – 95°C)
Variable or fixed self-refresh rate or any other DRAM power

consumption reduction control for the reduced temperature range.
1 0 Reduced User is required to ensure 45°C DRAM Tcasemax is not exceeded to (0°C – 45°C)
avoid any risk of data loss.
ASR Mode Enabled. Self-Refresh power consumption and data
Auto Self retention are optimized for any given operating temperature
1 1 All of the above
Refresh conditions

π PM Tech
Power down Mode
PMA212816C
Power-Down Entry and Exit

Power-down is synchronously entered when CKE is registered low (along with Deselect command). CKE is
not allowed to go low while mode register set command, MPR operations, ZQCAL operations, DLL locking or
read / write operation are in progress. CKE is allowed to go low while any of other operations such as row
activation, precharge or auto-precharge and refresh are in progress, but power-down IDD spec will not be
applied until finishing those operations. Timing diagrams are shown in following figures with details for entry
and exit of Power-Down.
The DLL should be in a locked state when power-down is entered for fastest power-down exit timing. If the
DLL is not locked during power-down entry, the DLL must be reset after exiting power-down mode for proper
read operation and synchronous ODT operation. DRAM design provides all AC and DC timing and voltage
specification as well as proper DLL operation with any CKE intensive operations as long as DRAM controller
complies with DRAM specifications.
During Power-Down, if all banks are closed after any in-progress commands are completed, the device will
be in precharge Power-Down mode; if any bank is open after in-progress commands are completed, the
device will be in active Power-Down mode.
Entering power-down deactivates the input and output buffers, excluding CK_t, CK_c, CKE and RESET_n. In
power-down mode, DRAM ODT input buffer deactivation is based on MR5 bit A5. If it is configured to 0b,
ODT input buffer remains on and ODT input signal must be at valid logic level. If it is configured to 1b, ODT
input buffer is deactivated and DRAM ODT input signal may be floating and DRAM does not provide
Rtt_Nom termination. Note that DRAM continues to provide Rtt_Park termination if it is enabled in DRAM
mode register MR5 bit A8:A6 To protect DRAM internal delay on CKE line to block the input signals, multiple
Deselect commands are needed during the CKE switch off and cycle(s) after, this timing period are defined
as tCPDED. CKE_low will result in deactivation of command and address receivers after tCPDED has
expired.
Power-Down Entry Definitions
Status of DRAM DLL PD Exit Relevant Parameters

Active
On Fast tXP to any valid command
(A bank or more Open)
Precharged
On Fast tXP to any valid command.
(All banks Precharged)
DLL is kept enabled during precharge power-down or active power-down. In power-down mode, CKE low,
RESET_n high, and a stable clock signal must be maintained at the inputs of the DDR4 SDRAM, and ODT
should be in a valid state, but all other input signals are “Don’t Care.” (If RESET_n goes low during Power-
Down, the DRAM will be out of PD mode and into reset state.) CKE low must be maintained until tCKE has
been satisfied. Power-down duration is limited by 9 times tREFI of the device.
The power-down state is synchronously exited when CKE is registered high (along with a Deselect
command). CKE high must be maintained until tCKE has been satisfied. DRAM ODT input signal must be at
valid level when DRAM exits from power-down mode independent of MR5 bit A5 if Rtt_Nom is enabled in
DRAM mode register. If DRAM Rtt_Nom is disabled then ODT input signal may remain floating. A valid,
executable command can be applied with power-down exit latency, tXP after CKE goes high. Power-down
exit latency is defined in the AC specifications Table.

π PM Tech
Active Power-Down Entry and Exit Timing Diagram (MR5 bit A5 =0)
PMA212816C
NOTE
1. Valid commands at T0 are ACT, DES, or PRE with one bank remaining open after completion of the PRECHARGE
command.
2. ODT pin driven to a valid state; MR5 [5] = 0 (normal setting).
Active Power-Down Entry and Exit Timing Diagram (MR5 bit A5=1)
NOTE
1. Valid commands at T0 are ACT, DES, or PRE with one bank remaining open after completion of the PRECHARGE
command.
2. ODT pin driven to a valid state; MR5 [5] = 1.

Power-Down Entry after Read and Read with Auto Precharge
Power-Down Entry After Write with Auto Precharge
NOTE 1. tWR is programmed through MR0.

Power-Down Entry after Write
Precharge Power-Down Entry and Exit

π PM Tech PMA213816B
Refresh Command to Power-Down Entry
Activate Command to Power-Down Entry

Precharge/Precharge all Command to Power-Down Entry
MRS Command to Power-Down Entry

π PM Tech
Power-Down Clarifications
PMA213816B
When CKE is registered LOW for power-down entry, tPD (MIN) must be satisfied before CKE can be
registered HIGH for power-down exit. The minimum value of parameter tPD (MIN) is equal to the minimum
value of parameter tCKE (MIN) as shown in the Timing Parameters by Speed Bin table. A detailed example
of Case 1 is shown below.
Power-Down Entry/Exit Clarification

Maximum Power Saving Mode

This mode provides lowest power consuming mode which could be similar to the Self-Refresh status with no
internal refresh activity. When DDR4 SDRAM is in the maximum power saving mode, it does not need to
guarantee data retention nor respond to any external command (except maximum power saving mode exit
and asserting RESET_n signal LOW) to minimize the power consumption.
Maximum Power-Saving Mode Entry

Max power saving mode is entered through an MRS command. For devices with shared control/address
signals, a single DRAM device can be entered into the max power saving mode using the per DRAM
Addressability MRS command. Note that large CS_n hold time to CKE upon the mode exit may cause DRAM
malfunction, thus it is required that the CA parity, CAL and Gear Down modes are disabled prior to the max
power saving mode entry MRS command.
Maximum Power Saving mode Entry
Figure below illustrates the sequence and timing parameters required for the maximum power saving mode
with the per DRAM addressability (PDA).
Maximum Power Saving mode Entry with PDA
When entering Maximum Power Saving mode, only DES commands are allowed until tMPED is satisfied.
After tMPED period from the mode entry command, DRAM is not responsive to any input signals except
CS_n, CKE and RESET_n signals, and all other input signals can be High-Z. CLK should be valid for
tCKMPE period and then can be High-Z.

π PM Tech
CKE Transition during Maximum Power-Saving Mode
PMA213816B
CKE toggle is allowed when DRAM is in the maximum power saving mode. To prevent the device from
exiting the mode, CS_n should be issued ‘High’ at CKE ‘L’ to ’H’ edge with appropriate setup tMPX_S and
hold tMPX_HH timings.
CKE Transition Limitation to hold Maximum Power Saving Mode
Maximum Power-Saving Mode Exit

DRAM monitors CS_n signal level and when it detects CKE ‘L’ to ’H’ transition, and either exits from the
power saving mode or stay in the mode depending on the CS_n signal level at the CKE transition. Because
CK receivers are shut down during this mode, CS_n =’L’ is captured by rising edge of the CKE signal. If
CS_n signal level is detected ‘L’, then the DRAM initiates internal exit procedure from the power saving mode.
CK must be restarted and stable tCKMPX period before the device can exit the maximum power saving
mode. During the exit time tXMP, any valid commands except DES command is not allowed to DDR4
SDRAM and also tXMP_DLL, any valid commands requiring a locked DLL is not allowed to DDR4 SDRAM.
When recovering from this mode, the DRAM clears the MRS bits of this mode. It means that the setting of
MR4 [A1] is move to ’0’ automatically.
Maximum Power Saving Mode Exit Sequence

Timing parameter bin of Maximum Power Saving Mode for

DDR4-1600/1866/2133/2400/2666/3200
DDR4-1600/1866/2133/2400 DDR4-2666/2933/3200
Description Symbol Unit Note
Min Max Min Max
Command path disable delay tMOD(min) + tMOD(min) +

tMPED - - CK
upon MPSM entry tCPDED(min) tCPDED(min)
Valid clock requirement after tMOD(min) + tMOD(min) +

tCKMPE - - CK
MPSM entry tCPDED(min) tCPDED(min)
Valid clock requirement before

tCKMPX tCKSRX(min) - tCKSRX(min) - CK
MPSM exit
Exit MPSM to commands not

tXMP tXS(min) - tXS(min) - CK
requiring a locked DLL
Exit MPSM to commands tXMP(min) + tXMP(min) +

tXMPDLL - - CK
requiring a locked DLL tXSDLL(min) tXSDLL(min)
tISmin + tISmin +
CS setup time to CKE tMPX_S - - CK
tIHmin tIHmin
CS_n High hold time to CKE

tMPX_HH tXP(min) - tXP(min) - CK
rising edge
CS_n Low hold time to CKE

tMPX_LH 12 tXMP-10ns 12 tXMP-10ns ns 1
rising edge
NOTE 1 tMPX_LH(max) is defined with respect to actual tXMP in system as opposed to tXMP(min).

Connectivity Test Mode

The DDR4 memory device supports a connectivity test (CT) mode, which is designed to greatly speed up
testing of electrical continuity of pin interconnection on the PC boards between the DDR4 memory devices
and the memory controller on the SoC. Designed to work seamlessly with any boundary scan devices, the
CT mode is required for all x16 width devices independant of density and optional for all x8 and x4 width
devices with densities greater than or equal to 8Gb.
Contrary to other conventional shift register based test mode, where test patterns are shifted in and out of the
memory devices serially in each clock, DDR4’s CT mode allows test patterns to be entered in parallel into the
test input pins and the test results extracted in parallel from the test output pins of the DDR4 memory device
at the same time, significantly enhancing the speed of the connectivity check. RESET_n is registered to High
and VrefCA must be stable prior to entering CT mode. Once put in the CT mode, the DDR4 memory device
effectively appears as an asynchronous device to the external controlling agent; after the input test pattern is
applied, the connectivity check test results are available for extraction in parallel at the test output pins after a
fixed propagation delay. During CT mode, any ODT is turned off.
A reset of the DDR4 memory device is required after exiting the CT mode.
Pin Mapping
Only digital pins can be tested via the CT mode. For the purpose of connectivity check, all pins that are used
for the digital logic in the
DDR4 memory device are classified as one of the following types:
1. Test Enable (TEN) pin: when asserted high, this pin causes the DDR4 memory device to enter the CT
mode. In this mode, the normal memory function inside the DDR4 memory device is bypassed and the IO
pins appear as a set of test input and output pins to the external controlling agent. The TEN pin is
dedicated to the connectivity check function and will not be used during normal memory operation.
2. Chip Select (CS_n) pin: when asserted low, this pin enables the test output pins in the DDR4 memory
device. When de-asserted, the output pins in the DDR4 memory device will be tri-stated. The CS_n pin in
the DDR4 memory device serves as the CS_n pin when in CT mode.
3. Test Input: a group of pins that are used during normal DDR4 DRAM operation are designated test input
pins. These pins are used to enter the test pattern in CT mode.
4. Test Output: a group of pins that are used during normal DDR4 DRAM operation are designated test
output pins. These pins are used for extraction of the connectivity test results in CT mode.
5. RESET_n : Fixed high level is required during CT mode same as normal function.
Pin Classification of DDR4 Memory Device in Connectivity Test (CT) Mode

CT Mode Pins Pin Names during Normal Memory Operation
Test Enable TEN
Chip Select CS_n

BA0-1, BG0-1, A0-A9, A10/AP, A11, A12/BC_n, A13, WE_n/A14, CAS_n/A15, RAS_n/A16,
A
CKE, ACT_n, ODT, CLK_t, CLK_c, Parity
B DML_n, DBIL_n, DMU_n/DBIU_n, DM_n/DBI_n
Test Input
C Alert_n
D RESET_n
Test Output DQ0-DQ15, DQSU_t, DQSU_c, DQSL_t, DQSL_c

π PM Tech
Signal Description
PMA213816B

Connectivity Test Mode is active when TEN is HIGH and inactive when TEN is LOW.
TEN must be LOW during normal operation TEN is a CMOS rail-to-rail signal with
TEN Input
DC high and low at 80% and 20% of VDD, i.e, 960mV for DC high and 240mV for
DC low.

π PM Tech
Min Terms Definition for Logic Equations
PMA213816B
Min Term Equations

MTx is an inernal signal to be used to generate the signal to drive the output signals.
MT0 = XOR (A1, A6, PAR)
MT1 = XOR (A8, ALERT, A9)
MT2 = XOR (A2, A5, A15)
MT3 = XOR (A0 A7, A11)
MT4 = XOR (CK_c, ODT, CAS_n)
MT5 = XOR (Cke, RAS_n,/A16, A10/AP)
MT6 = XOR (ACT_n, A4, BA1)
MT7 = XOR (((x16 and DMU_n / DBIU_n) or (!x16 and BG1)), ((x8 or x16) and DML_n / DBIL_n), CK_t))
MT8 = XOR (WE_n / A14, A12 / BC, BA0)
MT9 = XOR (BG0, A3, (Reset_n and TEN))
Output equations for x4 devices

DQ0 = XOR(MT0, MT1)
DQ1 = XOR(MT2, MT3)
DQ2 = XOR(MT4, MT5
DQ3 = XOR(MT6, MT7)
DQS_t = MT8
DQS_c = MT9

DQ0 = MT0
DQ1 = MT1
DQ2 = MT2
DQ3 = MT3
DQ4 = MT4
DQ5 = MT5
DQ6 = MT6
DQ7 = MT7
DQS_t = MT8
DQS_c = MT9

DQ0 = MT0
DQ1 = !DQ0
DQ2 = MT1
DQ3 = !DQ2
DQ4 = MT2
DQ5 = !DQ4
DQ6 = MT3
DQ7 = !DQ6
DQ8 = MT4
DQ9 = !DQ8
DQ10 = MT5
DQ11 = ! DQ10
DQ12 = MT6
DQ13 = MT7
DQ14 = MT8
DQ15 = !DQ14
DQSL_t = MT9
DQSL_c = !DQ12
DQSU_t = !DQSL_t
DQSU_c = !DQ13

π PM Tech
CT Input Timing Requirement
PMA213816B
Prior to the assertion of the TEN pin, all voltage supplies, including VREFCA, must be valid and stable and
RESET_n registered high prior to entering CT mode. Upon the assertion of the TEN pin HIGH with RESET_n,
CKE, and CS_n held HIGH; CLK_t, CLK_c, and CKE signals become test inputs within tCTECT_Valid. The
remaining CT inputs become valid tCT_Enable after TEN goes HIGH when CS_n allows input to begin
sampling, provided inputs were valid for at least tCT_Valid. While in CT mode, refresh activities in the
memory arrays are not allowed; they are initiated either externally (auto refresh) or internally (self refresh).
The TEN pin may be asserted after the DRAM has completed power-on. After the DRAM is initialized and
VREFDQ is calibrated, CT mode may no longer be used. The TEN pin may be de-asserted at any time in CT
mode. Upon exiting CT mode, the states and the integrity of the original content of the memory array are
unknown. A full reset of the memory device is required.
After CT mode has been entered, the output signals will be stable within tCT_Valid after the test inputs have
been applied as long as TEN is maintained HIGH and CS_n is maintained LOW.
Connectivity Test Mode Entry
Symbol Min Max Unit

tCT_IS 0 - ns
tCT_Enable 200 - ns
tCT_Valid - 200 ns

π PM Tech
CT Input Levels
PMA213816B
CMOS rail to rail Input Levels for TEN
Symbol Parameter Min Max Unit Note

VIH(AC)_TEN TEN AC Input High Voltage 0.8 x VDD VDD V 1
VIH(DC)_TEN TEN DC Input High Voltage 0.7 x VDD VDD V
VIL(DC)_TEN TEN DC Input Low Voltage VSS 0.3 x VDD V
VIL(AC)_TEN TEN AC Input Low Voltage VSS 0.2 x VDD V 2
TF_input_TEN TEN Input signal Falling time – 10 ns
TR_input_TEN TEN Input signal Rising time – 10 ns
NOTE
1. Overshoot should not exceed the Vin Absolute Maximum Ratings.
2. Undershoot should not exceed the Vin Absolute Maximum Ratings.
TEN Input Slew Rate Definition
CT Type-A Input Levels
Parameter Symbol Min Mix Unit Note

1
CTipA AC Input High Voltage VIH(AC)_CTipA VREFCA + 0.2 Note V
CTipA DC Input High Voltage VIH(DC)_CTipA VREFCA + 0.15 VDD V
CTipA DC Input Low Voltage VIL(DC)_CTipA VSS VREFCA - 0.15 V
CTipA AC Input Low Voltage VIL(AC)_CTipA Note 1 VREFCA - 0.2 V
CTipA Input signal Falling time TF_input_CTipA - 5 ns
CTipA Input signal Rising time TR_input_CTipA - 5 ns
NOTE 1 See “Overshoot and Undershoot Specifications”.
CT Type-A Input Slew Rate Definition

π PM Tech
CT Type-B Input Levels
PMA213816B

CTipB AC Input High Voltage VIH(AC)_CTipB VREFDQ + 0.3 Note 2 V 1
CTipB DC Input High Voltage VIH(DC)_CTipB VREFDQ + 0.2 VDDQ V 1
CTipB DC Input Low Voltage VIL(DC)_CTipB VSSQ VREFDQ - 0.2 V 1
CTipB AC Input Low Voltage VIL(AC)_CTipB Note 2 VREFDQ - 0.3 V 1
CTipB Input signal Falling time TF_input_CTipB - 5 ns
CTipB Input signal Rising time TR_input_CTipB - 5 ns
NOTE
1. VREFDQ is VDDQ*0.5.
2. See "Overshoot and Undershoot Specifications".
CT Type-B Input Slew Rate Definition
CT Type-C Input Levels (CMOS)

CTipC AC Input High Voltage VIH(AC)_CTipC 0.8 × VDD VDD 1 V 2
CTipC DC Input High Voltage VIH(DC)_CTipC 0.7 × VDD VDD V 2
CTipC DC Input Low Voltage VIL(DC)_CTipC VSS 0.3 × VDD V 2
CTipC AC Input Low Voltage VIL(AC)_CTipC VSS 1 0.2 × VDD V 2
CTipC Input signal Falling time TF_input_CTipC - 10 ns 2
CTipC Input signal Rising time TR_input_CTipC - 10 ns 2
NOTE
1. Refer to Overshoot and Undershoot Specifications.
2. CT Type-C inputs: Alert_n.
CT Type-C Input Slew Rate Definition

π PM Tech
CT Type-D Input Levels
PMA213816B

CTipD AC Input High Voltage VIH(AC)_CTipD 0.8 × VDD VDD V 4
CTipD DC Input High Voltage VIH(DC)_CTipD 0.7 × VDD VDD V 2
CTipD DC Input Low Voltage VIL(DC)_CTipD VSS 0.3 × VDD V 1
CTipD AC Input Low Voltage VIL(AC)_CTipD VSS 0.2 × VDD V 5
Rising time tR_RESET - 1 μs 3
RESET pulse width - after power-up tPW_RESET_S 1 - μs 2
RESET pulse width - during power-up PW_RESET_L 200 - μs 2
NOTE
1. After RESET_n is registered LOW, the RESET_n level must be maintained below VIL(DC)_RESET during tPW_RESET,
otherwise, the DRAM may not be reset.
2. After RESET_n is registered HIGH, the RESET_n level must be maintained above VIH(DC)_RESET, otherwise, operation
will be uncertain until it is reset by asserting RESET_n signal LOW.
3. Slope reversal (ring-back) during this level transition from LOW to HIGH should be mitigated as much as possible.
4. Overshoot should not exceed the VIN values in the Absolute Maximum Ratings table.
5. Undershoot should not exceed the VIN values in the Absolute Maximum Ratings table.
6. CT Type-D inputs: RESET_n; same requirements as in normal mode.
CT Type-D Input Slew Rate Definition

π PM Tech
CLK to Read DQS Timing Parameters
PMA213816B
DDR4 supports DLLOFF mode. Following parameters will be defined for CK to read DQS timings.
Speed DDR4-1600/1866/2133/2400/2666/2933/3200
Unit Note
Parameter Symbol Min Max
refer to AC parameter refer to AC parameter
tDQSCK (DLL On) ps 1,3,8,9
DQS_t, DQS_c rising edge output tables tables
timing location from rising CK_t, CK_c
tDQSCK (DLL Off) vendor specific vendor specific ps 2, 3, 8
refer to AC parameter
tDQSCKi(DLL On) - ps 1,5,6,8,9
DQS_t, DQS_c rising edge output tables
variance window
tDQSCKi(DLL Off) - vendor specific ps 2,4,5,6,8
VDD sensitivity of tDQSCK (DLL

dTDQSCKdV - vendor specific ps/mV 2, 6
Off)
Temperature sensitivity of tDQSCK
dTDQSCKdT - vendor specific ps/oC 2,6
(DLL Off)
NOTE
1. These parameters are applied when DRAM is in DLLON mode.
2. These parameters are applied when DRAM is in DLLOFF mode.
3. Measured over full VDD and Temperature spec ranges.
4. Measured at fixed and constant VDD and Temperature condition.
5. Measured for a given DRAM part, and for each DQS/DQ pair in case of x16 (part variation is excluded).
6. These parameters are verified by design and characterization, and may not be subject to production test.
7. Assume no jitter on input clock signals to the DRAM.
8. Refer to “READ Timing Definitions”.
9. Refer to READ Timing Definitions READ Timing Definitions.
tDQSCK,Min
= Min {Min {tDQSCK(j), all VDD and Temperature ranges}, all DQS pairs and parts}
tDQSCK,Max
= Max {Max {tDQSCK(j), all VDD and Temperature ranges}, all DQS pairs and parts}
tDQSCKJ, Max
= Max {tDQSCK(j), fixed VDD and Temperature} - Min {tDQSCK(j), fixed VDD and Temperature}
tDQSCK Definition Difference between DLL ON and DLL OFF

π PM Tech
dTDQSCKTdT Definition
PMA213816B
dTDQSCKdT = l tDQSCK(Toper,max) - tDQSCK(Toper,min)l / lToper,max - Toper,minl
TDQSCKTdV Definition
dTDQSCKdV = l tDQSCK(VDD,max) - tDQSCK(VDD,min)l / lVDD,max - VDD,minl

π PM Tech
On-Die Termination (ODT)
PMA213816B
The on-die termination (ODT) feature enables the device to change termination resistance for each DQ, DQS,
and DM_n/DBI_n signal for x4 and x8 configurations (and TDQS for the x8 configuration when enabled via
A11 = 1 in MR1) via the ODT control pin, WRITE command, or default parking value with MR setting. For the
x16 configuration, ODT is applied to each UDQ, LDQ, UDQS, LDQS, UDM_n/UDBI_n, and LDM_n/LDBI_n
signal. The ODT feature is designed to improve the signal integrity of the memory channel by allowing the
DRAM controller to independently change termination resistance for any or all DRAM devices. If DBI read
mode is enabled while the DRAM is in standby, either DM mode or DBI write mode must also be enabled if
RTT(NOM) or RTT(Park) is desired. More details about ODT control modes and ODT timing modes can be
found further along in this document.
The ODT feature is turned off and not supported in self refresh mode.
Functional Representation of ODT
The switch is enabled by the internal ODT control logic, which uses the external ODT pin and other control
information. The value of RTT is determined by the settings of mode register bits (see Mode Register). The
ODT pin will be ignored if the mode register MR1 is programmed to disable RTT(NOM) [MR1[10,9,8] = 0,0,0]
and in self refresh mode.
ODT Mode Register and ODT State Table

The ODT Mode of DDR4 SDRAM has 4 states, Data Termination Disable, RTT_WR, RTT_NOM and
RTT_PARK. And the ODT Mode is enabled if any of MR1{A10,A9,A8} or MR2 {A10:A9} or MR5 {A8:A6} are
non zero. In this case, the value of RTT is determined by the settings of those bits.
After entering Self-Refresh mode, DRAM automatically disables ODT termination and set Hi-Z as termination
state regardless of these setting.
Application: Controller can control each RTT condition with WR/RD command and ODT pin
 RTT_WR: The rank that is being written to provide termination regardless of ODT pin status (either HIGH
or LOW)
 RTT_NOM: DRAM turns ON RTT_NOM if it sees ODT asserted (except ODT is disabled by MR1).
 RTT_PARK: Default parked value set via MR5 to be enabled and ODT pin is driven LOW.
 Data Termination Disable: DRAM driving data upon receiving READ command disables the termination
after RL-X and stays off for a duration of BL/2 + X + Y clock cycles.
X is 2 for 1tCK and 3 for 2tCK preamble mode.
Y is 0 when CRC is disabled and 1 when it’s enabled
 The Termination State Table is shown in Table 54.

π PM Tech
Those RTT values have following priority.
PMA213816B
1. Data Termination Disable

2. RTT_WR
3. RTT_NOM
4. RTT_PARK
which means if there is WRITE command along with ODT pin HIGH, then DRAM turns on RTT_WR not
RTT_NOM, and also if there is READ command, then DRAM disables data termination regardless of ODT
pin and goes into Driving mode.
Termination State Table

RTT_PARK MR5{A8:A6} RTT_NOM MR1 {A10:A9:A8} ODT pin DRAM termination state Note
HIGH RTT_NOM 1,2
Enabled
Enabled LOW RTT_PARK 1,2
Disabled Don’t care RTT_PARK 1,2,3
HIGH RTT_NOM 1,2
Enabled
Disabled LOW Hi-Z 1,2
Disabled Don’t care Hi-Z 1,2,3
NOTE
1. When read command is executed, DRAM termination state will be Hi-Z for defined period independent of ODT pin and MR
setting of RTT_PARK/RTT_NOM. This is described in section 1.2.3 ODT During Read.
2. If RTT_WR is enabled, RTT_WR will be activated by Write command for defined period time independent of ODT pin and
MR setting of RTT_PARK /RTT_NOM. This is described in section 1.3 Dynamic ODT.
3. If RTT_NOM MRS is disabled, ODT receiver power will be turned off to save power.
On-Die Termination effective resistance RTT is defined by MRS bits.

ODT is applied to the DQ, DM, DQS_T/DQS_C and TDQS_T/TDQS_C (x8 devices only) pins.
A functional representation of the on-die termination is shown in the figure below.
On Die Termination

π PM Tech
On die termination effective Rtt values supported are 240, 120, 80, 60, 48, 40, 34 ohms.
PMA213816B
ODT Electrical Characteristics RZQ=240Ω +/-1% entire temperature operation range; after proper ZQ
calibration.
RTT Vout Min Nom Max Unit Note
VOLdc= 0.5* VDDQ 0.9 1 1.25 RZQ 1,2,3
240 ohm VOMdc= 0.8* VDDQ 0.9 1 1.1 RZQ 1,2,3
VOHdc= 1.1* VDDQ 0.8 1 1.1 RZQ 1,2,3
VOLdc= 0.5* VDDQ 0.9 1 1.25 RZQ/2 1,2,3
120 ohm VOMdc= 0.8* VDDQ 0.9 1 1.1 RZQ/2 1,2,3
VOHdc= 1.1* VDDQ 0.8 1 1.1 RZQ/2 1,2,3
VOLdc= 0.5* VDDQ 0.9 1 1.25 RZQ/3 1,2,3
VOHdc= 1.1* VDDQ 0.8 1 1.1 RZQ/3 1,2,3
VOLdc= 0.5* VDDQ 0.9 1 1.25 RZQ/4 1,2,3
VOHdc= 1.1* VDDQ 0.8 1 1.1 RZQ/4 1,2,3
VOLdc= 0.5* VDDQ 0.9 1 1.25 RZQ/5 1,2,3
VOHdc= 1.1* VDDQ 0.8 1 1.1 RZQ/5 1,2,3
VOLdc= 0.5* VDDQ 0.9 1 1.25 RZQ/6 1,2,3
VOHdc= 1.1* VDDQ 0.8 1 1.1 RZQ/6 1,2,3
VOLdc= 0.5* VDDQ 0.9 1 1.25 RZQ/7 1,2,3
VOHdc= 1.1* VDDQ 0.8 1 1.1 RZQ/7 1,2,3
DQ-DQ
Mismatch within byte VOMdc= 0.8* VDDQ 0 - 10 % 1,2,4,5,6
NOTE
1. The tolerance limits are specified after calibration with stable voltage and temperature. For the behavior of the tolerance
limits if temperature or voltage changes after calibration, see following section on voltage and temperature sensitivity.
2. Pull-up ODT resistors are recommended to be calibrated at 0.8*VDDQ. Other calibration schemes may be used to achieve
the linearityspec shown above, e.g. calibration at 0.5*VDDQ and 1.1*VDDQ.
3. The tolerance limits are specified under the condition that VDDQ=VDD and VSSQ=VSS
4. DQ to DQ mismatch within byte variation for a given component including DQS_T and DQS_C (characterized)
5. RTT variance range ratio to RTTNominal value in a given component, including DQS_T and DQS_C
DQ-DQ Mismatch in a device =
6. This parameter of x16 device is specified for Upper byte and Lower byte.

π PM Tech
Synchronous ODT Mode
PMA213816B
Synchronous ODT mode is selected whenever the DLL is turned on and locked. Based on the power-down
definition, these modes are:
 Any bank active with CKE high
 Refresh with CKE high
 Idle mode with CKE high
 Active power down mode
 Precharge power down mode
In synchronous ODT mode, RTT_NOM will be turned on DODTLon clock cycles after ODT is sampled HIGH
by a rising clock edge and turned off DODTLoff clock cycles after ODT is registered LOW by a rising clock
edge. The ODT latency is tied to the Write Latency (WL = CWL + AL + PL) by: DODTLon = WL - 2;
DODTLoff = WL - 2.
When operating in 2tCK Preamble Mode, The ODT latency must be 1 clock smaller than in 1tCK Preamble
Mode; DODTLon =WL - 3; DODTLoff = WL - 3."(WL = CWL+AL+PL).
ODT Latency and Posted ODT

In Synchronous ODT Mode, the Additive Latency (AL) and the Parity Latency (PL) programmed into the
Mode Register MR1 applies to ODT Latencies as shown below:
ODT Latency
Applicable when write CRC is disabled
Symbol Parameter 1 tCK Preamble 2 tCK Preamble Unit
DODTLon Direct ODT turn on Latency CWL + AL + PL – 2 CK CWL + AL + PL – 3 CK
DODTLoff Direct ODT turn off Latency CWL + AL + PL – 2 CK CWL + AL + PL – 3 CK
Read command to internal ODT turn off

RODTLoff Latency CWL + AL + PL – 2 CK CWL + AL + PL – 3 CK
RODTLon4 Read command to RTT_PARK turn on Latency

in BC4 RODTLoff + 4 RODTLoff + 5 tCK
RODTLon8 Read command to RTT_PARK turn on Latency

in BC8/BL8 RODTLoff + 6 RODTLoff + 7
ODTH4 4 5
ODTH8 6 7

π PM Tech
Timing Parameters
PMA213816B
In synchronous ODT mode, the following timing parameters apply:

 DODTLon, DODTLoff, RODTLoff, RODTLon4, RODTLon8, tADC,min,max.
 tADC,min and tADC,max are minimum and maximum RTT change timing skew between different
termination values. Those timing parameters apply to both the Synchronous ODT mode and the Data
Termination Disable mode.
When ODT is asserted, it must remain HIGH until minimum ODTH4 (BL=4) or ODTH8 (BL=8) is satisfied.
Additionally, depending on CRC or 2tCK preamble setting in MRS, ODTH should be adjusted.
Synchronous ODT Timing with BL8
Example for CWL = 9, AL = 0, PL = 0; DODTLon = AL + PL +CWL -2 = 7; DODTLoff = AL + PL + CWL -2 = 7.

ODT must be held HIGH for at least ODTH8 after assertion (T1).
Synchronous ODT with BL4
Example for CWL = 9, AL = 10, PL = 0; DODTLon/off = AL + PL+ CWL -2 = 17; ODTcnw = AL + PL+ CWL -2 = 17.

π PM Tech
ODT must be held HIGH for at least ODTH4 after assertion (T1). ODTH is measured from ODT first
PMA213816B
registered HIGH to ODT first registered LOW, or from registration of Write command. Note that ODTH4
should be adjusted depending on CRC or 2tCK preamble setting.
ODT During Reads

Because the DRAM cannot terminate with RTT and drive with RON at the same time, RTT may nominally
not be enabled until the end of the postamble as shown in the example below. At cycle T25 the device turns
on the termination when it stops driving, which is determined by tHZ. If the DRAM stops driving early (that is,
tHZ is early), then tADC (MIN) timing may apply. If the DRAM stops driving late (that is, tHZ is late), then the
DRAM complies with tADC (MAX) timing.
Using CL = 11 as an example for the figure below: PL = 0, AL = CL - 1 = 10, RL = PL + AL + CL = 21, CWL=
9; RODTLoff = RL - 2 = 19, DODTLon = PL + AL + CWL - 2 = 17, 1tCK preamble.
Example: CL=11, PL=0; AL=CL-1=10; RL=AL+PL+CL=21; CWL=9;

DODTLon=AL+CWL-2=17; DODTLoff=AL+CWL-2=17;1tCK preamble)

π PM Tech
Dynamic ODT
PMA213816B
In certain application cases and to further enhance signal integrity on the data bus, it is desirable that the
termination strength of the DDR4 SDRAM can be changed without issuing an MRS command. This
requirement is supported by the “Dynamic ODT” feature as described as follows:
Functional Description
The Dynamic ODT Mode is enabled if bit A[9] or A[10] of MR2 is set to ’1’. The function is described as
follows:
 Three RTT values are available: RTT_NOM, RTT_PARK and RTT_WR.
– The value for RTT_NOM is preselected via bits A[10:8] in MR1
– The value for RTT_PARK is preselected via bits A[8:6] in MR5
– The value for RTT_WR is preselected via bits A[10:9] in MR2
 During operation without commands, the termination is controlled as follows;
– Nominal termination strength RTT_NOM or RTT_PARK is selected.
– RTT_NOM on/off timing is controlled via ODT pin and latencies DODTLon and DODTLoff and
RTT_PARK is on when ODT is LOW.
 When a write command (WR, WRA, WRS4, WRS8, WRAS4, WRAS8) is registered, and if Dynamic ODT
is enabled, the termination is controlled as follows:
– A latency ODTLcnw after the write command, termination strength RTT_WR is selected.
– A latency ODTLcwn8 (for BL8, fixed by MRS or selected OTF) or ODTLcwn4 (for BC4, fixed by MRS or
selected OTF) after the write command, termination strength RTT_WR is de-selected.
– 1 or 2 clocks will be added or subtracted into/from ODTLcwn8 and ODTLcwn4 depending on CRC
and/or 2tCK preamble setting.
The following table shows latencies and timing parameters which are relevant for the on-die termination
control in Dynamic ODT mode.
The dynamic ODT feature is not supported in DLL-off mode. MRS command must be used to set RTT_WR,
MR2[11:9] = 000, to disable dynamic ODT externally.
Dynamic ODT Latencies and Timing (1 tCK Preamble Mode and CRC Disabled)
1600/1866/
Name and Description Abbr. Defined from Define to 2133/2400 2666 2933/3200 Unit
ODT Latency for change Registering Change RTT strength

from RTT_PARK/ ODTLcnw external write from RTT_PARK/ ODTLcnw = WL - 2 tCK
RTT_NOM to RTT_WR command RTT_Nom to RTT_WR
ODT Latency for change
Registering Change RTT strength
from RTT_WR to
ODTLcwn4 external write from RTT_WR to ODTLcwn4 = 4 + ODTLcnw tCK
RTT_PARK/RTT_Nom
command RTT_PARK/ RTT_Nom
(BL = 4)
ODT Latency for change
registering Change RTT strength
from RTT_WR to
ODTLcwn8 external write from RTT_WR to ODTLcwn8 = 6 + ODTLcnw tCK(avg)
RTT_PARK/RTT_Nom
command RTT_PARK/ RTT_Nom
(BL = 8)

tADC(min) tADC(min) tADC(min)

ODTLcnw = 0.3 = 0.28 = 0.26
RTT change skew tADC RTT valid tADC(max) tCK(avg)
ODTLcwn tADC(max) tADC(max)
= 0.7 = 0.72 = 0.74

π PM Tech
Dynamic ODT Latencies and Timing with Preamble Mode and CRC Mode Matrix
PMA213816B
1tck Preamble 2tck Preamble

Symbol Unit
CRC off CRC on CRC off CRC on
ODTLcnw WL - 2 WL - 2 WL - 3 WL - 3
ODTLcwn4 ODTLcnw +4 ODTLcnw +7 ODTLcnw +5 ODTLcnw +8 tCK
ODTLcwn8 ODTLcnw +6 ODTLcnw +7 ODTLcnw +7 ODTLcnw +8
ODT Timing Diagrams
Dynamic ODT (1t CK Preamble; CL = 14, CWL = 11, BL = 8, AL = 0, CRC Disabled)
ODTLcnw = WL-2 (1tCK preamble), WL-3 (2tCK preamble).

ODTLcwn = WL+2 (BC4), WL+4(BL8) w/o CRC or WL+5,5 (BC4, BL8 respectively) when CRC is enabled.
Dynamic ODT Overlapped with RTT_NOM (CL=14, CWL=11, BL=8, AL=0, CRC Disabled)
Behavior with WR command is issued while ODT being registered high.

π PM Tech
Asynchronous ODT mode
PMA213816B
Asynchronous ODT mode is selected when DRAM runs in DLL off mode. In asynchronous ODT timing mode,
the internal ODT command is not delayed by either Additive Latency (AL) or the Parity Latency (PL) relative
to the external ODT signal (RTT_NOM).
In asynchronous ODT mode, two timing parameters apply: tAONAS (MIN/MAX), tAOFAS (MIN/MAX).
RTT_NOM turn-on time
 Minimum RTT_NOM turn-on time (tAONAS (MIN) is the point in time when the device termination circuit
leaves RTT_PARK and ODT resistance begins to turn on.
 Maximum RTT_NOM turn-on time (tAONAS [MAX]) is the point in time when the ODT resistance has
reached RTT_NOM.
 tAONAS (MIN) and tAONAS (MAX) are measured from ODT being sampled HIGH.
RTT_NOM turn-off time
 Minimum RTT_NOM turn-off time (tAOFAS [MIN]) is the point in time when the device's termination circuit
starts to leave RTT_NOM.
 Maximum RTT_NOM turn-off time (tAOFAS [MAX]) is the point in time when the on die termination has
reached RTT_PARK.
 tAOFAS (MIN) and tAOFAS (MAX) are measured from ODT being sampled LOW.
Asynchronous ODT Timings with DLL Off
Asynchronous ODT Timing Parameters for all Speed Bins
Description Symbol min max Unit

Asynchronous RTT turn-on delay tAONAS 1.0 9.0 ns
Asynchronous RTT turn-off delay tAOFAS 1.0 9.0 ns

π PM Tech
ODT buffer disabled mode for Power down
PMA213816B
DRAM does not provide Rtt_NOM termination during power down when ODT input buffer deactivation mode
is enabled in MR5 bit A5. To account for DRAM internal delay on CKE line to disable the ODT buffer and
block the sampled output, the host controller must continuously drive ODT to either low or high when
entering power down. The ODT signal may be floating after tCPDEDmin has expired. In this mode,
RTT_NOM termination corresponding to sampled ODT at the input after CKE is first registered low (and
tANPD before that) may not be provided. tANPD is equal to (WL-1) and is counted backwards from PDE.
ODT timing for power down entry with ODT buffer disable mode
When exit from power down, along with CKE being registered high, ODT input signal must be re-driven and
maintained low until tXP is met.
ODT timing for power down exit with ODT buffer disable mode

π PM Tech
ODT Timing Definitions
PMA213816B
The reference load for ODT timings is different than the reference load used for timing measurements.
ODT Timing Reference Load

Definitions for tADC, tAONAS and tAOFAS are provided in the Table and measurement reference settings
are provided in the subsequent. The tADC for the Dynamic ODT case and Read Disable ODT cases are
represented by tADC of Direct ODT Control case.
Symbol Begin Point Definition End Point Definition
Rising edge of CK_t,CK_c defined by the end point of DODTLoff Extrapolated point at VRTT_NOM
Rising edge of CK_t,CK_c defined by the end point of DODTLon Extrapolated point at VSSQ
tADC Extrapolated point at VRTT_NOM
Rising edge of CK_t,CK_c defined by the end point of ODTLcnw
Rising edge of CK_t,CK_c defined by the end point of ODTLcwn4
Extrapolated point at VSSQ
or ODTLcwn8
tAONAS Rising edge of CK_t,CK_c with ODT being first registered high Extrapolated point at VSSQ
tAOFAS Rising edge of CK_t,CK_c with ODT being first registered low Extrapolated point at VRTT_NOM
Reference Settings for ODT Timing Measurements
Measured Parameter RTT_PARK RTT_NOM RTT_WR Vsw1 Vsw2 Note
Disable RZQ/7 - 0.20V 0.40V 1,2

tADC
- RZQ/7 Hi-Z 0.20V 0.40V 1,3
tAONAS Disable RZQ/7 - 0.20V 0.40V

1,2
tAOFAS Disable RZQ/7 - 0.20V 0.40V
NOTE
1. MR setting is as follows.
-MR1 A10=1, A9=1, A8=1 (RTT_NOM_Setting)
-MR5 A8=0 , A7=0, A6=0 (RTT_PARK Setting)
-MR2 A11=0, A10=1, A9=1 (RTT_WR Setting)
2. ODT state change is controlled by ODT pin.
3. ODT state change is controlled by Write Command.

π PM Tech
Definition of tADC
PMA213816B
Definition of tAOFAS and tAONAS

π PM Tech
Absolute Maximum DC Ratings
PMA213816B
Symbol Parameter Min. Max Unit NOTE

VDD Voltage on VDD pin relative to Vss -0.3 1.5 V 1,3
VDDQ Voltage on VDDQ pin relative to Vss -0.3 1.5 V 1,3
VPP Voltage on VPP pin relative to Vss -0.3 3.0 V 4
VIN, VOUT Voltage on any pin relative to Vss -0.3 1.5 V 1
TSTG Storage Temperature -55 100 °C 1,2
NOTE
1. Stresses greater than those listed under “Absolute Maximum Ratings” may cause permanent damage to the device. This is
a stress rating only and functional operation of the device at these or any other conditions above those indicated in the
operational sections of this specification is not implied. Exposure to absolute maximum rating conditions for extended
periods may affect reliability.
2. Storage Temperature is the case surface temperature on the center/top side of the DRAM. For the measurement
conditions, please refer to JESD51-2 standard.
3. VDD and VDDQ must be within 300 mV of each other at all times;and VREFCA must be not greater than 0.6 x VDDQ,
When VDD and VDDQ are less than 500 mV; VREF may be equal to or less than 300 mV.
4. VPP must be equal or greater than VDD/VDDQ at all times.

π PM Tech
AC and DC Operating Conditions
PMA213816B
Recommended DC Operating Conditions

Rating
Symbol Parameter Unit NOTE
Min. Typ. Max.
VDD Supply Voltage 1.14 1.2 1.26 V 1,2,3
VDDQ Supply Voltage for Output 1.14 1.2 1.26 V 1,2,3
VPP 2.375 2.5 2.75 V 3
NOTE
1. Under all conditions VDDQ must be less than or equal to VDD.
2. VDDQ tracks with VDD. AC parameters are measured with VDD and VDDQ tied together.
3. The DC bandwidth is limited to 20MHz.
AC and DC Input Measurement Levels

AC & DC Logic input levels for single-ended signals
Single-ended AC & DC input levels for Command and Address
DDR4-1600/1866/2133/2400 DDR4-2666/2933/3200
min max min max
VIH.CA(DC75) DC input logic high VREF+0.075 VDD V
VIL.CA(DC75) DC input logic low VSS VREF-0.075 V
VIH.CA(DC65) DC input logic high - - VREF+0.065 VDD V
VIL.CA(DC65) DC input logic low - - VSS VREF-0.065 V
VIH.CA(AC100) AC input logic high VREF+0.1 Note 2 1

V
VIL.CA(AC100) AC input logic low Note 2 VREF - 0.1 V 1
VIH.CA(AC90) AC input logic high - - VREF+0.09 Note 2 V
VIL.CA(AC90) AC input logic low - - Note 2 VREF - 0.09 V
Reference Voltage V 2,3

VREFCA(DC) for ADD, CMD inputs 0.49*VDD 0.51*VDD 0.49*VDD 0.51*VDD
NOTE
1. See “Overshoot and Undershoot Specifications” .
2. The AC peak noise on VREFCA may not allow VREFCA to deviate from VREFCA(DC) by more than ± 1% VDD (for reference : approx.
± 12mV) 3. For reference : approx. VDD/2 ± 12mV

π PM Tech
VREF Tolerances
PMA213816B
The DC-tolerance limits and ac-noise limits for the reference voltages VREFCA is illustrated in Illustration of
VREF(DC) tolerance and VREF AC-noise limits. It shows a valid reference voltage VREF(t) as a function of
time. (VREF stands for VREFCA).
VREF(DC) is the linear average of VREF(t) over a very long period of time (e.g. 1 sec). This average has to
meet the min/max requirement in Single-ended AC & DC input levels for Command and Address.
Furthermore, VREF(t) may temporarily deviate from VREF(DC) by no more than ± 1% VDD.
Illustration of VREF(DC) tolerance and VREF AC-noise limits
The voltage levels for setup and hold time measurements VIH(AC), VIH(DC), VIL(AC) and VIL(DC) are
dependent on VREF. "VREF" shall be understood as VREF(DC), as defined in Illustration of V.
This clarifies, that DC-variations of VREF affect the absolute voltage a signal has to reach to achieve a valid
high or low level and therefore the time to which setup and hold is measured. System timing and voltage
budgets need to account for VREF(DC) deviations from the optimum position within the data-eye of the input
signals.
This also clarifies that the DRAM setup/hold specification and derating values need to include time and
voltage associated with VREF AC-noise. Timing and voltage effects due to AC-noise on VREF up to the
specified limit (+/-1% of VDD) are included in DRAM timings and their associated deratings.

π PM Tech
AC and DC Logic Input Levels for Differential Signals
PMA213816B
Differential signal definition
Definition of differential ac-swing and “time above ac-level” tDVAC
Differential swing requirements for clock (CK_t - CK_c)

Differential AC and DC Input Levels
DDR4-1600/1866/2133 DDR4-2400/2666
min max min max
VIHdiff Differential input high +0.150 NOTE 3 +0.135 NOTE 3 V 1
VILdiff Ddifferential input low NOTE 3 -0.150 NOTE 3 -0.135 V 1

2 × (VIH(AC) 2 × (VIH(AC)
VIHdiff(AC) Differential input high (AC) NOTE 3 NOTE 3 V 2
- VREF) - VREF)
2 × (VIL(AC) - 2 × (VIL(AC) -
VILdiff(AC) Differential input low (AC) NOTE 3 NOTE 3 V 2
VREF) VREF)
DDR4-2933 DDR4-3200
min max min max
VIHdiff Differential input high +0.125 NOTE 3 +0.110 NOTE 3 V 1
VILdiff Ddifferential input low NOTE 3 -0.125 NOTE 3 -0.110 V 1

2 × (VIH(AC) 2 × (VIH(AC)
VIHdiff(AC) Differential input high (AC) NOTE 3 NOTE 3 V 2
- VREF) - VREF)
2 × (VIL(AC) - 2 × (VIL(AC) -
VILdiff(AC) Differential input low (AC) NOTE 3 NOTE 3 V 2
VREF) VREF)
NOTE
1. Used to define a differential signal slew-rate.
2. For CK_t, CK_c use VIH(AC) and VIL(AC) of ADD/CMD and VREFCA.
3. These values are not defined; however, the differential signals (CK_t, CK_c) need to be within the respective limits, VIH(DC)max and
VIL(DC)min for single-ended signals as well as the limitations for overshoot and undershoot.

π PM Tech
Allowed time before ringback (tDVAC) for CK_t - CK_c
PMA213816B
tDVAC [ps] @ |VIH/Ldiff(AC)| = 200mV tDVAC [ps] @ |VIH/Ldiff(AC)| = TBDmV

Slew Rate [V/ns]
min max min max
> 4.0 120 - TBD -
4.0 115 - TBD -
3.0 110 - TBD -
2.0 105 - TBD -
1.8 100 - TBD -
1.6 95 - TBD -
1.4 90 - TBD -
1.2 85 - TBD -
1.0 80 - TBD -
< 1.0 80 - TBD -
Single-ended requirements for differential signals

Each individual component of a differential signal (CK_t, CK_c) has to comply with certain requirements for
single-ended signals. CK_t and CK_c have to reach approximately VSEHmin/VSEL,max, which are
approximately equal to the AC levels VIH(AC) and VIL(AC) for ADD/CMD signals in every half-cycle. The
applicable AC levels for ADD/CMD might differ per speed-bin, and so on. For example, if a value other than
100mV is used for ADD/CMD VIH(AC) and VIL(AC) signals, then these AC levels also apply for the
singleended signals CK_t and CK_c.
While ADD/CMD signal requirements are with respect to VREFCA, the single-ended components of differential
signals have a requirement with respect to VDD/2; this is nominally the same. The transition of single-ended
signals through the AC levels is used to measure setup time. For single-ended components of differential
signals the requirement to reach VSEL,max/VSEH,min has no bearing on timing, but adds a restriction on
the common mode characteristics of these signals.
Single-Ended Requirements for CK

π PM Tech
Single-ended levels for CK_t, CK_c
PMA213816B
DDR4-2133 DDR4-2400/2666 DDR4-2933/3200

Min Max Min Max Min Max
Single-ended
(VDD/2) (VDD/2) (VDD/2)
VSEH high-level for NOTE3 NOTE3 NOTE3 V 1, 2
+0.100 +0.095 +0.085
CK_t - CK_c
Single-ended
(VDD/2) (VDD/2) (VDD/2)
VSEL low-level for NOTE3 NOTE3 NOTE3 V 1, 2
-0.100 -0.095 -0.085
CK_t - CK_c
NOTE
1. For CK_t, CK_c use VIH(AC) and VIL(AC) of ADD/CMD and VREFCA.
2. ADDR/CMD VIH(AC) and VIL(AC) based on VREFCA.
3. These values are not defined; however, the differential signal (CK_t, CK_c) need to be within the respective limits, VIH(DC)max and
VIL(DC)min for single-ended signals as well as the limitations for overshoot and undershoot.
Address, Command and Control Overshoot and Undershoot Specifications
AC overshoot/undershoot specification for Address, Command and Control pins
Specification
Parameter DDR4- DDR4- DDR4- DDR4- DDR4- DDR4- DDR4- Unit
1600 1866 2133 2400 2666 2933 3200
Maximum peak amplitude above VDD
0.06 0.06 0.06 0.06 0.06 0.06 0.06 V
Absolute Max allowed for overshoot area
Delta value between VDD Absolute Max
0.24 0.24 0.24 0.24 0.24 0.24 0.24 V
and VDD Max allowed for overshoot area
Maximum peak amplitude allowed for
0.3 0.3 0.3 0.3 0.3 0.3 0.3 V-ns
undershoot area
Maximum overshoot area per 1tCK Above
0.0083 0.0071 0.0062 0.0055 0.0055 0.0055 0.0055 V-ns
Absolute Max
Maximum overshoot area per 1tCK
0.2550 0.2185 0.1914 0.1699 0.1699 0.1699 0.1699 V-ns
Between Absolute Max and VDD Max
Maximum undershoot area per 1tCK Below
0.2644 0.2265 0.1984 0.1762 0.1762 0.1762 0.1762 V-ns
VSS
(A0-A13,BG0-BG1,BA0-BA1,ACT_n,RAS_n,CAS_n/A15,WE_n/A14,CS_n,CKE,ODT,C2-C0)
Address, Command and Control Overshoot and Undershoot Definition

π PM Tech
Clock Overshoot and Undershoot Specifications
PMA213816B
AC overshoot/undershoot specification for Clock
Specification
1600 1866 2133 2400 2666 2933 3200
Maximum peak amplitude above VDD
0.06 0.06 0.06 0.06 0.06 0.06 0.06 V
Absolute Max allowed for overshoot area
Delta value between VDD Absolute Max
0.24 0.24 0.24 0.24 0.24 0.24 0.24 V
and VDD Max allowed for over-shoot area
Maximum peak amplitude allowed for
0.3 0.3 0.3 0.3 0.3 0.3 0.3 V-ns
undershoot area
Maximum overshoot area per 1UI Above
0.0038 0.0032 0.0028 0.0025 0.0025 0.0025 0.0025 V-ns
Absolute Max
Maximum overshoot area per 1UI Between
0.1125 0.0964 0.0844 0.0750 0.0750 0.0750 0.0750 V-ns
Absolute Max and VDD Max
Maximum undershoot area per 1UI Below
0.1144 0.0980 0.0858 0.0762 0.0762 0.0762 0.0762 V-ns
VSS
(CK_t, CK_c)
Clock Overshoot and Undershoot Definition

π PM Tech
Data, Strobe and Mask Overshoot and Undershoot Specifications
PMA213816B
AC overshoot/undershoot specification for Data, Strobe and Mask
Specification
1600 1866 2133 2400 2666 2933 3200
Maximum peak amplitude above Max absolute
0.06 0.06 0.06 0.06 0.06 0.06 0.06 V
level of Vin, Vout
Overshoot area Between Max Absolute level of
0.24 0.24 0.24 0.24 0.24 0.24 0.24 V
Vin, Vout and VDDQ Max
Undershoot area Between Min absolute level of
0.3 0.3 0.3 0.3 0.3 0.3 0.3 V
Vin, Vout and VSSQ
Maximum peak amplitude below Min absolute
0.10 0.10 0.10 0.10 0.10 0.10 0.10 V
level of Vin, Vout
Maximum overshoot area per 1UI Above Max
0.0150 0.0129 0.0113 0.0100 0.0100 0.0100 0.0100 V-ns
absolute level of Vin, Vout
Maximum overshoot area per 1UI Between
0.1050 0.0900 0.0788 0.0700 0.0700 0.0700 0.0700 V-ns
Max absolute level of Vin,Vout and VDDQ Max
Maximum undershoot area per 1UI Between
0.1050 0.0900 0.0788 0.0700 0.0700 0.0700 0.0700 V-ns
Min absolute level of Vin,Vout and VSSQ
Maximum undershoot area per 1UI Below Min
0.0150 0.0129 0.0113 0.0100 0.0100 0.0100 0.0100 V-ns
absolute level of Vin,Vout
DQS_t, DQS_n, LDQS_t, LDQS_n, UDQS_t, UDQS_n, DQ[0:15], DM/DBI, UDM/UDBI, LDM/LDBI
Data, Strobe and Mask Overshoot and Undershoot Definition

π PM Tech
Slew Rate Definitions
PMA213816B
Slew Rate Definitions for Differential Input Signals (CK)

Input slew rate for differential signals (CK_t, CK_c) are defined and measured as shown in Table & Figure
below.
Differential input slew rate definition
Measured
Description Defined by
from to
Differential input slew rate for rising edge(CK_t - CK_c) VILdiffmax VIHdiffmin [VIHdiffmin – VILdiffmax] / DeltaTRdiff
Differential input slew rate for falling edge(CK_t - CK_c) VIHdiffmin VILdiffmax [VIHdiffmin - VILdiffmax] / DeltaTFdiff
NOTE: The differential signal (i,e.,CK_t - CK_c) must be linear between these thresholds.
Differential Input Slew Rate Definition for CK_t, CK_c
1.

π PM Tech
Slew Rate Definition for Single-ended Input Signals (CMD/ADD)
PMA213816B
Single-ended Input Slew Rate definition for CMD and ADD
NOTE
1. Single-ended input slew rate for rising edge = { VIHCA(AC)Min - VILCA(DC)Max } / Delta TR single.
2. Single-ended input slew rate for falling edge = { VIHCA(DC)Min - VILCA(AC)Max } / Delta TF single.
3. Single-ended signal rising edge from VILCA(DC)Max to VIHCA(DC)Min must be monotonic slope.
4. Single-ended signal falling edge from VIHCA(DC)Min to VILCA(DC)Max must be monotonic slope.

π PM Tech
Differential Input Cross Point Voltage
PMA213816B
To guarantee tight setup and hold times as well as output skew parameters with respect to clock, each cross
point voltage of differential input signals (CK_t, CK_c) must meet the requirements in Table. The differential
input cross point voltage VIX is measured from the actual cross point of true and complement signals to the
midlevel between of VDD and VSS.
Vix Definition (CK)
Cross point voltage for differential input signals (CK)
DDR4-1600/1866/2133/2400
Symbol Parameter Area of VSEH, VSEL
min max
VSEL =< VDD/2 - 145mV -120 N/A
VDD/2 - 145mV =< VSEL =< –(VDD/2-VSEL)

Differential Input Cross Point N/A
VDD/2 - 100mV +25mV
VlX(CK) Voltage relative to VDD/2 for
CK_t, CK_c VDD/2 + 100mV =< VSEH =< (VSEH - VDD/2) -
N/A
VDD/ 2 + 145mV -25mV
VDD/2 + 145mV =< VSEH N/A 120
DDR4-2666/2933/3200
Symbol Parameter Area of VSEH, VSEL
min max
VSEL =< VDD/2 - 145mV -110 N/A
VDD/2 - 145mV =< VSEL =< –(VDD/2-VSEL)

Differential Input Cross Point N/A
VDD/2 - 90mV +30mV
VlX(CK) Voltage relative to VDD/2 for
CK_t, CK_c VDD/2 + 100mV =< VSEH =< (VSEH - VDD/2)
N/A
VDD/ 2 + 145mV -30mV
VDD/2 + 145mV =< VSEH N/A 110

π PM Tech
CMOS rail to rail Input Levels
PMA213816B
CMOS rail to rail Input Levels for RESET_n
Parameter Symbol Min Max Unit NOTE

AC Input High Voltage VIH(AC)_RESET 0.8*VDD VDD V 6
DC Input High Voltage VIH(DC)_RESET 0.7*VDD VDD V 2
DC Input Low Voltage VIL(DC)_RESET VSS 0.3*VDD V 1
AC Input Low Voltage VIL(AC)_RESET VSS 0.2*VDD V 7
Rising time TR_RESET - 1.0 us 4
RESET pulse width tPW_RESET 1.0 - us 3,5
NOTE :
1. After RESET_n is registered LOW, RESET_n level shall be maintained below VIL(DC)_RESET during tPW_RESET,
otherwise, SDRAM may not be reset.
2. Once RESET_n is registered HIGH, RESET_n level must be maintained above VIH(DC)_RESET, otherwise, SDRAM
operation will not be guaranteed until it is reset asserting RESET_n signal LOW.
3. RESET is destructive to data contents.
4. No slope reversal(ringback) requirement during its level transition from Low to High.
5. This definition is applied only “Reset Procedure at Power Stable”.
6. Overshoot might occur. It should be limited by the Absolute Maximum DC Ratings.
7. Undershoot might occur. It should be limited by Absolute Maximum DC Ratings.
RESET_n Input Slew Rate Definition

π PM Tech
AC and DC Logic Input Levels for DQS Signals
PMA213816B
Differential signal definition
Definition of differential DQS Signal AC-swing Level
Differential swing requirements for DQS (DQS_t - DQS_c)
Differential AC and DC Input Levels for DQS
DDR4-1600/1866/2133 DDR4-2400
Symbol Parameter Unit Note
min max min max
VIHDiffPeak VIH.DIFF.Peak Voltage 186 VDDQ 160 VDDQ mV 1,2
VILDiffPeak VIL.DIFF.Peak Voltage VSSQ -186 VSSQ -160 mV 1,2
DDR4-2666 DDR4-2933 DDR4-3200

VIHDiffPeak VIH.DIFF.Peak Voltage 150 VDDQ 150 VDDQ 140 VDDQ mV 1,2
VILDiffPeak VIL.DIFF.Peak Voltage VSSQ -150 VSSQ -150 VSSQ -140 mV 1,2
NOTE
1. Minimum and maximum limits are relative to single-ended portion and can be exceeded within allowed overshoot and
undershoot limits.
2. Minimum value point is used to determine differential signal slew-rate.

π PM Tech
Peak voltage calculation method
PMA213816B
The peak voltage of Differential DQS signals are calculated in a following equation.
VIH.DIFF.Peak Voltage = Max(f(t)) VIL.
DIFF.Peak Voltage = Min(f(t))
f(t) = VDQS_t - VDQS_c
Definition of differential DQS Peak Voltage

π PM Tech
Differential Input Cross Point Voltage
PMA213816B
To guarantee tight setup and hold times as well as output skew parameters with respect to strobe, the cross
point voltage of differential input signals (DQS_t, DQS_c) must meet the requirements in Ta ble 18. The
differential input cross point voltage VIX is measured from the actual cross point of true and complement
signals to the mid level that is VrefDQ.Vix Definition (DQS)
Vix Definition (DQS)
Cross point voltage for differential input signals (DQS)
DDR4-1600/1866/2133/2400/2666/2933/3200
min max
DQS Differential input crosspoint

VIX_DQS,ratio - 25 % 1,2,3
voltage ratio
NOTE :
1. The base level of Vix_DQS_FR/RF is VrefDQ that is DDR4 SDRAM internal setting value by Vref Training.
2. Vix_DQS_FR is defined by this equation : Vix_DQS_FR = |Min(f(t)) x Vix_DQS_Ratio.
3. Vix_DQS_RF is defined by this equation : Vix_DQS_RF = Max(f(t)) x Vix_DQS_Ratio.

π PM Tech
Differential Input Slew Rate Definition
PMA213816B
Input slew rate for diff erential signals (DQS_t, DQS_c) are defined and measured as shown in Figures below.
Differential Input Slew Rate Definition for DQS_t, DQS_c
NOTE
1. Differential signal rising edge from VILDiff_DQS to VIHDiff_DQS must be monotonic slope. 2. Differential signal falling edge
from VIHDiff_DQS to VILDiff_DQS must be monotonic slope.
Differential Input Slew Rate Definition for DQS_t, DQS_c
Measured
from to
Differential input slew rate for rising edge
VILDiff_DQS VIHDiff_DQS |VILDiff_DQS - VIHDiff_DQS|/DeltaTRdiff
(DQS_t - DQS_c)
Differential input slew rate for falling edge
VIHDiff_DQS VILDiff_DQS |VILDiff_DQS - VIHDiff_DQS|/DeltaTFdiff
(DQS_t - DQS_c)
Differential Input Level for DQS_t, DQS_c
DDR4-1600/1866/2133 DDR4-2400
min max min max
VIHDiff_DQS Differntial Input High 136 - 130 - mV 1,2
VILDiff_DQS Differntial Input Low - -136 - -130 mV 1,2

π PM Tech
Differential Input Level for DQS_t, DQS_c
PMA213816B
DDR4-2666 DDR4-2933 DDR4-3200

VIHDiff_DQS Differntial Input High 130 - 115 - 110 - mV 1,2
VILDiff_DQS Differntial Input Low - -130 - -115 - -110 mV 1,2
NOTE
1. Differential signal rising edge from VIL,diff,DQS to VIH,diff,DQS must be monotonic slope.
2. Differential signal falling edge from VIH,diff,DQS to VIL,diff,DQS must be monotonic slope.
Differential Input Slew Rate for DQS_t, DQS_c
DDR4-1600/1866/2133/2400 DDR4-2666/2933/3200
min max min max
SRIdiff Differential Intput Slew Rate 3 18 2.5 18 V/ns

π PM Tech
AC and DC output Measurement levels
PMA213816B
Output Driver DC Electrical Characteristics

The DDR4 driver supports two different Ron values. These Ron values are referred as strong(low Ron) and
weak mode(high Ron). A functional representation of the output buffer is shown in the figure below. Output
driver impedance RON is defined as follows:
The individual pull-up and pull-down resistors (RONPu and RONPd) are defined as follows:
under the condition that RONPd is off
under the condition that RONPu is off
Output driver

π PM Tech
Output Driver DC Electrical Characteristics, assuming RZQ=240ohm; entire operating temperature
PMA213816B
range; after proper ZQ calibration
RONNOM Resistor Vout Min Nom Max Unit NOTE

VOLdc= 0.5*VDDQ 0.73 1 1.1 RZQ/7 1,2
RON34Pd VOMdc= 0.8* VDDQ 0.83 1 1.1 RZQ/7 1,2
VOHdc= 1.1* VDDQ 0.83 1 1.25 RZQ/7 1,2
34Ω
VOLdc= 0.5* VDDQ 0.9 1 1.25 RZQ/7 1,2
RON34Pu VOMdc= 0.8* VDDQ 0.9 1 1.1 RZQ/7 1,2
VOHdc= 1.1* VDDQ 0.8 1 1.1 RZQ/7 1,2
VOLdc= 0.5*VDDQ 0.73 1 1.1 RZQ/5 1,2
RON48Pd VOMdc= 0.8* VDDQ 0.83 1 1.1 RZQ/5 1,2
VOHdc= 1.1* VDDQ 0.83 1 1.25 RZQ/5 1,2
48Ω
VOLdc= 0.5* VDDQ 0.9 1 1.25 RZQ/5 1,2
RON48Pu VOMdc= 0.8* VDDQ 0.9 1 1.1 RZQ/5 1,2
VOHdc= 1.1* VDDQ 0.8 1 1.1 RZQ/5 1,2
Mismatch between pull-up
and pull-down, MMPuPd VOMdc= 0.8* VDDQ -10 - 17 % 1,2,3,4
Mismatch DQ-DQ within byte

variation pull-up, MMPudd VOMdc= 0.8* VDDQ - - 10 % 1,2,4
Mismatch DQ-DQ within byte

variation pull-dn, MMPddd VOMdc= 0.8* VDDQ - - 10 % 1,2,4
NOTE
1. The tolerance limits are specified after calibration with stable voltage and temperature. For the behavior of the tolerance
limits if temperature or voltage changes after cal-ibration, see following section on voltage and temperature sensitivity.
2. Pull-up and pull-dn output driver impedances are recommended to be calibrated at 0.8 * VDDQ. Other calibration
schemes may be used to achieve the linearity spec shown above, e.g. calibration at 0.5 * VDDQ and 1.1 * VDDQ.
3. 3. Measurement definition for mismatch between pull-up and pull-down, MMPuPd : Measure RONPu and RONPD both at
0.8*VDD separately; Ronnom is the nominal Ron value.
4. RON variance range ratio to RON Nominal value in a given component, including DQS_t and DQS_c.
5. This parameter of x16 device is specified for Uper byte and Lower byte.

π PM Tech
Output Driver Temperature and Voltage Sensitivity
PMA213816B
If temperature and/or voltage change after calibration, the tolerance limits widen according to the equations
and tables below.
Δ T = T - T(@calibration); Δ V = VDDQ - VDDQ(@ calibration); VDD = VDDQ
Output Driver Sensitivity Definitions

Symbol Min Max Unit
RONPU@ VOH(DC) 0.6 - dRONdTH × |ΔT| - dRONdVH × |ΔV| 1.1 _ dRONdTH × |ΔT| + dRONdVH × |ΔV| RZQ/6
RON@ VOM(DC) 0.9 - dRONdTM × |ΔT| - dRONdVM × |ΔV| 1.1 + dRONdTM × |ΔT| + dRONdVM × |ΔV| RZQ/6
RONPD@ VOL(DC) 0.6 - dRONdTL × |ΔT| - dRONdVL × |ΔV| 1.1 + dRONdTL × |ΔT| + dRONdVL × |ΔV| RZQ/6
Output Driver Voltage and Temperature Sensitivity
Voltage and Temperature Range

Symbol Unit
Min Max
dRONdTM 0 1.5 %/°C
dRONdVM 0 0.15 %/mV
dRONdTL 0 1.5 %/°C
dRONdVL 0 0.15 %/mV
dRONdTH 0 1.5 %/°C
dRONdVM 0 0.15 %/mV

π PM Tech
Alert_n output Drive Characteristic
PMA213816B
A functional representation of the output buffer is shown in the figure below. Output driver impedance RON is
defined as follows:
Resistor Vout Min Max Unit NOTE

VOLdc= 0.1* VDDQ 0.3 1.2 34Ω 1
RONPd VOMdc = 0.8* VDDQ 0.4 1.2 34Ω 1
VOHdc = 1.1* VDDQ 0.4 1.4 34Ω 1
Note: 1. VDDQ voltage is at VDDQ(DC).

π PM Tech
Output Driver Characteristic of Connectivity Test ( CT ) Mode
PMA213816B
Following Output driver impedance RON will be applied Test Output Pin during Connectivity Test ( CT )
Mode. The individual pull-up and pull-down resistors (RONPu_CT and RONPd_CT) are defined as follows:
Output Driver
RONNOM_CT Resistor Vout Max Unit NOTE

VOBdc = 0.2 x VDDQ 1.9 34Ω 1
VOLdc = 0.5 x VDDQ 2.0 34Ω 1
RONPd_CT
VOMdc = 0.8 x VDDQ 2.2 34Ω 1
VOHdc = 1.1 x VDDQ 2.5 34Ω 1
34Ω
VOBdc = 0.2 x VDDQ 2.5 34Ω 1
VOLdc = 0.5 x VDDQ 2.2 34Ω 1
RONPu_CT
VOMdc = 0.8 x VDDQ 2.0 34Ω 1
VOHdc = 1.1 x VDDQ 1.9 34Ω 1
NOTE : 1. Connectivity test mode uses un-calibrated drivers, showing the full range over PVT. No mismatch between pull up
and pull down is defined.

π PM Tech
Single-ended AC & DC Output Levels
PMA213816B
DDR4-
1600/1866/2133/2400/2666/2933/3200
DC output high measurement level
VOH(DC) 1.1 x VDDQ V
(for IV curve linearity)
DC output mid measurement level
VOM(DC) 0.8 x VDDQ V
DC output low measurement level
VOL(DC) 0.5 x VDDQ V
AC output high measurement level
VOH(AC) (0.7 + 0.15) x VDDQ V 1
(for output SR)
AC output low measurement level
VOL(AC) (0.7 - 0.15) x VDDQ V 1
(for output SR)
NOTE
1. The swing of ± 0.15 × VDDQ is based on approximately 50% of the static single-ended output peak-to-peak swing with a
driver impedance of RZQ/7Ω and an effective test load of 50Ω to VTT = VDDQ.
Differential AC & DC Output Levels

DDR4-
1600/1866/2133/2400/2666/2933/3200
AC differential output high measurement level
VOHdiff(AC) +0.3 x VDDQ V 1
(for output SR)
AC differential output low measurement level
VOLdiff(AC) -0.3 x VDDQ V 1
(for output SR)
NOTE
1. The swing of ± 0.3 × VDDQ is based on approximately 50% of the static differential output peak-to-peak swing with a driver
impedance of RZQ/7Ω and an effective test load of 50Ω to VTT = VDDQ at each of the differential outputs.
Single-ended Output Slew Rate

With the reference load for timing measurements, output slew rate for falling and rising edges is defined and
measured between VOL(AC) and VOH(AC) for single ended signals as shown below.
Single-ended output slew rate definition
Measured
From To
Single ended output slew rate for rising edge VOL(AC) VOH(AC) [VOH(AC)-VOL(AC)] / Delta TRse
Single ended output slew rate for falling edge VOH(AC) VOL(AC) [VOH(AC)-VOL(AC)] / Delta TFse
NOTE
1. Output slew rate is verified by design and characterization, and may not be subject to production test.

π PM Tech
Single-ended Output Slew Rate Definition
PMA213816B
Single-ended output slew rate

DDR4-
DDR4-2666 DDR4-2933/3200
Parameter Symbol 1600/1866/2133/2400 Units
Single ended output slew rate SRQse 4 9 4 9 4 9 V/ns
Description: SR: Slew Rate

Q: Query Output (like in DQ, which stands for Data-in, Query-Output)
se: Single-ended Signals
For Ron = RZQ/7 setting
NOTE :
1. In two cases, a maximum slew rate of 12 V/ns applies for a single DQ signal within a byte lane.
– Case 1 is defined for a single DQ signal within a byte lane which is switching into a certain direction (either from high to low
or low to high) while all remaining DQ signals in the same byte lane are static (i.e. they stay at either high or low).
– Case 2 is defined for a single DQ signal within a byte lane which is switching into a certain direction (either from high to low
or low to high) while all remaining DQ signals in the same byte lane are switching into the opposite direction (i.e. from low
to high or high to low respectively). For the remaining DQ signal switching into the opposite direction, the regular maximum
limit of 9 V/ns applies.

π PM Tech
Differential Output Slew Rate
PMA213816B
With the reference load for timing measurements, output slew rate for falling and rising edges is defined and
measured between VOLdiff(AC) and VOHdiff(AC) for differential signals as shown in Table & Figure below.
Differential output slew rate definition
Measured
From To
Differential output slew rate for rising edge VOLdiff(AC) VOHdiff(AC) [VOHdiff(AC)-VOLdiff(AC)] / Delta TRdiff
Differential output slew rate for falling edge VOHdiff(AC) VOLdiff(AC) [VOHdiff(AC)-VOLdiff(AC)] / Delta TFdiff
NOTE : 1. Output slew rate is verified by design and characterization, and may not be subject to production test.
Differential output slew rate

DDR4-
DDR4-2666 DDR4-2933/3200
Parameter Symbol 1600/1866/2133/2400 Units
Differential output slew rate SRQdiff 8 18 8 18 8 18 V/ns
Description:
SR: Slew Rate
Q: Query Output (like in DQ, which stands for Data-in, Query-Output)
diff: Differential Signals
For Ron = RZQ/7 setting

π PM Tech
Single-ended AC & DC Output Levels of Connectivity Test Mode
PMA213816B
Following output parameters will be applied for DDR4 SDRAM Output Signal during Connectivity Test Mode.
Symbol Parameter DDR4-1600 to DDR4-3200 Unit Note
VOH(DC) DC output high measurement level (for IV curve linearity) 1.1 x VDDQ V
VOM(DC) DC output mid measurement level (for IV curve linearity) 0.8 x VDDQ V
VOL(DC) DC output low measurement level (for IV curve linearity) 0.5 x VDDQ V
VOB(DC) DC output below measurement level (for IV curve linearity) 0.2 x VDDQ V
VOH(AC) AC output high measurement level (for output SR) VTT + (0.1 x VDDQ) V 1
VOL(AC) AC output below measurement level (for output SR) VTT - (0.1 x VDDQ) V 1
NOTE 1. The effective test load is 50Ω terminated by VTT = 0.5 * VDDQ.
Single-ended output slew rate of Connectivity Test Mode

DDR4-
DDR4-2666 DDR4-2933/3200
Parameter Symbol 1600/1866/2133/2400 Unit
Output signal Falling time TF_output_CT - 10 - 10 - 10 ns/V
Output signal Rising time TR_output_CT - 10 - 10 - 10 ns/V

π PM Tech
Test Load for Connectivity Test Mode Timing
PMA213816B
The reference load for ODT timings is defined in Figure below.
Connectivity Test Mode Timing Reference Load

π PM Tech
Speed Bin
PMA213816B
DDR4-1600 Speed Bins and Operations
Speed Bin DDR4-1600
CL-nRCD-nRP 11-11-11 Unit NOTE
Parameter Symbol min max

12
13.75
Internal read command to first data tAA 18.00 ns 12
(13.50) 5,10
Internal read command to first data with read
tAA_DBI tAA(min)+2nCK tAA(max)+2nCK ns 12
DBI enabled
13.75 12
ACT to internal read or write delay time tRCD - ns 12
(13.50) 5,10
12
13.75
PRE command period tRP - ns 12
(13.50) 5,10
ACT to PRE command period tRAS 35 9 x tREFI ns 12
48.75
ACT to ACT or REF command period tRC - ns 12
(48.50) 5,10
Normal Read DBI
1.5
1,2,3,4,11,
CL = 9 CL = 11 tCK(AVG) 1.6 ns
15
CWL = 9 (Optional) 5,10
CL = 10 CL = 12 tCK(AVG) Reserved ns 1,2,3,4,11
CL = 10 CL = 12 tCK(AVG) Reserved ns 1,2,3,4
CWL = 9,11 CL = 11 CL = 13 tCK(AVG) 1.25 <1.5 ns 1,2,3,4
CL = 12 CL = 14 tCK(AVG) 1.25 <1.5 ns 1,2,3
Supported CL Settings 9,11,12 nCK 12,14
Supported CL Settings with read DBI 11,13,14 nCK 14
Supported CWL Settings 9,11 nCK

π PM Tech
PMA213816B
Speed Bin DDR4-1866

13.92 12
(13.50) 5,10
DBI enabled
13.92 12
(13.50) 5,10
12
13.92
(13.50) 5,10
47.92
(47.50) 5,10
Normal Read DBI
1.5 1,2,3,4,11,
CWL = 9 (Optional) 5,10 14

CL = 10 CL = 12 tCK(AVG) Reserved ns 4
1.25 <1.5
CWL = 9,11 CL = 11 CL = 13 tCK(AVG) ns 1,2,3,4,6
5,10
(Optional)
CL = 12 CL = 14 tCK(AVG) 1.25 <1.5 ns 1,2,3,6
CL = 12 CL = 14 tCK(AVG) Reserved 1,2,3,4
CWL = 10,12 CL = 13 CL = 15 tCK(AVG) 1.071 <1.25 1,2,3,4
CL = 14 CL = 16 tCK(AVG) 1.071 <1.25 1,2,3
Supported CL Settings 9,11,12,13,14 nCK 12,14
Supported CL Settings with read DBI 11,13,14,15,16 nCK 14
Supported CWL Settings 9,10,11,12 nCK

π PM Tech
PMA213816B
Speed Bin DDR4-2133

14.06 12
(13.75) 5,10
DBI enabled
14.06 12
(13.75) 5,10
14.06 12
(13.75) 5,10
47.06
(46.75) 5,10
Normal Read DBI
1.5 1,2,3,4,11,
CWL = 9 (Optional) 5,10 13

1.25 <1.5
CL = 11 CL = 13 tCK(AVG) ns 1,2,3,4,7
5,10
CWL = 9,11 (Optional)
CL = 12 CL = 14 tCK(AVG) 1.25 <1.5 ns 1,2,3,7
1.071 <1.25
CL = 13 CL = 15 tCK(AVG) ns 1,2,3,4,7
5,10
CWL = 10,12 (Optional)
CL = 14 CL = 16 tCK(AVG) 1.071 <1.25 ns 1,2,3,7
CL = 16 CL = 19 tCK(AVG) 0.937 <1.071 ns 1,2,3
Supported CL Settings 9,11,12,13,14,15,16 nCK 13,14
Supported CL Settings with read DBI 11,13,14,15,16,18,19 nCK
Supported CWL Settings 9,10,11,12,14 nCK

π PM Tech
PMA213816B
Speed Bin DDR4-2400

14.16
(13.75) 5,10
DBI enabled
14.16
(13.75) 5,10
14.16
(13.75) 5,10
46.16
(45.75) 5,10
Normal Read DBI
CWL = 9
CL = 10 CL = 12 tCK(AVG) 1.5 1.6 ns 1,2,3,4,11
1.25 <1.5
5,10
(Optional)
CL = 12 CL = 14 tCK(AVG) 1.25 <1.5 ns 1,2,3,8
1.071 <1.25 1,2,3,8
CWL = 10,12 CL = 13 CL = 15 tCK(AVG) ns
5,10
(Optional)
CL = 14 CL = 16 tCK(AVG) 1.071 <1.25 ns 1,2,3,8
0.937 <1.071 1,2,3,4,8

CWL = 11,14 CL = 15 CL = 18 tCK(AVG) ns
(Optional) 5,10
CL = 16 CL = 19 tCK(AVG) 0.937 <1.071 ns 1,2,3,8
CWL = 12,16
CL = 17 CL = 20 tCK(AVG) 0.833 <0.937 ns
CL = 18 CL = 21 tCK(AVG) 0.833 <0.937 ns 1,2,3
Supported CL Settings 10,11,12,13,14,15,16,17,18 nCK 13
Supported CL Settings with read DBI 12,13,14,15,16,18,19,20,21 nCK
Supported CWL Settings 9,10,11,12,14,16 nCK

π PM Tech
PMA213816B
Speed Bin DDR4-2666

14.25
(13.75) 5,11
DBI enabled
14.25
(13.75) 5,11
14.25
(13.75) 5,11
46.25
(45.75) 5,11
Normal Read DBI
CWL = 9
CL = 10 CL = 12 tCK(AVG) 1.5 1.6 ns 1,2,3,12
1.25 <1.5
5,11
(Optional)
CL = 12 CL = 14 tCK(AVG) 1.25 <1.5 ns 1,2,3,9
1.071 <1.25
(Optional) 5,11
CL = 14 CL = 16 tCK(AVG) 1.071 <1.25 ns 1,2,3,9
0.937 <1.071
5,11
(Optional)
CL = 16 CL = 19 tCK(AVG) 0.937 <1.071 ns 1,2,3,9
CWL = 12,16 0.833 <0.937

CL = 17 CL = 20 tCK(AVG) ns 1,2,3,4,9
(Optional) 5,11
CL = 18 CL = 21 tCK(AVG) 0.833 <0.937 ns 1,2,3

CWL = 14,18
CL = 19 CL = 22 tCK(AVG) 0.75 <0.833 ns 1,2,3,4
CL = 20 CL = 23 tCK(AVG) 0.75 <0.833 ns 1,2,3
Supported CL Settings 10,11,12,13,14,15,16,17,18,19,20 nCK 13
Supported CL Settings with read DBI 12,13,14,15,16,18,19,20,21,22,23 nCK
Supported CWL Settings 9,10,11,12,14,16,18 nCK

π PM Tech
PMA213816B
Speed Bin DDR4-2933

Internal read command to first data tAA 13.64 18.00 ns 12

DBI enabled
ACT to internal read or write delay time tRCD 13.64 - ns 12
PRE command period tRP 13.64 - ns 12
ACT to ACT or REF command period tRC 45.64 ns 12

-
Normal Read DBI
CWL = 9
CL = 10 CL = 12 tCK(AVG) 1.5 1.6 ns 1,2,3,11
CWL = 9,11 CL = 11 CL = 13 tCK(AVG) 1.25 <1.5 ns 1,2,3,4,13
CL = 12 CL = 14 tCK(AVG) 1.25 <1.5 ns 1,2,3,9
CL = 14 CL = 16 tCK(AVG) 1.071 <1.25 ns 1,2,3,15
CL = 16 CL = 19 tCK(AVG) 0.937 <1.071 ns 1,2,3,15

CWL = 12,16
CL = 17 CL = 20 tCK(AVG) 0.833 <0.937 ns 1,2,3,4,15
CL = 18 CL = 21 tCK(AVG) 0.833 <0.937 ns 1,2,3,15

CWL = 14,18
CL = 19 CL = 22 tCK(AVG) 0.75 <0.833 ns 1,2,3,4,15
CL = 20 CL = 23 tCK(AVG) 0.75 <0.833 ns 1,2,3,15
CL = 20 CL = 24 tCK(AVG) 0.682 <0.75 ns 1,2,3,4
CWL = 16,20
CL = 21 CL = 26 tCK(AVG) 0.682 <0.75 ns 1,2,3,4
CL = 22 CL = 26 tCK(AVG) 0.682 <0.75 ns 1,2,3
Supported CL Settings 10,11,12,13,14,15,16,17,18,19,20,21,22 nCK 13
Supported CL Settings with read DBI 12,13,14,15,16,18,19,20,21,22,23,24,25,2 nCK
6
Supported CWL Settings 9,10,11,12,14,16,18,20 nCK

π PM Tech
PMA213816B
Speed Bin DDR4-3200

Internal read command to first data tAA 13.75 18.00 ns 12

DBI enabled
ACT to internal read or write delay time tRCD 13.75 - ns 12
PRE command period tRP 13.75 - ns 12
ACT to ACT or REF command period tRC 45.75 ns 12

-
Normal Read DBI
CWL = 9
CL = 10 CL = 12 tCK(AVG) 1.5 1.6 ns 1,2,3,11
CL = 12 CL = 14 tCK(AVG) 1.25 <1.5 ns 1,2,3,10
CL = 14 CL = 16 tCK(AVG) 1.071 <1.25 ns 1,2,3,10
CL = 16 CL = 19 tCK(AVG) 0.937 <1.071 ns 1,2,3,4,10

CWL = 12,16
CL = 17 CL = 20 tCK(AVG) 0.833 <0.937 ns 1,2,3,4,10
CL = 18 CL = 21 tCK(AVG) 0.833 <0.937 ns 1,2,3,10

CWL = 14,18
CL = 19 CL = 22 tCK(AVG) 0.75 <0.833 ns 1,2,3,4,10
CL = 20 CL = 23 tCK(AVG) 0.75 <0.833 ns 1,2,3,10
CL = 24 CL = 28 tCK(AVG) 0.682 <0.75 ns 1,2,3
Supported CL Settings 10,11,12,13,14,15,16,17,18,19,20,22,24 nCK 13
Supported CL Settings with read DBI 12,13,14,15,16,18,19,20,21,22,23,24,26,2 nCK
8
Supported CWL Settings 9,10,11,12,14,16,18,20 nCK

π PM Tech
Speed Bin Table Note
PMA213816B
Absolute Specification
– VDDQ = VDD = 1.20V +/- 0.06 V
– VPP = 2.5V +0.25/-0.125 V
– The values defined with above-mentioned table are DLL ON case.
– DDR4-1600, 1866, 2133, 2400, 2666, 2933 and 3200 Speed Bin Tables are valid only when Geardown
Mode is disabled.
1. The CL setting and CWL setting result in tCK(avg).MIN and tCK(avg).MAX requirements. When making a
selection of tCK(avg), both need to be fulfilled: Requirements from CL setting as well as requirements
from CWL setting.
2. tCK(avg).MIN limits: Since CAS Latency is not purely analog - data and strobe output are synchronized
by the DLL - all possible intermediate frequencies may not be guaranteed. An application should use the
next smaller JEDEC standard tCK(avg) value (1.5, 1.25, 1.071, 0.938 or 0.833 ns) when calculating CL
[nCK] = tAA [ns] / tCK(avg) [ns], rounding up to the next ‘Supported CL’, where tAA = 12.5ns and tCK(avg)
= 1.3 ns should only be used for CL = 10 calculation.
3. tCK(avg).MAX limits: Calculate tCK(avg) = tAA.MAX / CL SELECTED and round the resulting tCK(avg)
down to the next valid speed bin (i.e. 1.5ns or 1.25ns or 1.071 ns or 0.938 ns or 0.833 ns). This result is
tCK(avg).MAX corresponding to CL SELECTED.
4. ‘Reserved’ settings are not allowed. User must program a different value.
5. 'Optional' settings allow certain devices in the industry to support this setting, however, it is not a
mandatory feature. Refer to supplier's data sheet and/or the DIMM SPD information if and how this
setting is supported.
6. Any DDR4-1866 speed bin also supports functional operation at lower frequencies as shown in the table
which are not subject to Production Tests but verified by Design/ Characterization.
10. Any DDR4-2933/3200 speed bin also supports functional operation at lower frequencies as shown in the
table which are not subject to Production Tests but verified by Design/ Characterization.
11. 10 DDR4-1600 AC timing apply if DRAM operates at lower than 1600 MT/s data rate.
12. Parameters apply from tCK(avg)min to tCK(avg)max at all standard JEDEC clock period values as stated
in the Speed Bin Tables.
13. CL number in parentheses, it means that these numbers are optional.
14. DDR4 SDRAM supports CL=9 as long as a system meets tAA(min).
15. Each speed bin lists the timing requirements that need to be supported in order for a given DRAM to be
JEDEC compliant. JEDEC compliance does not require support for all speed bins within a given speed.
JEDEC compliance requires meeting the parameters for a least one of the listed speed bins.

π PM Tech
IDD and IDDQ Specification Parameters and Test conditions
PMA213816B
IDD, IPP and IDDQ Measurement Conditions

In this chapter, IDD, IPP and IDDQ measurement conditions such as test load and patterns are defined. F
igure below shows the setup and test load for IDD, IPP and IDDQ measurements.
 IDD currents (such as IDD0, IDD0A, IDD1, IDD1A, IDD2N, IDD2NA, IDD2NL, IDD2NT, IDD2P, IDD2Q,
IDD3N, IDD3NA, IDD3P, IDD4R, IDD4RA, IDD4W, IDD4WA, IDD5B, IDD5F2, IDD5F4, IDD6N, IDD6E,
IDD6R, IDD6A, IDD7 and IDD8) are measured as time-averaged currents with all VDD balls of the DDR4
SDRAM under test tied together. Any IPP or IDDQ current is not included in IDD currents.
 IPP currents have the same definition as IDD except that the current on the VPP supply is measured.
 lDDQ currents (such as IDDQ2NT and IDDQ4R) are measured as time-averaged currents with all VDDQ
balls of the DDR4 SDRAM under test tied together. Any IDD current is not included in IDDQ
currents.Attention: IDDQ values cannot be directly used to calculate IO power of the DDR4 SDRAM. They
can be used to support correlation of simulated IO power to actual IO power as outlined in Figure 22. In
DRAM module application, IDDQ cannot be measured separately since VDD and VDDQ are using one
merged-power layer in Module PCB.
For IDD, IPP and IDDQ measurements, the following definitions apply:
 “0” and “LOW” is defined as VIN <= VILAC(max).
 “1” and “HIGH” is defined as VIN >= VIHAC(min).
 “MID-LEVEL” is defined as inputs are VREF = VDD / 2.
 Timings used for IDD, IPP and IDDQ Measurement-Loop Patterns are provided in Table 36.
 Basic IDD, IPP and IDDQ Measurement Conditions are described in Table 37.
 Detailed IDD, IPP and IDDQ Measurement-Loop Patterns are described in Table 38 through Table 46.
 IDD Measurements are done after properly initializing the DDR4 SDRAM. This includes but is not limited
to setting
– RON = RZQ/7 (34 Ohm in MR1);
– RTT_NOM = RZQ/6 (40 Ohm in MR1);
– RTT_WR = RZQ/2 (120 Ohm in MR2);
– RTT_PARK = Disable;
– Qoff = 0B (Output Buffer enabled) in MR1;
– TDQS_t disabled in MR1;
– CRC disabled in MR2;
– CA parity feature disabled in MR5;
– Gear down mode disabled in MR3;
– Read/Write DBI disabled in MR5;
– DM disabled in MR5
 Attention: The IDD, IPP and IDDQ Measurement-Loop Patterns need to be executed at least one time
before actual IDD or IDDQ measurement is started.
 Define D = {CS_n, ACT_n, RAS_n, CAS_n, WE_n } := {HIGH, LOW, LOW, LOW, LOW} ; apply BG/BA
changes when directed.
 Define D# = {CS_n, ACT_n, RAS_n, CAS_n, WE_n } := {HIGH, HIGH, HIGH, HIGH, HIGH} apply invert of
BG/BA changes when directed above.

π PM Tech
Measurement Setup and Test Load for IDD, IPP and IDDQ Measurements
PMA213816B
NOTE: 1. DIMM level Output test load condition may be different from above
Correlation from simulated Channel IO Power to actual Channel IO Power supported by IDDQ
Measurement

π PM Tech
Timings used for IDD, IPP and IDDQ Measurement-Loop Patterns
PMA213816B
DDR4-1600 DDR4-1866 DDR4-2133 DDR4-2400 DDR4-2666 DDR4-2933 DDR4-3200
10-10-10
11-11-11
12-12-12
12-12-12
13-13-13
14-14-14
14-14-14
15-15-15
16-16-16
15-15-15
16-16-16
18-18-18
17-17-17
18-18-18
19-19-19
20-20-20
21-21-21
22-22-22
23-23-23
24-24-24
25-25-25
Symbol Unit
tCK 1.25 1.071 0.938 0.833 0.75 0.685 0.625 ns

CL 10 11 12 12 13 14 14 15 16 15 16 18 17 18 19 20 21 22 20 22 24 CK
CWL 9 11 11 10 12 12 11 14 14 12 16 16 14 18 18 14 18 18 16 20 20 CK
nRCD 10 11 12 12 13 14 14 15 16 15 16 18 17 18 19 20 21 22 20 22 24 CK
nRC 38 39 40 44 45 46 50 51 52 54 55 57 60 61 62 66 67 68 72 74 76 CK
nRAS 28 32 36 39 43 47 52 CK
nRP 10 11 12 12 13 14 14 15 16 15 16 18 17 18 19 20 21 22 20 22 24 CK
x4 16 16 16 16 16 16 16 CK
nFAW x8 20 22 23 26 28 31 34 CK
x16 28 28 32 36 40 44 48 CK
x4 4 4 4 4 4 4 4 CK
nRRDS x8 4 4 4 4 4 4 4 CK
x16 5 5 6 7 7 8 9 CK
x4 5 5 6 6 7 8 8 CK
nRRDL x8 5 5 6 6 7 8 8 CK
x16 6 6 7 8 9 10 11 CK
tCCD_S 4 4 4 4 4 4 4 CK
tCCD_L 5 5 6 6 7 8 8 CK
tWTR_S 2 3 3 3 4 4 4 CK
tWTR_L 6 7 8 9 10 11 12 CK
nREFI 6,240 7,283 8,324 9,364 10,400 11,437 12,480 CK
nRFC 2Gb 128 150 171 193 214 235 256 CK
nRFC 4Gb 208 243 278 313 347 382 416 CK
nRFC 8Gb 280 327 374 421 467 514 560 CK
nRFC 16Gb 440 514 587 660 734 807 880 CK
NOTE: 1. 1KB based x4 use same numbers of clocks for nFAW as the x8.

π PM Tech
Basic IDD, IPP and IDDQ Measurement Conditions
PMA213816B
Symbol Description
Operating One Bank Active-Precharge Current (AL=0) CKE: High; External clock: On; tCK, nRC, nRAS,
CL: see ”Timings used for IDD, IPP and IDDQ Measurement-Loop Patterns”; BL: 8 1; AL: 0; CS_n: High
between ACT and PRE; Command, Address, Bank Group Address, Bank Address Inputs: partially
IDD0 toggling (see IDD0, IDD0A and IPP0 Measurement-Loop Pattern); Data IO: VDDQ; DM_n: stable at 1; Bank
Activity: Cycling with one bank active at a time: 0,0,1,1,2,2,... (see IDD0, IDD0A and IPP0 Measurement-
Loop Pattern); Output Buffer and RTT: Enabled in Mode Registers2; ODT Signal: stable at 0; Pattern
Details: see “IDD0, IDD0A and IPP0 Measurement-Loop Pattern”
IDD0A Operating One Bank Active-Precharge Current (AL=CL-1) AL = CL-1, Other conditions: see IDD0
IPP0 Operating One Bank Active-Precharge IPP Current Same condition with IDD0
Operating One Bank Active-Read-Precharge Current (AL=0) CKE: High; External clock: On; tCK, nRC,
nRAS, nRCD, CL : see “Timings used for IDD, IPP and IDDQ Measurement-Loop Patterns”; BL: 8 1; AL: 0;
CS_n: High between ACT, RD and PRE; Command, Address, Bank Group Address, Bank Address
IDD1 Inputs, Data IO: partially toggling (see IDD1, IDD1A and IPP1 Measurement-Loop Pattern); DM_n: stable at
1; Bank Activity: Cycling with one bank active at a time: 0,0,1,1,2,2,... (see IDD1, IDD1A and IPP1
Measurement-Loop Pattern); Output Buf-fer and RTT: Enabled in Mode Registers2; ODT Signal: stable at
0; Pattern Details: see “IDD1, IDD1A and IPP1 Measurement-Loop Pattern”
IDD1A Operating One Bank Active-Read-Precharge Current (AL=CL-1) AL = CL-1, Other conditions: see IDD1
IPP1 Operating One Bank Active-Read-Precharge IPP Current Same condition with IDD1
Precharge Standby Current (AL=0) CKE: High; External clock: On; tCK, CL: see “Timings used for IDD,
IPP and IDDQ Measurement-Loop Patterns”; BL: 8 1; AL: 0; CS_n: stable at 1; Command, Address, Bank
Group Address, Bank Address Inputs: partially toggling (see IDD2N, IDD2NA, IDD2NL, IDD2NG, IDD2ND,
IDD2N IDD2N_par, IPP2,IDD3N, IDD3NA and IDD3P Measurement-Loop Pattern); Data IO: VDDQ; DM_n: stable at
1; Bank Activity: all banks closed; Output Buffer and RTT: Enabled in Mode Registers2; ODT Signal:
stable at 0; Pattern Details: see IDD2N, IDD2NA, IDD2NL, IDD2NG, IDD2ND, IDD2N_par, IPP2,IDD3N,
IDD3NA and IDD3P Measurement-Loop Pattern
IDD2NA Precharge Standby Current (AL=CL-1) AL = CL-1, Other conditions: see IDD2N
IPP2N Precharge Standby IPP Current Same condition with IDD2N
Precharge Standby ODT Current CKE: High; External clock: On; tCK, CL: see “Timings used for IDD, IPP
and IDDQ Measurement-Loop Patterns”; BL: 8 1; AL: 0; CS_n: stable at 1; Command, Address, Bank
Group Address, Bank Address Inputs: partially toggling (see IDD2NT and IDDQ2NT Measurement-Loop
IDD2NT
Pattern); Data IO: VSSQ; DM_n: stable at 1; Bank Activity: all banks closed; Output Buffer and RTT:
Enabled in Mode Registers2; ODT Signal: toggling (see IDD2NT and IDDQ2NT Measurement-Loop Pattern);
Pattern Details: see (see IDD2NT and IDDQ2NT Measurement-Loop Pattern)
IDDQ2NT Precharge Standby ODT IDDQ Current Same definition like for IDD2NT, however measuring IDDQ current
(optional) instead of IDD current
IDD2NL Precharge Standby Current with CAL enabled Same definition like for IDD2N, CAL enabled 3,5
Precharge Standby Current with Gear Down mode enabled Same definition like for IDD2N, Gear Down
IDD2NG
mode enabled 3,5
IDD2ND Precharge Standby Current with DLL disabled Same definition like for IDD2N, DLL disabled 3
IDD2N_par Precharge Standby Current with CA parity enabled Same definition like for IDD2N, CA parity enabled 3
Precharge Power-Down Current CKE: Low; External clock: On; tCK, CL : see “Timings used for IDD, IPP
and IDDQ Measurement-Loop Patterns”; BL: 8 1; AL: 0; CS_n: stable at 1; Command, Address, Bank
IDD2P
Group Address, Bank Address Inputs: stable at 0; Data IO: VDDQ; DM_n: stable at 1; Bank Activity: all
banks closed; Output Buffer and RTT: Enabled in Mode Registers2; ODT Signal: stable at 0
IPP2P Precharge Power-Down IPP Current Same condition with IDD2P

π PM Tech
Symbol Description
PMA213816B
Precharge Quiet Standby Current CKE: High; External clock: On; tCK, CL: see “Timings used for IDD,
IPP and IDDQ Measurement-Loop Patterns”; BL: 8 1; AL: 0; CS_n: stable at 1; Command, Address, Bank
IDD2Q
Group Address, Bank Address Inputs: stable at 0; Data IO: VDDQ; DM_n: stable at 1;Bank Activity: all
banks closed; Output Buffer and RTT: Enabled in Mode Registers2; ODT Signal: stable at 0
Active Standby Current CKE: High; External clock: On; tCK, CL: see “Timings used for IDD, IPP and
IDDQ Measurement-Loop Patterns”; BL: 8 1; AL: 0; CS_n: stable at 1; Command, Address, Bank Group
Address, Bank Address Inputs: partially toggling see “IDD2N, IDD2NA, IDD2NL, IDD2NG, IDD2ND,
IDD3N IDD2N_par, IPP2,IDD3N, IDD3NA and IDD3P Measurement-Loop Pattern”; Data IO: VDDQ; DM_n: stable at
1;Bank Activity: all banks open; Output Buffer and RTT: Enabled in Mode Registers2; ODT Signal: stable
at 0; Pattern Details: see “IDD2N, IDD2NA, IDD2NL, IDD2NG, IDD2ND, IDD2N_par, IPP2,IDD3N, IDD3NA
and IDD3P Measurement-Loop Pattern”
IDD3NA Active Standby Current (AL=CL-1) AL = CL-1, Other conditions: see IDD3N
IPP3N Active Standby IPP Current Same condition with IDD3N

Active Power-Down Current CKE: Low; External clock: On; tCK, CL: see “Timings used for IDD, IPP and
IDDQ Measurement-Loop Patterns”; BL: 8 1; AL: 0; CS_n: stable at 1; Command, Address, Bank Group
IDD3P
Address, Bank Address Inputs: stable at 0; Data IO: VDDQ; DM_n: stable at 1; Bank Activity: all banks
open; Output Buffer and RTT: Enabled in Mode Registers2; ODT Signal: stable at 0
IPP3P Active Power-Down IPP Current Same condition with IDD3P
Operating Burst Read Current CKE: High; External clock: On; tCK, CL : see “Timings used for IDD, IPP
and IDDQ Measurement-Loop Patterns”; BL: 82; AL: 0; CS_n: High between RD; Command, Address,
Bank Group Address, Bank Address Inputs: partially toggling according to “IDD4R, IDDR4RA, IDD4RB
and IDDQ4R Measurement-Loop Pattern”; Data IO: seamless read data burst with different data between
IDD4R one burst and the next one according to “IDD4R, IDDR4RA, IDD4RB and IDDQ4R Measurement-Loop
Pattern”; DM_n: stable at 1; Bank Activity: all banks open, RD commands cycling through banks:
0,0,1,1,2,2,... (see IDD4R, IDDR4RA, IDD4RB and IDDQ4R Measurement-Loop Pattern); Output Buffer and
RTT: Enabled in Mode Registers2; ODT Signal: stable at 0; Pattern Details: see “IDD4R, IDDR4RA,
IDD4RB and IDDQ4R Measurement-Loop Pattern”
IDD4RA Operating Burst Read Current (AL=CL-1) AL = CL-1, Other conditions: see IDD4R
IDD4RB Operating Burst Read Current with Read DBI Read DBI enabled3, Other conditions: see IDD4R
IPP4R Operating Burst Read IPP Current Same condition with IDD4R
IDDQ4R Operating Burst Read IDDQ Current Same definition like for IDD4R, however measuring IDDQ current
(optional) instead of IDD current
IDDQ4RB Operating Burst Read IDDQ Current with Read DBI Same definition like for IDD4RB, however measuring
(optional) IDDQ current instead of IDD current
Operating Burst Write Current CKE: High; External clock: On; tCK, CL : see “Timings used for IDD, IPP
and IDDQ Measurement-Loop Patterns”; BL: 8 1; AL: 0; CS_n: High between WR; Command, Address,
Bank Group Address, Bank Address Inputs: partially toggling according to “IDD4W, IDD4WA, IDD4WB
and IDD4W_par Measurement-Loop Pattern”; Data IO: seamless write data burst with different data between
IDD4W one burst and the next one according to “IDD4W, IDD4WA, IDD4WB and IDD4W_par Measurement-Loop
Pattern”; DM_n: stable at 1; Bank Activity: all banks open, WR commands cycling through banks:
0,0,1,1,2,2,... (see IDD4W, IDD4WA, IDD4WB and IDD4W_par Measurement-Loop Pattern); Output Buffer
and RTT: Enabled in Mode Registers2; ODT Signal: stable at HIGH; Pattern Details: see IDD4W, IDD4WA,
IDD4WB and IDD4W_par Measurement-Loop Pattern
IDD4WA Operating Burst Write Current (AL=CL-1)
IDD4WB Operating Burst Write Current with Write DBI Write DBI enabled3, Other conditions: see IDD4W
IDD4WC Operating Burst Write Current with Write CRC Write CRC enabled3, Other conditions: see IDD4W
IDD4W_par Operating Burst Write Current with CA Parity CA Parity enabled3, Other conditions: see IDD4W
IPP4W Operating Burst Write IPP Current Same condition with IDD4W

π PM Tech
Symbol Description
PMA213816B
Burst Refresh Current (1X REF) CKE: High; External clock: On; tCK, CL, nRFC : see “Timings used for
IDD, IPP and IDDQ Measurement-Loop Patterns”; BL: 8 1; AL: 0; CS_n: High between REF; Command,
Address, Bank Group Address, Bank Address Inputs: partially toggling according to “IDD5B
IDD5B
Measurement-Loop Pattern”; Data IO: VDDQ; DM_n: stable at 1; Bank Activity: REF command every nRFC
(see IDD5B Measurement-Loop Pattern); Output Buffer and RTT: Enabled in Mode Registers2; ODT
Signal: stable at 0; Pattern Details: see “IDD5B Measurement-Loop Pattern”
IPP5B Burst Refresh Write IPP Current (1X REF) Same condition with IDD5B
IDD5F2 Burst Refresh Current (2X REF) tRFC=tRFC_x2, Other conditions: see IDD5B
IPP5F2 Burst Refresh Write IPP Current (2X REF) Same condition with IDD5F2
IDD5F4 Burst Refresh Current (4X REF) tRFC=tRFC_x4, Other conditions: see IDD5B
IPP5F4 Burst Refresh Write IPP Current (4X REF) Same condition with IDD5F4
Self Refresh Current: Normal Temperature Range TCASE: 0 - 85°C; Low Power Array Self Refresh (LP
ASR) : Normal4; CKE: Low; External clock: Off; CK_t and CK_c#: LOW; CL: see “Timings used for IDD, IPP
IDD6N and IDDQ Measurement-Loop Patterns”; BL: 8 1; AL: 0; CS_n#, Command, Address, Bank Group
Address, Bank Address, Data IO: High; DM_n: stable at 1; Bank Activity: Self-Refresh operation; Output
Buffer and RTT: Enabled in Mode Registers 2; ODT Signal: MID-LEVEL
IPP6N Self Refresh IPP Current: Normal Temperature Range Same condition with IDD6N
Self-Refresh Current: Extended Temperature Range) TCASE: 0 - 95°C; Low Power Array Self Refresh
(LP ASR) : Extended4; CKE: Low; External clock: Off; CK_t and CK_c: LOW; CL: see “Timings used for
IDD6E IDD, IPP and IDDQ Measurement-Loop Patterns”; BL: 8 1; AL: 0; CS_n, Command, Address, Bank Group
Address, Bank Address, Data IO: High; DM_n:stable at 1; Bank Activity: Extended Temperature Self-
Refresh operation; Output Buffer and RTT: Enabled in Mode Registers 2; ODT Signal: MID-LEVEL
IPP6E Self Refresh IPP Current: Extended Temperature Range Same condition with IDD6E
Self-Refresh Current: Reduced Temperature Range TCASE: 0 - 45°C; Low Power Array Self Refresh
(LP ASR) : Reduced4; CKE: Low; External clock: Off; CK_t and CK_c#: LOW; CL: see “Timings used for
IDD6R IDD, IPP and IDDQ Measurement-Loop Patterns”; BL: 81; AL: 0; CS_n#, Command, Address, Bank Group
Address, Bank Address, Data IO: High; DM_n:stable at 1; Bank Activity: Extended Temperature Self-
Refresh operation; Output Buffer and RTT: Enabled in Mode Registers 2; ODT Signal: MID-LEVEL
IPP6R Self Refresh IPP Current: Reduced Temperature Range Same condition with IDD6R
Auto Self-Refresh Current TCASE: 0 - 95°C; Low Power Array Self Refresh (LP ASR) : Auto4;CKE: Low;
External clock: Off; CK_t and CK_c#: LOW; CL: see “Timings used for IDD, IPP and IDDQ Measurement-
IDD6A Loop Patterns”; BL: 8 1; AL: 0; CS_n#, Command, Address, Bank Group Address, Bank Address, Data
IO: High; DM_n:stable at 1; Bank Activity: Auto Self-Refresh operation; Output Buffer and RTT: Enabled in
Mode Registers2; ODT Signal: MID-LEVEL
IPP6A Auto Self-Refresh IPP Current Same condition with IDD6A
Operating Bank Interleave Read Current CKE: High; External clock: On; tCK, nRC, nRAS, nRCD, nRRD,
nFAW, CL : see “Timings used for IDD, IPP and IDDQ Measurement-Loop Patterns”; BL: 8 1; AL: CL-1;
CS_n: High between ACT and RDA; Command, Address, Bank Group Address, Bank Address Inputs:
partially toggling according to ”IDD7 Measurement-Loop Pattern”; Data IO: read data bursts with different
IDD7
data between one burst and the next one according to ”IDD7 Measurement-Loop Pattern”; DM_n: stable at 1;
Bank Activity: two times interleaved cycling through banks (0, 1, ...7) with different addressing, see ”IDD7
Measurement-Loop Pattern”; Output Buffer and RTT: Enabled in Mode Registers2; ODT Signal: stable at 0;
Pattern Details: see ”IDD7 Measurement-Loop Pattern”
IPP7 Operating Bank Interleave Read IPP Current Same condition with IDD7
IDD8 Maximum Power Down Current TBD
IPP8 Maximum Power Down IPP Current Same condition with IDD8

π
NOTE :
PM Tech PMA213816B
1. Burst Length: BL8 fixed by MRS: set MR0 [A1:0=00].

2. Output Buffer Enable
- set MR1 [A12 = 0] : Qoff = Output buffer enabled
- set MR1 [A2:1 = 00] : Output Driver Impedance Control = RZQ/7
RTT_Nom enable
- set MR1 [A10:8 = 011] : RTT_NOM = RZQ/6
RTT_WR enable
- set MR2 [A10:9 = 01] : RTT_WR = RZQ/2
RTT_PARK disable
- set MR5 [A8:6 = 000].
3. CAL enabled : set MR4 [A8:6 = 001] : 1600MT/s
010] : 1866MT/s, 2133MT/s
011] : 2400MT/s
Gear Down mode enabled :set MR3 [A3 = 1] : 1/4 Rate DLL disabled : set MR1 [A0 = 0]
CA parity enabled :set MR5 [A2:0 = 001] : 1600MT/s,1866MT/s, 2133MT/s
010] : 2400MT/s
Read DBI enabled : set MR5 [A12 = 1] Write DBI enabled : set :MR5 [A11 = 1]
4. Low Power Array Self Refresh (LP ASR) : set MR2 [A7:6 = 00] : Normal
01] : Reduced Temperature range
10] : Extended Temperature range
11] : Auto Self Refresh
5. IDD2NG should be measured after sync pulse(NOP) input.

π PM Tech
IDD0, IDD0A and IPP0 Measurement-Loop Pattern1
PMA213816B
RAS_n/ A16
CAS_n/ A15
CK_t /CK_c
A[17,13,11]
WE_n/ A14
Command
A12/BC_n
Sub-Loop
A[10]/AP
BG[1:0]2
Number
BA[1:0]
C[2:0]3
ACT_n
A[9:7]
A[6:3]
A[2:0]
Cycle
CS_n
CKE
ODT
Data4
0 ACT 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 -
1,2 D, D 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 -
3,4 D_#, D_# 1 1 1 1 1 0 0 32 3 0 0 0 7 F 0 -
0
... repeat pattern 1...4 until nRAS - 1, truncate if necessary
nRAS PRE 0 1 0 1 0 0 0 0 0 0 0 0 0 0 0 -
... repeat pattern 1...4 until nRC - 1, truncate if necessary
1 1*nRC repeat Sub-Loop 0, use BG[1:0]2 = 1, BA[1:0] = 1 instead
repeat Sub-Loop 0, use BG[1:0]2 = 0, BA[1:0] = 1 instead
Static High
4 4*nRC
toggling

11 11*nRC repeat Sub-Loop 0, use BG[1:0]2 = 3, BA[1:0] = 3 instead For x4 and
12 12*nRC repeat Sub-Loop 0, use BG[1:0]2 = 2, BA[1:0] = 1 instead x8 only

NOTE :
1. DQS_t, DQS_c are VDDQ.
2. BG1 is don’t care for x16 device.
3. C[2:0] are used only for 3DS device.
4. DQ signals are VDDQ.

π PM Tech
IDD1, IDD1A and IPP1 Measurement-Loop Pattern1
PMA213816B
CK_t, CK_c
RAS_n/A16
CAS_n/A15
A[17,13,11]
WE_n/A14
Command
A12/BC_n
Sub-Loop
A[10]/AP
BG[1:0]2
Number
BA[1:0]
C[2:0]3
ACT_n
A[9:7]
A[6:3]
A[2:0]
Cycle
CS_n
ODT
CKE
Data4
0 ACT 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 -
1, 2 D, D 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 -
3, 4 D#, D# 1 1 1 1 1 0 0 b 3 0 0 0 7 F 0 -
3
... repeat pattern 1...4 until nRCD - AL - 1, truncate if necessary
D0=00, D1=FF
0 D2=FF, D3=00
nRCD -AL RD 0 1 1 0 1 0 0 0 0 0 0 0 0 0 0
D4=FF, D5=00
D6=00, D7=FF
nRAS PRE 0 1 0 1 0 0 0 0 0 0 0 0 0 0 0 -
... repeat pattern 1...4 until nRC - 1, truncate if necessary
1*nRC + 0 ACT 0 0 0 1 1 0 0 1 1 0 0 0 0 0 0 -
1*nRC + 1, 2 D, D 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 -
1*nRC + 3, 4 D#, D# 1 1 1 1 1 0 0 b 3 0 0 0 7 F 0 -
3
... repeat pattern nRC + 1...4 until 1*nRC + nRAS - 1, truncate if necessary
1 D0=FF, D1=00
D2=00, D3=FF
1*nRC + nRCD - AL RD 0 1 1 0 1 0 0 1 1 0 0 0 0 0 0
D4=00, D5=FF
D6=FF, D7=00
Static High
toggling

1*nRC + nRAS PRE 0 1 0 1 0 0 0 0 0 0 0 0 0 0 0 -
... repeat nRC + 1...4 until 2*nRC - 1, truncate if necessary
12 12*nRC repeat Sub-Loop 0, use BG[1:0]2 = 3, BA[1:0] = 3 instead For x4 and x8
2 only
13 13*nRC repeat Sub-Loop 1, use BG[1:0] = 2, BA[1:0] = 1 instead
NOTE :
1. DQS_t, DQS_c are used according to RD Commands, otherwise VDDQ
2. BG1 is don’t care for x16 device
3. C[2:0] are used only for 3DS device 4. Burst Sequence driven on each DQ signal by Read Command. Outside burst operation, DQ
signals are VDDQ.

π PM Tech
IDD2N, IDD2NA, IDD2NL, IDD2NG, IDD2ND, IDD2N_par, IPP2,IDD3N, IDD3NA and IDD3P
PMA213816B
CK_t, CK_c Measurement-Loop Pattern1
RAS_n/A16
CAS_n/A15
A[17,13,11]
Command
WE_n/A14
A12/BC_n
Sub-Loop
A[10]/AP
BG[1:0]2
Number
BA[1:0]
C[2:0]3
ACT_n
A[9:7]
A[6:3]
A[2:0]
Cycle
CS_n
CKE
ODT
Data4
0 D, D 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 -
1 D, D 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 -
0 2
2 D#, D# 1 1 1 1 1 0 0 3 3 0 0 0 7 F 0 -
3 D#, D# 1 1 1 1 1 0 0 2 3 0 0 0 7 F 0 -
3
1 4-7 2
repeat Sub-Loop 0, but ODT = 1 and BG[1:0] = 1, BA[1:0] = 1 instead
2 8-11 repeat Sub-Loop 0, but ODT = 0 and BG[1:0]2 = 0, BA[1:0] = 2 instead
Static High
toggling


11 44-47 repeat Sub-Loop 0, but ODT = 1 and BG[1:0]2 = 3, BA[1:0] = 3 instead For x4 and
2 x8 only
12 48-51 repeat Sub-Loop 0, but ODT = 0 and BG[1:0] = 2, BA[1:0] = 1 instead
NOTE :

π PM Tech
IDD2NT and IDDQ2NT Measurement-Loop Pattern1
PMA213816B
CK_t, CK_c
RAS_n/A16
CAS_n/A15
A[17,13,11]
Command
WE_n/A14
A12/BC_n
Sub-Loop
A[10]/AP
BG[1:0]2
Number
BA[1:0]
C[2:0]3
ACT_n
A[9:7]
A[6:3]
A[2:0]
Cycle
CS_n
CKE
ODT
Data4
0 D, D 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 -
1 D, D 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 -
0 2
2 D#, D# 1 1 1 1 1 0 0 3 3 0 0 0 7 F 0 -
3 D#, D# 1 1 1 1 1 0 0 32 3 0 0 0 7 F 0 -

Static High
toggling


11 44-47 repeat Sub-Loop 0, but ODT = 1 and BG[1:0]2 = 3, BA[1:0] = 3 instead For x4 and
x8 only
NOTE :

π PM Tech
IDD4R, IDDR4RA, IDD4RB and IDDQ4R Measurement-Loop Pattern1
PMA213816B
CK_t, CK_c
RAS_n/A16
CAS_n/A15
A[17,13,11]
Command
WE_n/A14
A12/BC_n
Sub-Loop
A[10]/AP
BG[1:0]2
Number
BA[1:0]
C[2:0]3
ACT_n
A[9:7]
A[6:3]
A[2:0]
Cycle
CS_n
CKE
ODT
Data4
D0=00, D1=FF
D2=FF, D3=00
0 RD 0 1 1 0 1 0 0 0 0 0 0 0 0 0 0
D4=FF, D5=00
0 D6=00, D7=FF
1 D 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 -
2,3 D#, D# 1 1 1 1 1 0 0 2 3 0 0 0 7 F 0 -
3
D0=FF, D1=00
D2=00, D3=FF
4 RD 0 1 1 0 1 0 0 1 1 0 0 0 7 F 0
D4=00, D5=FF
1 D6=FF, D7=00
5 D 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 -
6,7 D#, D# 1 1 1 1 1 0 0 32 3 0 0 0 7 F 0 -
2 8-11 repeat Sub-Loop 0, use BG[1:0]2 = 0, BA[1:0] = 2 instead

Static High
toggling


For x4 and x8 only
NOTE :
1. DQS_t, DQS_c are used according to RD Commands, otherwise VDDQ.
4. Burst Sequence driven on each DQ signal by Read Command.

π PM Tech
IDD4W, IDD4WA, IDD4WB and IDD4W_par Measurement-Loop Pattern1
PMA213816B
CK_t, CK_c
RAS_n/A16
CAS_n/A15
A[17,13,11]
Command
WE_n/A14
A12/BC_n
Sub-Loop
A[10]/AP
BG[1:0]2
Number
BA[1:0]
C[2:0]3
ACT_n
A[9:7]
A[6:3]
A[2:0]
Cycle
CS_n
CKE
ODT
Data4
D0=00, D1=FF
D2=FF, D3=00
0 WR 0 1 1 0 1 1 0 0 0 0 0 0 0 0 0
D4=FF, D5=00
0 D6=00, D7=FF
1 D 1 0 0 0 0 1 0 0 0 0 0 0 0 0 0 -
2,3 D#, D# 1 1 1 1 1 1 0 2 3 0 0 0 7 F 0 -
3
D0=FF, D1=00
D2=00, D3=FF
4 WR 0 1 1 0 1 1 0 1 1 0 0 0 7 F 0
D4=00, D5=FF
1 D6=FF, D7=00
5 D 1 0 0 0 0 1 0 0 0 0 0 0 0 0 0 -
6,7 D#, D# 1 1 1 1 1 1 0 32 3 0 0 0 7 F 0 -

Static High
toggling


For x4 and x8 only
NOTE :
1. DQS_t, DQS_c are used according to WR Commands, otherwise VDDQ.
4. Burst Sequence driven on each DQ signal by Write Command.

π PM Tech
IDD4WC Measurement-Loop Pattern1
PMA213816B
CK_t, CK_c
RAS_n/A16
CAS_n/A15
A[17,13,11]
Command
WE_n/A14
A12/BC_n
Sub-Loop
A[10]/AP
BG[1:0]2
Number
BA[1:0]
C[2:0]3
ACT_n
A[9:7]
A[6:3]
A[2:0]
Cycle
CS_n
CKE
ODT
Data4
D0=00, D1=FF
D2=FF, D3=00
0 WR 0 1 1 0 1 1 0 0 0 0 0 0 0 0 0 D4=FF, D5=00
D6=00, D7=FF
0 D8=CRC
1,2 D, D 1 0 0 0 0 1 0 0 0 0 0 0 0 0 0 -
3,4 D#, D# 1 1 1 1 1 1 0 32 3 0 0 0 7 F 0 -
D0=FF, D1=00
D2=00, D3=FF
5 WR 0 1 1 0 1 1 0 1 1 0 0 0 7 F 0 D4=00, D5=FF
D6=FF, D7=00
1 D8=CRC
6,7 D, D 1 0 0 0 0 1 0 0 0 0 0 0 0 0 0 -
8,9 D#, D# 1 1 1 1 1 1 0 2 3 0 0 0 7 F 0 -
3
Static High
repeat Sub-Loop 1, use BG[1:0]2 = 1, BA[1:0] = 3 instead

toggling
3 15-19


For x4 and x8 only
NOTE :
4. Burst Sequence driven on each DQ signal by Write Command.

π PM Tech
IDD5B Measurement-Loop Pattern1
PMA213816B
CK_t, CK_c
RAS_n/A16
CAS_n/A15
A[17,13,11]
Command
WE_n/A14
A12/BC_n
Sub-Loop
A[10]/AP
BG[1:0]2
Number
BA[1:0]
C[2:0]3
ACT_n
A[9:7]
A[6:3]
A[2:0]
Cycle
CS_n
CKE
ODT
Data4
0 0 REF 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 -
1 D 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 -
2 D 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0
3 D#, D# 1 1 1 1 1 0 0 2 3 0 0 0 7 F 0 -
3
4 D#, D# 1 1 1 1 1 0 0 2 3 0 0 0 7 F 0 -
3
4-7 repeat pattern 1...4, use BG[1:0]2 = 1, BA[1:0] = 1 instead
Static High
toggling
1 24-27 repeat pattern 1...4, use BG[1:0]2 = 0, BA[1:0] = 3 instead


44-47 repeat pattern 1...4, use BG[1:0]2 = 3, BA[1:0] = 3 instead For x4 and
x8 only
48-51 2
repeat pattern 1...4, use BG[1:0] = 2, BA[1:0] = 1 instead
2 64 ... nRFC - 1 repeat Sub-Loop 1, Truncate, if necessary
NOTE :

π PM Tech
IDD7 Measurement-Loop Pattern1
PMA213816B
CK_t, CK_c
RAS_n/A16
CAS_n/A15
A[17,13,11]
Command
WE_n/A14
A12/BC_n
Sub-Loop
A[10]/AP
BG[1:0]2
Number
BA[1:0]
C[2:0]3
ACT_n
A[9:7]
A[6:3]
A[2:0]
Cycle
CS_n
CKE
ODT
Data4
0 ACT 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 -
D0=00, D1=FF
D2=FF, D3=00
1 RDA 0 1 1 0 1 0 0 0 0 0 1 0 0 0
D4=FF, D5=00
0 D6=00, D7=FF
2 D 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 -
3 D# 1 1 1 1 1 0 0 2 3 0 0 0 7 F 0 -
3
... repeat pattern 2...3 until nRRD - 1, if nRCD > 4. Truncate if necessary
nRRD ACT 0 0 0 0 0 0 0 1 1 0 0 0 0 0 0 -
D0=FF, D1=00
D2=00, D3=FF
1 nRRD + 1 RDA 0 1 1 0 1 0 1 1 0 0 1 0 0 0
D4=00, D5=FF
D6=FF, D7=00
... repeat pattern 2 ... 3 until 2*nRRD - 1, if nRCD > 4. Truncate if necessary
2 2*nRRD repeat Sub-Loop 0, use BG[1:0]2 = 0, BA[1:0] = 2 instead
3 3*nRRD repeat Sub-Loop 1, use BG[1:0]2 = 1, BA[1:0] = 3 instead

4 4*nRRD repeat pattern 2 ... 3 until nFAW - 1, if nFAW > 4*nRCD. Truncate if necessary
5 nFAW repeat Sub-Loop 0, use BG[1:0]2 = 0, BA[1:0] = 1 instead

Static High
toggling
6 nFAW + nRRD repeat Sub-Loop 1, use BG[1:0]2 = 1, BA[1:0] = 2 instead
7 nFAW + 2*nRRD repeat Sub-Loop 0, use BG[1:0]2 = 0, BA[1:0] = 3 instead
8 nFAW + 3*nRRD repeat Sub-Loop 1, use BG[1:0]2 = 1, BA[1:0] = 0 instead

9 nFAW + 4*nRRD repeat Sub-Loop 4
10 2*nFAW repeat Sub-Loop 0, use BG[1:0]2 = 2, BA[1:0] = 0 instead
11 2*nFAW + nRRD repeat Sub-Loop 1, use BG[1:0]2 = 3, BA[1:0] = 1 instead
12 2*nFAW + 2*nRRD repeat Sub-Loop 0, use BG[1:0]2 = 2, BA[1:0] = 2 instead

14 2*nFAW + 4*nRRD repeat Sub-Loop 4
For x4 and x8
only
15 3*nFAW repeat Sub-Loop 0, use BG[1:0]2 = 2, BA[1:0] = 1 instead
16 3*nFAW + nRRD repeat Sub-Loop 1, use BG[1:0]2 = 3, BA[1:0] = 2 instead

19 3*nFAW + 4*nRRD repeat Sub-Loop 4
20 4*nFAW repeat pattern 2 ... 3 until nRC - 1, if nRC > 4*nFAW. Truncate if necessary
NOTE :
4. Burst Sequence driven on each DQ signal by Read Command. Outside burst operation, DQ signals are VDDQ.

π PM Tech
IDD Specification
PMA213816B
IDD and IPP values are for typical operating range of voltage and temperature unless otherwise noted.
DDR4-2666
Symbol Width Unit
19-19-19
x4, x8 71 mA
IDD0: One bank ACTIVATE-to-RECHARGE current
x16 75 mA
IPP0: One bank ACTIVATE-to-RECHARGE IPP current ALL 6 mA
x4, x8 90 mA
IDD1: One bank ACTIVATE-to-READ-to- PRECHARGE current
x16 106 mA
IDD2N: Precharge standby current ALL 53 mA
IDD2NT: Precharge standby ODT current ALL 76 mA
IDD2P: Precharge powerdown current ALL 29 mA
IDD2Q: Precharge quiet standby current ALL 39 mA
IDD3N: Active standby current ALL 76 mA
IPP3N: Active standby IPP current ALL 2 mA
IDD3P: Active power-down current ALL 43 mA
x4 155 mA
IDD4R: Burst read current x8 193 mA
x16 268 mA
x4 174 mA
IDD4W: Burst write current x8 202 mA
x16 264 mA
x4 146 mA
IDD5R: Burst refresh current
x8 151 mA
(1X REF)
x16 156 mA
IPP5R: Burst refresh IPP current
ALL 37 mA
(1X REF)
IDD6N: Self refresh current;
ALL 29 mA
0–85°C1
IDD6E: Self refresh current;
ALL 32 mA
0–95°C 2
IDD6R: Self refresh current;
ALL 25 mA
0–45C 3, 4
IDD6A: Auto self refresh current ALL 30 mA
23 ALL 4.1 mA
IPP6x: Auto self refresh current

DDR4-2666
Symbol Width Unit
19-19-19
x4 235 mA
IDD7: Bank interleave read current x8 245 mA
x16 275 mA
x4, x8 50 mA
IPP7: Bank interleave read IPP current
x16 34 mA
IDD8: Maximum powerdown current ALL 22 mA
NOTE
1. Applicable for MR2 settings A7 = 0 and A6 = 0; manual mode with normal temperature range of operation (0–85°C).
2. Applicable for MR2 settings A7 = 1 and A7 = 0; manual mode with extended temperature range of operation (0–95°C).
3. Applicable for MR2 settings A7 = 0 and A7 = 1; manual mode with reduced temperature range of operation (0–45°C).
4. IDD6R and IDD6A values are typical.
5. When additive latency is enabled for IDD0, current changes by approximately 9%.
6. When additive latency is enabled for IDD1 , current changes by approximately +14% (x4/x8), +14% (x16).
7. When additive latency is enabled for IDD2N , current changes by approximately +0.0%.
8. When DLL is disabled for IDD2N, current changes by approximately 1%.
9. When CAL is enabled for IDD2N, current changes by approximately –34%.
10. When gear-down is enabled for IDD2N, current changes by approximately 0%.
11. When CA parity is enabled for IDD2N, current changes by approximately +15%.
12. When additive latency is enabled for IDD3N, current changes by approximately +9%.
13. When additive latency is enabled for IDD4R, current changes by approximately +6%.
14. When read DBI is enabled for IDD4R, current changes by approximately -8%.
15. When additive latency is enabled for IDD4W, current changes by approximately +6% (x4/8), +4% (x16).
16. When write DBI is enabled for IDD4W, current changes by approximately 13%.
17. When write CRC is enabled for IDD4W, current changes by approximately 4%.
18. When CA parity is enabled for IDD4W, current changes by approximately +15% (x4/x8), +10% (x16).
19. When 2X REF is enabled for IDD5R, current changes by approximately –16%.
20. When 4X REF is enabled for IDD5R, current changes by approximately –35%.
21. IPP0 test and limit is applicable for IDD0 and IDD1 conditions.
22. IPP3N test and limit is applicable for all IDD2x, IDD3x, IDD4x and IDD8 conditions; that is, testing IPP3N should satisfy the IPPs for
the noted IDD tests.
23. IPP6x is applicable to IDD6N, IDD6E, IDD6R and IDD6A conditions.

π PM Tech
Input/Output Capacitance
PMA213816B
Silicon Pad Input/Output Capacitance

DDR4- DDR4-
DDR4-2933 DDR4-3200
Symbol Parameter 1600,1866,2133 2400,2666 Unit Note
min max min max min max min max
CIO Input/output capacitance 0.55 1.4 0.55 1.15 0.55 1.15 0.55 1.00 pF 1,2,3
Input capacitance, CK_t

CCK 0.2 0.8 0.2 0.7 0.2 0.6 0.2 0.7 pF 2,3
and CK_c
Input capacitance delta
CDCK 0 0.05 0 0.05 0 0.05 0 0.05 pF 2,3,6
CK_t and CK_c
Input/output capacitance
CDDQS 0 0.05 0 0.05 0 0.05 0 0.05 pF 2,3,5
delta
Input capacitance
CI 0.2 0.8 0.2 0.7 0.2 0.6 0.2 0.55 2,3,4
(CTRL, ADD, CMD pins pF
Input capacitance delta
CDI_ CTRL -0.1 0.1 -0.1 0.1 -0.1 0.1 -0.1 0.1 2,3,8,9
(All CTRL pins only) pF
Input capacitance delta 1,2,10,
CDI_ ADD_CMD -0.1 0.1 -0.1 0.1 -0.1 0.1 -0.1 0.1
(All ADD/ CMD pins only) pF 11
Input/output capacitance
CDIO -0.1 0.1 -0.1 0.1 -0.1 0.1 -0.1 0.1 pF 1,2,3,4
delta
Input/output capacitance of
CALERT 0.5 1.5 0.5 1.5 0.5 1.5 0.5 1.5 pF 2,3
ALERT
CZQ - 2.3 - 2.3 - 2.3 - 2.3 pF 2,3,12
ZQ
CTEN 0.2 2.3 0.2 2.3 0.2 2.3 0.2 2.3 pF 2,3,13
TEN
NOTE
1. Although the DM, TDQS_t, and TDQS_c pins have different functions, the loading matches DQ and DQS.
2. This parameter is not subject to a production test; it is verified by design and characterization.
3. The capacitance is measured with VDD, VDDQ, VSS, and VSSQ applied and all other pins floating (except the pin under test, CKE,
RESET_n and ODT, as necessary). VDD = VDDQ = 1.5V, VBIAS = VDD/2 and on-die termination off.
4. This parameter applies to monolithic die, obtained by de-embedding the package L and C parasitics.
5. CDIO = CIO(DQ, DM) - 0.5 × (CIO(DQS_t) + CIO(DQS_c)).
6. Absolute value of CIO (DQS_t), CIO (DQS_c)
7. Absolute value of CCK_t, CCK_c
8. CI applies to ODT, CS_n, CKE, A[17:0], BA[1:0], BG[1:0], RAS_n, CAS_n, ACT_n, PAR and WE_n.
9. CDI_CTRL applies to ODT, CS_n, and CKE.
10. CDI_CTRL = CI(CTRL) - 0.5 × (CI(CLK_t) + CI(CLK_c)).
11. CDI_ADD_CMD applies to A[17:0], BA1:0], BG[1:0], RAS_n, CAS_n, ACT_n, PAR and WE_n.
12. CDI_ADD_CMD = CI(ADD_CMD) - 0.5 × (CI(CLK_t) + CI(CLK_c)).
13. Maximum external load capacitance on ZQ pin: 5pF.
14. Only applicable if TEN pin does not have an internal pull-up.

π PM Tech
DRAM package electrical specifications (x4/x8)
PMA213816B
DDR4-
1600/1866/2133/ DDR4-2933 DDR4-3200
Symbol Parameter 2400,2666 Unit Note

ZIO Input/output Zpkg 45 85 48 85 48 85 Ω 1
TdIO Input/output Pkg Delay 14 42 14 42 14 40 ps 1
Lio Input/Output Lpkg - 3.3 - 3.3 - 3.3 nH 1,2
Cio Input/Output Cpkg - 0.78 - 0.78 - 0.78 pF 1,3
ZDIO DQS DQS_t, DQS_c Zpkg 45 85 48 85 48 85 Ω 1
TdDIO DQS DQS_t, DQS_c Pkg Delay 14 42 14 42 14 40 ps 1
Lio DQS DQS Lpkg - 3.3 - 3.3 - 3.3 nH 1,2
Cio DQS DQS Cpkg - 0.78 - 0.78 - 0.78 pF 1,3
Delta Zpkg DQSU_t and DQSU_c - 10 - 10 - 10 Ω -
DZDIO DQS
Delta Zpkg DQSL_t, DQSL_c - 10 - 10 - 10 Ω -
Delta Delay DQSU_t and DQSU_c - 5 - 5 - 5 ps -
DTdDIO DQS
Delta Delay DQSL_t, DQSL_c - 5 - 5 - 5 ps -
ZI CTRL Input- CTRL pins Zpkg 50 90 50 90 50 90 Ω 1
TdI_ CTRL Input- CTRL pins Pkg Delay 14 42 14 42 14 40 ps 1
Li CTRL Input CTRL Lpkg - 3.4 - 3.4 - 3.4 nH 1,2
Ci CTRL Input CTRL Cpkg - 0.7 - 0.7 - 0.7 pF 1,3
ZIADD CMD Input- CMD ADD pins Zpkg 50 90 50 90 50 90 Ω 1
TdIADD_ CMD Input- CMD ADD pins Pkg Delay 14 45 14 40 14 40 ps 1
Li ADD CMD Input CMD ADD Lpkg - 3.6 - 3.6 - 3.6 nH 1,2
Ci ADD CMD Input CMD ADD Cpkg - 0.74 - 0.74 - 0.74 pF 1,3
ZCK CK_t and CK_c Zpkg 50 90 50 90 50 90 Ω 1
TdCK CK_t and CK_c Pkg Delay 14 42 14 42 14 42 ps 1
Li CLK Input CLK Lpkg - 3.4 - 3.4 - 3.4 nH 1,2
Ci CLK Input CLK Cpkg - 0.7 - 0.7 - 0.7 pF 1,3
DZDCK Delta Zpkg CK_t and CK_c - 10 - 10 - 10 Ω -
DTdCK Delta Delay CK_t and CK_c - 5 - 5 - 5 ps -
ZOZQ ZQ Zpkg - 100 - 100 - 100 Ω -
TdO ZQ ZQ Delay 20 90 20 90 20 90 ps -
ZO ALERT ALERT Zpkg 40 100 40 100 40 100 Ω -
TdO ALERT ALERT Delay 20 55 20 55 20 55 ps -
NOTE
1. Package implementations shall meet spec if the Zpkg and Pkg Delay fall within the ranges shown, and the maximum Lpkg and Cpkg
do not exceed the maximum value shown.
2. It is assumed that Lpkg can be approximated as Lpkg = Zo*Td.
3. It is assumed that Cpkg can be approximated as Cpkg = Td/Zo.

π PM Tech
DRAM package electrical specifications (x16)
PMA213816B
DDR4-
1600/1866/2133/ DDR4-2933 DDR4-3200
Symbol Parameter 2400,2666 Unit Note

ZIO Input/output Zpkg 45 85 48 85 48 85 Ω 1
TdIO Input/output Pkg Delay 14 45 14 45 14 45 ps 1
Lio Input/Output Lpkg - 3.4 - 3.4 - 3.4 nH 1,2
Cio Input/Output Cpkg - 0.82 - 0.82 - 0.82 pF 1,3
ZDIO DQS DQS_t, DQS_c Zpkg 45 85 48 85 48 85 Ω 1
TdDIO DQS DQS_t, DQS_c Pkg Delay 14 45 14 45 14 45 ps 1
Lio DQS DQS Lpkg - 3.4 - 3.4 - 3.4 nH 1,2
Cio DQS DQS Cpkg - 0.82 - 0.82 - 0.82 pF 1,3
Delta Zpkg DQSU_t and DQSU_c - 10 - 10 - 10 Ω -
DZDIO DQS
Delta Zpkg DQSL_t, DQSL_c - 10 - 10 - 10 Ω -
Delta Delay DQSU_t and DQSU_c - 5 - 5 - 5 ps -
DTdDIO DQS
Delta Delay DQSL_t, DQSL_c - 5 - 5 - 5 ps -
ZI CTRL Input- CTRL pins Zpkg 50 90 50 90 50 90 Ω 1
TdI_ CTRL Input- CTRL pins Pkg Delay 14 42 14 42 14 42 ps 1
Li CTRL Input CTRL Lpkg - 3.4 - 3.4 - 3.4 nH 1,2
Ci CTRL Input CTRL Cpkg - 0.7 - 0.7 - 0.7 pF 1,3
ZIADD CMD Input- CMD ADD pins Zpkg 50 90 50 90 50 90 Ω 1
TdIADD_ CMD Input- CMD ADD pins Pkg Delay 14 52 14 52 14 52 ps 1
Li ADD CMD Input CMD ADD Lpkg - 3.9 - 3.9 - 3.9 nH 1,2
Ci ADD CMD Input CMD ADD Cpkg - 0.86 - 0.86 - 0.86 pF 1,3
ZCK CK_t and CK_c Zpkg 50 90 50 90 50 90 Ω 1
TdCK CK_t and CK_c Pkg Delay 14 42 14 42 14 42 ps 1
Li CLK Input CLK Lpkg - 3.4 - 3.4 - 3.4 nH 1,2
Ci CLK Input CLK Cpkg - 0.7 - 0.7 - 0.7 pF 1,3
DZDCK Delta Zpkg CK_t and CK_c - 10 - 10 - 10 Ω -
DTdCK Delta Delay CK_t and CK_c - 5 - 5 - 5 ps -
ZOZQ ZQ Zpkg - 100 - 100 - 100 Ω -
TdO ZQ ZQ Delay 20 90 20 90 20 90 ps -
ZO ALERT ALERT Zpkg 40 100 40 100 40 100 Ω -
TdO ALERT ALERT Delay 20 55 20 55 20 55 ps -
NOTE
1. Package implementations shall meet spec if the Zpkg and Pkg Delay fall within the ranges shown, and the maximum Lpkg and Cpkg
do not exceed the maximum value shown.
2. It is assumed that Lpkg can be approximated as Lpkg = Zo*Td.
3. It is assumed that Cpkg can be approximated as Cpkg = Td/Zo.

π PM Tech
Electrical Characteristics & AC Timing
PMA213816B
Reference Load for AC Timing and Output Slew Rate

Figure below represents the effective reference load of 50 ohms used in defining the relevant AC timing
parameters of the device as well as output slew rate measurements.
It is not intended as a precise representation of any particular system environment or a depiction of the
actual load presented by a production tester. System designers should use IBIS or other simulation tools to
correlate the timing reference load to a system environment. Manufacturers correlate to their production test
conditions, generally one or more coaxial transmission lines terminated at the tester electronics.
tREFI
Average periodic Refresh interval (tREFI) of DDR4 SDRAM is defined.
tREFI by device density
Parameter Symbol 2Gb 4Gb 8Gb 16Gb Unit
All Bank Refresh to active/refresh cmd time tRFC 160 260 350 550 ns
0°C≤TCASE ≤85°C 7.8 7.8 7.8 7.8 μs
Average periodic refresh interval tREFI -40°C≤TCASE ≤85°C 7.8 7.8 7.8 7.8 μs
85°C≤TCASE ≤95°C 3.9 3.9 3.9 3.9 μs
NOTE:
Users should refer to the DRAM supplier data sheet and/or the DIMM SPD to determine if the devices support these options or
requirements.

π PM Tech
Timing Parameters by Speed Grade
PMA213816B
Timing Parameters by Speed Bin for DDR4-1600 to DDR4-2400

Speed DDR4-1600 DDR4-1866 DDR4-2133 DDR4-2400
Unit NOTE
Parameter Symbol MIN MAX MIN MAX MIN MAX MIN MAX
Clock Timing
Minimum Clock Cycle Time tCK

8 20 8 20 8 20 8 20 ns
(DLL off mode) (DLL_OFF)
Average Clock Period tCK(avg) 1.25 1.9 1.071 1.9 0.937 1.9 0.833 1.9 ps 35,36
Average high pulse width tCH(avg) 0.48 0.52 0.48 0.52 0.48 0.52 0.48 0.52 tCK(avg)
Average low pulse width tCL(avg) 0.48 0.52 0.48 0.52 0.48 0.52 0.48 0.52 tCK(avg)
MIN = tCK (AVG) MIN + tJITper_tot MIN;

Absolute Clock Period tCK(abs) tCK(avg)
MAX = tCK (AVG) MAX + tJITper_tot MAX
Absolute clock HIGH pulse width tCH(abs) 0.45 - 0.45 - 0.45 - 0.45 - tCK(avg) 23
Absolute clock LOW pulse width tCL(abs) 0.45 - 0.45 - 0.45 - 0.45 - tCK(avg) 24
Clock Period Jitter- total JIT(per)_tot -63 63 -54 54 -47 47 -42 42 ps 23
Clock Period Jitter- deterministic JIT(per)_dj -31 31 -27 27 -23 23 -21 21 ps 26
Clock Period Jitter during DLL

tJIT(per, lck) -50 50 -43 43 -38 38 -33 33 ps
locking period
Cycle to Cycle Period Jitter tJIT(cc)_total - 125 - 107 - 94 - 83 ps 25
Cycle to Cycle Period Jitter

tJIT(cc)_dj - 63 - 54 - 47 - 42 ps 26
deterministic
Cycle to Cycle Period Jitter during

tJIT(cc, lck) - 100 - 86 - 75 - 67 ps
DLL locking period
Cumulative error across 2 cycles tERR(2per) -92 92 -79 79 -69 69 -61 61 ps
Cumulative error across 3 cycles tERR(3per) -109 109 -94 94 -82 82 73 73 ps
Cumulative error across 11 cycles tERR(11per) -164 164 -141 141 -123 123 110 110 ps
Cumulative error across n = 13, tERRnper MIN = (1 + 0.68ln[n]) × tJITper_tot MIN

tERR(nper) UI
14 . . . 49, 50 cycles tERRnper MAX = (1 + 0.68ln[n]) × tJITper_tot MAX
Command and Address setup time

tIS(base,
to CK_t, CK_c referenced to Vih(ac) 115 - 100 - 80 - 62 - ps
AC100)
/ Vil(ac) levels
Command and Address setup time
to CK_t, CK_c referenced to Vref tIS(Vref) 215 - 200 - 180 - 162 - ps
levels
Command and Address hold time to
tIH(base,
CK_t, CK_c referenced to Vih(dc) / 140 - 125 - 105 - 87 - ps
DC75)
Vil(dc) levels
Command and Address hold time to

tIH(Vref) 215 - 200 - 180 - 162 - ps
CK_t, CK_c referenced to Vref levels
Control and Address Input pulse

tIPW 600 - 525 - 460 - 410 - ps
width for each input
Command and Address Timing
CAS_n to CAS_n command delay max(5nCK max(5nCK max(5nCK max(5nCK

tCCD_L ,6.250 ns)
-
,5.355 ns)
-
,5.355 ns)
-
, 5 ns)
- nCK 34
for same bank group
CAS_n to CAS_n command delay

tCCD_S 4 - 4 - 4 - 4 - nCK 34
for different bank group

π PM Tech
PMA213816B
Unit NOTE
ACTIVATE to ACTIVATE Command

tRRD_S Max(4nCK Max(4nCK Max(4nCK Max(4nCK
delay to different bank group for 2KB ,6ns)
-
,5.3ns)
-
,5.3ns)
-
, 5.3ns)
- nCK 34
(2K)
page size
tRRD_S Max(4nCK Max(4nCK Max(4nCK Max(4nCK
delay to different bank group for 2KB ,5ns)
-
,4.2ns)
-
,3.7ns)
-
, 3.3ns)
- nCK 34
(1K)
page size
Max(4nCK Max(4nCK Max(4nCK Max(4nCK
delay to different bank group for tRRD_S(1/2K) ,5ns)
-
,4.2ns)
-
,3.7ns)
-
, 3.3ns)
- nCK 34
1/2KB page size
delay to same bank group for 2KB tRRD_L(2K) ,7.5ns)
-
,6.4ns)
-
,6.4ns)
-
, 6.4ns)
- nCK 34
page size
delay to same bank group for 1KB tRRD_L(1K) ,6ns)
-
,5.3ns)
-
,5.3ns)
-
, 4.9ns)
- nCK 34
page size
delay to same bank group for 1/2KB tRRD_L(1/2K) ,6ns)
-
,5.3ns)
-
,5.3ns)
-
, 4.9ns)
- nCK 34
page size
Four activate window for 2KB page Max(28nC Max(28nC Max(28nC Max(28nC
tFAW_2K K,35ns)
-
K,30ns)
-
K,30ns)
-
K,30ns)
- ns 34
size
Four activate window for 1KB page Max(20nC Max(20nC Max(20nC Max(20nC
tFAW_1K K,25ns)
-
K,23ns)
-
K,21ns)
-
K ,21ns)
- ns 34
size
Four activate window for 1/2KB page Max(16nC Max(16nC Max(16nC Max(16nC
tFAW_1/2K K,20ns)
-
K,17ns)
-
K,15ns)
-
K,13ns)
- ns 34
size
Delay from start of internal write

max(2nCK max(2nCK max(2nCK max(2nCK
transaction to internal read tWTR_S ,2.5ns)
-
,2.5ns)
-
,2.5ns)
-
, 2.5ns)
- ns 1,2,34
command for different bank group
Delay from start of internal write max
max(4nCK max(4nCK max(4nCK
transaction to internal read tWTR_L ,7.5ns)
-
,7.5ns)
-
,7.5ns)
- (4nCK,7.5 - ns 1,3,4
command for same bank group ns)
Internal READ Command to max(4nCK max(4nCK max(4nCK max(4nCK

tRTP ,7.5ns)
-
,7.5ns)
-
,7.5ns)
-
,7.5ns)
- ns 34
PRECHARGE Command delay
WRITE recovery time tWR 15 - 15 - 15 - 15 - ns 1
tWR+max tWR+max tWR+max tWR+max

Write recovery time when CRC and tWR_CRC
(4nCK,3.7 - (5nCK,3.7 - (5nCK,3.7 - (5nCK,3.7 - ns 1,28
DM are enabled _DM 5ns) 5ns) 5ns) 5ns)
delay from start of internal write

tWTR_S+ tWTR_S+ tWTR_S+ tWTR_S+
transaction to internal read tWTR_S_C 2, 29,
max(4nCK - max(5nCK - max(5nCK - max(5nCK - ns
command for different bank group RC_DM 34
, 3.75ns) , 3.75ns) , 3.75ns) , 3.75ns)
with both CRC and DM enabled
tWTR_L+ tWTR_L+ tWTR_L+ tWTR_L+
transaction to internal read tWTR_L_C
max(4nCK - max(5nCK - max(5nCK - max(5nCK - ns 3,30,34
command for same bank group with RC_DM , 3.75ns) , 3.75ns) , 3.75ns) , 3.75ns)
both CRC and DM enabled
DLL locking time tDLLK 597 - 597 - 768 - 768 - nCK
Mode Register Set command cycle

tMRD 8 - 8 - 8 - 8 - nCK
time
Mode Register Set command update max(24nC max(24nC max(24nC max(24nC

tMOD K,15ns)
-
K,15ns)
-
K,15ns)
-
K, 15ns)
- nCK
delay
Multi-Purpose Register Recovery

tMPRR 1 - 1 - 1 - 1 - nCK 33
Time
tMOD tMOD tMOD tMOD

Multi Purpose Register Write
tWR_MPR (min) - (min) - (min) - (min) - nCK
Recovery Time + AL + PL + AL + PL + AL + PL + AL + PL
Auto precharge write recovery +

tDAL(min) Programmed WR + roundup ( tRP / tCK(avg)) nCK
precharge time
DQ0 or DQL0 driven to 0 set-up time

tPDA_S 0.5 - 0.5 - 0.5 - 0.5 - UI 45,47
to first DQS rising edge
DQ0 or DQL0 driven to 0 hold time

tPDA_H 0.5 - 0.5 - 0.5 - 0.5 - UI 46,47
from last DQS fall-ing edge


Unit NOTE
CS_n to Command Address Latency

CS_n to Command Address Latency tCAL ,3.748 ns)
-
,3.748 ns)
-
,3.748 ns)
-
,3.748 ns)
- nCK
Mode Register Set cyce time in CAL tMOD+ tMOD+ tMOD+ tMOD+
tMRD_tCAL tCAL
-
tCAL
-
tCAL
-
tCAL
- nCK
mode
Mode Register Set update delay in tMOD+ tMOD+ tMOD+ tMOD+

tMOD_tCAL tCAL
-
tCAL
-
tCAL
-
tCAL
- nCK
CAL mode
DRAM Data Timing
DQS_t,DQS_c to DQ skew, per tCK(avg)/ 13,18,3

tDQSQ - 0.16 - 0.16 - 0.16 - 0.17 2 9,49
group, per access
DQ output hold per group, per tCK(avg)/ 13,17,1

tQH 0.76 - 0.76 - 0.76 - 0.74 - 2 8,39,49
access from DQS_t,DQS_c
Data Valid Window per device: (tQH

17,18,3
- tDQSQ) of each UI on a given tDVWd 0.63 - 0.63 - 0.64 - 0.64 - UI
9,49
DRAM
Data Valid Window , per pin per UI :
17,18,3
(tQH - tDQSQ) each UI on a pin of a tDVWp 0.66 - 0.66 - 0.69 - 0.72 - UI
9,49
given DRAM
DQ low impedance time from CK_t,

tLZ(DQ) -450 225 -390 195 -360 180 -330 175 ps 39
CK_c
DQ high impedance time from CK_t,

tHZ(DQ) - 225 - 195 - 180 - 175 ps 39
CK_c
Data Strobe Timing
0.9 NOTE 44 0.9 NOTE 44 0.9 NOTE 44 0.9 NOTE 44 tCK 39,40
DQS_t, DQS_c differential READ
tRPRE
Preamble
NA NA NA NA NA NA 1.8 NOTE 44 tCK 39,41

tRPST 0.33 NOTE 45 0.33 NOTE 45 0.33 NOTE 45 0.33 NOTE 45 tCK 39
Postamble
DQS_t,DQS_c differential output

tQSH 0.4 - 0.4 - 0.4 - 0.4 - tCK 21,39
high time
DQS_t,DQS_c differential output low

tQSL 0.4 - 0.4 - 0.4 - 0.4 - tCK 20,39
time
0.9 - 0.9 - 0.9 - 0.9 - tCK 42

DQS_t, DQS_c differential WRITE
tWPRE
Preamble
NA - NA - NA - 1.8 - tCK 43

tWPST 0.33 - 0.33 - 0.33 - 0.33 - tCK
Postamble
DQS_t and DQS_c low-impedance

tLZ(DQS) -450 225 -390 195 -360 180 330 175 ps 39
time (Referenced from RL-1)
DQS_t and DQS_c high-impedance

tHZ(DQS) - 225 - 195 - 180 - 175 ps 39
time (Referenced from RL+BL/2)
DQS_t, DQS_c differential input low

tDQSL 0.46 0.54 0.46 0.54 0.46 0.54 0.46 0.54 tCK
pulse width
DQS_t, DQS_c differential input high

tDQSH 0.46 0.54 0.46 0.54 0.46 0.54 0.46 0.54 tCK
pulse width
DQS_t, DQS_c rising edge to CK_t,

tDQSS -0.27 0.27 -0.27 0.27 -0.27 0.27 -0.27 0.27 tCK
CK_c rising edge (1 clock preamble)
DQS_t, DQS_c falling edge setup

tDSS 0.18 - 0.18 - 0.18 - 0.18 - tCK
time to CK_t, CK_c rising edge
DQS_t, DQS_c falling edge hold

tDSH 0.18 - 0.18 - 0.18 - 0.18 - tCK
time from CK_t, CK_c rising edge
DQS_t, DQS_c rising edge output

tDQSCK 37,38,3
timing lo-catino from rising CK_t, -225 225 -195 195 -180 180 175 175 ps
9
(DLL On)
CK_c with DLL On mode
DQS_t, DQS_c rising edge output tDQSCKI 37,38,3

- 370 - 330 - 310 - 290 ps
9
variance window per DRAM (DLL On)


Unit NOTE
MPSM Timing
tMOD(min) tMOD(min) tMOD(min) tMOD(min)

Command path disable delay upon + + + +
tMPED tCPDED
-
tCPDED
-
tCPDED
-
tCPDED
-
MPSM entry
(min) (min) (min) (min)
tMOD(min) tMOD(min) tMOD(min) tMOD(min)

Valid clock requirement after MPSM + + + +
tCKMPE tCPDED
-
tCPDED
-
tCPDED
-
tCPDED
-
entry
Valid clock requirement before tCKSRX tCKSRX tCKSRX tCKSRX

tCKMPX (min)
-
(min)
-
(min)
-
(min)
-
MPSM exit
Exit MPSM to commands not

tXMP tXS(min) - tXS(min) - tXS(min) - tXS(min) -
tXMP(min) tXMP(min) tXMP(min) tXMP(min)
Exit MPSM to commands requiring a + + + +
tXMPDLL tXSDLL
-
tXSDLL
-
tXSDLL
-
tXSDLL
-
locked DLL
tISmin + tISmin + tISmin + tISmin +

CS setup time to CKE tMPX_S tIHmin
-
tIHmin
-
tIHmin
-
tIHmin
-
Calibration Timing
Power-up and RESET calibration

tZQinit 1024 - 1024 - 1024 - 1024 - nCK
time
Normal operation Full calibration

tZQoper 512 - 512 - 512 - 512 - nCK
time
Normal operation Short calibration

tZQCS 128 - 128 - 128 - 128 - nCK
time
Reset/Self Refresh Timing

Exit Reset from CKE HIGH to a valid
tXPR ,tRFC(min) - ,tRFC(min) - ,tRFC(min) - ,tRFC(min) - nCK
command +10ns) +10ns) +10ns) +10ns)
Exit Self Refresh to commands not tRFC(min) tRFC(min) tRFC(min) tRFC(min)

tXS +10ns
-
+10ns
-
+10ns
-
+10ns
- nCK
SRX to commands not requiring a tXS_ABORT tRFC4(min tRFC4(min tRFC4(min tRFC4(min

- - - - nCK
locked DLL in Self Refresh ABORT (min) )+ 10ns )+ 10ns )+ 10ns )+ 10ns
Exit Self Refresh to ZQCL,ZQCS

tXS_FAST tRFC4(min tRFC4(min tRFC4(min tRFC4(min
and MRS (CL,CWL,WR,RTP and )+10ns
-
)+ 10ns
-
)+ 10ns
-
)+ 10ns
- nCK
(min)
Gear Down)
Exit Self Refresh to commands tDLLK tDLLK tDLLK tDLLK

tXSDLL (min)
-
(min)
-
(min)
-
(min)
- nCK
Minimum CKE low width for Self tCKE(min) tCKE(min) tCKE(min) tCKE(min)
tCKESR + 1nCK
-
+ 1nCK
-
+ 1nCK
-
+ 1nCK
- nCK
refresh entry to exit timing
Minimum CKE low width for Self tCKE(min) tCKE(min) tCKE(min) tCKE(min)
refresh entry to exit timing with CA tCKESR_PAR + - + - + - + - nCK
Parity enabled 1nCK+PL 1nCK+PL 1nCK+PL 1nCK+PL
Valid Clock Requirement after Self

Refresh Entry (SRE) or Power-Down tCKSRE ,10ns)
-
,10ns)
-
,10ns)
-
,10ns)
- nCK
Entry (PDE)
Refresh Entry (SRE) or Power-Down tCKSRE_PAR ,10ns)+PL -
,10ns)+PL
-
,10ns)+PL
-
,10ns)+PL
- nCK
when CA Parity is enabled
Valid Clock Requirement before Self
Refresh Exit (SRX) or Power-Down tCKSRX ,10ns)
-
,10ns)
-
,10ns)
-
,10ns)
- nCK
Exit (PDX) or Reset Exit


Unit NOTE
Power Down Timing
Exit Power Down with DLL on to any

valid command;Exit Precharge
Power Down with DLL frozen to tXP ,6ns)
-
,6ns)
-
,6ns)
-
,6ns)
- nCK
commands not requiring a locked
DLL
CKE minimum pulse width tCKE , 5ns)
-
, 5ns)
-
, 5ns)
-
, 5ns)
- nCK 31,32
Command pass disable delay tCPDED 4 - 4 - 4 - 4 - nCK
Power Down Entry to Exit Timing tPD tCKE(min) 9*tREFI tCKE(min) 9*tREFI tCKE(min) 9*tREFI tCKE(min) 9*tREFI nCK 6
Timing of ACT command to Power

tACTPDEN 1 - 1 - 2 - 2 - nCK 7
Down entry
Timing of PRE or PREA command

tPRPDEN 1 - 1 - 2 - 2 - nCK 7
to Power Down entry
Timing of RD/RDA command to

tRDPDEN RL+4+1 - RL+4+1 - RL+4+1 - RL+4+1 - nCK
Power Down entry
Timing of WR command to Power WL+4+(tW WL+4+(tW WL+4+(tW WL+4+(tW

Down entry (BL8OTF, BL8MRS, tWRPDEN R/tCK(avg) - R/tCK(avg) - R/tCK(avg) - R/tCK(avg) - nCK 4
C4OTF) ) ) ) )
Timing of WRA command to Power

WL+4+W WL+4+W WL+4+W WL+4+W
Down entry (BL8OTF, BL8MRS, tWRAPDEN R+1
-
R+1
-
R+1 - R+1 - nCK 5
BC4OTF)
WL+2+(tW WL+2+(tW WL+2+(tW WL+2+(tW
Timing of WR command to
tWRPBC4DEN R/tCK(avg) - R/tCK(avg) - R/tCK(avg) - R/tCK(avg) - nCK 4
Power Down entry (BC4MRS) ) ) ) )
Timing of WRA command to Power tWRAPBC4DE WL+2+W WL+2+W WL+2+W WL+2+W

R+1
-
R+1
-
R+1
- R+1
- nCK 5
Down entry (BC4MRS) N
Timing of REF command to Power

tREFPDEN 1 - 1 - 2 - 2 - nCK 7
Down entry
Timing of MRS command to Power tMOD tMOD tMOD tMOD

tMRSPDEN (min)
-
(min)
-
(min)
- (min)
- nCK
Down entry
PDA Timing
Mode Register Set command cycle max(16nC max(16nC max(16nC max(16nC

tMRD_PDA K, 10ns)
-
K, 10ns)
-
K, 10ns)
-
K, 10ns)
- nCK
time in PDA mode
Mode Register Set command update

tMOD_PDA tMOD tMOD tMOD tMOD nCK
delay in PDA mode
ODT Timing
Asynchronous RTT turn-on delay

tAONAS 1.0 9.0 1.0 9.0 1.0 9.0 1.0 9.0 ns
(Power- Down with DLL frozen)
Asynchronous RTT turn-off delay

tAOFAS 1.0 9.0 1.0 9.0 1.0 9.0 1.0 9.0 ns
(Power- Down with DLL frozen)
RTT dynamic change skew tADC 0.3 0.7 0.3 0.7 0.3 0.7 0.3 0.7 tCK(avg)
Write Leveling Timing
First DQS_t/DQS_n rising edge after

tWLMRD 40 - 40 - 40 - 40 - nCK 12
write leveling mode is programmed
DQS_t/DQS_n delay after write

tWLDQSEN 25 - 25 - 25 - 25 - nCK 12
leveling mode is programmed
Write leveling setup time from rising

CK_t, CK_c crossing to rising tWLS 0.13 - 0.13 - 0.13 - 0.13 - tCK(avg)
DQS_t/DQS_n crossing
Write leveling hold time from rising
DQS_t/ DQS_n crossing to rising tWLH 0.13 - 0.13 - 0.13 - 0.13 - tCK(avg)
CK_t, CK_ crossing
Write leveling output delay tWLO 0 9.5 0 9.5 0 9.5 0 9.5 ns
Write leveling output error tWLOE 0 2 0 2 0 2 0 2 ns

π PM Tech
PMA213816B
Unit NOTE
CA Parity Timing
Commands not guaranteed to be tPAR_

- PL - PL - PL - PL nCK
executed during this time UNKNOWN
Delay from errant command to tPAR_ALERT_

- PL+6ns - PL+6ns - PL+6ns - PL+6ns nCK
ALERT_n assertion ON
Pulse width of ALERT_n signal when tPAR_ALERT_

48 96 56 112 64 128 72 144 nCK
asserted PW
Time from when Alert is asserted till
controller must start providing DES tPAR_ALERT_
- 43 - 50 - 57 - 64 nCK
commands in Persistent CA parity RSP
mode
Parity Latency PL 4 4 4 5 nCK
CRC Error Reporting
CRC error to ALERT_n latency tCRC_ALERT 3 13 3 13 3 13 3 13 ns
CRC_ALERT_
CRC ALERT_n pulse width 6 10 6 10 6 10 6 10 nCK
PW
Geardown timing (Not Supported Below DDR4-2666)
Exit RESET from CKE HIGH to a

tXPR_GEAR NA NA NA NA
valid MRS geardown (T2/Reset)
CKE High Assert to Gear Down

tXS_GEAR NA NA NA NA
Enable time(T2/CKE)
MRS command to Sync pulse

tSYNC_GEAR NA NA NA NA 27
time(T3)
Sync pulse to First valid

tCMD_GEAR NA NA NA NA 27
command(T4)
Geardown setup time tGEAR_setup NA - NA - NA - NA - nCK
Geardown hold time tGEAR_hold NA - NA - NA - NA - nCK
tREFI
2Gb 160 - 160 - 160 - 160 - ns 34
4Gb 260 - 260 - 260 - 260 - ns 34

tRFC1 (min)
8Gb 350 - 350 - 350 - 350 - ns 34
16Gb 550 - 550 - 550 - 550 - ns 34
2Gb 110 - 110 - 110 - 110 - ns 34
4Gb 160 - 160 - 160 - 160 - ns 34

tRFC2 (min)
8Gb 260 - 260 - 260 - 260 - ns 34
16Gb 350 - 350 - 350 - 350 - ns 34
2Gb 90 - 90 - 90 - 90 - ns 34
4Gb 110 - 110 - 110 - 110 - ns 34

tRFC4 (min)
8Gb 160 - 160 - 160 - 160 - ns 34
16Gb 260 - 260 - 260 - 260 - ns 34

π PM Tech
Timing Parameters by Speed Bin for DDR4-2666 to DDR4-3200
PMA213816B
Speed DDR4-2666 DDR4-2933 DDR4-3200

Unit NOTE
Parameter Symbol MIN MAX MIN MAX MIN MAX
Clock Timing
Minimum Clock Cycle Time tCK

8 20 8 20 8 20 ns
(DLL off mode) (DLL_OFF)
Average Clock Period tCK(avg) 0.750 <0.833 0.682 <0.750 0.625 <0.682 ps 35,36
Average high pulse width tCH(avg) 0.48 0.52 0.48 0.52 0.48 0.52 tCK(avg)
Average low pulse width tCL(avg) 0.48 0.52 0.48 0.52 0.48 0.52 tCK(avg)
MIN = tCK (AVG) MIN + tJITper_tot MIN;

Absolute Clock Period tCK(abs) tCK(avg)
MAX = tCK (AVG) MAX + tJITper_tot MAX
Absolute clock HIGH pulse width tCH(abs) 0.45 - 0.45 - 0.45 - tCK(avg) 23
Absolute clock LOW pulse width tCL(abs) 0.45 - 0.45 - 0.45 - tCK(avg) 24
Clock Period Jitter- total JIT(per)_tot -38 38 -34 34 -32 32 ps 23
Clock Period Jitter- deterministic JIT(per)_dj -19 19 -17 17 -16 16 ps 26
Clock Period Jitter during DLL locking

tJIT(per, lck) -30 30 -27 27 -25 25 ps
period
Cycle to Cycle Period Jitter tJIT(cc)_total - 75 - 68 - 62 ps 25
Cycle to Cycle Period Jitter deterministic tJIT(cc)_dj - 38 - - - - ps 26
Cycle to Cycle Period Jitter during DLL

tJIT(cc, lck) - 60 - 55 - 50 ps
locking period
Cumulative error across 2 cycles tERR(2per) -55 55 -50 50 -46 46 ps
Cumulative error across n = 13, 14 . . . 49, tERRnper MIN = (1 + 0.68ln[n]) × tJITper_tot MIN
tERR(nper) UI
50 cycles tERRnper MAX = (1 + 0.68ln[n]) × tJITper_tot MAX
Command and Address setup time to

CK_t, CK_c referenced to Vih(ac) / Vil(ac) tIS(base, AC90) 55 - 48 - 40 - ps
levels
Command and Address setup time to

tIS(Vref) 145 - 138 - 130 - ps
CK_t, CK_c referenced to Vref levels
Command and Address hold time to CK_t,

tIH(base, DC65) 80 - 73 - 65 - ps
CK_c referenced to Vih(dc) / Vil(dc) levels
Command and Address hold time to CK_t,

tIH(Vref) 145 - 138 - 130 - ps
CK_c referenced to Vref levels
Control and Address Input pulse width for

tIPW 385 - 365 - 350 - ps
each input
CAS_n to CAS_n command delay for max(5nCK , max(5nCK , max(5nCK ,

tCCD_L 5ns)
-
5ns)
-
5ns)
- nCK 34
same bank group
CAS_n to CAS_n command delay for

tCCD_S 4 - 4 - 4 - nCK 34
different bank group
ACTIVATE to ACTIVATE Command delay tRRD_S Max(4nCK,5 Max(4nCK,5 Max(4nCK,5

- - - nCK 34
to different bank group for 2KB page size (2K) .3ns) .3ns) .3ns)

π PM Tech
PMA213816B
Unit NOTE
ACTIVATE to ACTIVATE Command delay tRRD_S Max(4nCK,3 Max(4nCK,2 Max(4nCK,2

- - - nCK 34
to different bank group for 2KB page size (1K) ns) .7ns) .5ns)
ACTIVATE to ACTIVATE Command delay

Max(4nCK,3 Max(4nCK,2 Max(4nCK,2
to different bank group for 1/2KB page tRRD_S(1/2K) ns)
-
.7ns)
-
.5ns)
- nCK 34
size
ACTIVATE to ACTIVATE Command delay Max(4nCK,6 Max(4nCK,6 Max(4nCK,6

tRRD_L(2K) .4ns)
-
.4ns)
-
.4ns)
- nCK 34
to same bank group for 2KB page size

tRRD_L(1K) .9ns)
-
.9ns)
-
.9ns)
- nCK 34
to same bank group for 1KB page size

tRRD_L(1/2K) .9ns)
-
.9ns)
-
.9ns)
- nCK 34
to same bank group for 1/2KB page size
Max(28nCK, Max(28nCK, Max(28nCK,

Four activate window for 2KB page size tFAW_2K 30ns)
-
30ns)
-
30ns)
- ns 34

Four activate window for 1KB page size tFAW_1K 21ns)
-
21ns)
-
21ns)
- ns 34

Four activate window for 1/2KB page size tFAW_1/2K 12ns)
-
10.875ns)
-
10ns)
- ns 34

max(2nCK,2 max(2nCK,2 max(2nCK,2
transaction to internal read command for tWTR_S .5ns)
-
.5ns)
-
.5ns)
- ns 1,2,34
different bank group
transaction to internal read command for tWTR_L .5ns)
-
.5ns)
-
.5ns)
- ns 1,34
same bank group
Internal READ Command to PRECHARGE max(4nCK,7 max(4nCK,7 max(4nCK,7

tRTP .5ns)
-
.5ns)
-
.5ns)
- ns 34
Command delay
WRITE recovery time tWR 15 - 15 - 15 - ns 1
tWR+max tWR+max tWR+max

Write recovery time when CRC and DM
tWR_CRC _DM (4nCK,3.75n - (5nCK,3.75n - (5nCK,3.75n - ns 1,28
are enabled s) s) s)

tWTR_S+ tWTR_S+ tWTR_S+
transaction to internal read command for tWTR_S_C
max(4nCK, - max(5nCK, - max(5nCK, - ns 2,29,34
different bank group with both CRC and RC_DM 3.75ns) 3.75ns) 3.75ns)
DM enabled
tWTR_L+ tWTR_L+ tWTR_L+
transaction to internal read command for tWTR_L_C
max(5nCK, - max(5nCK, - max(5nCK, - ns 3,30,34
same bank group with both CRC and DM RC_DM 3.75ns) 3.75ns) 3.75ns)
enabled
DLL locking time tDLLK 1024 - 1024 - 1024 - nCK
Mode Register Set command cycle time tMRD 8 - 8 - 8 - nCK
max(24nCK, max(24nCK, max(24nCK,

Mode Register Set command update delay tMOD 15ns)
-
15ns)
-
15ns)
- nCK
Multi-Purpose Register Recovery Time tMPRR 1 - 1 - 1 - nCK 33
Multi Purpose Register Write Recovery tMOD (min) tMOD (min) tMOD (min)
tWR_MPR + AL + PL
-
+ AL + PL
-
+ AL + PL
- nCK
Time
Auto precharge write recovery + precharge

tDAL(min) Programmed WR + roundup ( tRP / tCK(avg)) nCK
time
DQ0 or DQL0 driven to 0 set-up time to

tPDA_S 0.5 - 0.5 - 0.5 - UI 45,47
first DQS rising edge
DQ0 or DQL0 driven to 0 hold time from

tPDA_H 0.5 - 0.5 - 0.5 - UI 46,47
last DQS fall-ing edge


Unit NOTE

CS_n to Command Address Latency tCAL .748 ns)
-
.748 ns)
-
.748 ns)
- nCK
tMOD+ tMOD+ tMOD+

Mode Register Set cyce time in CAL mode tMRD_tCAL tCAL
-
tCAL
-
tCAL
- nCK
Mode Register Set update delay in CAL tMOD+ tMOD+ tMOD+

tMOD_tCAL tCAL
-
tCAL
-
tCAL
- nCK
mode
DRAM Data Timing
DQS_t,DQS_c to DQ skew, per group, per tCK(avg)/ 13,18,39,

tDQSQ - 0.18 - 0.19 - 0.22 2 49
access
DQ output hold per group, per access from tCK(avg)/ 13,17,18,

tQH 0.74 - 0.74 - 0.74 - 2 39,49
DQS_t,DQS_c
Data Valid Window per device: (tQH - 17,18,39,

tDVWd TBD - TBD - TBD - UI
49
tDQSQ) of each UI on a given DRAM
Data Valid Window , per pin per UI : (tQH - 17,18,39,

tDVWp 0.72 - 0.72 - 0.72 - UI
49
tDQSQ) each UI on a pin of a given DRAM
DQ low impedance time from CK_t, CK_c tLZ(DQ) -310 170 -280 165 -250 160 ps 39
DQ high impedance time from CK_t, CK_c tHZ(DQ) - 170 - 165 - 160 ps 39
Data Strobe Timing
0.9 NOTE 44 0.9 NOTE 44 0.9 NOTE 44 tCK 39,40

tRPRE
Preamble
1.8 NOTE 44 1.8 NOTE 44 NA NA tCK 39,41

tRPST 0.33 NOTE 45 0.33 NOTE 45 0.33 NOTE 45 tCK 39
Postamble
DQS_t,DQS_c differential output high time tQSH 0.4 - 0.4 - 0.4 - tCK 21,39
DQS_t,DQS_c differential output low time tQSL 0.4 - 0.4 - 0.4 - tCK 20,39
0.9 - 0.9 - 0.9 - tCK 42

tWPRE
Preamble
1.8 - 1.8 - 1.8 - tCK 43

tWPST 0.33 - 0.33 - 0.33 - tCK
Postamble
DQS_t and DQS_c low-impedance time

tLZ(DQS) -310 170 -280 165 -250 160 ps 39
(Referenced from RL-1)
DQS_t and DQS_c high-impedance time

tHZ(DQS) - 170 - 165 - 160 ps 39
(Referenced from RL+BL/2)
DQS_t, DQS_c differential input low pulse

tDQSL 0.46 0.54 0.46 0.54 0.46 0.54 tCK
width
DQS_t, DQS_c differential input high pulse

tDQSH 0.46 0.54 0.46 0.54 0.46 0.54 tCK
width
DQS_t, DQS_c rising edge to CK_t, CK_c

tDQSS -0.27 0.27 -0.27 0.27 -0.27 0.27 tCK
rising edge (1 clock preamble)
DQS_t, DQS_c falling edge setup time to

tDSS 0.18 - 0.18 - 0.18 - tCK
CK_t, CK_c rising edge
DQS_t, DQS_c falling edge hold time from

tDSH 0.18 - 0.18 - 0.18 - tCK
CK_t, CK_c rising edge
DQS_t, DQS_c rising edge output timing

tDQSCK
lo-catino from rising CK_t, CK_c with DLL 170 170 -165 165 -160 160 ps 37,38,39
(DLL On)
On mode
DQS_t, DQS_c rising edge output tDQSCKI

- 270 - 265 - 260 ps 37,38,39
variance window per DRAM (DLL On)


Unit NOTE
MPSM Timing
tMOD(min) + tMOD(min) + tMOD(min) +

Command path disable delay upon MPSM
tMPED tCPDED - tCPDED - tCPDED -
entry (min) (min) (min)
tMOD(min) + tMOD(min) + tMOD(min) +

Valid clock requirement after MPSM entry tCKMPE tCPDED - tCPDED - tCPDED -
(min) (min) (min)
tCKSRX tCKSRX tCKSRX

Valid clock requirement before MPSM exit tCKMPX (min)
-
(min)
-
(min)
-
Exit MPSM to commands not requiring a

tXMP tXS(min) - tXS(min) - tXS(min) -
locked DLL
tXMP(min) + tXMP(min) + tXMP(min) +

Exit MPSM to commands requiring a
tXMPDLL tXSDLL - tXSDLL - tXSDLL -
locked DLL (min) (min) (min)
tISmin + tISmin + tISmin +

CS setup time to CKE tMPX_S tIHmin
-
tIHmin
-
tIHmin
-
Calibration Timing
Power-up and RESET calibration time tZQinit 1024 - 1024 - 1024 - nCK
Normal operation Full calibration time tZQoper 512 - 512 - 512 - nCK
Normal operation Short calibration time tZQCS 128 - 128 - 128 - nCK
Reset/Self Refresh Timing
Max(5nCK,t Max(5nCK,t Max(5nCK,t

Exit Reset from CKE HIGH to a valid
tXPR RFC(min)+1 - RFC(min)+1 - RFC(min)+1 - nCK
command 0ns) 0ns) 0ns)
Exit Self Refresh to commands not tRFC(min)+1 tRFC(min)+1 tRFC(min)+1

tXS 0ns
-
0ns
-
0ns
- nCK
SRX to commands not requiring a locked tXS_ABORT tRFC4(min)+ tRFC4(min)+ tRFC4(min)+

- - - nCK
DLL in Self Refresh ABORT (min) 10ns 10ns 10ns
Exit Self Refresh to ZQCL,ZQCS and MRS tRFC4(min)+ tRFC4(min)+ tRFC4(min)+

tXS_FAST (min) 10ns
-
10ns
-
10ns
- nCK
(CL,CWL,WR,RTP and Gear Down)
Exit Self Refresh to commands requiring a tDLLK tDLLK tDLLK

tXSDLL (min)
-
(min)
-
(min)
- nCK
locked DLL
Minimum CKE low width for Self refresh tCKE(min)+ tCKE(min)+ tCKE(min)+
tCKESR 1nCK
-
1nCK
-
1nCK
- nCK
entry to exit timing
Minimum CKE low width for Self refresh tCKE(min)+ tCKE(min)+ tCKE(min)+
tCKESR_PAR 1nCK+PL
-
1nCK+PL
-
1nCK+PL
- nCK
entry to exit timing with CA Parity enabled

Refresh Entry (SRE) or Power-Down Entry tCKSRE 0ns)
-
0ns)
-
0ns)
- nCK
(PDE)
Refresh Entry (SRE) or Power-Down when tCKSRE_PAR 0ns)+PL
-
0ns)+PL
-
0ns)+PL
- nCK
CA Parity is enabled
Valid Clock Requirement before Self
Refresh Exit (SRX) or Power-Down Exit tCKSRX 0ns)
-
0ns)
-
0ns)
- nCK
(PDX) or Reset Exit
Power Down Timing
Exit Power Down with DLL on to any valid

command;Exit Precharge Power Down max(4nCK,6 max(4nCK,6 max(4nCK,6
tXP ns)
-
ns)
-
ns)
- nCK
with DLL frozen to commands not
max(3nCK, max(3nCK, max(3nCK,

CKE minimum pulse width tCKE 5ns)
-
5ns)
-
5ns)
- nCK 31,32
Command pass disable delay tCPDED 4 - 4 - 4 - nCK
Power Down Entry to Exit Timing tPD tCKE(min) 9*tREFI tCKE(min) 9*tREFI tCKE(min) 9*tREFI nCK 6
Timing of ACT command to Power Down

tACTPDEN 2 - 2 - 2 - nCK 7
entry


Unit NOTE
Power Down Timing
Timing of PRE or PREA command to

tPRPDEN 2 - 2 - 2 - nCK 7
Power Down entry
Timing of RD/RDA command to Power

tRDPDEN RL+4+1 - RL+4+1 - RL+4+1 - nCK
Down entry
Timing of WR command to Power Down WL+4+(tWR WL+4+(tWR WL+4+(tWR

tWRPDEN /tCK(avg))
-
/tCK(avg))
-
/tCK(avg)) - nCK 4
entry (BL8OTF, BL8MRS, C4OTF)
Timing of WRA command to Power Down WL+4+WR+ WL+4+WR+ WL+4+WR+

tWRAPDEN 1
-
1
-
1 - nCK 5
entry (BL8OTF, BL8MRS, BC4OTF)
Timing of WR command to WL+2+(tWR WL+2+(tWR WL+2+(tWR

tWRPBC4DEN /tCK(avg))
-
/tCK(avg))
-
/tCK(avg))
- nCK 4
Power Down entry (BC4MRS)
Timing of WRA command to Power Down WL+2+WR+ WL+2+WR+ WL+2+WR+

tWRAPBC4DEN 1
-
1
-
1
- nCK 5
entry (BC4MRS)
Timing of REF command to Power Down

tREFPDEN 2 - 2 - 2 - nCK 7
entry
Timing of MRS command to Power Down tMOD tMOD tMOD

tMRSPDEN (min)
-
(min)
-
(min)
- nCK
entry
PDA Timing
Mode Register Set command cycle time in max(16nCK, max(16nCK, max(16nCK,

tMRD_PDA 10ns)
-
10ns)
-
10ns)
- nCK
PDA mode
Mode Register Set command update delay

tMOD_PDA tMOD tMOD tMOD nCK
in PDA mode
ODT Timing
Asynchronous RTT turn-on delay (Power-

tAONAS 1.0 9.0 1.0 9.0 1.0 9.0 ns
Down with DLL frozen)
Asynchronous RTT turn-off delay (Power-

tAOFAS 1.0 9.0 1.0 9.0 1.0 9.0 ns
Down with DLL frozen)
RTT dynamic change skew tADC 0.28 0.72 0.26 0.74 0.26 0.74 tCK(avg)
Write Leveling Timing
First DQS_t/DQS_n rising edge after write

tWLMRD 40 - 40 - 40 - nCK 12
leveling mode is programmed
DQS_t/DQS_n delay after write leveling

tWLDQSEN 25 - 25 - 25 - nCK 12
mode is programmed
Write leveling setup time from rising CK_t,

CK_c crossing to rising DQS_t/DQS_n tWLS 0.13 - 0.13 - 0.13 - tCK(avg)
crossing
Write leveling hold time from rising DQS_t/
DQS_n crossing to rising CK_t, CK_ tWLH 0.13 - 0.13 - 0.13 - tCK(avg)
crossing
Write leveling output delay tWLO 0 9.5 0 9.5 0 9.5 ns
Write leveling output error tWLOE 0 2 0 2 0 2 ns
CA Parity Timing
Commands not guaranteed to be executed tPAR_

- PL - PL - PL nCK
during this time UNKNOWN
Delay from errant command to ALERT_n

tPAR_ALERT_ON - PL+6ns - PL+6ns - PL+6ns nCK
assertion
Pulse width of ALERT_n signal when tPAR_ALERT_P

80 160 88 176 96 192 nCK
asserted W
Time from when Alert is asserted till

tPAR_ALERT_RS
controller must start providing DES - 71 - 78 - 85 nCK
P
commands in Persistent CA parity mode
Parity Latency PL 5 6 6 nCK


Unit NOTE
CRC Error Reporting
CRC error to ALERT_n latency tCRC_ALERT 3 13 3 13 3 13 ns
CRC ALERT_n pulse width CRC_ALERT_PW 6 10 6 10 6 10 nCK
Geardown timing (Not Supported Below DDR4-2666)
Exit RESET from CKE HIGH to a valid

tXPR_GEAR tXPR tXPR tXPR
MRS geardown (T2/Reset)
CKE High Assert to Gear Down Enable

tXS_GEAR tXS tXS tXS
time(T2/CKE)
MRS command to Sync pulse time(T3) tSYNC_GEAR tMOD + 4CK tMOD + 4CK tMOD + 4CK 27
Sync pulse to First valid command(T4) tCMD_GEAR tMOD tMOD tMOD 27
Geardown setup time tGEAR_setup 2CK - 2CK - 2CK - nCK
Geardown hold time tGEAR_hold 2CK - 2CK - 2CK - nCK
tREFI
2Gb 160 - 160 - 160 - ns 34
4Gb 260 - 260 - 260 - ns 34

tRFC1 (min)
8Gb 350 - 350 - 350 - ns 34
16Gb 550 - 550 - 550 - ns 34
2Gb 110 - 110 - 110 - ns 34
4Gb 160 - 160 - 160 - ns 34

tRFC2 (min)
8Gb 260 - 260 - 260 - ns 34
16Gb 350 - 350 - 350 - ns 34
2Gb 90 - 90 - 90 - ns 34
4Gb 110 - 110 - 110 - ns 34

tRFC4 (min)
8Gb 160 - 160 - 160 - ns 34
16Gb 260 - 260 - 260 - ns 34

π
NOTE :
PM Tech PMA213816B
1. Start of internal write transaction is defined as follows : For BL8 (Fixed by MRS and on-the-fly) : Rising clock edge 4 clock cycles
after WL. For BC4 (on-the-fly) : Rising clock edge 4 clock cycles after WL. For BC4 (fixed by MRS) : Rising clock edge 2 clock
cycles after WL.
2. A separate timing parameter will cover the delay from write to read when CRC and DM are simultaneously enabled.
3. Commands requiring a locked DLL are: READ (and RAP) and synchronous ODT commands.
4. tWR is defined in ns, for calculation of tWRPDEN it is necessary to round up tWR/tCK to the next integer.
5. WR in clock cycles as programmed in MR0.
6. tREFI depends on TOPER.
7. CKE is allowed to be registered low while operations such as row activation, precharge, autoprecharge or refresh are in progress,
but power-down IDD spec will not be applied until finishing those operations.
8. For these parameters, the DDR4 SDRAM device supports tnPARAM[nCK]=RU{tPARAM[ns]/tCK(avg)[ns]}, which is in clock cycles
assuming all input clock jitter specifications are satisfied.
9. When CRC and DM are both enabled, tWR_CRC_DM is used in place of tWR.
10. When CRC and DM are both enabled tWTR_S_CRC_DM is used in place of tWTR_S.
11. When CRC and DM are both enabled tWTR_L_CRC_DM is used in place of tWTR_L.
12. The max values are system dependent.
13. DQ to DQS total timing per group where the total includes the sum of deterministic and random timing terms for a specified BER.
BER spec and measurement method are TBD.
14. The deterministic component of the total timing.
15. DQ to DQ static offset relative to strobe per group.
16. This parameter will be characterized and guaranteed by design.
17. When the device is operated with the input clock jitter, this parameter needs to be derated by the actual tjit(per)_total of the input
clock. (output deratings are relative to the SDRAM input clock).
18. DRAM DBI mode is off.
19. DRAM DBI mode is enabled. Applicable to x8 and x16 DRAM only.
20. tQSL describes the instantaneous differential output low pulse width on DQS_t - DQS_c, as measured from on falling edge to the
next consecutive rising edge
21. tQSH describes the instantaneous differential output high pulse width on DQS_t - DQS_c, as measured from on falling edge to the
next consecutive rising edge
22. There is no maximum cycle time limit besides the need to satisfy the refresh interval tREFI
23. tCH(abs) is the absolute instantaneous clock high pulse width, as measured from one rising edge to the following falling edge
24. tCL(abs) is the absolute instantaneous clock low pulse width, as measured from one falling edge to the following rising edge
25. Total jitter includes the sum of deterministic and random jitter terms for a specified BER. BER target and measurement method are
TBD.
26. The deterministic jitter component out of the total jitter. This parameter is characterized and gauranteed by design.
27. This parameter has to be even number of clocks.
28. When CRC and DM are both enabled, tWR_CRC_DM is used in place of tWR.
29. When CRC and DM are both enabled tWTR_S_CRC_DM is used in place of tWTR_S.
30. When CRC and DM are both enabled tWTR_L_CRC_DM is used in place of tWTR_L.
31. After CKE is registered LOW, CKE signal level shall be maintained below VILDC for tCKE specification (Low pulse width).
32. After CKE is registered HIGH, CKE signal level shall be maintained above VIHDC for tCKE specification (HIGH pulse width).
33. Defined between end of MPR read burst and MRS which reloads MPR or disables MPR function.
34. Parameters apply from tCK(avg)min to tCK(avg)max at all standard JEDEC clock period values as stated in the Speed Bin Tables.
35. This parameter must keep consistency with Speed-Bin Tables.
36. DDR4-1600 AC timing apply if DRAM operates at lower than 1600 MT/s data rate. UI=tCK(avg).min/2
37. applied when DRAM is in DLL ON mode.
38. Assume no jitter on input clock signals to the DRAM.
39. Value is only valid for RZQ/7 RONNOM = 34 ohms.

π PM Tech
40. 1tCK toggle mode with setting MR4:A11 to 0.
PMA213816B
41. 2tCK toggle mode with setting MR4:A11 to 1, which is valid for DDR4-2400/2666/3200 speed grade.
42. 1tCK mode with setting MR4:A12 to 0.
43. 2tCK mode with setting MR4:A12 to 1, which is valid for DDR4-2400/2666/3200 speed grade.
44. The maximum read preamble is bounded by tLZ(DQS)min on the left side and tDQSCK(max) on the right side.
45. DQ falling signal middle-point of transferring from High to Low to first rising edge of DQS diff-signal cross-point.
46. last falling edge of DQS diff-signal cross-point to DQ rising signal middle-point of transferring from Low to High.
47. VrefDQ value must be set to either its midpoint or Vcent_DQ(midpoint) in order to capture DQ0 or DQL0 low level for entering PDA
mode.
48. The maximum read postamble is bound by tDQSCK(min) plus tQSH(min) on the left side and tHZ(DQS)max on the right side.
49. Reference level of DQ output signal is specified with a midpoint as a widest part of Output signal eye which should be approximately
0.7 * VDDQ as a center level of the static single-ended output peak-to-peak swing with a driver impedance of 34 ohms and an
effective test load of 50 ohms to VTT = VDDQ.

π PM Tech
The DQ input receiver compliance mask for voltage and timing
PMA213816B
The DQ input receiver compliance mask for voltage and timing is shown in the figure below. The receiver
mask (Rx Mask) defines area the input signal must not encroach in order for the DRAM input receiver to be
expected to be able to successfully capture a valid input signal; it is not the valid data-eye.
DQ Receiver(Rx) compliance mask
Across pin Vref DQ voltage variation
The Vref_DQ voltage is an internal reference voltage level that shall be set to the properly trained setting,
which is generally Vcent_DQ(midpoint), in order to have valid Rx Mask values. Vcent_DQ is defined as the
midpoint between the largest Vref_DQ voltage level and the smallest Vref_DQ voltage level across all DQ
pins for a given DDR4 DRAM component. Each DQ pin Vref level is defined by the center, i.e. widest
opening, of the cumulative data input eye as depicted in Figure “DQ Receiver(Rx) compliance mask”. This
clarifies that any DDR4 DRAM component level variation must be accounted for within the DDR4 DRAM Rx
mask.The component level Vref will be set by the system to account for Ron and ODT settings.

π PM Tech
DQS to DQ and DQ to DQ Timings at DRAM Balls
PMA213816B
DQS, DQs Data-in at DRAM Ball DQS, DQs Data-in at DRAM Ball
Rx Mask Rx Mask - Alternative View
NOTE : NOTE :
DQx represents an optimally centered mask. DRAMa represents a DRAM without any DQS/DQ skews.
DQy represents earliest valid mask. DRAMb represents a DRAM with early skews (negative tDQS2DQ).
NOTE : Figures show skew allowed between DRAM to DRAM and DQ to DQ for a DRAM. Signals assume data centered aligned at
DRAM Latch. TdiPW is not shown; composite data-eyes shown would violate TdiPW. VCENT DQ(midpoint) is not shown but is
assummed to be midpoint of VdiVW.
All of the timing terms in the previous figure are measured at the VdIVW_total voltage levels centered around
Vcent_DQ(midpoint) and are referenced to the DQS_t/DQS_c center aligned to the DQ per pin.

The rising edge slew rates are defined by srr1 and srr2. The slew rate measurement points for a rising edge
are shown in Figure 5A below: A low to high transition tr1 is measured from 0.5*VdiVW(max) below
Vcent_DQ(midpoint) to the last transition through 0.5*VdiVW(max) above Vcent_DQ(midpoint) while tr2 is
measured from the last transition through 0.5*VdiVW(max) above Vcent_DQ(midpoint) to the first transition
through the 0.5*VIHL_AC(min) above Vcent_DQ(midpoint).
Rising edge slew rate equations:
srr1 = VdIVW(max) / tr1
srr2 = (VIHL_AC(min) – VdIVW(max)) / (2*tr2)
Slew Rate Conditions For Rising Transition
The falling edge slew rates are defined by srf1 and srf2. The slew rate measurement points for a falling edge
are shown in Figure below: A high to low transition tf1 is measured from 0.5*VdiVW(max) above
Vcent_DQ(midpoint) to the last transition through 0.5*VdiVW(max) below Vcent_DQ(midpoint) while tf2 is
measured from the last transition through 0.5*VdiVW(max) below Vcent_DQ(midpoint) to the first transition
through the 0.5*VIHL_AC(min) below Vcent_DQ(pin mid).
Falling edge slew rate equations:
srf1 = VdIVW(max) / tf1
srf2 = (VIHL_AC(min) – VdIVW(max)) / (2*tf2)
Slew Rate Conditions For Falling Transition

π PM Tech
DRAM DQs In Receive Mode; * UI=tck(avg)min/2
PMA213816B
DDR4-1600/
DDR4-2400 DDR4-2666 DDR4-2933 DDR4-3200
Symbol Parameter 1866/2133 Unit Note
Min Max Min Max Min Max Min Max Min Max
VdiVW Rx Mask voltage - pk-pk - 136 - 130 - 120 - 115 - 110 mV 2,3
TdiVW Rx timing window - 0.2 - 0.2 - 0.22 - 0.23 - 0.23 UI 2,3
VIHL_AC DQ AC input swing pk-pk 186 - 160 - 150 - 145 - 140 - mV 4,5
TdiPW DQ input pulse width 0.58 - 0.58 - 0.58 - 0.58 - 0.58 - UI 6
tDQS2DQ Rx Mask DQS to DQ offset -0.17 0.17 -0.17 0.17 -0.19 0.19 -0.22 0.22 -0.22 0.22 UI 7
tDQ2DQ Rx Mask DQ to DQ offset - TBD - TBD - 0.115 - 0.115 - 0.125 UI 8
Input Slew Rate over

1 9 1 9 1 9 1 9 1 9 V/ns 9
Vdlvw if tCK >=0.935ns
srr1, srf1
Input Slew Rate over
- - 1.25 9 1.25 9 1.25 9 1.25 9 V/ns 9
Vdivw if tCK>=0.625ns
Rising Input Slew Rate 0.2 x 0.2 x 0.2 x 0.2 x 0.2 x

srr2 9 9 9 9 9 V/ns 10
over 1/2 VIHL_AC srr1 srr1 srr1 srr1 srr1
Falling Input Slew Rate 0.2 x 0.2 x 0.2 x 0.2 x 0.2 x

srf2 9 9 9 9 9 V/ns 10
over 1/2 VIHL_AC srf1 srf1 srf1 srf1 srf1
NOTE
1. All Rx mask specifications must be satisfied for each UI. For example, if the minimum input pulse width is violated when satisfying
TdiVW (MIN), VdiVW,max, and minimum slew rate limits, then either TdiVW (MIN) or minimum slew rates would have to be
increased to the point where the minimum input pulse width would no longer be violated.
2. Data Rx mask voltage and timing total input valid window where VdiVW is centered around VCENTDQ,midpoint after VREFDQ
training is completed. The data Rx mask is applied per bit and should include voltage and temperature drift terms. The input buffer
design specification is to achieve at least a BER =1e- 16 when the Rx mask is not violated.
3. Defined over the DQ internal VREF range 1.
4. Overshoot and undershoot specifications apply.
5. DQ input pulse signal swing into the receiver must meet or exceed VIHL(AC)min. VIHL(AC)min is to be achieved on an UI basis
when a rising and falling edge occur in the same UI (a valid TdiPW).
6. DQ minimum input pulse width defined at the VCENTDQ,midpoint.
7. DQS-to-DQ Rx mask offset is skew between DQS and DQ within a nibble (x4) or word (x8, x16 [for x16, the upper and lower bytes
are treated as separate x8s]) at the SDRAM balls over process, voltage, and temperature.
8. DQ-to-DQ Rx mask offset is skew between DQs within a nibble (x4) or word (x8, x16) at the SDRAM balls for a given component
over process, voltage, and temperature.
9. Input slew rate over VdiVW mask centered at VCENTDQ,midpoint. Slowest DQ slew rate to fastest DQ slew rate per transition edge
must be within 1.7V/ns of each other.
10. Input slew rate between VdiVW mask edge and VIHL(AC)min points.

π PM Tech
DDR4 Function Matrix
PMA213816B
DDR4 SDRAM has several features supported by ORG and also by Speed. The following Table is the
summary of the features.
Function Matrix (By ORG. V:Supported, Blank:Not supported)
Functions X4 X8 X16
Write Leveling V V V
Temperature controlled Refresh V V V
Low Power Auto Self Refresh V V V
Fine Granularity Refresh V V V
Multi Purpose Register V V V
Data Mask V V
Data Bus Inversion V V
TDQS V
ZQ calibration V V V
DQ Vref Training V V V
Per DRAM Addressability V V V
Mode Register Readout V V V
CAL V V V
WRITE CRC V V V
CA Parity V V V
Control Gear Down Mode V V V
Programmable Preamble V V V
Maximum Power Down Mode V V
Boundary Scan Mode V
Additive Latency V V

π PM Tech
Revision History
PMA213816B
Revision Date Page Notes

1.0 April, 2020 - Official release
1.1 September, 2020 4 Industrial grade added

PM Tech: Document Title

Uploaded by

Copyright:

Available Formats

PM Tech: Document Title

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

PM Tech: Document Title

Uploaded by

Copyright:

Available Formats

π PM Tech PMA212816C

4GBIT DDR4 DRAM

1600 1866 2133 2400 2666 2933 3200

tCK(min) 1.25 1.071 0.937 0.833 0.75 0.682 0.625 ns

CAS Latency 11 13 15 17 19 20 22 nCK

tRCD(min) 13.75 13.92 14.06 14.16 14.25 13.64 13.75 ns

tRP(min) 13.75 13.92 14.06 14.16 14.25 13.64 13.75 ns

tRC(min) 48.75 47.92 47.06 46.16 46.25 45.64 45.75 ns

Rev 1.1 1 September 2020

Organization 1024Mbx4 512Mb x 8 256Mb x 16

Bank group address BG[1:0] BG[1:0] BG0

Bank per group 4 4 4

Bank address BA[1:0] BA[1:0] BA[1:0]

Row address 64K (A[15:0]) 32K (A[14:0]) 32K (A[14:0])

Column address 1K (A[9:0]) 1K (A[9:0]) 1K (A[9:0])

Page Size 512B 1KB 2KB

tREFI Tc≦ 85°C:7.8μs, Tc>85°C:3.9μs

4Gb Dimension Ball pitch

Rev 1.1 2 September 2020

Rev 1.1 3 September 2020

Package Ballout and Addressing

C VDDQ DQ0 DQS_t VDD VSS VDDQ C

D VSSQ NC DQ2 DQ3 NC VSSQ D

E VSS VDDQ NC NC VDDQ VSS E

F VDD NC ODT CK_t CK_c VDD F

G VSS NC CKE CS_n NC TEN G

N VSS A11 PAR NC A13 VDD N

Rev 1.1 4 September 2020

C VDDQ DQ0 DQS_t VDD VSS VDDQ C

D VSSQ DQ4 DQ2 DQ3 DQ5 VSSQ D

E VSS VDDQ DQ6 DQ7 VDDQ VSS E

F VDD NC ODT CK_t CK_c VDD F

G VSS NC CKE CS_n NC TEN G

N VSS A11 PAR NC A13 VDD N

Rev 1.1 5 September 2020

A VDDQ VSSQ DQU0 DQSU_c VSSQ VDDQ A

B VPP VSS VDD DQSU_t DQ9 VDD B

C VDDQ DQ12 DQ10 DQ11 DQ13 VSSQ C

D VDD VSSQ DQU6 DQ15 VSSQ VDDQ D

G VDDQ DQ0 DQSL_t VDD VSS VDDQ G

H VSSQ DQ4 DQ2 DQ3 DQ5 VSSQ H

T VSS A11 PAR NC A13 VDD T

Rev 1.1 6 September 2020

Rev 1.1 7 September 2020

Symbol Type Function

On Die Termination: ODT (registered HIGH) enables RTT_NOM termination

Command Inputs: RAS_n/A16, CAS_n/A15 and WE_n/A14 (along with CS_n)

Rev 1.1 8 September 2020

Symbol Type Function

Auto-precharge: A10 is sampled during Read/Write commands to determine

Termination Data Strobe: TDQS_t/TDQS_c is applicable for x8 DRAMs only.

Rev 1.1 9 September 2020

Symbol Type Function

NC No Connect: No internal electrical connection is present.

VDDQ Supply DQ Power Supply: 1.2 V +/- 0.06 V

VSSQ Supply DQ Ground

VDD Supply Power Supply: 1.2 V +/- 0.06 V

VSS Supply Ground

VREFCA Supply Reference voltage for CA