1 10 Optimization PDF

Optimization
Training objectives:
• You are able to optimize the
current- and speed controller
- in the time range
- in the frequency range
SINAMICS Date:
File:
06.03.2007
ABU DHABI: Page 10 - 1
SITRAIN Training for
Automation and Drives
Siemens AG 2007. All rights reserved.
Optimization
Content Page
Scheme of closed loop control (example: feed drive) ....................................................................... 4
Structure of a speed control loop...................................................................................................... 5
Main objectives of (speed-)controller optimization............................................................................. 6
Cascade control and its evaluation................................................................................................... 7
Quick-optimization of a PI controller................................................................................................. 8
Transient function / step response.................................................................................................... 9
Application-specific optimization schemes........................................................................................ 10
Current controller optimization.......................................................................................................... 11
Optimization of a speed controller with PI-characteristic................................................................... 12
Transfer function and frequency response........................................................................................ 13
Amplitude- and phase-response....................................................................................................... 14
Bode diagram of the setpoint value frequency response................................................................... 15
Recording the transfer function of the open control loop.................................................................... 16
Phase- and Amplitude-reserve (open control loop)........................................................................... 17
Evaluation of the control property (closed control loop)..................................................................... 18
Step response in the frequency and time range................................................................................ 19
Bode diagram of a 2-mass system (ONLY controlled system).......................................................... 20
continued on page 3
SINAMICS Date:
File:
06.03.2007
Optimization
Content Page
Frequency response (multi-mass system / ONLY controlled system)................................................ 21
Multi-mass system / closed control loop........................................................................................... 22
Characteristics of low pass and band stop filters.............................................................................. 26
Evaluation of the filter criteria........................................................................................................... 27
Current setpoint filter / at resonance of the speed control loop.......................................................... 30
Setpoint frequency response in the speed control loop..................................................................... 31
Speed setpoint filter / at resonance of the position control loop......................................................... 32
Measuring functions......................................................................................................................... 33
Modes of application of measuring functions.................................................................................... 34
Current controller optimization.......................................................................................................... 35
Speed controller optimization........................................................................................................... 36
Reference model for the speed control loop (Servo + Vector)........................................................... 37
Settings for the reference model....................................................................................................... 38
Reference model and symmetry filter............................................................................................... 39
SINAMICS Date:
File:
06.03.2007
Scheme of closed loop control (example: feed drive)

Torque pre control
Speed pre control
Power unit
Position Speed Current Motor Encoder
xset nset Iset PWM
M G
- xact - nact - Iact

Current actual value
IU
Iq Iα
Motor- 2 IV ωM
Id Iβ
model 3 IW
ϕM
Speed Rotor position Encoder evaluation

Position
KV...Gain KP... Gain KP... Gain

TN... Integral time TN... Integral time
Position Speed Current Current Torque Speed Angle

(Motor + Load)
nv Iv cM, KT
1/L ML 1/GGes
Xset nset Iset uA iR MM ωM ϕM
– – – –
Xact nact Iact EMF
IR
KE
ωM
ϕM
SINAMICS Date:
File:
06.03.2007
Structure of a speed control loop
Controller Controlled system
Disturbance
Setpoint value
Manipulated I.set
n.set value
Speed
controller Current Power
controller unit
n.act
Measuring
Control loop transducer
Measuring Controlled value

transducer T Motor Load
Feedback
SINAMICS Date:
File:
06.03.2007
Main objectives of (speed-)controller optimization

Dynamics is defined by
• short rise time in time range
• large bandwidth in frequency range
is obtained by
• large controller gain k.P
• short cycle time t.cycle
Stability is defined by
• small overshoot in time range
• large phase reserve in frequency range
is obtained by
• small controller gain k.P
• large rise time T.N
Accuracy is defined by
• small following error in time range
• large controller gain at low frequencies in frequency range
is obtained by
• large controller gain k.P
• dynamically acting integral action controller property; small rise time T.N
Robustness Is the ability to provide both stability and dynamic response despite changes of
the controlled system (aging, change of temperature).
Robustness requires to compromise dynamic performance, stability and accuracy.
SINAMICS Date:
File:
06.03.2007
Cascade control and its evaluation
Advantages of cascade control:
• Step by step commissioning of the control loops from the inside to the outside (Current Æ Speed Æ Position) ;
independent settings.
• Lower-level controllers act as substitute delay time for the superposed system and in this way affect the dynamic
properties of the whole system in respect to setpoint response.
This disadvantage can be compensated by implementing pre control on the lower-level control loops.
• Easy to do limiting of the lower-level controlled value by limiting the superposed control setpoint value.
• Effects of non-linearity are limited. The superposed controller features reduced non-linearity.
• Reduction of delay times via compensation of the lower-level control loop.
• In the lower-level controller disturbances are immediately corrected and compensated very quickly.
There are many possibilities to evaluate the control response of a control system:
Methods to describe the dynamic response characteristic
Time range Frequency range
Description by means of Description by means of frequency response

differential equation
Description by means of Frequency response Bode diagram

step response equation (formulas) (graphical)
SINAMICS Date:
File:
06.03.2007
Quick-optimization of a PI controller
Input (Setpoint) Output (Actual value)

PI controller
Manip. Actual Oscillation
Setpoint value value
System
KP... Gain
TN... Integral time
t t
Set the integral time to a very Reduce KP to half

Increase KP
large value (i.e. 1000ms) of that value
until controller Set TN to the double
Input a constant input signal oscillates of that value
Now reduce TN
(1..5% of maximum value) to (evaluate the
until the controller
optimize at a friction-optimal manipulated
oscillates again
operating point value)
VERY rough procedure
SINAMICS Date:
File:
06.03.2007
Transient function / step response
x
Overshoot
speed
Tolerance band
Trise t torque
Trecovery
Output (actual value)

too Increase the gain KP
slow and repeat measurement
t
1 Set integral time TN to a Output (actual value) 1) KP is optimal
large value, i.e. 1000ms
2) Now decrease TN
(or define P-controller)
Kp Tn (TN ≈ rise time)
2 Input a step of reference only one undershoot;
to allow only one
3 The manipulated value in consequence an
t undershoot overshoot of 20–50 %
must not go into limit
Tan t
too
Decrease the gain KP
dynamic
and repeat measurement
example speed controller t
SINAMICS Date:
File:
06.03.2007
Application-specific optimization schemes

X [%]
Symmetrical optimum (SO)

43,4%
4,3% Value optimum (BO, Betragsoptimum)

100%
Linear optimum, LO
SO: preferred for speed control loops
BO: preferred for current control loops
LO: preferred for position control loops
3,1 4,7 X ... Controlled value [%] σ

σ ... Sum of small delays
SINAMICS Date:
File:
06.03.2007
Current controller optimization
Siemens drives
If the motor was selected via the menu “motor selection..." and once the motor data
have been downloaded to the drive, you can do without the current controller
optimization.
Along with the actual motor data the motor selection sets optimized controller settings.
A recording of the current actual value step response in the time range allows a reliable
evaluation of these standard settings.
In the operating mode Vector, the commissioning tool STARTER offers an automatic
motor identification for the self-optimization of induction motors.
In a number of test runs relevant electrical data of the induction motor are determined
automatically.
Based on these values the current controller is optimized automatically.
Third party drives

For third party drives, the operating mode Servo requires an individual optimization of
the current controller.
To do this a small step of setpoint is programmed (to initially maintain the standard
parameter settings is a good start condition) and the step response of the current actual
value is analyzed.
Aim to get the shortest possible rise time combined with a good damping, i.e. the
overshoot should not be larger than max. 10% (the step response should come close to
the Value Optimum).
SINAMICS Date:
File:
06.03.2007
Optimization of a speed controller with PI-characteristic
speed
torque
SINAMICS Date:
File:
06.03.2007
Transfer function and frequency response
Linear measuring system
(direct MS)
Encoder Table (Load)
Coupling
Motor
Ball screw “infinitely large mass“
(not affected by the
Machine base movement of the load)
Spring
Response to the
Noise signal mLoad
noise signal
Damper
Numerical control
xMotor xLoad
Response to the noise signal
Transfer function = Frequency response =
Noise signal
The processing of this transient function within a

large frequency band results in the amplitude and
phase response of the BODE diagram.
SINAMICS Date:
File:
06.03.2007
Amplitude- and phase-response
Input value e Amplitude response

aˆ
A =
eˆ
Spring e a eˆ aˆ
System
Damper
Phase delay
Δt
ϕ = 360° ⋅
Mass Output value a Δt T
Eingang Ausgang
f1 low f2 middle f3 high
The output can follow The output oscillates with large The output hardly moves. The damper
the input amplitude ⇒ resonance operates optimally ⇒ shock absorber
A( f1 ) = 1 A( f 2 ) >> 1 A( f 3 ) << 1
ϕ ( f1 ) = 0° ϕ ( f 2 ) = − 90 ° ϕ ( f 3 ) = −180°
SINAMICS Date:
File:
06.03.2007
Bode diagram of the setpoint value frequency response
XLoad
XMotor
SINAMICS Date:
File:
06.03.2007
Recording the transfer function of the open control loop
noise-
generator
transfer function of
speed- current the mechanical system
controller control loop
m.set
n.set n.act
k.P T.N t.sub.
n.act Æ measuring function “Mode 2“

m.set
Response to the noise signal

Transfer function = Frequency response = Bode - Diagram
Noise signal
SINAMICS Date:
File:
06.03.2007
Phase- and Amplitude-reserve (open control loop)
Amplitude [dB]
45
30
15
0
Amplitude reserve
-15
-30
-45
Criteria of evaluation
-60
A speed control loop optimized for
-75 stability and robustness has to
satisfy following conditions in an
Phase [degr] analysis using the Bode-diagram:
0
- Amplitude reserve > 12 dB
-45
-
600 > Phase reserve > 300
-90
-135
Phase reserve
-180
-225
-270
-315
-360
1 10 100 Frequency [ Hz ]
SINAMICS Date:
File:
06.03.2007
Evaluation of the control property (closed control loop)

Amplitude [ dB ] The gain peak (max. 5 dB) defines the overshoot
Amplitude-
0 margin
- 3dB
-20 Natural frequency as high as possible
Gain peak to a maximum of +5dB
-40 At a phase more negative than -180 degree

Æ amplitude < 0 dB
(amplitude reserve required)
Phase [ Grad ]
-90
resonance frequency
(at gain peak)
-180
The natural frequency (-3 dB) defines the rise time
SINAMICS Date:
File:
06.03.2007
Step response in the frequency and time range
Natural Natural
frequency: frequency:
40 Hz 40 Hz
Gain peak: Gain peak:

+ 6dB + 12dB
The natural frequency defines the rise time
The gain peak defines the overshoot
SINAMICS Date:
File:
06.03.2007
Bode diagram of a 2-mass system (ONLY controlled system)

Spring
mMotor mLoad
Damper
xMotor xLoad
Pole position = Resonance frequency

Æ critical if control loop is closed
Pole position via motor encoder
Zero position
Zero position = Natural frequency

Æ possibly critical if control loop is
closed via machine encoder
SINAMICS Date:
File:
06.03.2007
Frequency response (multi-mass system / ONLY controlled system)
n.act
m.set
SINAMICS Date:
File:
06.03.2007
Multi-mass system / closed control loop
Amplitude-
margin
For a phase more negative than -180 degree, the

Amplitude has to be < 0 dB! Æ the large amplitude-
margin allows to increase the gain k.P
n.act
n.set
An increase of the gain k.P reduces the amplitude margin
SINAMICS Date:
File:
06.03.2007
Danger !
Pole position is no longer damped (positive dB)

Requirement of amplitude reserve is violated !
SINAMICS Date:
File:
06.03.2007
Filter not yet activated
Filter as band stop to compensate the pole position
SINAMICS Date:
File:
06.03.2007
Filter activated
Result
Amplitude < 0 dB Æ sufficient Amplitude reserve

Æ gain k.P can again be increased
SINAMICS Date:
File:
06.03.2007
Characteristics of low pass and band stop filters
SINAMICS Date:
File:
06.03.2007
Evaluation of the filter criteria
Notch frequency
Notch depth
50%
Bandwidth
50%
SINAMICS Date:
File:
06.03.2007

k.P is set to its optimum value as the
resonance peak starts to show up
To achieve a higher natural frequency

(shorter rise time) T.N is reduced
SINAMICS Date:
File:
06.03.2007
As the natural frequency increases (shorter rise time)

the gain peak increases as well (speed overshoot)
If this rise time is required (dynamic response), but no

speed overshoot can be accepted, the reference model
has to be activated and optimized
SINAMICS Date:
File:
06.03.2007
Current setpoint filter / at resonance of the speed control loop

Speed controller Filter 1 Filter 2 Filter 3 Filter 4 Current controller
nset Iset
KP... Gain Low pass Band stop KPi... Gain

nact TN... Integral time Natural frequency Lock frequency Iact TNi... Integral time
damping Band width
Filter 1: preset for Motor ÅÆ Encoder coupling
Amplitude response without band stop Current setpoint filter
Amplitude response with band stop
SINAMICS Date:
File:
06.03.2007
Setpoint frequency response in the speed control loop
By analyzing the setpoint frequency response, the gain and integral time can be set to values that
ensure a high natural frequency with sufficient stability reserve. For the optimization and frequency
response measurements in the speed control loop, the speed setpoint filters have to be “switched
off“. They have no effect on the speed control loop stability, but they substantially affect the result of
the measurement.
kp=0.03 Tn=50,0 kp=0.3 Tn=5,0
The “more dynamic“ settings of Kp and Tn result in

Æ higher natural frequency / Æ shorter rise time
Æ larger gain peak / Æ larger speed overshoot
SINAMICS Date:
File:
06.03.2007
Speed setpoint filter / at resonance of the position control loop
Without influence on the dynamics of the speed control loop
To dampen resonance in the in the position control loop
To avoid the static following error when using setpoint ramps
For interpolating axes, the axis-specific system delay has

to be considered as it will differ with differently set speed
setpoint filters!
SINAMICS Date:
File:
06.03.2007
Measuring functions
• Controller optimization with Pre-configured measuring functions:
automatic suppression of
superposed control loops. 1 Speed controller setpoint frequency response (after speed setpoint filter)
• Easy to do parameterization 2 Speed controlled system (signal injection after current setpoint filter)
• Immediately ready to run: 3 Speed controller disturbance frequency response (after current setpoint filter)
• function generator Speed controller setpoint frequency response (before speed setpoint filter)
• trace channels
5 Speed controller step of setpoint (after speed setpoint filter)
6 Speed controller step of disturbance variable (after current setpoint filter)
7 Current controller setpoint frequency response (after current setpoint filter)
8 Current controller step of setpoint (after current setpoint filter)
step of speed setpoint
frequency response
SINAMICS Date:
File:
06.03.2007
Modes of application of measuring functions
1 5 2 7 8
3 6
Current Identification current control loop frequency range mode 7

control loop
Verification response current controller - setpoint time range mode 8
Speed Identification speed control loop frequency range mode 2

control loop
Verification response speed controller – setoint value frequency range mode 1
response speed controller – setpoint value time range mode 5
response speed controller – disturbance value frequency range mode 3
response speed controller – disturbance value time range mode 6
SINAMICS Date:
File:
06.03.2007
Current controller optimization
Signals to be recorded: 1. r77: torque generating current setpoint

2. r78: torque generating current actual value
SINAMICS Date:
File:
06.03.2007
Speed controller optimization
2 3
Signals to be recorded: 1. r62: speed setpoint after the filter

2. r61: speed actual value (motor encoder)
3. r80: torque actual value
SINAMICS Date:
File:
06.03.2007
Reference model for the speed control loop (Servo + Vector)
r62[D] n.set P m.set

Kp
speed-
controller
I
Tn
r1436[D]
n.act
p1433[D]
r61[D]
p1434[D]
p1435[D]
reference model
r61 off
Vm on
The reference model delays the set-actual-deviation for r62
the integral component of the speed controller. In
consequence, transient overshoot is noticeably reduced.
Steps of load are controlled with the high dynamic t
property set via proportional gain. Trise
SINAMICS Date:
File:
06.03.2007
Settings for the reference model
1) Set the damping in respect to the currently valid response

criteria of the speed control loop (refer to diagrams below)
2) Experimentally identify the natural frequency to have the

transfer function of the reference model (*) match the transfer
function of the closed speed control loop as good as possible (*)
3) Evaluate the dead time from the step response (see diagram)
(*) trace of:
input = r0062 ; output = r1436
Tt
SINAMICS Date:
File:
06.03.2007
Reference model and symmetry filter
setpoint control
n.set m.set
k.P
reference- speed-
Reference model model controller
k.P T.N control of

disturbance
variable
n.act
setpoint control
x.set n.set
symmetry- position-
Symmetry filter filter controller
k.V control of
disturbance
variable
x.act
SINAMICS Date:
File:
06.03.2007

1 10 Optimization PDF

Uploaded by

Copyright:

Available Formats

1 10 Optimization PDF

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

1 10 Optimization PDF

Uploaded by

Copyright:

Available Formats

Optimization

Scheme of closed loop control (example: feed drive)

- xact - nact - Iact

Speed Rotor position Encoder evaluation

KV...Gain KP... Gain KP... Gain

Position Speed Current Current Torque Speed Angle

Controller Controlled system

Measuring Controlled value

Main objectives of (speed-)controller optimization

Time range Frequency range

Description by means of Description by means of frequency response

Description by means of Frequency response Bode diagram

Input (Setpoint) Output (Actual value)

Set the integral time to a very Reduce KP to half

VERY rough procedure

Output (actual value)

Application-specific optimization schemes

Symmetrical optimum (SO)

4,3% Value optimum (BO, Betragsoptimum)

SO: preferred for speed control loops

BO: preferred for current control loops

LO: preferred for position control loops

3,1 4,7 X ... Controlled value [%] σ

Third party drives

Optimization of a speed controller with PI-characteristic

The processing of this transient function within a

Amplitude- and phase-response

Input value e Amplitude response

f1 low f2 middle f3 high

Recording the transfer function of the open control loop

n.act Æ measuring function “Mode 2“

Response to the noise signal

Evaluation of the control property (closed control loop)

-20 Natural frequency as high as possible

Gain peak to a maximum of +5dB

-40 At a phase more negative than -180 degree

Gain peak: Gain peak:

The natural frequency defines the rise time

The gain peak defines the overshoot

Bode diagram of a 2-mass system (ONLY controlled system)

Pole position = Resonance frequency

Zero position = Natural frequency

Multi-mass system / closed control loop

For a phase more negative than -180 degree, the

An increase of the gain k.P reduces the amplitude margin

Pole position is no longer damped (positive dB)

Multi-mass system / closed control loop

Filter not yet activated

Filter as band stop to compensate the pole position

Amplitude < 0 dB Æ sufficient Amplitude reserve

Characteristics of low pass and band stop filters

Multi-mass system / closed control loop

To achieve a higher natural frequency

As the natural frequency increases (shorter rise time)

If this rise time is required (dynamic response), but no

Current setpoint filter / at resonance of the speed control loop

KP... Gain Low pass Band stop KPi... Gain

Amplitude response without band stop Current setpoint filter

Amplitude response with band stop