Accurate Sensors for Electric Power Testing

Current, Voltage, Torque, Speed and more….

Klaus Lang
Business Development Manager
Electric Power Testing
 Sensors for accurate electric power testing
1. The measurement chain for electric power measurement

2. The individual pieces

• Current
• Voltage
• Torque
• Speed
• Other sensors

3. The link to the Power analyzer

4. Some remarks an measurement uncertainty

2 UNRESTRICTED
 The measurement chain for electric power testing

3 UNRESTRICTED
 The measurement chain for electric power testing
Automation
Measurement
Power source and PC
PC
inverter temperatures

Power Source

Inverter
Voltage and current
of power source

Electric
Voltage and current Machine eMachine
of inverter output vibration

eMachine and
Torque and speed at Inverter noise
eMachine output

eMachine (winding)
temperature
4 UNRESTRICTED
 Current Measurement

5 UNRESTRICTED
 Sensors for current measurement – Technologies overview
Ohm’s Law V = I R Ampere’s Circuital Law I = ∮H ds
Non-
Isolated
Isolated

AC only AC + DC

Magnetic Air-core
Hall effect Fluxgate
core (no core)

Current Rogowski Closed With air Single Double

Shunt Open loop Balanced
Transformer coil loop gap core core

Intrusive Non-Intrusive

6 UNRESTRICTED
 Sensors for current measurement – Pros and Cons in a nutshell

Technologies Bandwidth Accuracy EMC Size Power Drift Isolation Typical

immunity application
Power semiconductors
Shunt +/- + + +/- +/- +/- - and pulse power tests

Hall effect / Open loop - - - + + - + Power measurement

Power and
Hall effect / Closed loop + +/- +/- +/- +/- +/- + Harmonic Analysis

Fluxgate / Air Gap + +/- +/- +/- +/- + + Power semiconductors

Fluxgate / Single Core - +/- + +/- +/- +/- + Power measurement

Power and
Fluxgate / Double Core + + + - +/- +/- + Harmonic Analysis

Power and
Fluxgate / Balanced + + + - +/- + + Harmonic Analysis

HBM´s current sensors are Balanced Fluxgate type

7 UNRESTRICTED
 Current sensors from HBM – Main specs
• Balanced closed loop fluxgate based
• Full-aluminum body for superior EMI shielding
• High bandwidth up to 1000 kHz
• Superior dynamic response
• Extremely accurate up to 0,01%
• 1 or 2 ppm accuracy, pending from type
• Wide operating temp range - 40°C to + 85°C
• Industry standard 9 pin connector
• Large aperture size
• ASPC - advanced circuit for sensor protection
• Large overload capacity

8 UNRESTRICTED
 Current sensors from HBM - Family
• Broad range from 50 A rms to 1200 A rms
• Complete solution with power supply and all cables

9 UNRESTRICTED
 HBM CTs – superior step response
Step response of different current transducers
Test circuit:
DC-Link Half Bridge

(low inductance)
R=1Ω
CT
(DUT)

Very low inductive design in

order to realize a high di/dt-ratio

Note: Good step response is crucial to capture

high frequency switching phenomena in power
source and inverter outs current accurately
Note: CT2 is HBK type

Source: Prof. Dr.-Ing. J. Teigelkötter, THAB

10 U N R E S T R I C T E D
 Current sensors - Summary

 Zero-Flux CT´s are standard for accurate power measurement

• Established and accurate
• Some installation effort
˗ needs current cable to be opened
˗ current cable shielding needs to be opened
˗ Often CTs are installed in shielded metal box
• Connection to PA is straight forward using internal shunts of PA´s

 For Efficiency, low bandwidth is sufficient > 20 x fundamental freq 20-100 kHz

 For Harmonic analysis

• Harmonics of fundamental frequency >100 x fundamental freq 100-500 kHz
• Harmonics of switching frequency >50 x switching freq up to MHz range
-> other current sensors (Rogowski coils) might be used

11 U N R E S T R I C T E D
 Voltage Measurement

12 U N R E S T R I C T E D
 Voltage

 In the LV range up to 1 kV, voltages are measured directly

• Most PAs offer direct inputs up to 1 kV,
so no sensors (probes) needed
˗ Note: Inputs need to comply to overvoltage
categories (CAT rating) be used in electrical testing

 For MV and HV, probes are used

• Accuracy is challenging
• Safety is even more challenging
• BOTH is very rare

13 U N R E S T R I C T E D
 HBM solutions for accurate and safe voltage measurement

 1 kV power analyzer card GN310B

and 1500 V DC
Up to 1 kV rms

˗ Build in solution for most applications in grid and small / mid size drives
˗ Input 1000 Vrms and +/- 1500 V DC
˗ Highest accuracy 0.015% reading + 0.02% range
˗ Highest overvoltage category available: 1000 V CAT IV

 5 kV passive differential probe for GN310B

˗ External probe for large drives, railway, aerospace, wind energy MV levels
Up to 5 kV

˗ Accuracy 0.1%
˗ 1500 DC CAT IV, 5 kV CAT 0
˗ Certified safety / protection mechanism

 Isolated digitizer 6600 with HV probe

Virtually any voltage

˗ Fibre-optical solution: 100% SAFE

˗ Ultra-large drives, HV inverters, DC-/Smartgrids
˗ Range / Accuracy pending from probe: ~ 0.5 to 1 %
˗ No standard for overvoltage available Digital, fibre optic
transmission

14 U N R E S T R I C T E D
 Overvoltage categories in a nutshell

 Defined in IEC 61010-2-30

 To protect the user, each measurement device
needs to provide overvoltage protection
• Level of protection depends from WHERE
the instrument is used
• The closer to the grid the higher the needed protection

Source: Mouser Germany

• A test rig – as always connected to power - Test rig
requires test systems certified to CAT III

• Next edition of IEC 61010 (now in draft) defines a new

DC class of protection: 1500 V DC, introduced for EV´s
Notes:
* From a safety standpoint, 1000 V rms is NOT equivalent to +/- 1415 V DC
* Latest ref of IEC610 replaces CAT I with CAT 0 or „no CAT“ Source: Wikipedia

15 U N R E S T R I C T E D
 Competitive comparison – Overvoltage categories
Manufacturer Model CAT IV CAT III CAT II open Remark

Vendor 1 Model a 600 1000 -- --

Vendor 1 HV extension for a -- -- -- 10 kV Manual: Do not use above 60V!
Vendor 2 Model b -- 600 1000 1250 DC
Vendor 3 Model c -- 300 600 1000 DC
Vendor 5 Model e -- 600 1000 --
Vendor 6 Model f -- -- 1000 -- Manual: Do not use for CAT III and IV
Vendor 7 Model g -- -- 1000 --
Vendor 8 Model h -- -- 600 --
Vendor 9 Model j -- 600 1000 --
HBM GN310B 1000 1500 DC* -- --
HBM HVD50R-5kV 1000 -- -- 5 kV With protective ground
1500 DC* monitoring system
* pending from latest IEC 61010 rev to become active

Application example: An EV with 800 V battery voltage must only be testing with 2 PA´s from above list

16 U N R E S T R I C T E D
 Voltage sensors - summary
 Low Voltage application < 1 kV small & medium drives, automotive, EVs
• Can be addressed with a direct inputs
• Variable input spans allow best match to input signal - > good for MU, see later under MU
• PA should meet the overvoltage category required by IEC safety standard
˗ Test rig usage needs 1 kV CAT III minimum
˗ Only few products certify for test rig use without extra safety measures

 Medium Voltage: large drives, train, aerospace, ships

• Various entry level probes available with limited accuracy (2-3% // 500 € - 1500 €)
• Accurate, safe and certified probes are available but at a certain cost (0.1% // 5 k€)

 High voltage: very large drives, subsea pumps, switchgear

• Safety first
• Fibre optic isolation and proper divider (~ 0.5% - 1 % // 10 k€ - 15 k€)

17 U N R E S T R I C T E D
 Torque Measurement

18 U N R E S T R I C T E D
 Measurement principle for in-line measurement

Strain gage
Change in resistance
Strain proportional to strain

Torque sensor

Torque puts stress on Output signal

housing with embedded proportional to Torque
strain gages

19 U N R E S T R I C T E D
 Accuracy Class

HBK definition: The biggest single deviation determines the accuracy class

The following specifications are included:

Linearity deviation incl. hysteresis (dlh )
Relative standard deviation of repeatability (σrel)
Temperature coefficient - zero signal (TC0)
Temperature coefficient - sensitivity (TCS)

Do not mix up with:  „Total accuracy“

 Classification according to DIN51309

23 U N R E S T R I C T E D
 Accuracy Class - Example

Sensor 1 Sensor 2

 Linearity deviation incl. hysteresis dlh 0,02 % 0,05 %

 Relative standard deviation of repeatability σrel 0,02 % 0.03 %
 Temperature coefficient - zero signal TC0 0,02 % 0.05 %
 Temperature coefficient - sensitivity TCS 0,05 % 0,05 %

Same Accuracy Class!

24 U N R E S T R I C T E D
 Outputs of the torque transducer: example T40B

Voltage / Torque

Frequency / Torque (x Hz +/- y Hz)

Frequency / Speed (0 to x Hz)

Shunt ON/OFF

Energy

Which output to use?

25 U N R E S T R I C T E D
 Frequency output for Torque

Torque as Frequency (x Hz +/- y Hz)

 Advantages:
• Most accurate output signal (no noise, no drift)
• Differential transmission to PA immune to noise and EMC
• Easy to T-off for PA and Automation system

 Problem in analysing this signal:

• Electronics (Counter) needed to convert frequency back to torque
˗ Counters have a systematic error based on time limits (+/- 1 pulse error per counting cycle)
˗ Short counting time gives dynamic torque but larger error
˗ Long counting time gives accurate torque but less dynamic signal
˗ Higher centre frequency (for 0 Nm) gives higher accuracy and better dynamic range

26 U N R E S T R I C T E D
 Typical test rig setup and associated problems
Parasitic Loads

Temperature,
Environmental influences

Dyno eMachine under test

Torque flange transducer Coupling Shaft

Influence from sensor and DAQ Influence from design and process
• Linearity
• Temperature
• Hysteresis
• Parasitic Loads
• Temperature drift (Bending moment, radial and axial forces)
• Mechanical mounting • Dynamic
• Stiffness • Other environmental influence
• Measurement electronics errors • ...

27 U N R E S T R I C T E D
 Torque Sensors - Summary

 For medium to high torque a flange type delivers best accuracy

• HBK offers best in class to reference type torque flanges

 Most errors are not introduced by the torque sensor but by

• mounting issues and
• environmental influences

 Frequency / digital out offers high accuracy and easy handling

• This needs proper measurement electronics
˗ Should be same accuracy class or better
˗ Use high centre frequency to achieve smallest error (….in measurement electronics)

29 U N R E S T R I C T E D
 Speed Measurement

30 U N R E S T R I C T E D
 Measurement principle

 The rotation of the shaft is measured with an incremental encoder.

 This device generates pulses as it turns.

So strictly speaking this is measuring angle.

 Two phase shifted pulses are used to indicate direction.

 These pulses can be created in various ways:

Magnetically or optically being the most common.

31 U N R E S T R I C T E D
 Speed Sensor: Optical system

light source

condenser lens

diaphragm
Signal A

Signal B

Reference pulse

slotted disc
receiver

32 U N R E S T R I C T E D
 Magnetic speed measurement

Reference pulse

Pole ring with magnets

Speed sensor (Coils)

33 U N R E S T R I C T E D
 Speed Sensor: Magnetic System
Pole ring
The magnets are inducing
Angle-dependent, analog signals two sin/cos voltage
Bridge output signal 1 signals into coils

Bridge output signal 2

These analogue voltage
signal are then converted
into a binary pulse train

Pulse series
Binary signal

34 U N R E S T R I C T E D
 Comparison of the methods

Optical system Magnetic system

360 – 720 increments / turn 128 – 1024 increments / turn
Uncertainty = 0,03% of range Uncertainty = 0,03% of range

High accuracy
Advantages
High accuracy
Small group delay of 5 µs Immune to dust, dirt, fog, …
Simple, cheap Larger distance between sensor and elements
Lower slowest speed Robust against boundary conditions

Disadvantages
High group delay of 150 µs
Fragile speed disc More complex and more expensive
Slotted disc sensitive to boundary conditions Magnetic field limited to 100 A/m
like moisture, mechanical load, dust, dirt…. Does not work to stand still

35 U N R E S T R I C T E D
 Speed Sensor – how to transfer the signal
 Differential transmission ensure noise immunity
RS422

F1A

F1B

F2A

F2B

Zero Index A

Zero Index B

36 U N R E S T R I C T E D
 Frequency output for Speed

Speed as Frequency

 Advantages:
• Most accurate output signal (no noise, no drift)
• Differential transmission to PA immune to noise and EMC
• Easy to T-off for PA and Automation system

 Problem in analysing this signal:

• Electronics (Counter) needed to convert frequency back to speed
˗ Counters have a systematic error based on time limits (+/- 1 pulse error per counting cycle)
˗ Short counting time gives dynamic speed but larger error
˗ Long counting time gives accurate speed but less dynamic signal

37 U N R E S T R I C T E D
 Speed Sensors - Summary

 Most common way to measure speed is with an incremental encoder

• This measures angle, not speed
• Differential transmission ensures noise immunity
• Electronics needs to convert angle to speed

 Incremental encoders might be magnetic or optical – both with pros and cons

 HBM offers both speed sensor technologies

Note: Resolvers are not covered here as very special

38 U N R E S T R I C T E D
 Other sensors to complete the picture

39 U N R E S T R I C T E D
 Other sensors
PT100
 Temperature sensors: TCs or PTxxx or Optical
• Motor winding temperature, inverter and battery temperatures
• Problems:
˗ Potentially temp point is on high voltage potential Thermocouple
˗ Very low level temp signals are subject to noise from inverter

 Acceleration sensor HBK´s Optical temp sensor

• Torque ripple
• Torsional vibration
• Durability

 Microphones
• Switching noise
• Noise caused by mechanical vibration

40 U N R E S T R I C T E D
 The link to the Power analyser

41 U N R E S T R I C T E D
 Current measurement – from the sensor to the PA
 Current inputs should offer build in shunts (included in the accuracy specification)
• Verify these shunts are included in the accuracy specification

 Current inputs should offer different spans to adapt to the actual level
• Especially during an efficiency mapping the current level is changing significantly
• Best to adapt the input range to the actual level
• Better to have AUTO-Range doing this always

 PA should allow to use Multi-Point calibration certificates for the current sensor
• This improves MU big time due to elimination of offset error

 Bandwidth / sample rate:

• Should be the same for current & voltage input to avoid phase shift in electrical power computation
• Should be the same for electrical and mechanical inputs to avoid different averaging cycles for
efficiency calculation
42 U N R E S T R I C T E D
 Voltage measurement
 Voltage inputs should comply with safety standards required for the installation
• Test rigs are CAT III
• EV´s require new class 1500 V DC

 Voltage inputs should offer different spans to adapt to the actual level
• During eDrive tests the voltage typically does not change but other test might require different levels
• Best to adapt the input range to the actual level
• Better to have AUTO-Range doing this always

 Bandwidth and sample rate -> same remarks as for current inputs

43 U N R E S T R I C T E D
 Torque measurement – from the sensor to the PA
 Differential signal transmission enables noise free torque measurement

 Somewhere in the chain between sensor and PA the signal needs a T-off for AuSy

 PA should allow to use Multi-Point calibration certificate for the torque sensor
• This improves MU big time due to elimination of offset error

 Electronics need to convert frequency to torque using counting and averaging

• Long averaging has small error but looses all dynamic content
• Short averaging shows dynamic content like torque ripple but is less accurate
Long averaging cycle,
accurate steady torque for
• Solution: efficiency
“DualTorque” technology
Short averaging cycle,
dynamic torque shoing
ripple and vibrations

44 U N R E S T R I C T E D HBM´s power analyser compute both torque signals in parallel in real time
 Speed measurement – from the sensor to the PA

 Differential signal transmission enables noise free torque measurement

 Somewhere in the chain between sensor and PA the signal needs a T-off for AuSy

 Electronics need to convert frequency to torque

• Most PA´s do this with counting and averaging introducing errors or loosing all dynamic content

• Solution:

HBM´s power analyser first integrate the frequency to get to the angle.
Then they do a special differentiation (suppressing jump 360 ->0) to obtain instantaneous speed.
Thus no counter timer or averaging needed: speed value is dynamic and accurate

45 U N R E S T R I C T E D
 Some remarks on measurement uncertainty

46 U N R E S T R I C T E D
 Some remarks on measurement uncertainty

 Current and Voltage

• Inputs should have programmable ranges to adapt to real signal
˗ As error is always % of reading + % of range, the smallest possible range deliver best MU
˗ Systems with AUTO-Range automatically achieve best MU

 Torque and speed

• Parasitic loads are critical and need to be assessed
• Frequency measurement of Torque should be accurate
• Speed should be derived from angle to avoid counter error

 Power loss and efficiency

• Power loss is the difference of two values – but errors add up !! – see next slide

47 U N R E S T R I C T E D
 Measurement uncertainty and different accuracies - simplified

Input Output
Power Power • 500 kW inverter
• 95% efficiency
• Power losses = Input power – Output power

Measurement chain error @ 1% Measurement chain error @ 0,1%

“Truth” Measured Error Measured Error
Input 500 kW 500 kW +/- 5 kW 1% 500 kW +/- 0.5 kW 0,1%
Output 475 kW 475 kW +/- 4.75 kW 1% 475 kW +/- 0.475 kW 0,1%
Losses 25 kW 25 kW +/- 9.75 kW 39,0% 25 kW +/- 0.975 kW 3,9%

If you want to achieve reasonable Measurement Uncertainty for derived values like Power Loss
or Efficiency you need the most accurate sensors (and the most accurate PA) you can get

48 U N R E S T R I C T E D
 Summary

49 U N R E S T R I C T E D
 Summary – Accurate sensors for electric power measurement
Current: Zero-Flux sensors are established and easy to use
• Some attention needed for mounting, cabling, shielding

Voltage: Most applications do not need sensor but can be measured directly
• Make sure your signal input fulfils legal requirements (CAT)

Torque: Flange sensors dominate the market due to accuracy and handling
• Mounting and introduced parasitic loads by this carry most problems

Speed: Various sensors available; optical, mechanical, resolvers, encoders

• Accuracy is dominated not by sensor, but by electronics

 TCs, Accels and Mics used for complete testing, so a PA should allow to use these

 Measurement uncertainty is dominated by weakest piece of the chain

• Errors add up in derived results like power loss or efficiency !!!
50 U N R E S T R I C T E D
Do you have a question for the
Presenter? Visit the Guest Speakers
Virtual booth within the next hour
for an interactive Q&A session.