2. What is Current Sensing?
• To measure current flowing through a
conductor
• Typical Approaches:
– Shunt resistor
– Current Transformer
– Hall Effect
Shunt resistor approach:
Measure voltage drop across a small
resistor.
Step down primary current
Current => Flux => Voltage
3. Hall Effect Current Sensing
• Physics:
– Lorentz Force
– Uneven Distribution of Charges
– Electric Field & Voltage
• Closed Loop Hall Effect Sensor
– Feedback circuit to drive flux to 0
– IF related to Iin by turn ratio
4. Why is Current Sensing Important?
• Field Oriented Control: Control of flux-producing
(d-axis) and torque-producing(q-axis) currents.
• Sensor quality directly impacts output dq
currents hence machine torque.
• Also used in sensorless control to determine
rotor position
5. Flux Concentrators (Cores/Toroids)
• Function: Concentrate flux by having a high
permeability (B (flux density) = u (permeability) * H (field strength) )
– Permeability of Air ~= 1
– Permeability of 4% SiFe ~= 2000 - 35000
Current in BussBar Magnetic Filed (H) Flux Density (B) Hall Voltage
Ampere’s Law Permeability Hall Effect
6. Core Properties
• Electrical:
– Permeability (Gain)
– Saturation
– Hysteresis
Desired Linear
Relationship
Cores I investigate:
O.D = 31.07 mm
I.D = 17.4 mm
Gap = 5.14mm
7. Core Saturation
• In saturation, core permeability decreases
significantly
• Can cause major torque ripple issues
-2500
-2000
-1500
-1000
-500
0
500
1000
1500
2000
2500
-1.50E-02 -1.00E-02 -5.00E-03 0.00E+00 5.00E-03 1.00E-02 1.50E-02
1400Arms 13MM gap core saturation
Actual(red) vs Measured(green)
8. Core Hysteresis
• Generally a nonlinear effect
• For our test results, a delay model can
approximate the effect
• More on this later
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
0.8
-6.00E-04 -4.00E-04 -2.00E-04 0.00E+00 2.00E-04 4.00E-04 6.00E-04
Hysteresis : Actual (green) and Measured (red)
9. My Project: Core Evaluations
• Goal: Compare the saturation and hysteresis
characteristics of Tape Wound (TW) cores and
Stacked Lamination(SL) cores
Note: Comparison is for 38533 PN Core
Gap = 5.15 ± 0.4 mm
Rated current: up to 250 Arms (354 A peak)
Cost Saving:
TW 38533 $0.99
Stacked Lam cores $0.348
Annual volume ~150,000
Potential Annual Saving from
cores ~= 96,300 !
10. Benchtop Core Comparison
Old Setup:
- Larger Center Conductor (0.62 inch d)
- Difficult to position cores and sensors =>
flux to current conversion gain changes due
to mechanical vibration
New Setup:
- Smaller Center Conductor (0.5 inch d)
- Sensor position stable => more consistent
gain
Core I.D = 0.68 in
11. Benchtop Core Comparison
• Main Comparisons between cores:
– Hysteresis vs frequency
– Hysteresis vs current
– Saturation
• Other Considerations:
– Sensor Type: Lakeshore Gaussmeter vs Hall Bridge vs
Allegro/Melexsis
– Setups
• Center conductor diameter – affects B field
• Mechanical – position affects results
– Gain error (calibration)
– Variances between cores
12. Hysteresis vs Freq
• Experiment:
1. Put ~250Arms current through center
conductor at 100Hz, 250Hz, 500Hz.
2. Measure gap flux with Lakeshore
gaussmeter.
3. Convert gap flux to current measured.
Gain = Irms / Brms.
4. Plot Current Sensing Error vs Current
in conductor
13. Hysteresis vs Freq Results
SLTW
- Spikes due to LEM problem
- Differences between cores insignificant at this current level
14. Hysteresis vs Current Level
- Stacked Lam core slightly worse hysteresis at each current level
- At 500Arms @ 500Hz, SL cores have worse hysteresis with max current error ~= 10A.
TW cores have max current error ~= 5A.
This current level is x2 rated current for this core!
100 Arms – 500 Hz 200 Arms – 500 Hz 500 Arms – 500 Hz
15. Core Saturation Comparison
(800Arms 500Hz)
TW SL
- SL slightly more error before saturation
- SL softer saturation (limited operating range)
- Does not appear to be an issue for the 38533 part at rated current levels
- Gap tolerances for SL are significantly tighter
- Prudent to further evaluate any additional higher current replacements
16. Hall Sensor Evaluation
• Compare LEM(closed loop hall effect), Hall
bridge, Lakeshore Guassmeter, Allegro and
Melexsis
LEM
Lakeshore
Hall Bridge
Allegro
Melexsis
17. LEM vs Lakeshore Gauss meter vs Hall Bridge
LEM
Lakeshore Hall Bridge
Significant delay in the hall
bridge signal output.
Hysteresis appears similar to
a delay under steady state
conditions.
Difficult to distinguish
between sensor delay and
core hysteresis.
18. Lakeshore(L) vs Hall Bridge(R) SL cores
- Current sensing error due to a fixed delay is proportional to freq of
operation.
19. Lakeshore vs Allegro
• Allegro sensor performance close to Lakeshore
LEM
Lakeshore
Allegro
21. Core Evaluation Conclusion
• PML Core samples are substantially similar
during normal operating conditions
• Recommend tooling production candidates
with PML
– Requires some further mechanical and electrical
evaluation
22. Further Work
Even with existing current sensors, Iq/Id currents can be quite noisy!
• Explore quantization effects and other noise in:
– Position sensor
– Current sensors
– Rotor harmonics (PMAC)
Currents:
20A/Div
23. Simulation Results
No Theta_e error. No
sampling
w/ current quantizer
2V encoder output
5V 10-bit ADC
w/ better quantizer
2V encoder output
3.3V 12-bit ADC