This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

DRV8323R: current sense amplifier output mismatch external measurement

Part Number: DRV8323R
Other Parts Discussed in Thread: DRV832X

Hello guys,

I am implementing FOC on a PMSM motor using DRV8323RS. I have a problem with the current values that I sample using an MCU+ADC, it does not match the current clamp probe that I assume to be my truth reference.

To understand the problem, I measured on oscilloscope the following signals :

  • phase B current with clamp probe (YELLOW)
  • Low-side gate GLB signal (BLUE)
  • Amplifier output SOB (PURPLE), gain = 10

The DRV8323RS is set in bidirectional current mod, VDD = 3.3V and Rsense = 2 mOhm.

On the picture bellow, you can see, while motor running, the phase current descreasing (yellow), the amplifier output (purple) transiting from zero to non-zero when the low-side gate switches ON (blue).

At the measurement point, AY = 40 mV means the current value I = 40mV / 10 = 4 A
At the measurement point, BY = 1.764V means the current value I = (VDD/2 - 1.764) / (G_CSA * Rsense) = (3.3/2 - 1.764) / (10 * 0.002) = 5.7 A

The difference between probe measurement and above amplifier conversion is significative !

On the picture bellow, I increased the motor load and redid the same experimentation.
You can see, while motor running, the phase current descreasing (yellow), the amplifier output (purple) transiting from zero to non-zero when the low-side gate switches ON (blue).

At the measurement point, AY = 90 mV means the current value I = 90mV / 10 = 9 A
At the measurement point, BY = 1.93V means the current value I = (VDD/2 - 1.93) / (G_CSA * Rsense) = (3.3/2 - 1.93) / (10 * 0.002) = 14 A


The difference is still huge.

Finally, the last picture shows even bigger load currents.

At the measurement point, AY = 120 mV means the current value I = 120mV / 10 = 12 A
At the measurement point, BY = 2.05V means the current value I = (VDD/2 - 2.05) / (G_CSA * Rsense) = (3.3/2 - 2.05) / (10 * 0.002) = 20 A

Following the above observations, here are my questions:

  1. Are the fluctuation on SOB signal (purple) part of an expected transient phase ?
  2. What could be the reason of difference of shape and magnitude between line current (yellow) and low-side amplifier output SOB (purple) ?
  3. As this prevents me from measuring accurately the current with my MCU, what would be a potential fix for this problem ?

[EDIT]

I redid the same experiment at a higher rotation speed, which means higher voltage, higher high-side duty-cycle and smaller low-side duty cycle. Consequently, the GLB logic high period is even smaller and we can observe than the SOB fluctuations barely converge during the time GLB is high. (see picture bellow)

Thanks a lot for your help !! 

  • Hi Victorien, 

    Thank you for your question!  

    I will look over your provided info and update you with any suggestions or further questions. 

    Best Regards, 

    -Joshua

  • Hi Victorien, 

    I am still looking into your issue and will aim to provide a suggestion early next week.

    Best Regards,

    -Joshua

  • Hi Joshua, 
    any update regarding this problem ?

    Thanks,
    Victorien.

  • Hi Victorien,

    We are still working to provide an update, and if you could provide your current schematic that would be helpful moving forward.

    We will keep you further updated on the matter and hope to have an answer/suggestion for you soon. 

    Best Regards,

    -Joshua

  • Hi Joshua,

    Please find bellow our schematics. Please not that this PCB design has been developed for testing purposes with compatibility requirements for both DRV8323RS and DRV8323RH chips. In our setup and tests the DRV8323RH pin tracks are not used and are marked with red cross in the screenshot bellow.

    Thank you in advance.
    Regards,
    Victorien.

  • Hi Victorien,  

    Thanks for the schematics and clarifications.  Let me review over them to better help reach a solution. 

    Best Regards, 

    -Joshua

  • Hi Joshua,

    I did a new experimentations in order to see if the unaccuracy of the current measurement was also visible before the amplification stage, so accross the sensing resistor. To be able to measure enough signal, I replaced the 2mOhm resistors for 20mOhm resistors. By software, I also changed the gain G_CSA from 10 to 5.
    Then I run again my motor at different loads and speeds and I noticed a much lower oscillation of the SOB signal.

    You can see on the picture bellow : 

    • Current probe measurement (yellow)
    • SOB signal (purple)
    • SPB voltage (light blue)
    • GLB gate (blue)

    We see that the SOB signal oscillates much less than on my previous experiment which is something good. In those conditions, I could measure a current of 11.4 A from the probe, 12.3A from SOB and 12.8A from SPB. Those results seems much more satisfying because they are grouped in a narrower range.

    As those results are much better for me, I wanted to know what was the cause of this improvement.
    I redid new experimentation where I put back sensing resistors of 2mOhm and I variate the G_CSA value.

    Bellow you can see measurement with G_CSA = 10

    And bellow the measurement taken in the same operating conditions except that G_CSA = 20.

    It seems that the bigger G_CSA is, the bigger and longer will be the oscillations.
    Is this an expected behaviour of the amplifier ? Can I do something with it ?

    Thank you in advance for your help,
    Victorien.

  • Hi Victorien, 

    It is actually expected that oscillations on the CSA stage would scale with the gain and sense resistor values, and that is a reason we have certain equations to ensure that the CSA configuration is optimized for each application as detailed in section 9.2.1.2.4 of the DRV832x datasheet:

    I would recommend double-checking that the CSA configurations you are using when this behavior is seen are suitable for your current design parameters using the above example.

    Regarding the ringing seen on the CSA outputs, I did not notice any implementation of the following suggestions on the schematic:  I strongly recommend implementing filtering capacitors on the CSA inputs between SPx and SNx as well as RC filters on the outputs, SOx. If necessary, you can implement RC filters on the inputs if a filtering capacitor alone is not sufficient enough for each SP/SNx.

    Let me know if I can help clarify anything further.

    Best Regards,

    -Joshua

  • Hello Joshua,

    We have followed the proposed methodology to choose the gain and sensing resistor but unfortunately we still don't have satisfying results.

    In addition I tried several couples of (Rsense, G_CSA) but none of them is completely good.
    If (Rsense * G_CSA) is too big then my measurement saturates when the current is high.
    If (Rsense * G_CSA) is too small, then my SNR is too small for low current and my system becomes unstable.
    For some in between couples with G_CSA>= 10, the current is quite good for low values but I still have the ringing problem on higher values making the system unstable at high torque.

    Is RC filter commonly applied where you suggest to ? I am surprised because it is never mentionned in the datasheet. 

  • Hey Victorien, 

    We do often recommend additional filtering on both the CSA inputs and outputs, but this is optional and so may not have been included in this device's datasheet. However, if you look into the DRV8323Rx EVM schematic (found in the design files on the product page here), you will see these filters actually implemented:

    I recommend implementing these filters with the Rsense + gain combination that gave you the best results, as you mentioned below:

    For some in between couples with G_CSA>= 10, the current is quite good for low values but I still have the ringing problem on higher values making the system unstable at high torque.

    Let me know if you have any additional questions.

    Best Regards,

    -Joshua