DRV8323R: Weird GHx waveform

Hey James,

The gate drive voltage seems pretty small, could you please provide a waveform of VM, VCP, VGS and SHx?

Best,

Akshay

Hi, Akshay

Waveforms as shown below,

The are math waveforms in pictures because we don't have a differential probe.

Hey James,

Thank you for the waveforms. I will aim to provide feedback next week.

Best,

Akshay

James Wang said:

Does anyone know how it is generated? 

Could be due to shut off protection. Can you include the nFault signal in the waveform?

Brian

Hi Brian,

The nFault waveforms are shown below.

It seems the nFualt is not pulled down due to ringing interference or something others.

I think the weird peak of GHB is caused by the instability of the INHB, but I am not sure.

Thank you very much.

Hi James,

James Wang said:

I think the weird peak of GHB is caused by the instability of the INHB, but I am not sure.

I don't think the GHB is caused by the noise on the INHB, because if it is the case then we should see a normal voltage relationship between GHB and SHB, regardless, namely GHB - SHB = boost voltage VCP = 11v or so. 

Pic below shows VCP is less than zero in the left highlighted circle where SHB is higher than GHB; I mean how can the FET turned on with VGS less than 0v? Can you have both waveforms HGB and SHB lined up at the 0v level for easy to compare their values?

The right high-lighted circle shows VCP is much less than 11v during FET turning on, then later VCP is recovered to proper 11v.

I think there is a high impedance connection from the driver chip GHB to the FET Gate pin, and this resistance limit the Gate charging current, and this caused the dip to the GHB signal.

Brian

Brian Dang said:

the driver chip GHB to the FET Gate pin,

Hi Brian,

I capture some waveforms in single mode (pic 1 to 3) and run mode (pic 4).

The abnormal crest does not occur every cycle.

Then I change IDRIVE to 120mA and 240mA, the waveforms are shown below. The rise time of VGS is smaller, but the abnormal wave is still there.

When I increase IDRIVE to a higher value (330mA/660mA) and try to make the motor move, the board seems to stop working or reboot.

James

Hey James,

There seems to be a lot of ringing present in the motor outputs, have you tried implementing an RC snubber in the schematic to mitigate this?

I also suggest taking a look at the our note on high power system design considerations. 48V is very high voltage for a 60V driver so the system must be very well designed to account for any parasitics.

https://www.ti.com/lit/an/slvaf66/slvaf66.pdf

https://e2e.ti.com/support/motor-drivers-group/motor-drivers/f/motor-drivers-forum/991693/faq-proper-rc-snubber-design-for-motor-drivers

Best,

Akshay

Hi Akshay,

Does it means that this weird crest is may also caused by ringing?

The new board which is being soldered has an RC snubber, and we will test it soon.

James

Hey James,

It is possible that the ringing is responsible and is coupling into other signals. Please let me know if the RC snubber is resolving this. 

I would also like to know if you saw any sort of damage on the IC or the MOSFET. Also were these GHX spikes seen during any specific tests, for example at startup only, or during high torque/current? Or were they always present?

Best,

Akshay

Hi Akshay,

These spikes are always present when the motor is spining. The signals are worse when using a more powerful motor.

Can you provide us some advice on choosing MOSFETs? We use a 48V motor with a maximum bus current of 90A or more.

James

Hey James,

So was there MOSFET damage/short?

From a BLDC perspective, when selecting a MOSFET, our major concerns are:

1. Voltage rating (VDS): This should be able to account for maximum drain voltage with some room for safety.
2. Drain current (ID): Be suited for the expected motor current to prevent damage.
3. Q_GD: The rise/fall time of the MOSFET is dependent on QGD. Trise/fall = QGD/ Isource/sink.
4. Gate to source threshold voltage (VGS (th)): The gate drive voltage of driver should be able to turn on the FET.

Best,

Akshay

Hi Akshay,

Thanks for your advice.

"So was there MOSFET damage/short?"

- Yes, The MOSFET was damaged during high torque and I raised some questions in this thread. We use another motor now while the signals are worse and weird.

---------------------------

We tested the new borad which designed with some capacitors,  RC snubbers, gate resistors and so on according to slvaf66 note.

We have not yet calculated the exact values of RC snubbers and gate resistors yet, but the signals has improved just not perfect. 

There was a lot of noise on INHA and I got some information suggesting single-point grounding. Should we use it in this case?

James

Hi James,

Sorry for the delayed response, the team has been on on holiday vacation. Akshay should be able to provide a response next week.

Regards,

Anthony Lodi

Hi Anthony,

Coool, have a good vacation.

James

Thanks!

Regards,

Anthony Lodi

ok, next we will try a partitioning ground.

James

Hey James,

Understood, but please understand that this would be helpful only if the noise is coming from the motor.

 I will close the thread for now. Please feel free to reopen the thread if you have more questions or start a new thread for new issues.

Best,

Akshay

James Wang said:

I capture some waveforms in single mode (pic 1 to 3) and run mode (pic 4).

The abnormal crest does not occur every cycle.

Hi James,

From waveform below, VGS is zero during highlighted in the circle, so what had driven the FET to turn on with SHB was more than 20V? Impossible!!! So I really don't think that the FET was actually turned on during the highlighted circle, but maybe the signal SHB was driven by the other phase of the motor, not the B FET. IOW, if the motor phases are disconnected then we don't see the pulse in the highlighted circle. I assume you have Hall sensor signals for commutation in this application, if so, can you disconnect the motor phase wires and then manually rotate the rotor and observe the same signal waveforms?

Brian

Hi Brian,
We use an angle sensor with spi communication. Do you mean to measure the motor phase signal directly instead of the SHx signal in board?
The new new board has just been soldered and we are going to test it.

Hey James, 

Yes please let us know how the testing goes with the new boards. In order to see where the spike is coming from lets capture a waveform of GHB, SHB, SHA and SHC.

best,

Akshay

Hi James,

James Wang said:

We use an angle sensor with spi communication. Do you mean to measure the motor phase signal directly instead of the SHx signal in board?

In my previous post, I pointed out that the highlighted circle shown SHB and GHB were going high the same time and so their delta = VGS = 0v, and this makes no sense as VGS=0 cannot turn on a FET to have SHB going high. Therefore, I don't think the driver is causing the highlighted signals. To confirm this, you need to disconnect the motor 3 phase wires off the board, then manually turning the motor shaft so the MCU reads the rotor position -- via Hall signals or angle sensor signals according to your design -- and MCU output the COMMUTATION signals (the PWM) to the driver to control the outputs. Then by displaying the driver output during this motor rotor turning and see if the said weird pulses change or not. If it looks clean normal then the weird pulse was not output by the driver but it was from something else.

Brian

Understood. Thanks for your detailed explanation, I will try it.

James

Hey James,

Good luck on testing!

Best,

Akshay

Hi Brian,

We tried your method and the driver output was clean normal. 

The new borad was also tested. The waveforms of GHx and SHx have improved and we found they can be better by reducing the dead time.

James

Hi Akshay,

No more questions, thanks for your help.

James

Hi Brian,

Thanks for your help and suggestion. We will do more tests especially at high load.

James