TMS320F28069M: Custom Board Design failing Lab 01c

 Please try to follow the steps to verify the hardware, you may check if step 2 and step 3 are right since you can run the lab01b.

1. Change the PWM and ADC configuration in hal.c refer to the guide, motorware_hal_tutorial.pdf.

C:\ti\motorware\motorware_1_01_00_18\docs\tutorials

2. Set correct parameters based on the hardware board in user.h

#define USER_IQ_FULL_SCALE_VOLTAGE_V           (xx)

#define USER_VOLTAGE_FILTER_POLE_Hz               (xx)

#define USER_ADC_FULL_SCALE_VOLTAGE_V       (xx)

#define USER_IQ_FULL_SCALE_CURRENT_A           (xx)

#define USER_ADC_FULL_SCALE_CURRENT_A       (xx)

3. make sure that the sign of the current coefficient in HAL_readAdcData() in hal.h matches the current sensing circuit. Refer to chapter 5.2.2 Current Feedback Polarity in InstaSPIN user's guide (SPRUHJ1, https://www.ti.com/lit/spruhj1) to set the sign of the current scale factor.

static inline void HAL_readAdcData(HAL_Handle handle,HAL_AdcData_t *pAdcData)

{

   _iq current_sf = (+/-)HAL_getCurrentScaleFactor(handle);

}

4. Follow the instaSPIN lab guide to use lab01b and lab01c to verify your own hardware since you are not using the TI EVM kits and then run the subsequent labs.

5. Using the lab02b or lab02c to identify the motor parameters, and use the identified parameters to run the motor if the current and voltage sensing signals are verified and good.

Thanks Yanming. I'll try everything you've listed.

Best Regards,

Dave

Hi Yanming,

I've been spinning these motors for some time, using the DRV8301-HC-EVM evaluation kit. I'm pretty sure I have all the parameters in user.h at good values (may need to tweak some of them a little). The sign of the scale factor is correct (negative feedback, so offsets are positive). I can run the motor in Lab01b. The graphs generated don't bear much resemblance to those in the Lab guide. They're very "choppy" looking, as though the sampling rate just isn't fast enough. But that's true whether I perform the lab with my custom board or with the evaluation board.

I've noticed that the waveforms associated with the FETs are somewhat different on my board, as compared to the evaluation board. I've attached some screen shots from the scope. It looks like the EVM has a much softer turn on at the gates. Maybe the gate capacitance on the FETs I've chosen is too low? I haven't compared the actual values from the data sheets, but I'll do that next. Also, the transient ringing on the current sense line is much larger and wilder on my board than on the EVM, but that may be a consequence of the FETs snapping On and Off so rapidly.

Thanks for your help.

Best Regards,

Dave

You may have a look at the datasheet of DRV8301, and find some application notes on TI website. You can follow the recommendation to design the driver board for motor control.

Hi Yanming,

Thanks for the reply. However, I was hoping for some help regarding the information provided above.  Do you think that FETs that turn On / Off too quickly could cause the problems I described in the first item of this exchange? Does the ringing on the source pin of the low FET (high end of current sense resistor) look excessive to you? Is there anything I can do to mitigate this problem without entirely redesigning the board? That is, could I put a capacitor on the gate, or something like that - something that would get me through testing This board? Then I could make any required changes (perhaps selecting different FETs)  permanent on the next spin. Would adjusting dead-time help with these issues? What Application Notes do you suggest?

Thanks,

Dave

Thanks Yanming!

Is there some way I could transfer this whole conversation over to a DRV8301 forum?

Best Regards,

Dave

Thank you. So should I look in the Motor Drivers forum for this topic?

Best Regards,

Dave

Hi Aaron,

Thanks for the advice. I agree that the gate capacitance is much lower for the FETs I'm using, compared to the DRV8301-HC kit. They seem to be way too snappy in almost every way. Also, the Rds(on is considerably higher:

                     BSO150N03MDG (my FET)             SUM110N06-3m9H (used on EVM)
Rds(on)                    0.015 ohm                                              0.0039 ohm
Ciss                            970 pF                                                  15,800 pF

I will definitely try out your suggestions. By "gate drive setting," I assume you mean the peak gate drive current, which is adjustable on the DRV8301?

I've been reading your presentation How_to_do_BLDC_Schematic_Review_and_Debug.pptx, along with the commented mock schematic, etc... You've put together a very helpful package there - thank you. I noticed that you mentioned a couple of times

"Do NOT connect SNA, SNB, SNC directly to GND; this will sense GND noise rather than the voltage at the low-side of the shunt resistor." The EVM kit appears to make a direct connection to ground, but they may have taken other measures to mitigate noise.

In my case I'm connecting SP1, SP2 and SP3 directly to ground, since I'm using negative feedback. So, how should I avoid a direct ground connection? Should I put another resistor in series with the sense resistor? The sense resistor is 2 milli ohms. Would a duplicate resistor be the way to go?

Thanks,

Dave

Hi Aaron,

Ok - I see what you mean. I'm already using Kelvin connections to just "sip" off the main current carrier. But the diff-amp is not nearly as close to the source as shown in your drawing. I'm having trouble locating just how to change the peak current for the 8301 in the EVM code, so I'll probably start with increasing the gate resistance - I'm using 1 ohm right now. Thanks for the link and thanks for your help! Have a great weekend.

Dave

Hi Dave,

Sure thing! Please mark as resolved if this solved your issue. Have a good weekend!

THanks,
Aaron

Hi Aaron,

I did finally figure out how to change the peak gate current. There are evidently 3 values available 1.7A, 0.7A & 0.25A. It was set to 1.7A by default in the InstaSpin code. Tried both of the other two, to little, if any, effect. I also doubled the gate resistance from 1 ohm to 2. I think maybe the noise at the current sense amps was damped down a little, but my original problems still remain (see first entry in this discussion). Should I increaINFNS16227-1.pdfse the gate resistance even further? I've attached the data sheet for the FETs I'm using; perhaps you can suggest a gate resistance, or provide me with a procedure for determining what resistance I should use?

Thanks,

Dave

Thanks Aaron - just knowing a general range is a big help!

Dave

Sure thing Dave, let me know if any more help is needed. 

Thanks,
Aaron

Hi Aaron,

I've tried various values of Rbase. Here are the qualitative results:

100 ohm -  Supply current went up considerably and the system shut down after a few seconds.

                   Didn't capture any waveforms; I was too hung up on getting something to work.

                   Besides, they were gone in an instant.

                   From what I could see, though, the gate drive voltage climb was much more moderate, especially toward the peak,

                   which makes the behavior all the more puzzling. Maybe a large spike develops across the base resistor, since the R

                   is so large. Anyway, I didn't dare take it any higher.

47 ohm -  Pretty much the same as 100 ohm.

18 ohm  - Acts about the same as when a 2 ohm resistor in place - very snappy turn-on and turn-off.

                 Didn't think to capture any waveforms here either.

27 ohm - Climb of waveform at gate appears gentler than with 1 ohm or 2 ohm. Much more rounded at the top - rise time about

                a quarter of a microsecond.

               The resistor seems to have made less of an impact on turn-off, however. Looks like a bit less ringing on the current sense signal.

               The motor, however, is behaving as badly as ever - very shaky and haltingly turns a fraction of a revolution at a time.

               It also gets pretty hot after a few minutes. This is while running lab 01c of the InstaSpin Labs.

Thanks,

Dave

Hi Dave,

The GLx rise time looks good (~400ns). The ringing at the sense resistor does not look good, there is a lot of ringing that does not settle until after 1us. Which gain setting is used, and does it match with the table below?



If the motor is behaving worse, my guess is that current feedback is not good. I think that this was also the original issue on the DRV8301-HC-EVM kit. To confirm there's not an issue at the powerstage, monitoring GH_x, SH_x, GL_x, and SL_x for each phase is good to ensure the gate drive waveforms switch correctly and not switch too fast. 

Is deadtime being implemented? If so, can dead time be removed and does this help?

If the waveforms appear good with no known issues, I would plot SNx vs SPx to see if there is large ringing at the resistor when current is switching. Ringing can be caused to parasitic inductance in the layout of the board, grounding, or placement/size/package of the sense resistor(s). A filtering cap between SNx and SPx (~1nF) may help too with dampening ringing. Also applying an RC filter at the output of the CSA (SOx) can help filter noise from the signals from motor switching. 

I think investigating schematic and layout may also help, can you share with me? Feel free to send over PM as well if you would not like to share publicly. 

I will be out of office until 1/3/2022, so please be patient with any replies on this thread. 

Thanks,
Aaron

Hi Aaron,

Do you mean the gain of the CSA amps? They're set up with a gain of 20, so the rise time should be 600ns, as opposed to 450 or so.

Dead-time is being used in a very minimal way. There's a 1 ohm resistor from DTC to ground. Would it make much difference if I shorted that out?

The ringing shows up everywhere to some degree, including power and ground...

I have dummy locations for RC filtering at the CSA outputs. I'll give that a try.

I think you're right; it's probably a layout issue. At the very least, I'm probably going to have to bring the DRV8031 closer to the sensing/FET bridge, and route the SNx and SPx lines more carefully.

I'll send you the schematic & layout via PM.

Thanks for your help & Merry Christmas and a Happy New Year!

Best Regards,

Dave

Hi Dave,

Thanks for posting your question to e2e - please give us some time to investigate this further

Responses may be delayed a bit due to the team being out during holiday timeframe 

Best Regards, 
Andrew

Hi Andrew,

Aaron already explained that he'll be out 'til Jan 3rd. Starting tomorrow, I'm out until the 4th, so no problem.

Thanks,

Dave

Hi Dave, 

understood and hope you have a nice holiday break!

(f.y.i for e2e metrics, we have to have the latest response in the thread, so that the thread is marked as attended/resolved) 

Best Regards, 
Andrew 

Hi Aaron,

Happy New Year! Hope you had a good vacation. 

Just got back from xmas break today. I sent you the schematics, etc..., using the "Send a private message" selection when hovering over your photo, above.

If you didn't get it let me know, and I'll resend. I don't have the layout in front of me right now, but I'm pretty sure the gate drive runs are long and skinny.

Thanks for the advice and the App notes.

Beat Regards,

Dave

Hi Dave,

Can you resend? I do not have anything in my inbox at the moment. 

Thanks,
Aaron

Tried it again, and noticed that when I click on "Send a private message", it sends me to a page that doesn't refer to Aaron Barrera, instead "admin" is shown at the top.

Dave

Hi Aaron,

So is there some other step I'm missing? Should I email subscribe? Or request friendship?

Dave

Did you get it?

Thanks for all the advice! The first two I can definitely fix. For the third, could I use the same FETs, just widen the traces once I get a little ways out from the pin?