DRV8350R: Low Side FET on Phase C Hotter than other FETs

Lauren Sabra

Part Number: DRV8350R
Other Parts Discussed in Thread: DRV8350, CSD19536KTT, TIDA-010056

Hello,

I have designed a BLDC controller with DRV8350RH and after building up multiple PCBs, I am finding that the low-side MOSFET on Phase C is consistently hotter than all of the others (high and low). When I have the IDRIVE setting as low as it will go, the boards operate but low-side C is hot while the others are not. If I step up the IDRIVE, the motor runs for and a few minutes before stopping with a permanent fault. After further inspection the DRV8350 seems to be damaged; the resistance between the GLC and SLC pins drop to around 1K from 142K while it does not change on A or B. We really need to correct this issue promptly, I have attached the schematic and gerber files for the design below. Thanks!

Design Details:

FETs: CSD19536KTT

Mode: 1X PWM

1 FET per gate

BLDC Schematic.pdf 8831.BLDC.zip

over 5 years ago

0 Adam Sidelsky over 5 years ago

TI__Guru 65190 points

Lauren,

Is this design related to a TIDA? I see a TIDA number in the files you included.

The gerbers are not easy to read, is Phase C furthest away from the DRV/PWR/GND? Have you checked the Gate and source voltage with a scope comparing phase C to the other phases?

Regards,

-Adam

0 Lauren Sabra over 5 years ago in reply to Adam Sidelsky

Prodigy 125 points

Hi Adam,

Yes, the design is based off of TIDA-010056. Sorry about that, Phase C is actually the closest to the DRV/PWR/GND; Phase A is the furthest. I have attached some scope plots below:

There aren't any major differences sticking out to me when comparing C to the other phases. However, there are some large negative spikes on the low side gate signals of all the phases. Those negative spikes increase as the motor spins faster and faster.

0 Adam Sidelsky over 5 years ago in reply to Lauren Sabra

TI__Guru 65190 points

Lauren,

Can you confirm if the switch node SHx of the three phases is also dipping in voltage when the gates turn off? Can you share the schematic?

Regards,

-Adam

0 Lauren Sabra over 5 years ago in reply to Adam Sidelsky

Prodigy 125 points

The switch node voltage on SHx is spiking when the gates turn off:

I've attached the schematic, please let me know if you are unable to see it. Thanks!

5353.BLDC Schematic.pdf

0 Lauren Sabra over 5 years ago in reply to Lauren Sabra

Prodigy 125 points

Hi Adam, I've put together a PDF of the PCB layout, hopefully this is easier for you to view. Please let me know if you have any idea of what may be causing issues, thanks!BLDC_Layout.pdf

0 Adam Sidelsky over 5 years ago in reply to Lauren Sabra

TI__Guru 65190 points

Lauren,

Spiking on the gate and source imply there is too much parasitic inductance in your gate and source lines.

What is your source and gate trace width?

What is the length of your gate and source traces, are any of them longer than the rest?

Regards,

-Adam

0 Lauren Sabra over 5 years ago in reply to Adam Sidelsky

Prodigy 125 points

The trace widths are 10mils and the lengths do vary. The shortest path is GLC (~0.5in) and the longest path is SHA (~3.5 in). Is there any rule of thumb you would suggest following for the trace lengths? I just looked back at the TIDA-010056 reference design and it seems like the source and gate length vary from about (~0.25in to ~1.5in). The PCB I have created is over 2X the area and has room for parallel FETs for thermal benefits, so naturally the traces will need to reach further out but I can lessen the differences on the next spin. Thanks!

0 Adam Sidelsky over 5 years ago in reply to Lauren Sabra

TI__Guru 65190 points

Lauren,

Your traces may be too thin to carry the gate current needed. Please take a look at our layout guide here: https://e2e.ti.com/support/motor-drivers/f/38/t/746299

Trace length is less important than width but obviously the more similar, the better.

Are you currently using parallel FETs? If so gate resistance, even if small (<20 ohm) is needed so that the parallel FETs don't fight with each other.

Regards,

-Adam

Motor drivers

Motor drivers forum

DRV8350R: Low Side FET on Phase C Hotter than other FETs