TMS320F28027F: FETs get REALLY hot even with no motor connected

Asif Ahmed

Part Number: TMS320F28027F
Other Parts Discussed in Thread: MOTORWARE, DRV8301

Hello -

I have a new PCB and it -should- be much better from an EMI standpoint then my previous PCB. All gate drive and gate sense traces are routed in thick pairs, differential kelvin signals from current shunts, dedicated ground plane (4-layer), etc....

However, the board gets REALLY hot (over 150 F) within 1 minute of powering it up and starting Motorware. This is even without a motor connected....

Motor ID runs fine, and I can spin the motor under speed control and torque control.

I've scoped the gates - and something that seems strange - the high-side fet is being driven with +/- 30 V and the low-side is being driven with +/- 5V. Is this normal?

Any helpful tips on what to look for here?

Thanks much.

-asifjahmed

over 9 years ago

0 Asif Ahmed over 9 years ago

Expert 1570 points

An update on this: I probed my other (no heat issue) PCB and realized that it also uses the same mentioned gate drive voltages. So i suspect that is not the issue.

A few more clues:
-The motor makes a humming noise even while idle with the problematic board. It seems like it is driving too much current
-I am using EPWM4 with this board (phase A-PWM4, phase-B-PWM2, phase C-PWM1). This is different than the other board. I have made the necessary changes in hal.c
-This board gets 3.3V from the buck regulator in the DRV8301 directly. The other board outputs 5V and goes through an LDO to get 3.3V

Please advise what else to check!

Best,
-asifjahmed

0 Yanming Luo over 9 years ago

TI__Guru** 117270 points

1. Why you said motor ID runs fine and motor spin well since the board hot without connecting motor? Where is heat source on the board? Power supply? Or FET?
2. Is the dead time enough for driver and FET? If no, increase the dead time, if yes, you have to check the gate driver circuit.

0 Asif Ahmed over 9 years ago in reply to Yanming Luo

Expert 1570 points

Heat is from the FETs. As I explained in my post.

Dead time is set using 1.0 ohm resistor. This is same as ref board and same as my other PCB without the heat issue.

0 Asif Ahmed over 9 years ago in reply to Asif Ahmed

Expert 1570 points

Just learned something:

When I change the DRV8301 gate drive current to 0.25A (from 0.70A) the heat issue goes away...

On my other layout (same circuit!) this is not a problem.

What could cause this?

PS. There is a slight difference in this board - I use 10 ohm gate resistors. Other board uses 4.3 ohm

0 Asif Ahmed over 9 years ago in reply to Asif Ahmed

Expert 1570 points

Another update:

The high-side gate drive voltage varies with my input voltage. I believe this is causing my issue at 50V, as the high-side gate drive is +/- 30V. At 25V input, the high-side gate voltage is more like +/- 15-16V

0 Rick Duncan over 9 years ago in reply to Asif Ahmed

TI__Guru** 114905 points

Hi Asif,

This sounds like you could have some type of deadtime issue in the FETs.

Please compare the high and low side gate voltages (the voltage between the 10 Ohm resistor and the gate of the FETs.

0 Asif Ahmed over 9 years ago in reply to Rick Duncan

Expert 1570 points

Thanks Rick. Per my post on the Motor Drives forum, the High-side gate voltage is around 60V and the Low-side around 10.5V. Or are you referring to something else?

0 Rick Duncan over 9 years ago in reply to Asif Ahmed

TI__Guru** 114905 points

Hi Asif,

Yes, I am referring to something else. It is important to examine the voltages and confirm that only one FET is on.

In some cases this can be due to dead time issues. In other cases it could be due to noise coupling, where turning on the high side causes the low side FET to also turn on.

0 Asif Ahmed over 9 years ago in reply to Rick Duncan

Expert 1570 points

OK here is a scope capture of my gates (probes are directly on FET legs). All 3 phases looked the same:

This isn't quite as clean as the signal on the ref board, but pretty close. Your thoughts?

0 Rick Duncan over 9 years ago in reply to Asif Ahmed

TI__Guru** 114905 points

Hi Asif,

What does the ref board look like? Can you zoom in further on the falling/rising edges?

0 Asif Ahmed over 9 years ago in reply to Rick Duncan

Expert 1570 points

Absolutely - will get zoomed captures of both my board and the ref board in the next hour or so.

Really appreciate your help Rick!

0 Asif Ahmed over 9 years ago in reply to Rick Duncan

Expert 1570 points

Rick -

Here are the better scope captures. All captures were done with same motor at 1Krpm with very low load. Yellow is high-side and blue is low-side:

Ref board - rising edge (run/single):

Ref board - falling edge (run/single):

Custom board - rising edge (run/single):

Custom board - falling edge (run/single):

0 Rick Duncan over 9 years ago in reply to Asif Ahmed

TI__Guru** 114905 points

Hi Asif,

Is the series resistor 4.3 Ohms for both ref board and custom in the scope captures?

0 Asif Ahmed over 9 years ago in reply to Rick Duncan

Expert 1570 points

For the scope captures I used the DRV8301_REVD board and for the custom i used my latest PCB.

Ref board has 1.0 ohm gate resistors, custom board has 10.0 ohm

Thanks Rick!

-asifjahmed

0 Asif Ahmed over 9 years ago in reply to Asif Ahmed

Expert 1570 points

Bump!

0 Harry Kroonen over 9 years ago in reply to Asif Ahmed

Intellectual 260 points

I have never used the drv8301, but datasheet says with 1 Ohm resistor deadtime 50nsec, and your screenshot showing ~400nsec of rampup / FET switching time could be a problem / bad combination. drv8301 mentions current shoot-through protection so maybe deadtime is not your direct problem / protected by other feature. But slow FET switching could mean long time in linear region and much dissipation.

C2000™︎ microcontrollers

C2000 microcontrollers forum

TMS320F28027F: FETs get REALLY hot even with no motor connected