Author Topic: Failed electronics from customer  (Read 783 times)

0 Members and 1 Guest are viewing this topic.

Offline mattko

  • Contributor
  • Posts: 22
  • Country: 00
Failed electronics from customer
« on: October 07, 2019, 06:45:42 am »
Hello all,

In our company we just had a returned electronics from our customer. It is a three phase inverter for motor control. The problem is that high and low side MOSFET are in short circuit, which means 0,5 Ohm between VDC and GND input connections. Gate resistor is 10 Ohms and driven from internal driver in the MCU. Due to company policy, I can't give every information there is.

We allready performed CT scan, thermal picture, potential root cause testing etc. But we did not found the problem.

Any suggestions on how to find the root cause would be greatly appreciated.

Best regards!
 

Offline Siwastaja

  • Super Contributor
  • ***
  • Posts: 2010
  • Country: fi
Re: Failed electronics from customer
« Reply #1 on: October 07, 2019, 06:58:24 am »
Look at a working unit instead.

Test and verify that your overcurrent protection kicks in properly.

Measure switch node ringing (peak voltage) using high bandwidth (at least 100MHz?) oscilloscope with full load. Is it close to the MOSFET ratings?

Measure gate voltages with full load. Does it approach the limits, even for very short times? How much negative does it go? Some gate drivers die at slightly negative gate voltages, say, just -5V will kill some.
 
The following users thanked this post: mattko

Offline moffy

  • Frequent Contributor
  • **
  • Posts: 340
  • Country: au
Re: Failed electronics from customer
« Reply #2 on: October 07, 2019, 08:33:59 am »
You might not find out, if it was say an overvoltage event/spike(externally induced). Just repair it. If you get multiple similar failures then you have a problem.
 
The following users thanked this post: Kilrah, Jacon

Online MagicSmoker

  • Super Contributor
  • ***
  • Posts: 1178
  • Country: us
Re: Failed electronics from customer
« Reply #3 on: October 07, 2019, 12:44:33 pm »
In our company we just had a returned electronics from our customer. It is a three phase inverter for motor control. The problem is that high and low side MOSFET are in short circuit...

I bet you are using one of those shitty level-shifting driver ICs that require a bootstrap charge pump for the upper MOSFET, aren't you? Those things are notoriously sensitive to high dV/dt leading to all sorts of mysterious failures at frustratingly random times.

 

Online Psi

  • Super Contributor
  • ***
  • Posts: 7280
  • Country: nz
Re: Failed electronics from customer
« Reply #4 on: October 07, 2019, 01:22:32 pm »
Gate resistor is 10 Ohms and driven from internal driver in the MCU.

Is this a specific "gate driver" built into the mcu designed to drive mosfets? or just a GPIO ?

Do you have TVS/Zener diodes between the gates and sources to prevent Vgs spikes from exceeding the max and damaging the mosfets?
Greek letter 'Psi' (not Pounds per Square Inch)
 

Online MagicSmoker

  • Super Contributor
  • ***
  • Posts: 1178
  • Country: us
Re: Failed electronics from customer
« Reply #5 on: October 07, 2019, 03:34:19 pm »
Gate resistor is 10 Ohms and driven from internal driver in the MCU.

Is this a specific "gate driver" built into the mcu designed to drive mosfets? or just a GPIO ?
...

Oh, I totally missed that... I sure hope the OP isn't trying to directly drive a MOSFET from an MCU GPIO pin (well, not one bigger than a 2n7000, anyway)!
 

Online Psi

  • Super Contributor
  • ***
  • Posts: 7280
  • Country: nz
Re: Failed electronics from customer
« Reply #6 on: October 08, 2019, 01:43:22 am »
It's fine as long as the frequency is low and you have enough dead time.
Greek letter 'Psi' (not Pounds per Square Inch)
 

Offline mattko

  • Contributor
  • Posts: 22
  • Country: 00
Re: Failed electronics from customer
« Reply #7 on: October 08, 2019, 06:43:20 am »
Test and verify that your overcurrent protection kicks in properly.

Measure switch node ringing (peak voltage) using high bandwidth (at least 100MHz?) oscilloscope with full load. Is it close to the MOSFET ratings?

Measure gate voltages with full load. Does it approach the limits, even for very short times? How much negative does it go? Some gate drivers die at slightly negative gate voltages, say, just -5V will kill some.
Overcurrent protection is OK, but we are not sure if in some "special cases" it can cause a problem.
Switch node voltage under normal operation is well below MOSFET rating.
Gate voltage goes to -1 V. The driver alowed voltage is down to -8 V.

I bet you are using one of those shitty level-shifting driver ICs that require a bootstrap charge pump for the upper MOSFET, aren't you? Those things are notoriously sensitive to high dV/dt leading to all sorts of mysterious failures at frustratingly random times.
Correct. We use a bootstrap capacitor to double the voltage to open the high side MOSFET. Is there some documentation regarding bootstrap and dV/dt problems?

Is this a specific "gate driver" built into the mcu designed to drive mosfets? or just a GPIO ?

Do you have TVS/Zener diodes between the gates and sources to prevent Vgs spikes from exceeding the max and damaging the mosfets?
It is a gate driver in the MCU. MCU is specialized in motor control aplication.
We do not have a zener to protect Vgs over/under voltage due to cost reasons. But we have a TVS input diode in the circuit, so an overvoltage to the MOSFETs is unlikely.

It's fine as long as the frequency is low and you have enough dead time.
Frequency is 13kHz and dead time is 400 ns. I think this is enough.
 

Online MagicSmoker

  • Super Contributor
  • ***
  • Posts: 1178
  • Country: us
Re: Failed electronics from customer
« Reply #8 on: October 08, 2019, 09:25:10 am »
Correct. We use a bootstrap capacitor to double the voltage to open the high side MOSFET. Is there some documentation regarding bootstrap and dV/dt problems?

There's lots of documentation available such as this PDF from OnSemi which I pulled at random from a duckduckgo search and verified it discusses the issue: https://www.onsemi.com/pub/Collateral/AND9674-D.PDF
 
The following users thanked this post: mattko

Offline aiq25

  • Regular Contributor
  • *
  • Posts: 168
  • Country: us
Re: Failed electronics from customer
« Reply #9 on: October 08, 2019, 04:48:37 pm »
I bet you are using one of those shitty level-shifting driver ICs that require a bootstrap charge pump for the upper MOSFET, aren't you? Those things are notoriously sensitive to high dV/dt leading to all sorts of mysterious failures at frustratingly random times.
Correct. We use a bootstrap capacitor to double the voltage to open the high side MOSFET. Is there some documentation regarding bootstrap and dV/dt problems?

Is this a specific "gate driver" built into the mcu designed to drive mosfets? or just a GPIO ?

Do you have TVS/Zener diodes between the gates and sources to prevent Vgs spikes from exceeding the max and damaging the mosfets?
It is a gate driver in the MCU. MCU is specialized in motor control aplication.
We do not have a zener to protect Vgs over/under voltage due to cost reasons. But we have a TVS input diode in the circuit, so an overvoltage to the MOSFETs is unlikely.

I worked on a couple of 12V to 120/230 Vrms inverter module projects that initially used discrete charge pump circuits for the H-Bridge to drive the AC output. They did not work well under high inrush loads. They didn't have enough drive and the frequency response needed to turn OFF the MOSFET's under high loads. Our senior/lead engineer changed to isolated gate drives (not necessary but that's what he did) and I copied his design for the modules I worked on. That solved a lot of issues we were having with the H-Bridge.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf