DRV8704EVM: FET Q7 and Q5 are shorted

Hi Michael,

Did you have any loads connected to the output when you were spinning the POT? If so, what type of load?

Was there any irreversible damage to Q7 and Q5 or the driver?

What is the device power supply voltage?

Did you capture any waveforms showing the voltage rise? Analyzing those waveforms may help understand what is going on.

If Q7 and Q5 are shorted, then it could mean cross conduction or a shoot-through short occurred. Whether or not this is caused by having slow decay mode and spinning the POT quickly, that's something we'll have to look into.

Slow decay, as its name indicates, takes longer for the current through the inductor to decay while the current is recirculating through the bottom two FETs. You can learn more about the different types of decay mode in this app-note.

Hello Pablo,

I had one motor connected to side B while side a was open. Both outputs set to PWM. The motor is a standard off the shelf 775. The extent of damage is not clear but the two FETs are shorted measuring roughly 10ohm from D to S. Does not appear to go farther then that. The power supply voltage was 13.5V. I did not capture the scope plot but I saw it rise higher then my supply voltage observed across the motor terminals (MB1 and MB2). I assumed some kind of shoot-through. et me know if you need me to capture more data.

Thanks               

Hi Michael,

I would like to see a waveform showing the VM voltage and current if possible while attempting to drive the motor. If there is shoot-through, there will be a large current rise in the VM line. Can you provide this waveform?

How did you measure the resistance between D and S of the FETs? Did you use a multimeter? Was the measurement done while the device was active and driving the motor?

So far all the evidence is pointing to a shoot-through as the cause of the damage but I'll have to take a look at the waveforms to fully confirm.

Hello Pablo,

I wont be able to get a screenshot until the new year. What I can tell you is once power is applied, the fault status led is tripped. Also using a IR camera, the two FETs get really hot when sitting idle. Is it still possible to drive something with the fault LED tripped?     

The resistance measurements was taken while everything is off and disconnected.  

Michael,

The H-bridge is disable whenever a fault is triggered. If a UVLO or thermal shutdown is triggered, the fault will clear and the driver will resume operation once the voltage or temperature rise above the safe limits. For other faults, the device needs to be rebooted to clear the fault.

It seems like the FETs are damaged if the temperature is high while not driving. Can you replace the damaged FET with a new one? 

Did you perform the test when in fast decay using the EVM with the damaged FETs or was a different EVM used?

Hello Pablo,

The plan for the next year is to replace the FETs and resume testing. I did do the same test in fast decay using the same EVM prior to slow decay and did not have any issues.  

So thinking about this more when I noticed the PWM voltage on the motor leads (MB1 and B2) rising way above my supply voltage could it be possible the amplitude exceeded the FET 60V rating. Would the FETs short if 60V was exceeded? The reason why I think this might be the case because I noticed while in slow decay mode and spinning the pot quickly is it would shutdown my power supply. Unfortunately, I was not monitoring any signals at the time. This happened twice before I switched to a physically larger power supply. Immediately after switching power supplies and while in slow decay mode spinning the pot quickly was when it shorted but it did not shutdown the power supply.  

Best regards,

Hi Michael,

Please send over the test results once you have replaced the FETs. It does seem like slow decay might be causing the problem. 

Just to confirm, how are you modifying the input signals to set the decay mode to either slow or fast? are you following the truth table on the datasheet (below). So if you want slow decay, one of the INx signals should be HIGH while the other INx signal is toggled by the POT. On the other hand, to set Fast decay (or Coast), one of the INx signals should be LOW while the other toggles between 1 and 0.

In regards to the power supply, I think the voltage rise causes the current through VM to rise above the power supply current limit which would lower the VM voltage. When you changed to a larger supply, the current limit was much higher which prevented the power supply from shutting down. Looks like the smaller supply was actually protecting the FETs. I suggest lowering the current limit of the power supply to protect the FETs for future tests. 

The driver internally monitors the VDS of each FET. If the VDS exceeds the value set by the OCPTH bits (max=1V) for longer than the deglitch time (OCPDEG), then an OCP fault will be triggered disabling the device. If the voltage across the FETs rising suddenly to above 60V for less than OCPDEG, the FET could get damaged without OCP fault triggering. Have you tried setting the OCPDEG and OCPTH to the lowest value? this will ensure an OCP event is quickly detected. 

Hi Michael,

Any updates from your side?

Hello Pablo,

Sorry for the delay, I am now just getting back into the office. Looks like its going to be slow going to start the New Year. Probably won’t be back in the lab for another week or two. As of the truth table, I just used the jumpers that the board came with and followed the table as you have highlighted. I have not missed with the OCP, just ran tests right out of the box. Figured it would not have any overcurrent problems since the motor was unloaded. I will get the FETs replaced and lower the OCP threshold before running a test. Should I start a new forum or continue with the current one?

Thanks,     

Hi Michael,

Thank you for the update.

Since it will be quite some time before you can provide data, let's close this thread. Once you are ready to post, just ask a new related question.