Stepper Motor H Bridge High Side N-Mos Burn Problem

Question

Last year, i designed a stepper motor driver card with DRV8711 and CSD17506. I am driving my stepper motor with Vm=24 Volts and motor current Im= 2.14A. Texas Instruments gate driver drv8711 is driving both high and both sides. The design schematic values are same the as DRV8711EVM user guide. Here are my H-Bridge schematics.

My company is using this card in our products and at least 100 cards worked great and there was no problem until new Mosfets arrived. We get these mosfets from digikey.

The problem is, when I powered 'MOTOR', one of the high side Mosfets burn suddenly, some of them are working for 1 second and some of them are working for 2 hours, but all of them are burning in the end. Only high side mosfets, and mostly Mosfet1 in schematic.

I first thought about soldering, and solder mosfets with special care and problem decrease but not removed.

I thought about if there is a problem about components and ordered new ones from Digi-Key again and the same problem occurs. Resistance of drain and source of burned mosfets are nearly 0\$\Omega\$.

Next I thought about if there is a version change in CSD17506 but could not find any change in the datasheet or on Texas Instruments forums.

When I solder older ones they are working well, but the new ones are getting damaged.

Then I thought about using gate resistors.

Results:
Gate resistors at only at high side mosfets gates = 20\$\Omega\$. The problem continued but worked for only a little longer.

Gate resistor at all mosfets gates = 240 \$\Omega\$. Problem solved but mosfets generated a lot of heat and step motor is vibrating and working a bit noisy and vibrating. While motor driver enabled any driver keeps stepper motor at a position there is still noise.

I measure V_DS of mosfet while driving my motor, it sometimes reaches 30 V.

One of TI experts recommended that I change my mosfets to 60 V. I changed my mosfet and problem removed. And also he recommend me to use gate resistors only at high side no larger than 20 \$\Omega\$. And also extend the dead time but dead time is already at its highest value which is 800ns

Although I solved my problem with 60 V rated mosfets, the main problem is I have approximately 1500 Mosfets and each cost $3. I need to use them in some way.

I already used 20 \$\Omega\$ at high side but problem decreased but continues.

What can you recommend me?

What can be the main reason?

I also thought to use flyback diodes at mosfets but I dont have any space to use any additional diodes at mosfets on my PCBs.

Can you help me to find and solve the problem?

Answer 1

I guess the problem is the back EMF of the motor. Just make sure the power supply never goes near 30V and you will be OK. You can use additional MOSFET + zener to shunt the spikes within 28V for example.

You may try adding 15V zener between gate-source of all the MOSFETS, although I don't think the gate voltage is the problem.

Adding gate resistance makes MOSFETs switching a bit slower, which removes some voltage from inductive peaks here and there, but since the back emf is the real problem, the effect of it is marginal. DRV8711 has option for adjusting the gate current if I remember well - you may play with it instead.

Answer 2

There was a similar topic here recently. Use only 0R on gate. For EMI just rout your PCB according to best practice thumb rules. Your problem looks most like dead time issues.

Full bridge HVDC to modfied square sine - random half side MOSFET burn

Here, this is the other guy with resistors on gate.

Answer 3

The problem you are facing is related to the rating of the FETs. NexFets are probably the fastest ones you could find at the moment: this means that the 1st spike is quite high (around double VIN) ==> 60V rated mosfet must be used on both HS and LS of the bridge. You could still use 30V Fets, but you have to apply a huge gate resistor of the HS, and this will spoil your system efficiency; furthermore the high switching losses on the HS could also lead to thermal issues.