DRV8302: occasional MOSFET damage from high-speed motion

scott klar

Part Number: DRV8302

Hi, we have this drv8302 driver chip in hundreds of industrial applications used with <1hp motors. On occasion a customer will experience problems and return a drive. A quick measurement of the MOSFETs on the driver stage shows that they are blown, shorted drain to source. The problem seems to occur after a quick motor motion with a high acceleration to high speed then stops quickly typical during initial drive setup and tuning adjustment.

We were able to experience the same failure at our desk. The motor jumped quickly due to tuning and executing a motion and the FET were blown as well as taking out our 48V, 10A switching power supply. I cannot tell if the driver chip is damaged as it is difficult to replace the FETs. In this instance we did not have any voltage clap device on the power supply for protection. Under normal use case these power supplies work fine. Occasionally the motor regen energy will fault the supply but it will come back on after some time. Our bridge design has flyback diodes (ST STPS1H100AF) across each FET. They can handle 100V, and 50A surge for 10ms, and 10A RMS.

Our design uses: 5 Ohm resistor on gate, and 0.005 Ohm shunt resistor (we allow 20A max). Have 2 390uF bulk caps on Vbus (PVDD).

FETs are NXP P/N: PSMN011-80YS, ratings: VDS=80V, Id=67A, P=117W, drain-source avalanche energy 121mJ, Vgs min -20V, max 20V

How can I pin down the root cause of the FET failure? Could there be a cause from the DRV8302?

Here are my thoughts on possible causes. Please comment on if they are likely to happen or not:

Voltage supply spike above 80V exceeding VDS max of FET.
The dv/dt of the motor exceeds the avalanche ruggedness of the MOSFET.
From TI BLOG..."Switch-node ringing occurs due to high dV/dt (slew rates)...This ringing can cause the switch-node voltage to drop below ground or rise above the supply, often violating specifications of the gate driver or power MOSFETs and thus leading to catastrophic breakdowns." Figure 2 shows a -4V ringing example. This MOSFET is rated for -20V so its hard to imagine this is being exceeded.
Crosstalk on signals, PWM? Gate?
Driver chip violating deadtime?
Power supply dv/dt too fast for driver chip? Datasheet states "Maximum supply voltage ramp rate" is 1V/us. With all the capacitance I would not think this would be exceeded.

Any help would be appreciated. Thanks.

over 8 years ago

0 Rick Duncan over 8 years ago

TI__Guru** 114905 points

Hi Scott,

We understand it is difficult to replace the FETs. It would help further debug if you can remove them to determine if the DRV8302 is still operational.
Is there some other method you can use to determine if the DRV8302 is working?
At a minimum can you power up the device, and issue an EN_GATE command? This can provide an indication the internal regulators are operational.

If the DRV8302 is operational, can you then replace the FETs to determine complete operation?

How do you stop the motor quickly from high speed?

Most likely cause of the ones listed:
Voltage supply spike above 80V exceeding VDS max of FET. <- this could take out the DRV8302 with its abs max voltage of 65V
Switch node ringing <-

Other possible causes:
Abs min violations of Vlogic? If ground rises at the DRV8302 with respect to the mcu ground, this could violate the spec.
FET ground goes negative with respect to low side FET GL_x signal, creating shoot through.
Violation of SOA of the FETs? Do you monitor the thermals of the FETs?

0 scott klar over 8 years ago in reply to Rick Duncan

Prodigy 70 points

Thanks Rick,

The motor is stopped by controlled motion (strong current, 20A, in opposing direction). The MCU ground is same as ground to shunt resistors on low side, it should be solid.

On a damaged board, I removed power from the high-side FETs, measured EN_GATE going high, but see nothing on the FET gate signals. The internal voltage regulator is maintaining 5V powering all the electronics. Our firmware is throwing an Overcurrent fault. The OCTW pin is high at power up and stays high. I think this means the driver is bad?

I took some measurements on a running unit. I ran a square wave velocity move which is a velocity command back and forth. This drives full 20A to accelerate the motor until it reaches speed. there is a weighted disc on the motor for load, 5x the rotor inertia value. I have a voltage clamp protecting the power supply.

I measure Gate-Source voltage on the scope for the low side FET. I do see a negative spike up to -3.3V shown here in first image, not near the 20V required to damage the FET. But can this damage the driver output? I also captured an unexpected positive pulse at +2.6V in the second image. I also measured drain to source voltage on lower FET and it touched -4.6V, but technically I shouldn't place the ground clip from the probe on the source pin as this would ground out the signal to the shunt current measurement.

any thoughts?

0 Rick Duncan over 8 years ago in reply to scott klar

TI__Guru** 114905 points

Hi Scott,

The 5V regulator is not used to power the gate driver circuitry. AVDD, DVDD, and GVDD are used to power the gate driver circuitry.

Is nFAULT low prior to EN_GATE high? If so, the device may be damaged.
Once EN_GATE is set to high, does nFAULT go low and then back high? This is an indication AVDD, DVDD, and GVDD are at the expected voltages.
If nFAULT remains low, please check AVDD, DVDD, and GVDD voltages.
AVDD should be ~6.5V, DVDD should be ~3.3V and GVDD should be ~10.5V. Without these, nFAULT may be low.

nOTCW high is normal operation. Please refer to Table 5 of the datasheet for possible faults when nFAULT and nOCTW are asserted.

0 Murat_ over 8 years ago

Prodigy 150 points

I think your problem is related with regeneration. When you supply motor driver from a DC/DC converter such problems occur because usually DC/DC converters can't sink current.

The solution is add Rshunt resistor with suitable rating and burn the excess energy on it when the supply voltage rises above a limit.

You may read the following application notes.

0 scott klar over 8 years ago in reply to Rick Duncan

Prodigy 70 points

I measured two boards, both have no voltage at pins: AVDD, DVDD, GVDD. One board you can hear the motor energize with a faint high pitch noise. I believe the FETs are degraded on this board but not shorted. Even without internal regulators working on the drv8302 somehow the PWM still attempts to drive the gate of the FETs. I will include these screenshots below.

Vgs lower Fet: