L6470 "blows up" in certain operating mode

[c.schmidt]

We are using a L6470 to drive a stepper. We use the SW-Pin to hold the motor in stop condition. The SW-pin causes a hard stop on falling edge. To prevent the motor from running again also if the firmware would send a run command we provide a clock signal on SW-pin with roundabout 2,7kHz.

Now it happens twice that the driver burst into flames.

What happens exactly during the hard stop? From the view of the damage I think this is no overcurrent on the motor windings, but an cross-current through both high side and low side MOSFETs.

Normally I would have expected the driver itself is safe for cross-currents. Could it be, that our clock signal on the SW-pin can bring the driver to an undefined condition wher we have both mosfets turned on?

Do you have any other ideas, suggestions?

[Javier1]

>>Normally I would have expected the driver itself is safe for cross-currents

Why would you expect this

>>Could it be, that our clock signal on the SW-pin can bring the driver to an undefined condition wher we have both mosfets turned on?

You answered yourself.

Time to plug in some logic analyzer/oscilloscope in your gate driving signals

[c.schmidt]

Hi Javier,

thanks for your reply.

I have no access to the gate driving signals. The L6470 is fully integrated, therefore I expect that the chip itself should care that no cross currents happen.

Christoph

[Javier1]

sorry i didnt have experience with this L6470 chip, i tought it was a generic semiconductor bridge, i spoke too son.

did you ruled out the usual suspects?

• Over voltage on Dc bus
• Motor inductance too high and spining too fast

[c.schmidt]

I have to say no. But I don't think we have an overvoltage here. The system is powered by an ACDC-24V-power supply.

In the case of "accident" the motor does not spin at all.

Spinning is not the problem here. The motor runs reliable. The problem are the undefined or error states:

For example:

• µC is in debug state and stops communication or is resetted.
• "safety" clock signal is added o the driver to hold the driver in stop condition no matter what the µC does.

The second burning happened like this:

• motor not running
• I disconnect the connector to the µC. In this case, the driver gets no more commands from the µC and at the same time the communication watchdog triggers and applies the clock signal to the switch input (i hope this was working in this situation, but I was not able to measure the clock)

--> The L6470 burns immediately....