I've come across some weirdness in the motor control PWM module on the
dsPIC33fj128MC804 and I'm hoping for some useful suggestions.

I'm using the same processor in several different designs and I'm having
what appears to be the same issue, at least the symptoms are the same.

One of the designs is running a three-phase BLDC motor (about 300 watts)
and the other is running a smallish DC brushed motor (a couple of different
motors here, but both under 30 watts)

It's blowing up the FET drivers. The weird part is that every failure we've
had is associated with the same PIC output channel - PWM1. Four failures in
the three phase design and at least two on the brushed design and they've
all been on the same output channel. (yes, that's a smoking gun coming from
the software side, more on that in a bit...)

Much of the hardware designs are cut and paste of each other. The power is
coming from a 28V Li-Ion battery, the regulators are the same National/TI
LM2675 parts, the FET drivers are the same Fairchild FAN7382. The three
phase unit has massive TO-220 FETs and the other has some smaller SOIC 8
parts.

The failures seem to be happening in the mid-point of a run, at relatively
low loads.

The first failures had the high-side FET shorted, the current sense
resistor open and FET driver visibly damaged (bubbles, cracks and smoke)
The leading theory at the time was that a voltage spike was killing the
(40v rated) high side FET and the rest was a cascade failure. We built up a
test batch with 100V rated FETs and managed to trigger another failure -
this time the FETs are fine, but the FET driver is shorted (HO to VS) and
the current sense resistor is open again (takes more than 20 A to open that
resistor) So, the FET first -> cascade theory can't be.

The other thought was that we were exceeding the Vcc rail on the FET
drivers. So this last test unit had an added 10 ohm pass resistor and 16v
ZENER clamping the Vcc rail. Didn't stop the failure. So the exceeding FET
driver Vcc doesn't stand up either.

So, I'm down to software. Try as I might, I can't imagine a scenario where
software can destroy the FET drivers - The FETs sure, a little
shoot-through will let the smoke out. But just the driver? And the FETs are
fine...

Add to this some issues with the Microchip documentation. Section 14 of the
FRM was edited sometime before 2010 and the newer version shows a PWMLOCK
bit in the FOSCSEL configuration register - earlier versions do not. Though
the processor header file I downloaded with C30 version 3.31 has no
reference to the PWMLOCK bits or the associated PWMKEY functionality.

Suggestions greatly appreciated.

Thanks,
-Denny
--=20
http://www.piclist.com/techref/piclist PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist
.