All we did was prevent the code from sending the two extremes of the pwm signal within N duty cycles of each other. More specifically, we try to ensure that only a maximum of 12V potential is seen across the motor outputs on the Jaguar. This keeps the motor from going full forward to full reverse in a split second, which has led to some teams seeing a 24V difference (from -12V to 12V) for a split second across the motor output terminals. This trips a fault circuit on the Jaguar, which takes 3 seconds to reset. I suspect that it also severely shortens the life of the Jaguar. Further discussion on this topic by some very highly reputable sources is
here.
So I'd say you should blame software, not the hardware
.