Chris
Pic I got from here
http://abertoatedemadrugada.com/2013/04/o-que-e-o-pwm.html Top row, if you imagine the pulses are right to left in time. the LH edges are all the same.
So, the frequency is F = 1/t. This is the base frequency, otherwise known as the PRF. ( Pulse Repetition Frequency ).
To take an easy eg. PRF is 1 kHz, if I have a pulse length of 1mS, the trailing edge of pulse 1 will be co-incident with the leading edge of pulse 2. ie. It's on all the time. 100% modulated.
If your doo-dah is 1.2kHz, then that is well within the audible range.
I do PWM with pic 16F886, and you can get a PRF of several kHz which takes ( hopefully ) most of the squawk well above audible.
BTW that motor is not on a desk top is it? Not got a sounding board effect have you?
I have some steppers that kick up the most hideous clatter until I pick them up ..
Good Cap ??
OK
If you shove a sharp edge pulse through an inductor ( your armature winding ) you will get a back e.m.f. the voltage dependent on the rate of change of current.
ie. -V = L x ( di/dt )
Numbers ??
Inductance = 10mH
Pulse rises from 0 to 5A in 100uS
-V = 0.01 x ( 5 / .0001 )
Some 500V - ish of nasty voltage spike ??
You need a cap. that is made to tolerate such nonsense ... not you average electrolytic eh ??
I have no idea what your motor inductance is or what the rate of change of current is either, but the effect IS there to some extent or other ...
The preceding stuff is from memory BTW, but I think it's right ...
Poor 'owd codger ...
Dave BC