"Arny Krueger" wrote in message
...
The filters on PWM amplifiers are there to deal with the
thermal effects of
their high frequency hash, not necessarily its direct effects
on sound
quality.

There are also other issues. Three come to mind.

The first is FCC and CISPR emission limits. Sending a pulse
train, several volts in amplitude, with a carrier in the hundreds
of kHz, if not MHz, down ordinary speaker wire just can't be done
in a practical sense. With controlled slew rates (very lossy) it
might be possible to roll off the high end before the 30MHz
radiated template but the signal would probably still radiate
back into the power line and fail conducted EMI in the 150kHz and
up range.

The second issue is insulation erosion in the voice coil or
crossover networks due to corona or displacement currents . The
varnish typically used in very small gauge wire would have a
short lifetime due to displacement currents near the leadouts.
This was a severe problem with early motor drives. They switched
at low frequency (few kHz) and had slow rise/fall times. Post
filtering was unnecessary to meet EMI limits and would have added
additional loss. However, within a short time motor insulation
failures occurred even though the peak voltage stress was well
below the insulation rating. A simple LPF was all that was
needed.

The third issue is high frequency ringing due to the parasitics
of a long cable and the speaker networks. At the amp you may
have nice pulses but at the speaker lots of overshoot and ringing
which increases number two above.