[sdiy] PWM VCA was: CMOS Switch as VCA?

[Tom Wiltshire]

On 15 Feb 2010, at 14:22, cheater cheater wrote:

> Interesting, how fast can PWM choppers go - cheaply, as in, 'under 30
> dollars for the chopper'?

I guess a way of rephrasing that question is "how fast does a CMOs  
switch respond?"
I suspect the answer is "plenty fast enough". The datasheet for the  
74HC40xx series CMOS switches gives switch delays in nS (E.g. switch  
frequencies into MHz).

http://www.fairchildsemi.com/ds/MM%2FMM74HC4053.pdf

> Either way:
> wouldn't it be a bit unwise to just assume that the audio is the only
> thing that goes into the pwm chopper? I'm fairly sure that the
> preceding circuits will add at least *some* sidebands..
> I think even some rudimentary band limiting is in order? Any idea how
> this is addressed in the Sennheiser?

I'd certainly hope that any preceding circuits do not add any  
significant sidebands. If they did, I'd redesign them rather than use  
filtering to remove the problem.
That said, what you've got is basically a sampling scheme, so the  
normal rules about not feeding in frequencies above Nyquist apply,  
and if there's any danger of out-of-audio-band signals getting in,  
then you are going to need some filtering on the input. In practice,  
I'd doubt that you're going to be feeding powerfully ultrasonic  
sources through it, and I'd probably just make sure that all the  
input stages have an RC roll off above 20-25KHz - the rudimentary  
bandlimiting you mentioned. If you wanted to be sure, you can stick a  
dedicated anti-aliasing filter op-amp stage in there.

As an aside, I found this compressor using the PWM VCA technique:

http://www.barryrudolph.com/mix/trakker.html

T.