[sdiy] Help, I'm Desperate! (Charge Injection with DG408)

rburnett at richieburnett.co.uk rburnett at richieburnett.co.uk
Wed Dec 12 17:21:12 CET 2018


You connect each input to a summing amplifier via its own analogue 
switch.  Each analogue switch mimics the behaviour of a VCA by 
Pulse-Width-Modulating it's control signal, and then low-pass filtering 
the output from the summing amplifier.  A low-cost micro generates the 
PWM signals to control the crossfade...

One switch's control signal has duty ratio ramping down from 100% to 0%, 
whilst another switch's control signal has duty ratio ramping up from 0% 
to 100%.  (Use whatever control law you want for the crossfade:  Linear, 
Constant power, etc...)  Only two PWM streams are active and need to be 
generated at any given time so shouldn't be too demanding on a low-end 
micro.

Where PWM "VCAs" fall down is the dynamic range.  Since the control is 
linear it's hard to get perceptually fine control when the gain is down 
at -50dB or so.  For a volume control I can see that this is a 
significant shortfall.  However, for a crossfader it might be less of a 
problem because presumably the signal "coming in" perceptually masks the 
tail of the very quiet signal "going out" ?

-Richie,


On 2018-12-12 15:07, Guy McCusker wrote:
> On Wed, Dec 12, 2018 at 2:52 PM <rburnett at richieburnett.co.uk> wrote:
>> 
>> I haven't done the calculations, but I'm left wondering whether or not
>> you can do this "multi-input crossfade scanner" thing well enough 
>> using
>> just fast PWM of the analogue switches themselves, and ditch the VCA's
>> altogether.
> 
> Interesting idea. But how would we overcome the clicking issue?
> 
> Grant Richter's Electro-Optical Mixer used an LM3914 with a
> "dithering" i.e. PWM controller to create crossfades using vactrols.
> https://web.archive.org/web/20110720183735/http://www.musicsynthesizer.com/Corpse/VCEOM.html
> Somehow the PWM idea feels similar to that. Richter's circuit takes
> advantage of the slow response of the vactrols to smooth out the
> response. Presumably your idea would involve some explicit integration
> for the same purpose? Could that also reduce the clicking, without
> low-pass-filtering the audio too much?
> _______________________________________________
> Synth-diy mailing list
> Synth-diy at synth-diy.org
> http://synth-diy.org/mailman/listinfo/synth-diy



More information about the Synth-diy mailing list