FW: Less stupid questions...

[Fraser, Colin J]

> -----Original Message-----
> From: Harry Bissell [mailto:harrybissell at prodigy.net]
> Sent: 24 September 1999 07:38
> To: Martin Gustafsson
> Cc: synth-diy at mailhost.bpa.nl
> Subject: Re: FW: Less stupid questions...
> 
> 
> In practice the same A/D converter is multiplexed to all 
> pots, and refreshed
> so fast you don't notice... another plus is that in most cases we poor
> mortals can only turn 2-3 knobs at the same time...
> 
> The Prophet V had only 127 levels (7 bit) scanning... This 
> will definatly
> give (*gasp*) the dreaded "Zipper Noise" when you move a pot...
> 
> There is a neat way to scan pots without a formal A/D 
> converter... charge a
> cap to a known voltage, discharge through an unknown R (the 
> pot) and count
> clock cycles until a comparitor says you have reached a 
> certain discharge
> point... If the cap is only discharges a part of the way, it 
> can be somewhat
> linear...

Thats how the 'A to D' input channels on the SID chip work. OK for game
controllers but not too hot for precise controls - especially if you don't
want to wait around for a cap to discharge.
 
> IMHO Zipper noise is the killer of all hybrid synth 
> designs... The only one I
> have ever heard without this "feature" is the Studio 
> Electronics ATC (the one
> with the membrane frontpanel and one big knob....) In this 
> case its a rotary
> encoder with no stops... so it doesn't matter where its "preset"

I agree.
Zipper noise has two sources - low convertor resolution = noticeable steps
between values, and low scanning rate = not enough sample points on the
curve of pot movement to reproduce the curve rather than a staircase.

Very high ADC/DAC resolution, and a high scanning rate is the best way to
avoid zipper noise.
This should be easy in these days of ultra fast microcontrollers and cheap
high resoution convertors.
It just gets difficult if you want to pass your control over midi.

Zipper noice can be avoided another way.
The typical digitally controlled analogue synth has a load of pots, analogue
switches to route each pot to the ADC, the cpu to read and store the values,
the DAC, another load of analogue switches, s&h circuits, then the cv inputs
to the voices.

Say you build in an extra analogue switch that allows you to bypass the ADC,
cpu and DAC.
When you recall a patch, the cpu outputs all the voice cvs via the DAC. At
the same time, it also scans all the pots.
When a pot changes, that parameter switches to manual control - at it's turn
in the scanning cycle, the voltage from the pot is passed directly to voice
cv (via the same s&h circuit as the DAC output).

You now have digital patch storage, with infinite resolution for manually
adjusted pots, and if the refresh rate in the scanning circuit is fast
enough, no zipper noise.

I'm sure I've seen this done somewhere - can't remember exactly where.


Colin f