allpass & pwm

[Martin Czech]

I just remember another circuit idea 
out of an old elektor magazine
(actually, the circuit was the reasin to by this
very edition):

Below you see the wellknown allpas opamp-stage.
Now, use a 4066 CMOS switch (or better) with pulse width modulated
on-time to replace Rv. This is no usuall sc filter, since 
pulse width is modulated, not frequency.
The switching signal was derived from a 50kHz triangle
using a fast comparator, some logic ensured, that 0% and 100%
pulse width was avoided (anti krrrrk).
Today, we could use even higher clock speeds.

So, the whole thing was voltage controlled, "Rv"=Tcycle/Ton*Ron.
The authors claimed, that even a cascade of 32 of such stages
was pretty noiseless, compared to former jfet solutions and
oszi screenshots did not show any dirty signals...

Well, since this is a sampling system, you might also need
input/output lp-filters for Mr. Nyquist...


I never tryed this, any experience out there?
             
If there is an interest for this elektor circuit,
I could look for the date of edition or even
mail a fotocopy to someone to scan the pages,
so that it could be stored on some web site.

Just ask !
      
in               R1    
  ---/\/\/\-- -/\/\/--------
  |     R2   |             |
  |          |             |
  |          |             |   
  |          |  --\        |
  |          |  |  \       |   out 
  |          ---| - \-------
  |    Rv       |   /         
  -- /\/\/\-----| +/              
             |  | / OP1          
             |  --            
             |                  
            ---                 
            --- C
             |
             |
            gnd