AW: Dome Filter Question

[Martin Czech]

> 	>I'm just curious: did you simulate the phase difference between the
> two
> 	>branches? I mean how good is the quadrature approximation, or how
> far
> 	>can you get with six stages (theoretically)?  
> 
> I don't know the theory of this, sorry. You find a lot of design and 
> optimizing rules about LPFs in the textbooks, plus LPF->BPF
> and ->HPF transformation, but no all pass constand phase 
> difference stuff. It must be somewhere, sure, but I have not found
> it yet.

Well, this was a bit unprecise question. I just wanted to ask, if you
made a simulation (that's "theory"). I don't know any approximations
either. I know only all pass theory for flat group delay, ie. Bessel
(Tietze/Schenk).

> 
> As for simulation, yes. Don't have the results handy, but it was 
> extremely good, a fraction of a degree error over the audio
> range, if memory serves.

If this is the case, the polyphase network is no longer interesting,
because you need much more components (caps). 2x6 stages sounds easier
then 4x15 or even more stages. I think from the Moog manual the circuit is 
clear now. I'll try my own simulation.

> 
> I did a simple measurement with a cheap scope in XY mode.
> Set the horizontal scale to the same value as the vertical one,
> and connect the two dome filter outputs to either one.
> Connect a low distortion sine VCO (oscillating 4pole filter ...)
> to the input, and sweep thru the frequency range. You'll notice
> a perfect circle inside the audio band, and an ellipse outside
> the valid region.

Another test could be with your ears, to shift a simple sine wave. Is the
unwanted side frequency audible ? Could you hear something like that?
This works of course only, if both sidefreq.  are quite far away from
each other, otherwise the larger component will mask the smaller one. Of
course, this gives an impression of the total system error, you don't
know if it is the quadrature osc.  or the dome filter.

m.c.