Tau Systems 1010 Filter

[Martin Czech]

> Thank you for your message.  In your message, you wrote:
> 
> << Well, the Moog cascade is also four simple lowpass filters in series...
> If you do an lowpass-highpass transformation on the circuit, you'll
> get the Mogg Highpass.  >>
> 
> I'm afraid that I don't know what a "lowpass-highpass transformation" is.
> Could you try to explain?  Is it some kind of mathematical function, or are
> you referring to a particular circuit configuration?  I also must admit that I
> do not have much ability with mathematics, so if it is at all possible, please
> try to explain without using formulas.

Ok, if s is your complex frequency in your low pass transfer function,
replace it with 1/s, this means 0 will get to infinity and infinity to
0. This way you'll get a highpass transfer function out of an existing
lowpass. You simply change the way how frequency "runs" allong the
axis. There are other transforms that create e.g. bandpass shapes out
of lowpass.  This way you only need lowpass theory, the rest is done
via transformation.

In most circuits this 1/s transform will translate into exchanging R <-> C.


 
> 
> << It is not really differential but some kind of
> symmetry property has been preserved. But one needs matched pnp
> AND npn.  >>
> 
> Does this mean that the circuit will require both NPN and PNP transistors and
> that they must all be gain-matched to each other?  I do not have access to a
> specific transistor testing device, but I could use the transistor matching
> technique (using a DVM and a small biasing & current source circuit) that is
> described in the Minimoog service manual.  Are you familiar with the Moog
> technique?  If so, would it be useful for matching PNP devices with NPN?

Yes, as somebody pointed out, you have to replace the ladder rungs (which
are C in the Moog LP) with a resistive element and the trannys with Cs.
Doing this straight forward causes Problems, 8 Caps, and the resistive
element isn't differential any more, how to bias ? 
So the resistive element is split up into a PNP to one rail and a NPN
to another rail, and only 4 caps couple the resistive element knots. The
original moog modular schematics are on the web (music machines, I think).

So, for each of the four stages there is a PNP and NPN which work together
as a resistive element (they are in parallel if you think about an AC
analysis).  In order to keep control feedthrough low, they should have
identical collector currents (Kirchhoff`s law), so they should be matched
(by the Moog method, but under consideration of the special application
in the filter, ie. what Ube do both transistors see?).

>
 
> << The Moog cascade adds a lot of distortion, but this is not so relevant in
> a lowpass application. The additional sidebands may get dominant in a 
> highpass filter application, however. Perhaps this is not nice at all,
> more aggressive.  >>
> 
> <<  Feedback: If you have audio hp characteristic, what will happen if you add
> feedback?
> Depending on parasitics there could be trouble. HP filters don't offer a
> good sine wave, when resonance is max.  >>
> 
> Yes, you may be right, but aren't you curious to actually hear it?  I have
> always loved the sound of the MS20 highpass filter, especially with a bit of
> resonance added.  My main interest is in using it as a filter, not as a sine
> wave oscillator.  


Yes, interesting.
No, no time.

m.c.