[sdiy] Sources of noise and how to deal with them?

[David Ingebretsen]

I've just finished (well, sort of) a Moog 914 fixed filter bank clone. I
will admit right off that I have gained a real appreciation for the fact
that these magic vintage designs have problems, are not necessarily the best
designs, and have their own quirks, issues, and unwanted behavior which I
will call "charm". I did it as a learning experience and to make one that is
more euro rack friendly. As much as I love my YUSynth module, it's hard to
shoe horn it into a Frac or Euro rack.

That said, I did it two ways, one with real inductors thanks to Carsten
Tonnesman and one with simulated inductors. The real inductor version was
picking up so much hum from my power supply, I had to put it about 2 feet
from my cabinet. I abandoned that option, at least for now.

For the simulated inductors, I looked at several different methods but I
decided to use an Antoniou GIC in the Moog 914 fixed filter bank clone. I
studied what Yves Usson and Jurgen Haible did and decided I liked the idea
of a simulated inductor more than an active filter. 

I have two options for the output, a simple op-amp circuit and the Moog
three transistor output stage. Both have to have a gain of about 220 due to
the "passive" filtering stages. 

The problem is even with all the input attenuators to the filter cells shut
down and no input, there is some very apparent noise and just a little 60Hz
at higher output gain levels using either output stage. Jurgen warned "There
will be some noise - just like the original had some noise, even with all
the channel potentiometers set to zero." He was right apparently.

I will admit part of the problem might be because I turn the input/output
pots pretty much all the way up because the Moog signal is much lower than
in the other equipment I use. But even so, it seems the signal to noise
ratio in the 914 is not nearly as good as I am used to or expected.

I'm wondering a couple of things:

1. Has anyone heard an original 914? (even a 907) If so, was there
noticeable or objectionable noise? 
2. Does the Antoniou topology have an inherent noise problem that is somehow
worse than using real inductors? I used a resistor network, as did Jurgen,
in the GIC and the same 4558 op-amp. Would a different op-amp be better?
3. What are the typical sources for noise in a circuit like this? Are there
some design/build tips that would help minimize the noise? 
4. Jurgen mentioned the noise was because the attenuators come before the
filter stage, why does that matter?

Really, if noise is part of the "charm" of this filter, then I embrace the
noise, I cherish the noise, I will try in vain to ignore the noise :) But,
if the noise is not inherent in this type of design, I might go back to
square one and adopt a Salen-Key type of topology like Yves Usson did. I
compared my YUSynth 914 clone against the simulated inductor clone and the
YUSynth 914 is amazingly quiet. Kudos to you Yves, brilliant module.

David