Re: What type of filter does the Satellite have?

To: analogue@hyperreal.org,
From: Michael Caloroso MLCaloroso@worldnet.att.net,
cc:
Subject: Re: What type of filter does the Satellite have?
Date: Tue, 14 Oct 1997 19:31:16 -0700

> Also, while I'm on the subject, what Moog filter do people consider to > be the best? People talk about the Minimoog filter, but I have never
> heard one. I tend to think that the filters in the cheaper Moogs
> (MG-1, Rogue, Opus 3) sound better than the filter in my "nice" Moog
> (the Multimoog, which rocks in pretty much all other ways, but has a
> weaker filter than my Opus 3 or Rogue). Any technical types know what > makes the difference in the individual sounds? My favorite filters use > 1K resistors in the ladder; but this doesn't explain why the Rogue
> sounds different than the Multimoog (as they both use 1K resistors, as > well as a CA3046 array for the top and bottom transistor pairs).

One "technical type" reporting to duty....sorry for the long append.

Some time ago just for curiosity's sake, I got my schematics and drew up
the filter variations between the different Moog instruments (Mini,
Micro/Multi, Taurus, Rogue/Taurus II, Source, Memorymoog, and
Liberation). Unfortunately I did not have access to 904A or Satellite
schematics, although Bob Moog reported that the Mini filter was copied
from the 904A.

I then did a side by side comparision. The single factor that dominates
the differences is not the bias resistors in the ladder (1K), not
whether the CA3046 array is in the design, but it is the differential
stage, the feedback path between the top of the ladder and the input
transistor opposite the audio input at the bottom.

Feedback has more impact on a circuit's soronity than most people care
to acknowledge. When Dick Denny was designing the Vox AC30, arguably
the best sounding guitar amp ever made, he found this during his
research and used it to his advantage. The AC30 was not the most
powerful amp but it was sonically the most variable. Don Leslie (of
Leslie rotating speaker fame) found the same thing, and patented a
transistor circuit that sounds like tubes, its crucial element being the
feedback methodology.

The Mini (or 904A) is the simplest and is all discrete components (no
op-amps), and it sounds the best to most analogue die-hards. The
Micro/Multi is the most convoluted, way too many components in the audio
path, and it suffers as a result. For op-amp designs, the Taurus has
the simplest circuit and the beefiest sound. Some designs use FETs at
the differential stage, some use bipolars (CA3046 or otherwise), some
just couple directly to the op-amp input pins.

Here's how I personally rate the soronity of my Moogs, the last being
the best, with audio path notes on the feedback designs:

Micro/Multi DC coupled to FET pair
CA3080A
NPN Darlington pair
AC coupled to bottom of ladder
VCA: very strange coupling to CA3080A
Rogue/Taurus II AC coupled to 353 pair
4558 (yuk)
AC coupled to bottom of ladder
VCA: CA3080A taps off 353 outputs
Liberation AC coupled to differential Darlington pairs
4558
DC coupled to bottom of ladder
VCA: CA3080A taps off emitters of Darlington pair
Source DC coupled to FET pair
CA3080A
DC coupled to bottom of ladder
VCA: CA3080A taps off drains of FETs
Taurus I DC coupled to CA3080A
NPN transistor
AC coupled to CD4016 FET switch to bottom
of ladder
VCA: CA3080A taps off output of filter CA3080A
Memorymoog DC coupled to 353 pair
CA3080A (controls emphasis from CPU)
DC coupled to bottom of ladder
VCA: CA3080A taps off 353 outputs
Minimoog AC coupled to differential Darlington pairs
AC coupling to bottom of ladder
VCA: discrete transistors, taps off collector
of Darlington pair

The simpler the machine, the better, thus the more "open" the sound is.
BTW, the Mini, the Liberation, & the Source use 150 ohm bias resistors
(not the 1Ks), and the Liberation doesn't sound anything like the other
two. Also note that the transistor pairs at the top & bottom of the
Memorymoog filter ladder are IT122 packages and not the more common
CA3046 arrays.

Why the variations in the feedback designs? Moog engineers attempted to
extend the range of resonance in their VCFs. On the Mini & Taurus I, it
disappears on the low end; all the later designs resonate at either
extreme. Sweep a self oscillating VCF on a Mini and then on any later
Moog VCF and you'll hear the difference. Ultimately, these
"improvements" impacted the soronity.

Plug a Mini's output into the audio input of the Micro's VCF, and listen
to the difference between that and the straight Mini.

ARP did a better job improving the ladder filter (the 4012) but ceased
its use once Moog discovered that its patented design was being copied.
Note that the ARP Pro-Soloist contains a 4034 filter that is the Moog
transistor ladder, and it sounds pretty good too.

Why is this significant? The Minimoog is *the* classic bass synthesizer
because of its limited range at near-resonance of the VCF. Very few
people realize that the resonance on Minimoog VCFs are non-linear and
have a limited range. If you dial up a patch with a near-resonating VCF
that has a percussive envelope applied to it, as the cutoff approaches
the extreme low range, the resonance decreases. This explains some of
the Mini's beef.

I posted some findings on AH about the Mini's oscillators that also
contribute to its inimitable sound, but I won't repeat them here.

You wanted technical details, you got it. Don't ask about the math
behind this ladder filter.

MC