New Stuff

[Haible Juergen]

	>> This is a very nice VCA indeed, and worth to be added to any
modular system.
	>> I've built it twice, meanwhile, for my Minimoog clone and - a
slightly
	>> expoanded version - for my JH-4. 
	>> Opamp-less VCAs, designed for soft limiting rather than hard
clipping,
	>> can really make a difference.
	>
	>Please help me, I don't get it. 
	>
	>First of all, what is the reason to use as many as three
differential stages ?

Let me first answer with a question in return: Why does the widely used 
OTA / opamp buffer VCA circuit use as many as 6 gain stages ??

Counting: OTA: (1) differential pair, (2) and (3) current mirrors (for the
longer path),
Opamp: (4) differential pair, (5) voltage gain stage, (6) current gain
stage.

Let this sink in, what a waste or pn junctions is torturing the poor audio
signal in
such a simple circuit as a 3080 + opamp VCA.

To answer your question: In the Minimoog, one stage was used for VCA
envelope,
one stage was used for Pedal Volume control, and the last stage is a class
A buffer amplifier / line driver.
In my adaption, the first stage is used for crossfading between 4 VCF
outputs
(4 "first stages" in parallel), the second stage is used for VCA envelope
and VC pan (2 "second stages" in parallel), and the 3rd stage is used as 
semi-balanced line driver (2 ones for stereo).

	>BTW, I cannot see how the middle stage works in the absence of
external
	>loudness cv. I think, if Ib of Q6 is zero, the Q4+Q5 should realise
a
	>very high attenuation. Should probably the base of Q6 be grounded
through
	>the 33K ?

I think in the Minimoog it's normalized to a fixed value.

	>I am also very curious, why all the resistances are so small. 

Small resistance, low noise, low interference.
You have to attenuate the signal for any differential-pair-based VCA,
OTA or discrete. Normally you attenuate your signal down from
your usual 5V or 10V "opamp" levels, and amplify it again by feeding
the OTA current into a large resistor. And because of the high impedance,
you normally need a buffer stage. This makes sense when you really
need the high intermediate voltage (for other purposes), but it's completely
nuts to do this if you don't really need it. Just feed the OTA (or
differential
pair) output current into the low impedance of the next stage (the lower
leg of the resistor divider you would otherwise build ...). This has several
advantages: No buffer stages required, no dangerous loss of headroom
as with 5V or even 10V levels, low component count. Ahmm - and
at last pleasant sound (;->).
When I go from the output of my SSM2040 clone into the Minimoog-style
VCA, there is no opamp in between either. SSM did suggest a similar
thing on some of their later products. (SSM2044 VCF driving a SSM2024
VCA - should be in the data sheet.)

	>The 1K,
	>470R and 8R2 (?!!) inside the differential pairs are a little bit
mysterious,
	>too. My guess is that all these are designed to obtain that special
soft
	>saturation behaviour mentioned by Juergen, but I do not have enough
knowledge
	>about differential transistor pairs to explain it.

On the last stage (which is not used for VCA functions), you can afford a
little
bit of linearisation (compared to the normal tanh function of the diff
pair).
One way to explain the VCA function of a diff pair is the change of (the
internal)
emitter resistance re of the transistors with emitter current, thus changing
the
(local, internal) feedback and thus the gain of the transistors. An external
resistor RE adds to the internal re and provides more feedback, i.e. less
gain and more linearity. The 8R2 just give enough feedback to linearize
the transfer courve a little, but still featuring very rounded corners at
the end
of the linear range - there is no possible case where whis circuit would run
into a hard clipping (or even worse, feedback prolonged saturation delay
times as in satturated opamps).

	>Would some transistor guru please provide some explanation/analysis
?

Spice is very helpful for this kind of circuits. I couldn't have designed
something
like that from scratch, but looking at it and analyzing it is a very
rewarding
experience.

	JH.


PS: (related topic)

I finally found the circuit which inspired me to use small emitter resistors
in my combined multiplier / sine shaper of the Frequency Shifter. It's in an
old
electronotes issue, and it was submitted by listmember Terry Michaels.
I want to say a (sorry, late) *thanks* to Terry for this great little
circuit !