Compensating multi-stage feedback (was: RE: all tranny vca+ )

[Haible Juergen]

	>Juergen said I that I should explain that this is the basic circuit
	>and that a compensation capacitors and offset trimmers would
	>be a good idea....Well what he really asked was if it was stable.
	>LOL.....anyway have fun...
	>
	>oh....wait....I should really explain huh.....ok..... The fist pair
need 
	>a comp cap. Place a 47pF 
	[...]
	>The other two NPN pairs could also use a comp cap. 

Well, just to clarify, I really don't know much about compensating
discrete opamps.

I've tried it one time or the other, last time I had to deal with that was
for the Synthi Clone, when the original cap values didn't fit to the 
transistors I used.
So, I did get that stable, sort of, BUT in a traditional sense (which
the EMS oddities probably don't fit) the compensation was not
satisfactory, and surely not for general applications. What did I
do (same as EMS): add a capacitor across the base of the last stage.
Works nicely to avoid oscillations, but also slows down the slew
rate, so in the end it's not better than an integrated opamp in
terms of speed - rather the contrary. (Even when compared to
the old "industry standard" 0.5V/us 741)
I know that there are methods to better stabilize a discrete opamp,
in particular slewing other stages than the last one, using series
resistors with the compensation caps and so on, BUT in practice
I've always avoided it, because (and I'm not ashamed to say so)
lack of knowledge, and lack of straight forward design guides.
Even applying the usual rough rules, and doing extensive Spice
simulations afterwards was not satisfactory for me. So, if
someone has a straight forward design guide for stabilizing
3-stage or even 4-stage discrete amplifiers, I'd be interested.
In opamps, for instance they have invented things like multi 
emitter transistors to reduce the effective beta, thus achieving 
a better compromise between stability and and high speed /
high slew rate, and probably a lot more of such tricks, so it
would be hard to compete with that, using high tolerance
discrete transitors - in every discrete High End HiFi stuff I see
capacitors to be manually selected ...

Bottom line for *me* is that until I see any groundbreaking article
on that subject, I'm either using integrated opamps, or
(and with increasing pleasure) discrete circuits that don't use
multi stage feedback at all. It's astounding how many 
functions in a synth do not need the opamp circuit topology
(integrated or discrete) at all. 

JH.