[sdiy] Query about best cap type for '2164 stability network

Neil Johnson neil.johnson71 at gmail.com
Fri May 11 23:57:10 CEST 2018


David,

> We're not talking about opamp compensation (where one typically installs a
> small capacitor between the output and the virtual ground), but the 2164,
> where one installs a resistor and capacitor from the virtual ground to
> ground.

There are many ways of compensating op-amps, not just strapping a
capacitor across the feedback resistor ("feedback lead compensation").

In the context of the 2164, the front end _is_ very much like an
op-amp.  Or, as Doug Frey (inventor of the OVCE, of which the 2164
contains four of them) wrote:

"The OVCE structure may be described conveniently
with the help of Fig. 3. This structure consists of a
high-gain differential amplifier whose front end re-
sembles that of a standard operational amplifier, but
whose output is a current compatible with the gain
cell."
The Operational Voltage-Controlled Element: Generalizing the VCA
JAES Volume 39 Issue 10 pp. 775-784; October 1991

In the case of the 2164, the frequency of interest that we're
concerned about is around 5.8MHz.  To give enough room for at least 45
degrees of phase margin we drop down by a decade, to around 580kHz,
for applying compensation.  That explains the ratio of the
compensation components.  The resistor reduces the gain and thus
shifts the Bode plot down to give us phase margin and hence stability,
and because the gain also depends on the input resistor, as it is part
of the feedback network, then the compensation resistor must be
adjusted as the input resistor is changed.  The capacitor is there to
only bring in the compensation when it is needed - around 580kHz.

(Note: I'm hoping for some feedback on my analysis from the original
designer of the SSM2164).

> My understanding of the stability issues of opamps is that they stem from
> the very high loop gains and the use of feedback networks.  These things
> don't seem to apply to the 2164.  Are you saying that the issues are the
> same?  If so, then please elaborate.

Gain alone is not a problem.  Gain and phase combined is the problem.

Chapter 6, in Design with Operational Amplifiers and Analog Integrated
Circuits, Franco, McGraw-Hill.
Or if you don't have that, here's a TI paper on the subject:
http://www.ti.com/lit/an/snoa486b/snoa486b.pdf

> Concerning the textbooks: I don't believe that I have ever seen the 2164
> mentioned in any textbook.

It is unlikely that you will find the 2164 in a text book.  There are
some references on earlier versions of the OVCE (SSM2014, SSM2018),
for example:
http://www.gyraf.dk/schematics/VCAs_Ben_Duncan.pdf

There is also some discussion on VCAs in Handbook for Sound Engineers
(Glen Ballou, Focal Press) although the section on VCAs is about as
comprehensive as reading the datasheets and app.notes for the various
devices discussed.

If you really want to understand how the 2164 works you need to read
up about the OVCE - although it is mostly described in Frey's papers
published by the AES.  Your university library might be able to help
you there.

> The only source of information I have ever seen
> on the performance of the 2164 is the datasheet, which is woefully
> inadequate.

The SSM2164 datasheet is not that bad.  I've seen a lot worse.

> I'd be the first one to admit that, when I look at the internal
> structure of the 2164, I'm not enough of an electrical engineer to
> understand how it might suffer from the same stability issues as an opamp.

Then I suggest you read a book on how op-amps really work (down to the
transistor level), then correlate that with the schematic of the 2164
and then start drawing Bode plots to work out gain and phase margins.

> If you really wish to be helpful, please suggest a reference (title, author,
> page numbers) that will help me understand precisely what the stability
> issues are in the 2164 VCA.

*rolleyes*

Cheers
Neil
-- 
http://www.njohnson.co.uk


More information about the Synth-diy mailing list