I've been trying for some time to find a Cheap Automatic Gain Control
Circuit for some time now. Probably using one opamp, and a transistor
for the feedback.  Or even done using all transistors.  I've googled
many search terms but have come up empty handed.  I've simulated and
tested a lot of designs but nothing works.  I know this is the type of
thing that people did all the time 10-15 years ago so there should be
reference docs.

So here's what I have: I'm making a speaker phone using a cheap
electret mike so the input range is down around 10mv all the way to 1V
peak to peak.  I can only output in the range from 0V-3.3V.  My only
power supply voltage is 3.3V and I really want to do this on the
cheap, so maybe one opamp and a few transistors or all transistors
would be great.

If someone can find me a circuit that does this in some type of
publically available refernce, a book or even draw me one that works I
would really appreciate it.  I'm not opposed to buying a book that has
the answer but I do have most common EE reference books.

Thanks

You have a slight problem because of the low voltage supply. There has
never been a profusion of companders and the function is now generally
performed digitally. Digital potentiometers are available from many
manufacturers such as ?Analog Devices?, ?Maxim?, Xicor and Dallas,
which can be employed in compressors but you have to know how to drive
them. They are quite cheap and some will run of low voltage. A few
analogue companders are still available though, such as the AD devices
SSM2120, SSM2122, SSM2166 (which will run off 5 volts) and SSM2167
(which will run off 3 volts).

http://www.analog.com/en/prod/0%2C2877%2CSSM2166%2C00.html
http://www.analog.com/en/prod/0%2C2877%2CSSM2167%2C00.html

I have actually used the SSM2166 for the exact same application and it
worked fine but is not hi-fi. This is pretty general with analogue
companders. One or two others available but not a lot. The commonly
used SA571N seems to have become obsolete though.

You can also use transconductance amps, configured with a peak
detector for level clamping. Same of course is true for a simple FET
arrangement, but again, your low voltage is an inhibitor, especially
since the gate voltage of the usually used JFET is a bit variable. Not
so much a problem with MOSFETs but the maximum voltage swing is
limited by the parasitic drain / source diode. Nevertheless, I have
built several microphone compressors using a MOSFET which worked very
well  -- curiously with less distortion when upside down. Never got
round to figuring that one out.

If you want further stuff you might like to post an email address (in
cryptic form so it won?t be read by spider programs) and I can send
circuits.

Best