I decided that it was fine time to tackle an audio compression circuit.   I asked by buddy J what type of compression plug-in he is using the most because I would like to build something.  I asked for some sort of useful circuit with minimal controls, the type of unit that one might expect in a recording console.   He suggested the Universal Audio LA-3A.  He sent me a schematic immediately, but as of this project I have still not opened it.   Years ago, before I could tell you the difference between Thevenin and Norton I read the LA-2A schematic, and my circuit is not similar at all.  Here goes...

What we discussed was a unit with a fixed threshold into which a signal would be amplified and the output waveform would end up being peak-limited.  The unit was to be controlled optically.   

Years ago J and I had gone to Princess Auto and purchased some obsolete Radio Shack kits, and I had a couple of photoresistors and LEDs left.  Perfect!

The next step, to create the optocouplers.....nothing but some sandpaper, superglue and some SteelStick!

Speaking of sandpaper, well, I've got some left.   Sounds like it'll make a great proto-board!

So far, we've got a nice and complicated piece of wire...that inverts!!!   Perfect!!!

Right On!!!!   Ok, lets add the optocoupler!   We'll need some gain on the voltage amp, and a rectifier circuit for the LED signal.  I've got some 5532s and some 1N4148s in stock...and a really big 1uF Panasonic film cap.  ...we'll try it out...

Hmmm, seems like my fabricated optocoupler has a fantastically high OFF impedance, but the bright-white LEDs can only get the photo-resistor down to about 250-300 ohms.....a bit of a bummer.  I coulda used more Red or IR I'm sure!   An incandescent bulb achieved 20 ohms when I first tested the photoresistor, but the OFF resistance was a pitiful 18k in the room, at best!

All that said, 2N5551 gain stage is working perfectly, and the OpAmp/bridge/LED seems to be working too.  But, will it servo???

After making the necessary circuit additions and calibrations, you can see from the scope data that the circuit will indeed limit the peaks.  Shown are 20Hz and 500Hz.  The waveform stays pretty tasty up until about 3kHz.

I really just can't help myself.  I've gotta run some audio across this sheet of sandpaper and see what it sounds like.  The idea that sandpaper will be smoothing off the sharp edges is just a little bit too funny come to think of it!

The keen observer will have noticed that the YELLOW output waveform is 10x greater in magnitude with respect to the BLUE input waveform.  This is because the easiest way to compress a wave is to amplify it by the desired ratio and then cut it down to unity, hence, a ratio like 10:1.  The ratio of this compressor is somewhere around 2:1 due to the high ON resistance of the optocoupler.  Notice the slope differences at the zero-crossings, the ratio of this compressor decreases with increasing frequency. 

In this case we can see that with the scales adjusted to compensate for the gain factor difference, the output wave is about half the size of the input wave.  Therefore, an additional gain stage of Av=0.2 is needed in order to attenuate the signal down to something listenable.  It also appears that another gain stage of Av=2 may be needed on the front end, as the signal generator is set to +6dBV but the scope shows 0dBV.  I'll get back to that later on but for now a decision needs to be made between using a transistor or an Op amp.  Hmmm.......