Hello eveyone...
OK, you got my attention (fwiw)....
An aside... I built my first version of this circuit back in 1978. I had been working with a band called "Loggins and Messina" and after they broke up I did some work with each of them individually. At the time I designed an preamp for Jimmy's electric guitar rig and I wanted to have a compressor in it. This is how this side-chain circuit got started. Over the years I messed around with it in various configurations until I arrived at the one shown on my website. The part that took the longest to get correct was the gain reduction metering... specifically getting it to accurately indicate GR as the ratio is varied. If you doubt that this was difficult to do, go out and measure actual GR vs meter indication on almost all of the currently available opto based compressor that have variable ratios. You'll find that darn few of them track the ratio control very well, if at all.
Originally I had the ratio pot connected in series with LDR and tried various dual pot techniques to have the meter sensitivity track the ratio control. Basically it all sucked. When I hit on the idea for making the ratio control vary the current to the LED the whole thing clicked into place.
So, if you plan on using a variable ratio control (highly recommended, imo) you should use a 5k rev log (who groaned out there?) pot in this position (VR4). The current limiting resistor should be set for the max current rating of the LED, typically 100-200 ohm. I personally do not like using VU type meters for GR reduction, finding them too slow for attack indication. I typically use a LM3916 type IC with a 1k resistor to ground on the input (usually a pot). This shows GR from the Lowest to the highest LED, not from "zero" down like a VU meter, so you do not need VR5. And it's fast.
Also a 500 ohm trim pot (wiper to the output of R7) and a cell's LED connected to each end of the pot will allow you to trim the balance of the cells. Typically I drive the circuit into a measured 5 dB of GR and use the 500 ohm trim pot to get 5 dB GR on the meter. I hope that make sense.
Other than the ratio control (VR4) Tamas values are very close to what I ended up using with this design. I used...
VR1= 100k log
VR2= 500k linear
VR3= 500k linear
VR4= 5k Rev Log
R6= 49.9k
R5= 2M
Gain of U1 set so that threshold is reached when pot is set at 2 o'clock position with +4 dBu feeding side chain.
Now, having said all of that, I recommend that anyone building this circuit pay very careful attention to the connection of the side chain and LDR to the audio signal path. The schematic shows a series R, then the LDR in shunt to ground. That's cool, but where to pick off the feed to the side-chain? There are two choices (three actually)... feed forward (side chain picked off in front of the series R) and feedback (side chain after the series R). For use as a compressor, I do NOT recommend a feed forward approach. It is far better to have the pick off point after the series R. The only down side to this approach is that you will be asking your side chain amplifiers to put out lots of voltage if you want to have very high amounts of GR. So the third approach is to use two series connected resistors, pick off the side-chain input at the junction of the two Rs and place the LDR in shunt at output of the second R. You've seen this in the LA-2a circuit and in other places. It is, imo, the best of both worlds. If you only want small amount of compression ratios you can opt for the single resistor approach.
Also, if you use the feed forward connection and place the LDR in series with the main audio path with a shunt resistor to ground, you can use this thing for a gate. Making the shunt resistor variable will allow you to control the amount of "off" attenuation.
Good luck fellow DIY'ers. Let me know if I can answer any other questions about this circuit.
Fred
This is just an example and it has not been tested yet:
VR1=500k
VR2=1M
VR3=500k
VR4=500k
C1=1.5u
C2=0.47u
PR1=PR2=NSL-32R3
Attack:
VR3*C2=250ms max
Release:
VR2*C1=1.5s max
R3=10k (unsure how much gain one needs here)
R4=1k
R8=1.5k
