"JOE CHEEP" AUDIO COMPRESSOR PARTS LIST AND CONSTRUCTION HINTS http://dt.prohosting.com/hacks/compressor.html David B. Thomas dt@dt.prohosting.com PARTS LIST C1,4 1 uF (tantalum or electrolytic) C2,5 10 uF (tantalum or electrolytic) C7-8 470 uF 50V electrolytic (35v ok) C9-28 0.1 uF D1-5,7,9-13,15 1N914 diode D6,8,14,16 yellow leds D17-18 1N4003 rectifier diode Q1,2 2N3906 or equiv. PNP R1-2,11,20,45 100K R3,12,21,34 10K pot, audio taper R4,13,23,26,33, 42 1K R5,9-10,14-18, 27,31-32,35-39 10K R6-8,19,28-30, 40 20K R22,41 100 ohms R24,43 46.4K (47K ok) R25,44 1.21K (1.2K ok) V1-2 EG&G Vactec Vactrol VTL5C9 (LED and photoresistor together) S1 DPDT switch U1-8 MC1458 or equiv. dual op amp U9 7815 +15v regulator 1A U10 7915 -15v regulator 1A CONSTRUCTION HINTS The only part of this circuit's construction that is remotely crucial is the signal path. Keep the path from the signal input (through the 1.2K resistor, and the photoresistor) to the output as short as reasonably possible. If those parts end up being far apart in your design, use shielded wire for the quietest operation. Everything else is involved only in sidechain processing and it's hard to mess it up. This circuit works well in a breadboard. J1 through J6 are drawn that way only because my input and output jacks weren't mounted directly to the PC board. So long as you get input signal to J1 and ground, and take output from J2 and ground, you're doing fine. Same goes for the other channel. The photoresistor/led combination V1-2 is an important part and might be hard to find. It's made by EG&G. If you absolutely cannot find any, let me know. You can, of course, make your own with a photocell and a yellow LED, but the EG&G part is likely to perform better, and the stereo channels are more likely to match. You can use just about any op amp at all for U1-8. If you use one that isn't internally compensated, add a 680pF capacitor across R5 and R27. The 1458 op amp that I used is super cheap and should be easy to find. D6 and D14 never light up. You can mount them wherever it makes sense. D8 and D16 do light up, and should be visible. They indicate how much gain reduction the compressor is applying at any given time. All LEDs really do have to be yellow, as different colored LEDs have different forward voltages. The LED in V1-2 is yellow, since photoresistors respond best to yellow LEDs. Thus, the others are chosen for us. S1 is a DPDT switch. Its two sections are connected with a dotted line in the schematic. Note that one section is shown upside-down relative to the other one. So long as you go by the pin numbers or labels, you'll wire it right. Q1 and Q2 are PNP high gain silicon transistors. 2N3906 is a common type but there are many others that could work. Note that the collectors really do connect to the negative power supply. You might want to install heat sinks on U9 and U10 to keep them cool. If you don't, the circuit probably won't burn up. It will just run hot. Are all those decoupling capacitors really necessary? Probably not. I'm just in the habit of peppering them liberally about a circuit to keep the power supply religiously clean. The circuit may operate acceptably with all the .1uF caps removed and it would certainly work okay as long as C9-12 are installed. Naturally, if you're planning to build yourself a piece of serious studio gear, I'd install them all. Each should be connected as closely as possible to the IC leads. If you're just building this for fun, especially in a breadboard, you can afford to be sloppy. R24 and R43 might not provide completely optimal biasing of the diodes in V1-2. Only mess with this if you're very fussy or have reason to believe the circuit isn't performing well at low volumes. To get the bias just right, follow this procedure. Remove R25 and R44 and connect nothing to the output jack. Place an ohmmeter across the output jack, so you're measuring the photoresistor. With the power on but no audio signal applied, you should see something in between 200K and 8K ohms. Provided that both photoresistors are approximately equal to each other (say, within 20%) and both are in this range, the bias is fine. To fix a biasing problem, lowering R24/43 will cause the photoresistor's resistance to drop. It's not a linear relationship, so you'll just have to experiment. It's unlikely that values outside the range of 10K to 100K will be useful. C2, C5, C1 and C4 have the negative lead grounded (the rounded plate). Build the power supply first, and test it. Or, use an existing +/- 15V power supply for experimenting. Build up a single channel and test that before attempting to make a stereo unit. You can go ahead and leave off the second channel and stereo switch completely if you only want a mono compressor. Happy compressing!