Super Simple Opto Comp

[leigh]

I'm splitting off a design variation of the SS Opto comp that I've been working on over in this Drawing Room thread: http://www.groupdiy.com/index.php?topic=29986

EDIT: I slighly modified this today, from what I posted yesterday. Sorry it's looking a bit cramped now.

Circuit Explanation

The goal of this variation was to get interesting compression with a bare minimum of parts. There is no makeup gain on board - you'll have to use a mixer channel or something else to bring the signal level back up.

For R3, I was using a 100 ohm linear pot. A reverse log taper might be preferable, as might a value of 200 or 250 ohms - I did not have these parts handy for testing, however.

R4 (placed in series with R3) helps to keep the voltage reaching the Vactrols higher - since the Vactrol and panel indicator LED's in series have a combined voltage drop of roughly 3 volts, it doesn't make sense to "waste" the lowest part of the pot's throw that would just be throttling between 0 and 3 volts. If the value of R3 was raised, and/or changed to a reverse log taper, R4 might not be necessary.

R5 limits the current going to the LEDs. Without it, the loud signals at the heaviest compression settings were getting crackly.

R6 discharges C2 once signal has stopped - otherwise, the LEDs can light up briefly when the threshold pot is lowered.

And, last but not least, I added C3 to cut down on some crazy noise problems. This was not a subtle change! I got lazy with supply bypassing since I just built this on the bench, and I wasn't going to be listening to the output of the 386 anyways. But it makes a huge difference in performance.

Removed - was causing crackling of the output when threshold was in mid-range of its throw: R2 and C1 form the Zobel network recommended by the LM386's datasheet. It is supposed to dampen high frequency oscillation of the '386. I don't have a real oscilloscope here, so I was not able to determine if it's actually necessary in this circuit.

I like the sound of this compressor, especially for having such a low parts count. With the highish voltage drop of the two LEDs in series, however, the onset of its compression action is a bit sudden. I've got something a bit more nuanced planned, expanding upon this idea, but it will double or triple the parts count.

Parts List:

R1 = 5k
R3 = 100 ohm pot
R4 = 50 ohm
R5 = 100 ohm
R6 = 10k
C2 = 100 µF
C3 = 0.05 µF
U1 = LM386
U2, U3 = VTL5C2 vactrols
LED1, LED2 = LEDs for front panel display - or use a single dual-element LED


Thanks again to PRR for all the feedback on this design.

Cheers,
Leigh

 

[okgb]

Thank you

 

[buschfsu]

have you built it? what does it sound like?
thanks

 

[leigh]

[quote author="buschfsu"]have you built it? what does it sound like?[/quote]

Yes, I built this on the bench. Mostly been testing it with tones and drum loops. It's fast enough to get some pumping-style compression on drums, at heavier levels of compression.

Other than that extreme setting, it doesn't "sound" like much - it's pretty smooth and transparent. I like it enough that, after developing a slightly more complex design, I plan on boxing this as a stereo unit for studio use.

I'd want to test this Super Simple design on a wider array of musical sources, before vouching for it's awesomeness, but I like it so far. And I like the idea of an entry-level compressor project that can be built with only a dozen parts (well, not counting the power supply).

cheers,
Leigh

 

[rascalseven]

Very cool!

I scored several VTL5C2's at the local surplus for $0.95 each, so this is perfect. Thanks for sharing. I look forward to seeing what it'll do.

Peace,

JC

 

[TheJames]

I notice that there are 2 Vactrols in parallel. I would imagine that would "average out" any performance variations in the vactrols and provide a slightly smoother response characteristic overall. Since the resistive element is paralleled the effective resistance shunting the signal to ground is half of what a single vactrol would do.

For something that's down and dirty I can see this being a pretty cool tool to have around. I would imagine the attack and release is a little on the sluggish side for heavy compression thus resulting in some heavy pumping/breathing.

In a more "complete" version, providing a simple make-up gain section and balanced input/output drivers would be really easy to make. I'd suggest the line receiver/driver parts from THAT and the LM4562 Opamp in a simple 0 to somewhere around +20db gain configuration.

Parts count would still be pretty low. The THAT 1243/6 receiver really just needs 5 caps to work, a basic opamp gain circuit would be another 5 components or so, then the THAT 1646 and a handful of caps to drive the output.

I would imagine that without compression it would sound pristine and during heavy crush just pump and breathe like an angry monster.

For those that like side-chain filtering, sticking a TLO72 opamp configured as an adjustable high-pass (or maybe even band-pass) filter in front of the 386 might be pretty cool as well. A tunable bandpass filter might make for a nice wide-band de-esser. Although you might need some gain between the filter and 386 to make it hit the vactrols hard enough.

On second thought...The vactrol might be just a bit too slow for that app.

James

 

[leigh]

[quote author="TheJames"]In a more "complete" version, providing a simple make-up gain section and balanced input/output drivers would be really easy to make. I'd suggest the line receiver/driver parts from THAT and the LM4562 Opamp in a simple 0 to somewhere around +20db gain configuration.

...

For those that like side-chain filtering, sticking a TLO72 opamp configured as an adjustable high-pass (or maybe even band-pass) filter in front of the 386 might be pretty cool as well.[/quote]

Thanks for the suggestions. Both makeup gain and sidechain filtering are part of the next design version.

For makeup gain, I was leaning towards using an MC33078. I haven't heard the LM4562 yet, though.

See the original thread in the Drawing Board for how I was doing sidechain filtering at first - just switching out different values for the series cap. I will probably take a similar approach again, but this time the filter (and the threshold pot) will reside between a TL072 and the LM386. That will give the filter a consistent rolloff frequency (rather than the LEDs forming a variable-impedance path to ground, post-cap).

Also, adding a "line level" buffered point before the "power level" (LED driver) stage will allow for stereo linking. It may not track well, but as you pointed out, using vactrols in parallel should help to average out the differences per channel. Only one way to find out...

cheers,
Leigh

 

[3nity]

hey i have all the parts needed just wondering what can i expect from this design?

Thanks