Disitortion in compressors?

[MetalGuy]

Hi,

Last weekend I decided to build the Flatline compressor but I wasn't satisfied with the result  because it  always produced some amount of audible distortion which is really annoying. I went through all the previous posts on this  compressor and tried all kinds of things without much improvement - the thing is working but a tiny amount of distortion is still audible and annoying. Then I built DOD's compressor as on Tonepad and I got the same result - compression + distortion. Tried several types of opamps - same result.
So before moving to MXR compressor my question is: Do all compressors produce some distortion along with the compression or I'm missing something?

[Mark Hammer]

No, and yes you're missing something.

Compressors and limiters all use a sidechain to control the main audio path.  The sidechain contains a rectifier circuit, which essentially tracks the "average" signal amplitude.  That extracted voltage is then used to control a gain or attenuation cell of some kind.

But herein lies the problem: How accurate is that "average"?

Imagine a tachomete on your car that showed you, with an update every 10msec, how fast your engine was revving.  Could you even BEGIN to read the tach?  probably not.  Okay, let's fix that and have the tach circuit update that readout with an average derived every 10 seconds.  Smoother ans easier to read, yes.  But sensitive enough for the performance driver taking a test vehicle around a race track?  Not likely.

So, clearly, the averaging that takes place needs to occur fast enough to reflect moment to moment changes, but not SO fast that you can'tmake heads or tails of the data/meter.

Okay, back to the sidehain.  The rectifier circuit used to derive this voltage representing/tracking average signal amplitude needs to be slow enough to not be susceptible to "the jitters", but needs to be fast enough that it accurately follows notes and adjusts amplitude even when the notes come fast and furious.  That CAN be done but it ain't simple, or cheap.  What stiompboxes often aim for IS simple and cheap.  As a result, the rectifier/averaging circuit in most compressors will tend to err on the side of jitteriness by using a half-wave rectifier, or err o the side of sluggishness by having a very slow release time.  Most are of the former type, though.

Unless you are simply feeding your compressors with signal way out of proportion to what they can cleanly handle, the chanes are very good that the "distortion" you hear is during the decay/sustain portion of a strummed chord or plucked note.  What you are hearing is not clipping-type distortion, but rather what ammounts to an extremely fast and shallow tremolo, produced by the "envelope ripple"; i.e., the lack of sufficient smoothing of the envelope voltage derived by the sidechain.

There are two strategies for dealing with it without overhauling the design.  One is to make the attack and decay so fast that the compressor action doesn't last long enough for ripple to be heard.  The Orange Squeezer tends to be an example of that tactic.  The other strategy is to stretch out the decay of the envelope voltage and smooth it out so that any ripple is electronically "sanded down".  The common feature of these two strategies involves the same two components: being the averaging cap and whatever lets it discharge or recharge faster and slower.  For example, in the Dynacomp/Ross, there is a 150k fixed resistor that sets the rate at which this cap charges up again.  Make the resistance bigger and it takes longer to charge up.  In the case of the Flatline, the 100nf cap in the midst of all those diodes sets the rate at which the LED illumination fades and the LDR changes value.  Increasing its value to, say, 220nf will result in it taking longer to fade out, and produce less ripple.

One thing I should note isthat while optical compressors are overall more immune to ripple, because of the built-in sluggishness of LDRs, LDRs DO vary in speed and can still respond to any ripple still in the control signal.

[MetalGuy]

Thanks for your reply Mark! Informative and detailed as usual!
I have the DOD compressor assembled on my board now -  if I got it right I'll have to focus on that 47uF-LED-4.7k combo?
Also I'm using BC546B - is the transisitor of big importance?
I'm feeding the compressor directly from the guitar.

[MR COFFEE]

Well, Mark said

Unless you are simply feeding your compressors with signal way out of proportion to what they can cleanly handle

or something ELSE is causing distortion, which is my guess.

Power supply issues, oscillations, a zillion other issues. If you don't have the hottest pickups in town overdriving your compressor circuit, it's probably not just the inherent damn-near-inaudible distortion inherent in all compressors as Mark noted. If it's sounding really distorted, it's probably some other circuit problem.

Tell us more about what is going on. Replace the variable resistance element with a resistor and see if the distortion goes away. If it doesn't, the problem is definitely NOT the distortion inherent in compressor sidechains.

[petemoore]

  Could it be a lack of headroom problem.
  On my OS Comp. I was getting annoying noises using a TL082, I went to an NE5532 and it cleared that up, the 5532 draws more current but seems to handle the peaks better.
  Another alternative for testing that migth be to raise the supply voltage, minding the cap voltage ratings, using a voltage regulator for precise power supply voltage metering, I've noticed that jumping from 9v to 12v can make quite a difference on certain circuits, most noticable for those intended as 'clean.

[MetalGuy]

The compressor is definately overdriving  because even with the rectifier circuit completely removed at higher gain settings it's still  distorting at peaks  . No such problems at lower gain settings.
I'm feeding the circuit from a 9V regulated power supply which is adjustable to up to 12V.
My bridge humbucker is Seymour Duncan SH-12 /George Lynch Screaming Demon/. The neck pickup is SD SH-1. I'm getting most  overdrive  when strumming 5th and 6th cords when playing only with SH-12 on which is closer to the strings than SH-1. When playing bridge + neck there's less distortion. Only SH-1 /neck/ - almost no distortion.
I tried 4558, TL062, TL072 and NE5532 and got most headroom and least distortion with TL072. Switching to 12V had almost no effect on sound.
Lowering the SH-12 removed big deal of the distortion but when reconnecting the rectifier circuit  some of it, although now very distant, came back. However I think I'll be able to live with that.

[scaesic]

The compressor is definately overdriving  because even with the rectifier circuit completely removed at higher gain settings it's still  distorting at peaks  . No such problems at lower gain settings.
I'm feeding the circuit from a 9V regulated power supply which is adjustable to up to 12V.
My bridge humbucker is Seymour Duncan SH-12 /George Lynch Screaming Demon/. The neck pickup is SD SH-1. I'm getting most  overdrive  when strumming 5th and 6th cords when playing only with SH-12 on which is closer to the strings than SH-1. When playing bridge + neck there's less distortion. Only SH-1 /neck/ - almost no distortion.
I tried 4558, TL062, TL072 and NE5532 and got most headroom and least distortion with TL072. Switching to 12V had almost no effect on sound.
Lowering the SH-12 removed big deal of the distortion but when reconnecting the rectifier circuit  some of it, although now very distant, came back. However I think I'll be able to live with that.

sounds like humbuckeritis, try it with a guitar with single coils.