Optical Compressors...What's the deal??

[El Heisenberg]

So, the gain is up all the time and the LED is triggered by the signal and the LDR lowers the gain.

This is seems crappy to me. How is there less chance of distortion?

I've been messing with a Slow Gear on breadboard for a couple days and decided to try a compressor in front of it. I chose the DOD 280 cos I hadn't built an optical compressor before and I thought it might be quieter than the Ross or Orange Squeezer.

It compresses I guess. I know it's doing the same thing as any compressor, but It seems like it's really only limiting the signal, and then slowly letting off.

I don't like this cos I get noise. Like my guitar is next to a TV constantly. SO when I use this in front of the slow gear, you can hear the noise after the swell has happened.


Gunna try the Flatline now.  But I'm thinkin I might just go with an Orange squeezer. not really looking for that much compression here anyway.

[jacobyjd]

You will ALWAYS have noise with a compressor. You're not just limiting the signal--it's raising the gain a bit when the signal is quieter, thus amplifying the noise.

Do you have the compression control cranked way up? You shouldn't have an outrageous amount of noise unless the level of compression is really high.

Also, are you running it after an overdrive or other effects? They add to the noise.

[PRR]

> How is there less chance of distortion?

JFET, tube, 3080, and most other variable-gain elements have a narrow range between distortion and hiss. Limiter action narrows the range.

Max level is often less than 1 Volt, and gets gross quickly.

LDRs have a higher overload level, low THD near 1V and tolerable THD at several volts. And often with simpler circuitry than LEDs or tubes.

LDRs "can" also do the rectification and time-filtering needed. (Truly euphonic time-constants need magix mixes and careful selection; these days it is expedient to use a fast LDR and do the time-action with chips and caps.)

VCAs of the dbx heritage can have far better overload to hiss range. They have come way down in price, but still cost more than a beer. They need considerable support circuits. They are alien to guitar-stuff thinking.

> you can hear the noise after the swell has happened.

Any limiter can be over-done. IMHO it is folly to use a limiter without a GR meter; but that's because often when I use a limiter I don't really want to listen too close, yet still have good sound. At least have some indicator so you know when it is squooshing and when it isn't. Even a voltmeter on the test bench to lead you toward appropriate settings for performance.

Remember that "real" compressor/limiters have many knobs. Even if only to tune for various programs (football, sermon, drums, etc) and you only have one program (guitar), you need to tune those parameters as presets.

[El Heisenberg]

I know compressors are noisy. But this was crazy noise.

This is the noisiest compresser I've built. THe compression is cranked way up. I always do that. but the Ross and Dynacomps weren't as noisy and of course the orange squeezer wasn't.

I mentioned distortion because I read that optical compressors distorted less. But I'e never had much trouble with distorting compressors anyway, so I was asking what other befefit there was to an optical compressor. Simple to build?

After a bit of playing with it, I do sorta kinda like this compressor. But I'm not sure it's what I want for this project. It's going in before a Slow Gear to "fix" it. I'm also gunna put the DOD swell effect in with the metal pick. It'll always be on so I can just drop the metal pick and keep playing, or use an optional momentary foot switch. I think I want a more subtle compressor. Maybe less sustain, and a louder attack. Orange squeezer? You guys can guess what I'm trying to do to that Slow Gear circuit....



I could't get the flatline compressor to work on the breadboard before I broke the fragile glass GE NTE diodes I can get locally. My breadboard is freakin TRASH! Most of the rails are totally useless. No contact!

[Mark Hammer]

There's optical and optical.  Just because there are photons involved does not mean that a DOD 280 and an LA-2 or Joe Meek are identical.

PRR has addressed the chief benefits of optical compression elements, but I'll try and make it a little more intuitive.  You can drive a FET into distortion (as many proud owner of P90 phasers will attest), and you can drive an OTA (like a 13600 or 3080 into distortion), but you can't overdrive an LDR.  So, where one needs to carefully manage gain (at the forfeiture of S/N ratio) with FET or OTA-based designs, you can let it ALL hang out when using LDRs.

And as also described, LDRs do not respond as fast as FETs and OTAs, so they serve to provide additional envelope smoothing.  Imagine what your car speedometer would look like if it registered every single second-to-second change in speed; the speedomoter would never seem to sit still long enough to be legible.  Very often, with any circuit using a sidechain and envelope extractor (e.g., autowah, compressor, limiter, noise-gate, et al), people will sometimes report/describe the "chatter" that can occur during the decay phase of the envelope (when ripple starts to be much more noticeable) as distortion.  It's not really distortion, but it sounds like it.  The LDR in "optical" compressors provides that additional smoothing to remove the chatter/ripple.  In effect, the properties of the "ideal" envelope extractor are partly subcontracted out from the rectifier circuit to the LDR ("Here.  I've smoothed it as much as I can. You do the rest.")

[merlinb]

I know compressors are noisy. But this was crazy noise.

Assuming you're using this schem
http://www.tonepad.com/getFile.asp?id=40
The noise will be due to the extremely large resistances used in the feedback loop of IC1; essentially it's just not a great circuit design. You may be able to reduce the noise by scaling down the resistors and scaling up the cap proportionately. However, this depends on the light/dark range available from your LDR. Assuming you can get a range from a few hundred ohms to 1Meg or more, then the following values ought to work.

(A lower-value volume pot might also help.)

[pjwhite]

Are you getting noisy noise or hum?  Are there fluorescent lights in the room?  Is your LED/LDR coupling completely shielded from outside light sources?

Some things to think about...

[CynicalMan]

IMO, the Flatline is just as good as the 280 and simpler, so try that one out. You could also try the gain stage of the flatline with the 280 envelope generator.

Also, if you like high compression, try reducing the 22k resistor connected to the first stage (10k on the Flatline). It's lots of squishy fun. 

[El Heisenberg]

Merlinb ill give that a try.


Pj white. Im getting my normal noise just amplified. This compressor is reaaly doing its job.

i said it yesterday but even now i like it even more. Still noisier than my ross compressor but for what im usin it for it doesn matter much.


I think ill try the flatline but is it really much different? I gotta make a trip to get more ge diodes to breadboard that up first and im anxious to box this Slow Gear up.




I can see/hear the LED rippling a little. Thats a dowbside to the design too.

But using LDR/LEDs for vibrato and stuff shpuld be really clean eh??

[PRR]

> noise will be due to the extremely large resistances used in the feedback

Yes. Ouch. That's 5 or 10 times higher hiss than a guitar cord chain should have.

220K is better but still high. 5K-33K is a happier zone. (LA-2A uses 25K-50K.)

That will mean scaling the shunt network also, and may lead to an awkward cap value.

I also think the fixed resistor should be AT the opamp "-" pin, then the cap, then the pot to ground. Aside from less wiring, hanging long wires off opamp inputs invites instability, sometimes inaudible except as higher THD and supersonic hiss aliased down into the audio band.

[merlinb]

That will mean scaling the shunt network also, and may lead to an awkward cap value.

Indeed. I was keeping the impedance ratios the same with my modified values. Unfortunately, if you reduce them much further then the shunt resistors becomes too small for the opamp to drive reliably, when set to max compression!

[MikeH]

Maybe you've got something else going on... I built the 280 and it's actually pretty quiet.  It is a little on the subtle side; I'd say it's not a 'compy' as a ross/dyna, but there's way more squish there than in something like an OS.

*I may have used metal film resistors though, I forget.

[Gurner]

I was keeping the impedance ratios the same with my modified values. Unfortunately, if you reduce them much further then the shunt resistors becomes too small for the opamp to drive reliably, when set to max compression!

What levels where you going down to to witness problems with the opamp?

To my eyes, taking the resistors in the -ve feedback 'potential divider down' in value, shouldn't affect the opamp (unless you go crazy low), but it will likely impact the range of compression (unless your LDR happens to be well matched to your revised reistor values).

[merlinb]

What levels where you going down to to witness problems with the opamp?

A TL071 can drive 5Vp-p into 1k clipping, so that determines the minimum size of the shunt resistor that I would dare use in a pedal. Admittedly you don't get many signals of that level, and at 1Vp-p it will drive about 100 ohms before struggling, but I prefer design pedals to handle the largest possible signals.

A quick test shows that the TL061 will manage 5Vp-p into an impressive 500 ohms before clipping.
TL082 will do 5Vp-p into 1k.
A 4558 will do 5Vp-p into 700 ohms.

[Johan]

when guitarists want compression, we usually want a LOT more compression than most people would consider sane. we usually want it for effect and not transparent at all. my guess is that is why you don't see many guitar comps. using LDR. and why the ROSS and MXR comps using OTA's work so well for us.
for high ration compresion  with LDR's you need high impedance's and as allready stated, that means noise..or at least very carefull planing.. I'm not saying it cant be done, just that, for what most guitarists want, perhaps LDR isn't the best approach.
J

[wavley]

Yeah, but the beautiful part of optical compression, especially in something like an LA-2A is priming it.  As you play and warm it up the photocell it has a memory and it affects the release time and gets nice and comp-y

[El Heisenberg]

If it weren't for the noise I'd replace my Ross compressor with this in an instant.

But there is noise. It's not TOO BAD I suppose. I guess it's the noise you'd expect from so much compression.

This is really the most intsense compressor I've built tho. Longest sustain and best attack. There doesn't seem to be any delay from when I pick the string and when the signal gets clamped down!

[Gordo]

The noise will be due to the extremely large resistances used in the feedback loop of IC1; essentially it's just not a great circuit design. You may be able to reduce the noise by scaling down the resistors and scaling up the cap proportionately. However, this depends on the light/dark range available from your LDR. Assuming you can get a range from a few hundred ohms to 1Meg or more, then the following values ought to work.
(A lower-value volume pot might also help.)

I built a 280 quite some time ago and although it sounded OK it was pretty lack luster. Seeing this thread made me dig it out and revisit it.  I swapped a few parts out per Merlin and replaced the opto and I'm really liking this thing. It's on my pedal board for the first time and I've had it for years. Thanks!!!

[petemoore]

  Ok compressors reduce gain of high strength signals.
  But what is it intended for ?
  If intended for low noise and powerful gain reduction, start peeling onions with the thought that these two simply don't mix. noise is low strength signal, just the thing compressors boost by allowing larger gain to be applied.
  For just obvious compression or less than quick ramp-down that may contain some 'slice' of initial pick attack [gets through before the LDR response reacts like the 280].
  Photocells are fun if fast enough, slow has its advantages [such as before distorto-fuzz box, since these put hard ceiling clipping compression anyway, and the bit-slices of pick attack @ high gain produce cutting harmonics [requires other stuff and settings etc., one way to use comp, also a slow comp that is makes a long gain recovery cycle can produce thickening of harmonics of distortion, increasing level as the guitars output fades.
  For quick response Jfet, but the OS is quite subtle and distorts...that's just right or not obvious enough or doesn't fit someone else's needs.
  Love the OS myself, quick and slick it's hard to tell but it does compress and distort mildly, other Jfet-controlled comp schematics passed over because of this and thier increasing complexity...also I think the OS is just plain elegant and gets the Jfets setup about perfectly as can be for guitar comp.
  Or Dyna/Ross, a more substantial build and nice fairly powerful and quick response compression. I increased the possible slow response to where it gets 'sticky' in the low ramp areas [high input strength w/strong settings], also a 'ducky' slow recovery sound, but wobbly when doing this, may be someone elses cup of tea, otherwise plenty powerful gain reduction circuit, enough to be quite obviously a compressor this one.
  2 comps in a row's been done, but seems like a double baited gain invitation to any available noise.

[Gurner]

 Ok compressors reduce gain of high strength signals.
 But what is it intended for ?

I'd say a compressor is to reduce the dynamic range of a signal....thereby allowing you to sit it at a higher level in the overall mix.  

So it both reduces the level of high strength signals...and increases the strength of lower signals.

Personally, I'm somewhat surprised the DIY stomp box community doesn't make more use of $1.00 PICs .....this is possibly becuase they're considered digital, but that would be a shame ......basic PICs don't sit in the signal chain at all.

I use them as an envelope detector, JFET/switch controller, LED driver etcr. All the compressors I dabble with now, use a PIC to detect the guitar signal level ....& I then either have the PIC output PWM (as a Control voltage to an LED or IC with an IC with a CV Pin)  or have the PIC control a digipot. This keeps the component count very low (compared to the equivalent featured all analogue solution), yet gives amazing versatility.