Best optical compressor?

[PRR]

An incandescent lamp is NOT fast enough for decent audio limiting. (tho it has been tried.)

Green or red LEDs are sufficient. The opto like green, but red LEDs have been more sensitive. Six of one and half-dozen of the other.

LDR selection is FAR more critical, because ALL the timeconstants are in the opto.

THE "transistor LA-2A" is clearly the LA-4.

[fryingpan]

An incandescent lamp is NOT fast enough for decent audio limiting. (tho it has been tried.)

Green or red LEDs are sufficient. The opto like green, but red LEDs have been more sensitive. Six of one and half-dozen of the other.

LDR selection is FAR more critical, because ALL the timeconstants are in the opto.

THE "transistor LA-2A" is clearly the LA-4.

There are compressors (even guitar stompboxes) that use incandescent lamps. Of course, they won't do any limiting (although of the gain stage clips during the long attack phase, well, there's your limiting  ), but they will provide a lot of character (dip and swell, that kind of stuff).

[fryingpan]

An incandescent lamp is NOT fast enough for decent audio limiting. (tho it has been tried.)

Green or red LEDs are sufficient. The opto like green, but red LEDs have been more sensitive. Six of one and half-dozen of the other.

LDR selection is FAR more critical, because ALL the timeconstants are in the opto.

THE "transistor LA-2A" is clearly the LA-4.

Anyway, wanting to use a Vactrol or a LED/LDR pair: what kind of model? The LDR I'm seeing either have time constants the other way round (60ms rise, 25ms fall, when I would need something like 10ms rise and 60ms fall) or they don't state them at all.

[Mark Hammer]

Fall time can be easily adjusted with the time constants of the envelope rectifier; just extend the decay/release time.  As for an LDR that doesn't drop resistance fast enough, consider the clipping diode trick I mentioned earlier, that clamps output level instantly, doing what a VERY fast LDR might.  The duration of that role is very brief, and the degree of clipping and threshold can be tinkered with easily.  And remember that the 60ms spec is for it to reach minimum "On" resistance fully.  But dropping to half resistance within a few msec is nothing to sneeze at.  That reduction in resistance may well be the very amount that does what you need the circuit to do for you.

[fryingpan]

Fall time can be easily adjusted with the time constants of the envelope rectifier; just extend the decay/release time.  As for an LDR that doesn't drop resistance fast enough, consider the clipping diode trick I mentioned earlier, that clamps output level instantly, doing what a VERY fast LDR might.  The duration of that role is very brief, and the degree of clipping and threshold can be tinkered with easily.  And remember that the 60ms spec is for it to reach minimum "On" resistance fully.  But dropping to half resistance within a few msec is nothing to sneeze at.  That reduction in resistance may well be the very amount that does what you need the circuit to do for you.

The thing is, if you want proper LA-2A behaviour, you're not fiddling with filtering much, you're relying entirely on the optocoupler (with its memory effect and quirky behaviour). Of course you can't really replicate an LA-2A fully (it uses specific technology to do its optocoupling, namely an electroluminescent panel coupled with a hand-selected LDR) but to get in the whereabouts would be nice.

The NSL32-SR2 looks like the closest bet, with a 5ms attack time (which could be further slowed down) and an 80ms release time (nominal).

[PRR]

The timings on the LDR are not the timings you get on the audio. By basic physics the LA-* topology will reduce loud signal faster than it releases. The numbers depend heavily on light flux and reference resistor. In an LA-*, if the opto claims 50mS release, and you use a 5Meg series resistor, it will be a good part of a second.
I only know one guy so determined as to recreate the testing that gave the LA- its euphonious sound.

[ElectricDruid]

The LDR I'm seeing either have time constants the other way round (60ms rise, 25ms fall, when I would need something like 10ms rise and 60ms fall) or they don't state them at all.

That seems odd. I think *all* the ones I've looked at turn on faster than they turn off. And they're not *a bit* different, like 25 vs 60msec, they're at least *an order of magnitude* different.

[fryingpan]

The LDR I'm seeing either have time constants the other way round (60ms rise, 25ms fall, when I would need something like 10ms rise and 60ms fall) or they don't state them at all.

That seems odd. I think *all* the ones I've looked at turn on faster than they turn off. And they're not *a bit* different, like 25 vs 60msec, they're at least *an order of magnitude* different.

That's a full vactrol. Anyway, of course I got rise and fall wrong (fall is the fall in resistance, and rise, well...) so the figures are correct. Although 25ms is a bit slow for me.

Anyway, I have found a few optocouplers that can work. VTL5C2 (slow release), NSL32-SR2 (just about right, it seems).

In an LA-*, if the opto claims 50mS release, and you use a 5Meg series resistor, it will be a good part of a second.

By 5M series resistor do you mean, like, Input -> Rseries * LDR - ground? (Output at the *). But that also means you need to find an LDR with a suitable Roff value (at least tens of Mohms), unless you're fine with 5/6 signal reduction when idle. Also, a 5M series resistor will inject a lot of noise.

[PRR]

Just ignore me.

[Mark Hammer]

Fall time can be easily adjusted with the time constants of the envelope rectifier; just extend the decay/release time.  As for an LDR that doesn't drop resistance fast enough, consider the clipping diode trick I mentioned earlier, that clamps output level instantly, doing what a VERY fast LDR might.  The duration of that role is very brief, and the degree of clipping and threshold can be tinkered with easily.  And remember that the 60ms spec is for it to reach minimum "On" resistance fully.  But dropping to half resistance within a few msec is nothing to sneeze at.  That reduction in resistance may well be the very amount that does what you need the circuit to do for you.

The thing is, if you want proper LA-2A behaviour, you're not fiddling with filtering much, you're relying entirely on the optocoupler (with its memory effect and quirky behaviour). Of course you can't really replicate an LA-2A fully (it uses specific technology to do its optocoupling, namely an electroluminescent panel coupled with a hand-selected LDR) but to get in the whereabouts would be nice.

The NSL32-SR2 looks like the closest bet, with a 5ms attack time (which could be further slowed down) and an 80ms release time (nominal).

But again, I ask the question: what is this intended compressor for?  If the intended tasks are the sort of thing an LA-2A excels at, then go nuts trying to replicate it fully.  But if not, then it may be a waste to spend too much time and effort.

[fryingpan]

Fall time can be easily adjusted with the time constants of the envelope rectifier; just extend the decay/release time.  As for an LDR that doesn't drop resistance fast enough, consider the clipping diode trick I mentioned earlier, that clamps output level instantly, doing what a VERY fast LDR might.  The duration of that role is very brief, and the degree of clipping and threshold can be tinkered with easily.  And remember that the 60ms spec is for it to reach minimum "On" resistance fully.  But dropping to half resistance within a few msec is nothing to sneeze at.  That reduction in resistance may well be the very amount that does what you need the circuit to do for you.

The thing is, if you want proper LA-2A behaviour, you're not fiddling with filtering much, you're relying entirely on the optocoupler (with its memory effect and quirky behaviour). Of course you can't really replicate an LA-2A fully (it uses specific technology to do its optocoupling, namely an electroluminescent panel coupled with a hand-selected LDR) but to get in the whereabouts would be nice.

The NSL32-SR2 looks like the closest bet, with a 5ms attack time (which could be further slowed down) and an 80ms release time (nominal).

But again, I ask the question: what is this intended compressor for?  If the intended tasks are the sort of thing an LA-2A excels at, then go nuts trying to replicate it fully.  But if not, then it may be a waste to spend too much time and effort.

In my case, it would be for bass duties. Where the LA-2A excels (at least if preceded by another, faster, more aggressive compressor). A staple of studio production is a FET compressor in series with the LA-2A, with the former taming the peaks and the latter making it all into one thick, juicy paste. I don't want to replicate it 1:1 (it would be pointless, without all the stuff an LA-2A has: the input transformer, the tube circuitry, the particular optocoupler, etc.). But getting close would be nice. I was playing around with a few topologies for the gain section (which I would limit to probably 25-30dB of gain) and I'm experimenting with getting like 1-2% distortion (possibly higher than you would with the gain section of an LA-2A, but bass can take a lot).

[Fancy Lime]

Fall time can be easily adjusted with the time constants of the envelope rectifier; just extend the decay/release time.  As for an LDR that doesn't drop resistance fast enough, consider the clipping diode trick I mentioned earlier, that clamps output level instantly, doing what a VERY fast LDR might.  The duration of that role is very brief, and the degree of clipping and threshold can be tinkered with easily.  And remember that the 60ms spec is for it to reach minimum "On" resistance fully.  But dropping to half resistance within a few msec is nothing to sneeze at.  That reduction in resistance may well be the very amount that does what you need the circuit to do for you.

The thing is, if you want proper LA-2A behaviour, you're not fiddling with filtering much, you're relying entirely on the optocoupler (with its memory effect and quirky behaviour). Of course you can't really replicate an LA-2A fully (it uses specific technology to do its optocoupling, namely an electroluminescent panel coupled with a hand-selected LDR) but to get in the whereabouts would be nice.

The NSL32-SR2 looks like the closest bet, with a 5ms attack time (which could be further slowed down) and an 80ms release time (nominal).

But again, I ask the question: what is this intended compressor for?  If the intended tasks are the sort of thing an LA-2A excels at, then go nuts trying to replicate it fully.  But if not, then it may be a waste to spend too much time and effort.

In my case, it would be for bass duties. Where the LA-2A excels (at least if preceded by another, faster, more aggressive compressor). A staple of studio production is a FET compressor in series with the LA-2A, with the former taming the peaks and the latter making it all into one thick, juicy paste. I don't want to replicate it 1:1 (it would be pointless, without all the stuff an LA-2A has: the input transformer, the tube circuitry, the particular optocoupler, etc.). But getting close would be nice. I was playing around with a few topologies for the gain section (which I would limit to probably 25-30dB of gain) and I'm experimenting with getting like 1-2% distortion (possibly higher than you would with the gain section of an LA-2A, but bass can take a lot).

I have been fiddling with that particular purpose concept on and off for years. I have come to no definitive result but some fundamental observations about what is "really" needed for the kind of bass tone that I think you are describing:

1. You indeed want 2 "compressors", one fast for the attack phase and one slow and smooth for the rest.

2. The fast compressor needs to be an instantaneous hard limitar and can actually just as well be a clipper (as has been discussed before in this thread) because the note attack is so fast and harmonically chaotic that it dies not make a difference. Clipping is much easier to implement.

3. The slow compressor needs to be a proper compressor with an adjustable ratio in the 1:2 to 1:8 range. Therefore most "guitar compressors" with a simple back feeding side chain are out because they are actually limiters.

4. You want to be able to set the thresholds of both of the above separately.

5. Some pseudo-multiband trickery is usually necessary to keep highly compressed sound from devolving into a muddy mess.

I have an untested concept for this exact thing kicking around on a piece of paper somewhere but haven't even breadboarded it. If you are interested in testing it, let me know and I will do my best to digitize it over the weekend.

Andy

[Mark Hammer]

Have any of you tried the What compressor?  I have a PnP layout for it, but have never taken the steps to build one, having far too many compressors already.

http://wiki.diyrecordingequipment.com/projects/what-compressor/

http://dt.prohosting.com/hacks/what.html

[Fancy Lime]

Have any of you tried the What compressor?  I have a PnP layout for it, but have never taken the steps to build one, having far too many compressors already.

http://wiki.diyrecordingequipment.com/projects/what-compressor/

http://dt.prohosting.com/hacks/what.html

I always considered that one a little outside my definition of "as simple as possible, as complicated as necessary" as it pertains to the use with stringed instruments in a rock band.

[fryingpan]

Yeah, this is more of a studio comp. Which is fine¹! But overkill. My idea was to target a stompbox (9-18V supply, smallish footprint) and try to make it fully discrete (not a big problem, although that means either FETs or an optocoupler). The LA-2A is conceptually simple, and adding a fast limiter before it is not too hard either (and if you make it as a hard clipper, essentially one or two transistors).

¹ : although in the era of digital mixers and ITB production one might question the utility of a hardware compressor costing, even when DIYed, several times a plugin.