How interchangeable (if at all) are optocouplers with vactrols?

[BluffChill]

After a recent failure to build an optical compressor due to crappy LDR/LED combo result, I was wondering about optocouplers. I might be way off track here as I'm no experts, but can they be used in lieu of vactrols? How are the principles different?

[Mark Hammer]

The principles are identical but the specs can differ in ways that either complement what the design needs...or not.

For instance, it the LDR only varies between 500R and 10k, but the circuit needs the resistance to vary between 500R and 5meg, in order to do what it does, then you're usng the wrong LDR or vactrol.   As well, the time it takes for the LDR to settle down to full dark resistance can play havoc with the circuit.  Finally, all LDRs have a prefered light wavelength that they respond to most.  You can't expect a blue LED and LDR that is maxially sensitive to red to play ball with each other.  Find out what peak sensitivity is (in nanometers), and match LDR to LED.

FWIW, companies that make Vactrols have done all of that for you.

I've made a number of LED/LDR combos for myself, and one of the things I do is to file  the LED "dome" down (and buff it smooth once flat),so that the LDR sits flush against it.

[highwater]

The principles are identical but the specs can differ in ways that either complement what the design needs...or not.

As well, the time it takes for the LDR to settle down to full dark resistance can play havoc with the circuit.

In the case of compressors, this is likely to be a Good Thing... as long as you can find the right vactrol or LDR. The slowish dark-to-light transition sets a minimum attack time, and the very slow light-to-dark transition sets a minimum release time. As a result, the envelope detector can be *much* simpler -- sometimes nothing more than a means of driving the LED with the input signal.

The phototransistor-based optocouplers are a lot faster (the photodiode ones even more so), so even in a compressor that they are *electrically* compatible with, you'll have to tweak (or even completely re-design) the envelope section to get reasonable attack/release times.

The same applies to envelope filters, but they're much less likely to have the sort of bare-bones envelope detector that relies on the slow LDR response.

---

In something like a phaser or tremolo, at high-LFO rates the characteristic LDR response time can limit the available depth-of-modulation and/or create asymmetry in the shape of the modulation. The former can be convenient, but is only critical if you don't have a depth knob (even then, some LFO designs will do this on their own). The latter can be inconsequential, but it can make an easily noticeable difference with something like a square-wave-LFO tremolo at high speed and depth.

[anotherjim]

5mm flat top LED. A good match for 5mm LDR in heat shrink tube.

[R.G.]

The phototransistor-based optocouplers are a lot faster (the photodiode ones even more so), so even in a compressor that they are *electrically* compatible with, you'll have to tweak (or even completely re-design) the envelope section to get reasonable attack/release times.

It's worth noting that optocouplers come in several varieties. There are LED-LDR optocouplers, which is what I think you're calling a Vactrol.

LED-transistor optocouplers are the most common, but they have issues where you're relying on using the output (a bipolar transistor in this case) to be a variable resistor, like the use of a Vactrol. This comes out as a pretty severe limitation on the size of the signal they can act like a resistor for. Any bigger signal gets distorted badly, and you may not like that in your compressor, as it may never be clean.

LED-MOSFET output optocouplers MAY offer a smooth-resistance option, but they're not marketed or spec'd that way, as they're mostly intended for signal switching.

You might get decent smoothly varying resistance action from the switching kinds (transistor or MOSFET outputs) by turning them on and off at a very high rate of speed, pulse width modulating them, but this rapidly becomes more complicated than the entire compressor.

Bottom line: it's probably better in the short run to practice and get better at the mechanical construction of an LED/LDR than to rely on trying to adapt the more common optocouplers to linear audio processing.

[Mark Hammer]

Question: In those instances where a bipolar transistor is used AS a voltage-controlled resistance to ground, such as the venerable Doctor Q and its derivatives, is there any advantage to use of an LED-transistor optocoupler, as opposed to driving the base of the tyransistor directly?  Or would the speed of the LED be every bit as prone to audible envelope ripple as a direct-to-base connection?

Clearly, the lag of an LDR injects a bit of ripple rejection.  And if the source of illumination is an incandescent bulb - itself a source of lag that smudges any residual ripple - then there are two elements working in one's favour to reduce audible ripple.

But with an LED-transistor optocoupler, it would seem that both halves are fast enough in their response that such advantages would not really exist.  Or am I missing something?

[R.G.]

No, you're not missing something. Both LED and phototransistor response times are at least microseconds, maybe sub-microsecond. LDR and incandescent are in the tens-of-milliseconds for fast ones, out to seconds for slow LDRs. There's no advantage on a speed basis to using LED/photo transistors.

Well, maybe. the LED indirection makes it possible to use additional circuitry to actively slow down the change in current in the LED. The LED and phototransistor are both fast enough to follow the "slowness". A more-advantageous rectifier/filter, like the modified three-way peak detector you mentioned some time back would be capable of minimizing ripple, then additional slowness could be added by filtering or integrating.