BYOC Optical Compressor- issue?

[Mick Bailey]

PM sent. I'm pretty sure it's not working correctly. I've built plenty of optical compressors and none behave as you describe.

I'm in sunny Staffordshire. At least it's sunny for three days of the year - usually.

[Mark Hammer]

Sounds to me like it's working just fine. What you may be experiencing is simply a matter of a mis-set gain-recovery time. C6 gets charged up by the Q1 and Q2 outputs. It "averages" the amplitude of the signal to direct the photocell to go low and reduce gain. But since it also hangs onto that stored charge for a while and has nowhere to drain off, it holds gain low, until it has drained and allowed the LDR to come back up high and restore full gain.

Some compressors have an adjustment for decay/release time, that they usually erroneously refer to as "Attack". With faster gain-restoration, the unit will sound a little more transparent, and a tiny bit more like a limiter, rather than a compressor, since it will reduce the peak, but restore gain in time for the next peak. This design doesn't have such a feature. Try sticking a 220k-330k fixed resistor in parallel with C6, and see if you like that better.

[J M Fahey]

Dear Mark, what you describe is correct for a "normal" compressor: signal gets rectified somehow, voltage charges a capacitor, that voltage is used to control some kind of ... er .... voltage controlled gain cell (in the earliest ones it was the bias voltage of push pull pentodes) , a resistor in series with the cap defines the charging time constant, usually called "attack" , acapacitor in parallel with such cap (what you suggest) defines discharging time constant, usually called "release".

This one is completely different, to simplify circuit.

To begin with, the gain cell is current controlled, whatever current passes through the LED.

At turn on, some current passes through the LED , peak value: (9-1.9)V/4k7 until C6 charges, then it's essentially 0 mA.

On signal peaks Q1/Q2 discharge C6 , attack time is undefined but very fast anyway, there's no series resistor so it only depends on what transistors pull ... may be as high as 100mA ... or even more.

When signal stops driving transistors, they turn off and R10 starts charging C6 again.
As voltage rises current diminishes until eventually C6 reaches peak voltage and LED turns off.

As you see, exactly the opposite to a regular compressor or limiter.

R10 does 2 jobs: sets release time constant but most important, defines maximum compression available, so it's not prudent to mess with it as the schematic designer suggests.
A resistor in parallel with C6 would pass some current all the time, so applying a constant attenuation no matter what, not very useful.

In a nutshell, iy is a very crude compressor, works very well for what it is, not mod friendly and that's an understatement.

Only parameter/value which can be changed is C6 value, which changes attack and release at the same time.

And not even that much, LDRs are stubborn and have their own attack and release time, fwiw my built in power amp limiter uses one with no RC time constants at all, the LED is straight driven with narrow pulses coming from an input/output comparator ... works like a charm.

[Mark Hammer]

Thanks, Juan. You caught a few things I missed. What lies in common, though, is the matter of recovery time.

Reading over the fuzzy print, R10 seems to be as critical to the feel/behaviour of the circuit as C6. It is also analogous to the 150k resistor in the upper right hand corner, here, between V+ and the 10uf cap. It governs how quickly the unit "recovers" or returns to full gain.

The speed of return is dictated by the combination of both C6 and R10, and in the case of the Dynacomp circuit, the 150k resistor and 10uf cap. A faster recovery time can be created by using a smaller-value cap so that it charges and discharges faster, although faster discharge can be accomplished by giving the cap an alternate path to ground to discharge through, via a parallel resistor...which was the basis of my suggestion.

Of course, I neglected the connection of the LED to V+ and was thinking that the LED was being fed via a voltage from the rectifier portion, so that it would light up on peaks.

[nickb]

The attack time will (should!) be quite slow, tens of 10ms and is determined by the values of of the 0.047uF capacitors and the 220k bleed resistors. If you lower the resistor the attack is faster. The transistors are turned on for very short pulses where, as Juan says, fairly large currents flow, but the most of the time the 47nF is discharged by the 220k resistor and no current flows at all. The more charge that can be removed by the resistor so more can be pumped in on the next cycle.

[Mick Bailey]

From other people's observations and reviews of the pedal it seems to be a reasonably well-regarded design, despite the circuit limitations. I guess like many other compressors you have to sit and play through it to get the feel of the dynamics. Depends on what an individual player wants - whether just to even-out string response and level-off excessive peaks, or to use as a distinct effect in its own right.

I have funk and country customers who like a compressor to sound like it's there, and acoustic players who don't. So, maybe someone who plays 'Sultans of Swing' or 'Finish what ya started' wouldn't necessarily consider a Drawmer 1966 studio compressor to be a good bet.

[J M Fahey]

Oh, no doubt about that, it sounds very good, as shown in many demo videos

Personally I LOVE minimalistic designs, and use them all the time, my own amps and preamps look "empty" .

[Mick Bailey]

I got the pedal today and it certainly works to a degree, but doesn't work well. The main fault is the release time is much quicker than the natural string decay, so there's no real sustain increase. It also hikes the note decay to a much higher level than the inital attack level and sounds like a setting you'd get from a Slow Gear pedal. Very unnatural.

The 'compression' pot only acts on the last 1/5th rotation (at best) and needs to be C taper as Juan said. This is a common error with such circuits and in the absence of a C taper pot I usually wire an A taper backwards to give at least some graduation.

The attack time is a little too slow for my liking, though it's a matter of taste. Could be the circuit, or the vactrol transition time. Or a little bit of both.

Bass roll-off is excessive. Compression aside, the tonality shouldn't alter between effect on/off. The effected sound is very thin. Because the low-end is getting attenuated this is impacting on how the unit dynamically responds.

The unit is pretty noisy. May be better with a low-noise opamp.

The unit has been put together pretty well, nothing wrong or out-of-place. Just the way it's turned out. I've lifted a few components and wired it to a breadboard and it's already sounding better, though there are a few things I want to try out.

[Sea Chief]

I got the pedal today and it certainly works to a degree, but doesn't work well. The main fault is the release time is much quicker than the natural string decay, so there's no real sustain increase. It also hikes the note decay to a much higher level than the inital attack level and sounds like a setting you'd get from a Slow Gear pedal. Very unnatural.

The 'compression' pot only acts on the last 1/5th rotation (at best) and needs to be C taper as Juan said. This is a common error with such circuits and in the absence of a C taper pot I usually wire an A taper backwards to give at least some graduation.

The attack time is a little too slow for my liking, though it's a matter of taste. Could be the circuit, or the vactrol transition time. Or a little bit of both.

Bass roll-off is excessive. Compression aside, the tonality shouldn't alter between effect on/off. The effected sound is very thin. Because the low-end is getting attenuated this is impacting on how the unit dynamically responds.

The unit is pretty noisy. May be better with a low-noise opamp.

The unit has been put together pretty well, nothing wrong or out-of-place. Just the way it's turned out. I've lifted a few components and wired it to a breadboard and it's already sounding better, though there are a few things I want to try out.

Hi Mick- glad got you ok. Reading your above seems like it does work to a degree- difficult to ascertain for me/ defo no use for it as it is, so Id be interested to see what you think with it.

I often find the tone gets thinner with pedals, the chorus Ive made (tonepad) albeit great brightens alot- useful for me actually. The mxr phaser (tonepad) I made the tone gets considerably thinner, & the vol drops alot too/ only mediocre strength too even at the trimpots best setting.

This is the major caveat with DIY. If working and fires up ok, how one knows if its sounding correct/ if its in good health or not without substantial knowledge of the original circuit its cloned from whether its a full-on DR clone, comp or a phaser pedal clone.

Much appreciated Mick.

[g1]

I'm very interested to find out whether it is working as intended. Defective unit, or crappy design?

[J M Fahey]

I listened to a YT video where it worked very well, so much so that I got interested in building one.

[Mick Bailey]

The main issue turns out to be the Vacrol sensitivity on the LED side. After pretty much eliminating everything else I removed it and checked the response. At full LED brightness the LDR resistance was 10K Ohm. I temporarily clipped in a VTL5C9 and it works pretty much as the YT clip, except that at high compression levels this Vactrol is way too fast for the circuit design and the output flutters (like a tremelo).

Originally I couldn't figure out how the 4.7k resistor charging the 33uf cap could give a useable time-constant for a compressor. The answer lies in the recovery time of the original VTL5C4 Vactrol - 1500ms (the VTL5C9 is 50ms). The slow recovery time assists in getting a reasonable release-time for the compressor and helps to smooth out any flutter.

While I was at it I increased C3 to 'cure' the thin response. Just soldered a .1uf cap on the back of the PCB. Now the effected sound is pretty much the same as the direct.

So, it needs a new Vactrol, which I don't have to hand. I've carefully dremelled the old LED out and epoxied a replacement high-brightness green LED in place to see if this works. As the recovery time is largely down to the LDR side I'm pretty sure it ought to be fine. Just waiting for the epoxy to cure.

I also tried a pot to vary the release time. This works really well - far better than I would have thought. It does decrease the compression depth when full up, though this is only really noticable at high compression settings and can be adjusted to give a far more natural release (though I need to get the original Vactrol reinstalled to get a better idea).

Playing about with C6 to 'improve' the attack doesn't really improve the pedal. A smaller cap gives a quicker release time as a consequence of reducing the attack so R10 needs increasing too much to compensate for this and reduces the dynamics - it doesn't compress as well at higher settings and makes the effect too shallow. I think I'll leave the original value in there.

[Sea Chief]

Hi Mick- wow youve put in some effort, Im most grateful. Should I be looking at putting in a VTL5C9 then or shall we see if your epoxied addition works (interesting- proper DIY im all in favour of!). Im itching to try it already. Good learning a bit about this vactrol thingy then- a 2 ended 'roachy' thing like I put in my DR trem I guess using a simple led one end is it?

Im all in anticipation of results of the epoxy led idea.

Fab stuff! thanks, SC

[Steve A.]

I'm moving this to the Guitar Effects forum...

Steve Ahola

[J M Fahey]

FWIW I never ever used a real vactrol, unavailable here and if somebody brings a few he wants you to pay for his air ticket ... no thanks.

I always make my own out of 5mm LDR (easier to find than you'd think because they are used in colourTVs to sense ambient light and automatically adjust contrast and brightness, go figure) and high efficiency Red Leds (tried green too but Red is better by far) plus a little heat shrink tube.

Works like a charm, even better than original (I use some crazy efficient 5000mCd diodes) , and for peanuts.

I easily get a couple hundred ohms resistance and sometimes a few tens of ohms, how's that?