Another Optical Compressor

samhay · Post by **samhay** » 18 Feb 2013, 14:36

I posted this over at DIYSB a little while ago and thought I should also share it with you good folks.

Like similar designs - DOD 280, Hollis' Flatline, Mictester's Really Cheap Compressor, etc. - it is a one-knob (+ level) dual op-amp optical compressor. What does this do differently? The LED driver is a little different (increasing the resistance into the second op-amp (-) input decreases the current in the LEDs) and it is somewhat self compensating - when you crank up the compression (and it gets pretty squishy) the volume stays reasonably constant.

I rolled my own vactrol with one 3mm yellow LED and the LDR. This way, we are only using half-wave rectification to do the feed-back, but the other LED (which is required) can be used as a signal/compression indicator as it gets quite bright. The LDR I used is 8k-24k at 10 lux & 1M+ dark resistance. I got it from here: http://www.bitsbox.co.uk/sensors.html (half way down the page).

To adjust it: Plug the guitar in but don't play anything. Now turn the compression all the way up and increase R2 (you may need to use a 10k trimmer) until the LEDs are almost out. R1 can now be set to compensate for the volume increase as the compression is turned up.

While I would suggest you breadboard it first - it is pretty minimal after all, here is the vero layout I used. It was designed to replace something else in a 1590A so there were a number of space/layout constraints. It is verified.

Cheers,
Sam

mictester · Post by **mictester** » 18 Feb 2013, 20:42

Wow - an even cheaper compressor!

I like the threshold adjustments. There's two minor errors in your design - put 220µ from the 9V rail to ground and a 470n non-polarised capacitor across the LEDs to get rid of the noise that the switching of the LEDs introduces. As the LEDs go into conduction, there are tiny glitches on the supply rail which show up as little clicks in the audio.

samhay · Post by **samhay** » 18 Feb 2013, 21:53

Thanks mictester - this came about after playing with your Really Cheap Compressor for quite a while. It's definitely inspired-by and a work-a-like.
I spent quite a lot of time playing with filtering caps on the breadboard as the background noise can creeps up with the compression knob. A large cap from V+ to ground is a good idea, but not essential. In this case I struggled to fit it in the space I had.
Did you mean a 470p cap in the feedback loop of the LED op-amp? 470n will roll of too much treble. This I also tried and couldn't hear the difference - not that that is the best judge of anything. For anyone interested, there is room to add a ceramic cap in the vero layout.

skywise · Post by **skywise** » 19 Feb 2013, 02:21

Now that is pretty damn awesome.

samhay · Post by **samhay** » 20 Feb 2013, 22:12

Well I had another listen tonight with headphones. Ripple from the LEDs can be heard at max compression when driving a very low-impedance load. A cap across the second op-amp's feedback loop tames this, and a 10n seems to be a good value - schematic and layout updated. Thanks for the suggestion mictester.

elkulon · Post by **elkulon** » 30 Jul 2013, 15:09

Very interesting design...

Simple question...: What happens if I feed it with more than 9V???
Let's say 12v-15v...

~arph · Post by **~arph** » 30 Jul 2013, 15:17

Would it help if you point both LED' s to the LDR, this should lower the ripple for lower frequencies, or is the slow response of the LDR sufficient?

elkulon · Post by **elkulon** » 30 Jul 2013, 15:20

Very interesting design...

Simple question...: What happens if I feed it with more than 9V???
Let's say 12v-15v...

EDIT: I don't know why the message was sent twice, and I can't delete this last message...

deltafred · Post by **deltafred** » 30 Jul 2013, 17:50

elkulon wrote: Simple question...: What happens if I feed it with more than 9V???
Let's say 12v-15v...

It should work just the same as 9v, make sure that any electrolytics have a voltage rating a few volts higher than the supply you are going to use. 30v is the safe limit for the opamp.

elkulon · Post by **elkulon** » 31 Jul 2013, 10:34

No extra headroom with more voltage??? Maybe, in this case, plus ratio-less ratio??
I know that those Opamps can take more than 9v, peak-to-peak...

friedtransistor · Post by **friedtransistor** » 31 Jul 2013, 21:36

Does it sputter like other comps like it? Like the Orange Squeezer and the flatline? It looks promising, but I don't want to build it just to get a sputtering decay...

mictester · Post by **mictester** » 31 Jul 2013, 23:11

samhay wrote: I rolled my own vactrol with one 3mm yellow LED and the LDR. This way, we are only using half-wave rectification to do the feed-back, but the other LED (which is required) can be used as a signal/compression indicator as it gets quite bright. The LDR I used is 8k-24k at 10 lux & 1M+ dark resistance. I got it from here: http://www.bitsbox.co.uk/sensors.html (half way down the page).

To adjust it: Plug the guitar in but don't play anything. Now turn the compression all the way up and increase R2 (you may need to use a 10k trimmer) until the LEDs are almost out. R1 can now be set to compensate for the volume increase as the compression is turned up.

I added 100Ω from the output of the second op-amp to limit the current into the LEDs. This stops you blowing up LEDs at power-on when they can light brightly as the capacitors charge.

samhay · Post by **samhay** » 08 Aug 2013, 20:01

Wow. The thread has risen from the dead, and just when I'm on holiday.

skywise - Thanks.

elkulon - You will get more headroom with more voltage. The max current through the LEDs is set by R2, which you might have to increase slightly - but then, that's why it's a trimmer.

arph - I don't get any audible ripple, but them my LED/LDR combo is probably quite slow. If you use a faster LDR, then maybe point both LEDs at it, but if you are going to use a relatively fast vactrol, you still need the 2nd LED, which you might as well still use as a compression indicator.

friedtransistor - mine doesn't splutter or or pump noticeably. However, it is about as small a circuit as you will find, so I would heartily recommend you try it on the breadboard first.

mictester - can't say I have blown up any LEDs on startup, but I'm not a great sample size. Did you put a 100R from the output of U1b to ground or to Vref? I can see how the latter would be a good idea, but wonder if a slightly larger resistor might give the op-amp a bit of an easier time.

This has been my main compressor for the last few months. Two observations.
1. It can squish like a bstard, but you pay with this with noise at higher compression settings. This is not worse than the other optical compressors I have tried, and is only objectionable when the compression is turned up when placed before a really high gain pedal.
2. I would probably use an A100k compression pot if I was to build another one for myself - it gets pretty un-subtle pretty fast when you twist the knob.

friedtransistor · Post by **friedtransistor** » 08 Aug 2013, 23:24

Tried this comp, but it really doesn't do much. The only thing I did differently from you is use high brightness green led's and a 5mm ldr. It barely compresses.

samhay · Post by **samhay** » 09 Aug 2013, 06:51

friedtransistor wrote:Tried this comp, but it really doesn't do much. The only thing I did differently from you is use high brightness green led's and a 5mm ldr. It barely compresses.

Do the LEDs light up and how did you set the trim pots? Also - did you make the LED(s) and LDR light-tight or did you test in the dark?

samhay · Post by **samhay** » 09 Aug 2013, 09:18

There is a parallel thread over on DIYSB and I wrote this a while ago on how to set the trimmers:

Set up is with guitar plugged in, but no playing: Start by setting both trimmers to 0. The LED should get progressively brighter as you increase the compression. At about 95% compression, dial the 1k trimmer up until the LED is quite dim. You should find a sweet spot where it changes quite quickly - set it at the dim edge of this. Now set compression at noon and increase the 10k trimmer unit the LED is essentially out. It shouldn't take much. This should set a reasonable range of compression over the knobs range. Now tweak to taste.

Basically what is happening is that this is acting like a soft limiter. It shouldn't clip (if it does, decrease the 1k and/or 10k trimmers), but the max volume will be similar(ish) regardless of the compression setting. As you increase the compression, something like the threshold is decreased. You should be able to play a very gentle single note and hear it get progressively louder with the compression knob.

Mbas974 · Post by **Mbas974** » 09 Aug 2013, 11:36

samhay ..first thanks for this thread (and circuit too)

could you post your trimmer values for 1K and 10K... so it will be easier to prototype such circuit on breadboard with (let me say) "general setup" using standard resistors.

and using trimmers only on final board for customized pedal setup.
Thanks.

samhay · Post by **samhay** » 09 Aug 2013, 12:12

Mbas974 wrote:samhay ..first thanks for this thread (and circuit too)

could you post your trimmer values for 1K and 10K... so it will be easier to prototype such circuit on breadboard with (let me say) "general setup" using standard resistors.

and using trimmers only on final board for customized pedal setup.
Thanks.

Unfortunately, the trimmer values will vary signifcantly depending on the type of LED/LDR or vactrol you use, so posting values is unlikely to be very helpful. I guess I should add the disclaimer that this is not an effect for those who are trimmer-adverse.

However, the 10k trimmer is absolutely not mission critical as it is only there to give some volume compensation - as you turn up the compression, you raise the gain of the first op-amp (U1a), which makes everything louder. As you increase the gain, you also increase the current to the LEDs (because lugs 2,3 of the compression pot are in series with the second op-amp's input). More current through the LEDs = brighter LEDs, which in turn decreases the LDR's resistance, which decreases the gain and trys to keep the volume constant as you change the compression. This works reasonably well. If you set the trimmer to minimum, you will not have any volume compensation, but you will still have a compressor. This is how the other common optical compressors tend to work, so while you are missing a trick, you will still potentially have something worth playing with.

The 1k trimmer sets the maximum current through the LEDs. with the 10k trimmer at 0, you have a max of ~25 mA current though each LED if you overdrive U1a such that it clips on the rails. The duty cycle will be <50%, so this amount of current is probably ok. In order for the compressor to work at its best, you want the LEDs to be in regime where they vary their brightness most signifcantly as the current changes - this was almost certainly occur when there is signifcantly less current flowing through them. Also bear in mind that the LED brightness has to be compatible with the chosen LDR. This all gets a bit messy, so a trimmer is an easy way of compensating for the vagaries of LEDs and LDRs.

If you still don't want any trimmers and you wanted a 3 knob compressor, you could remove the 10k trimmer and bring the 1k trimmer out of the box (as an B or C 1k or 10k pot perhaps) and have individual control of gain/threshold (the 100k gain pot) and compression. In this case, you will have plenty of control options where the first op-amp will clip - high gain, low compression, so I would also add antiparallel LEDs to the feedback loop of U1a (think Tubescreamer or Lovesqueeze) in this case to make this clipping more graceful.

mictester · Post by **mictester** » 09 Aug 2013, 16:03

samhay wrote:
mictester - can't say I have blown up any LEDs on startup, but I'm not a great sample size. Did you put a 100R from the output of U1b to ground or to Vref? I can see how the latter would be a good idea, but wonder if a slightly larger resistor might give the op-amp a bit of an easier time.

The 100Ω goes from the output of the second op-amp to the LEDs - it just limits the current through the LEDs. You could make the value higher which will reduce the maximum LED current still further but shouldn't matter with modern high efficiency LEDs. A large supply capacitor across the rails is a good idea to prevent the current drawn as the LEDs go into conduction from causing noise on the supply line.

samhay · Post by **samhay** » 09 Aug 2013, 16:27

>The 100Ω goes from the output of the second op-amp to the LEDs
Gotcha. In principle, the 100R to the (-) input of U1b should do that, but I guess another one in the feedback loop wouldn't hurt. Did you try this and manage to blow up any LEDs?

I will concede that the power filtering is a little minimal in the above schematic. I guess it is probably time to draw up the revisions.