DIYstompboxes.com

Are you missing something in your life?
Is it a fatter bottom end? Tighter mids? A more present presence?
Do you want a tone that barks, then sits back in the mix and plays nice with others? A more transparent, yet nicely coloured tone that is both warm and bright.
Do you want tone that travels at the speed of light, and isn't RoHS compliant?
Then this may just be what you have been searching for.*

Quite simply, this is hands down the best compressor on my breadboard at the moment.

(https://dl.dropboxusercontent.com/u/11996927/Light_Speed_Compressor.png)
(click for larger version)

*I make no promises, but it is a handy little compressor/limiter.

And for a more measured description:

There have been quite a few recent threads about the John Hollis Flatline compressor. It seems to be a popular circuit, but I have always been slightly bothered by the side chain - it does full wave rectification, but then spoils the effect by doing so in a very asymmetric way. I figured I could do better, or at least different.

The above circuit uses a precision bridge rectifier to do a full-wave rectification without any diode drop. This is a fairly popular way of driving old moving coil meters and it is easy to calibrate - R7 determines the response, and will give 1mA/V(peak) as shown. It will also drive an LED, but this must be fairly efficient as most op-amps will crap out if you drop the value of R7 much more to push more current through the LED.

I have used a couple of vactrols in the circuit - a Silonex NSL-32S and the newer NSL-32SR3, which is available from Farnell for  about £2. These are relatively fast with 5 ms rise and 10 ms decay times.
http://www.farnell.com/datasheets/1674205.pdf
If you roll your own, or use a slower vactrol it should work fine, but if the light resistance is >> 1k, you may have to scale up R3 and the ratio pot. As it is, it can just about manage a ratio of 20 at max setting, thus putting it in limiter territory. There is some distortion from envelope ripple, especially with smaller values of C8-10, but if you are looking for a 'character' compressor then you might find this quite pleasant - I like it.

IC2A should be a fairly fast op-amp - I have tried NE5532 and TL072s.
D4-8 should have low forward voltage. I used 1N60 Schottky's.

The compression indicator is actually an envelope indicator, but for anyone that wants to add an indicator to the Flatline, this should work ok.

Edit - and it works nicely with Bass.

Edit 2 - here's some data showing how the ratio and threshold controls work. These were measured after the initial attack using a 440 Hz sine wave, a 1u cap across the vactrol LED and the level pot at max.

(https://dl.dropboxusercontent.com/u/11996927/coming_soon.png)
(click for larger version

cool...subbed.  8)

cheers sam.

Neat work Sam
I might have to try it on the breadboard when I get a spare corner :)
Thanks

Looks great, can't wait to add this one to the stable. Not too many controls, not too many chips, I wonder what size box it will fit in?

now I won't feel so bad about [still] not building your other opti compressor, sam.

Thanks all.

Russ - you could easily do away without the switchable attack/release time, which should make it a fairly straightforward build in a 15090A if you use 9 mm pots. As the ratio and threshold controls have some interactivity, you could replace one with a trimmer if you really wanted to too...

duck - this one is better in most respects, so consider any guilt absolved.

Light Speed Compressor is catchy and all, but if I build one I would have to name it "Threshold-ish." 8)

To each there own.
When I get around to finishing measuring/plotting the V(out) vs. V(in) curves you will see that there is some inter-dependence of the ratio and threshold(ish) controls. This is fairly typical of a feed-back style compressor, and they are both useful - at least to me - but I wouldn't want to be accused of trying to over-sell the design!

3 pots and a toggle are no problem to fit in the rather large 1590A. Unless you stuff a 9V battery in there too.  :icon_twisted:

>3 pots and a toggle are no problem to fit in the rather large 1590A.
It's all relative.

>Unless you stuff a 9V battery in there too.
Dare you.

In other news, I have added a couple of Vout vs. Vin plots to the second post to show off some of the compression characteristics - in this case, the ratio control. It takes a wee while to measure one of these, so I will update the file as/when I am motivated and able.
It also seems that it might be worthwhile increasing the max threshold further. This is easily done by reducing the value of R8 and/or increasing R7, so expect a little more tinkering...

Subbed cuz I love the flatline and like tweaks.  8)

Quote from: vigilante397 on March 23, 2016, 06:04:00 PM
Light Speed Compressor is catchy and all, but if I build one I would have to name it "Threshold-ish." 8)

mine can be called "the guilt absolver", he said himself.

(https://upload.wikimedia.org/wikipedia/en/9/93/Buddy_christ.jpg)

Hmmm, this or the Eng Thumb, that is the question.  ... Also been interested in a diode bridge based comp for the colour.

What's the attack /release switch do - does it actually affect both Attack and Release or is it just labelled that because some known pedals used to confuse those 2 (knob labelled Attack actually controled Release)?

>What's the attack /release switch do

Good question. Below is a simulation of the current throught the vactrol LED driven by a 100 Hz sine wave from 50-550 ms.

Top graph is with 100n (green), 1u (blue) and 10u (red) parallel to the LED - i.e. switched by the attack/release control - and a 100 mV peak signal.
Bottom graph is with 10u attack/release cap and the signal increased from 100 mV peak (red) to 300 mV (blue) and 1V (green).

Notice how the attack gets shorter with a stronger signal - i.e. the envelope detector is adaptive. I forgot to mention this is my earlier spiel, but it's a pretty neat feature.

(https://dl.dropboxusercontent.com/u/11996927/Light_Speed_Compressor_Attack-Release.png)
(click for bigger version)

If I am reading that right it appears to be a considerably interactive (with threshold/ratio) attack control. Thanks for that.

Really interesting how interactive all the controls are. 

The ratio sets the max gain of IC1A - anywhere from unity to about 21x. This feeds into the rectifier via the threshold pot, which can attenuate the signal by 0-90% (but this can be 100% if you jumper R8).
So, if you increase the ratio, you increase the signal going into the sidechain, which looks like the threshold is decreasing as you increase the ratio - thus threshold(ish). While it doesn't look perfect on paper, it works pretty well in practice and both controls are quite useful.

However, this circuit can be tweaked to work in a feed-forward topology where IC1B is fed dry signal. I tried this briefly and it will work, but will also need some optimisation. This is an option (and could be switchable) if you want complete independence of the controls.

Had another play with this in feed-forward topology - I moved the threshold pot to replace R4, and increased it to 1M. It works, but it is more prone to breathing - which makes sense for an open loop system - and probably has significantly less max compression.
I will be keeping mine as per the schematic in the OP, but have to finish something else before I start building it. Expect a vero layout at some stage in the coming weeks.

ehm, my take for samhay design to perfboarder style who want onboard pots led and switch
just add hole in bottom (negative side) C2 if you want reduce up to 470n for guitar only
top side
https://1.bp.blogspot.com/-yFLuO9W9D78/VweoBmKwemI/AAAAAAAAAJ8/ns-6KtoFmnErkgH92Se-y262XHnm4lb-w/s1600/light-speed-compressor-layout.png (https://1.bp.blogspot.com/-yFLuO9W9D78/VweoBmKwemI/AAAAAAAAAJ8/ns-6KtoFmnErkgH92Se-y262XHnm4lb-w/s1600/light-speed-compressor-layout.png)
bottom side
https://3.bp.blogspot.com/-DqUMe8ypNTk/VweoBk1MlaI/AAAAAAAAAJ4/flfcE68EvZULVF4Zmpw4sF2F208wnPAJw/s1600/light-speed-compressor-etch.png (https://3.bp.blogspot.com/-DqUMe8ypNTk/VweoBk1MlaI/AAAAAAAAAJ4/flfcE68EvZULVF4Zmpw4sF2F208wnPAJw/s1600/light-speed-compressor-etch.png)
hope you all like this, jumper isn't crime. Right? :D

Thanks Febri

Quote from: samhay on April 08, 2016, 09:23:12 AM
Thanks Febri

youre welcome Sam
but the thanks is to you, the designer and for valuable share (y)

Nice design Sam, I will put that on breadboard as soon as I get some free time. :)

Cool - thanks Mick.

I also have a theoretical version with a combined ratio and threshold-ish pot for those that want to squash it into a 1590A. I will post a schematic when I have the chance, but it works much the same was as my old AOK compressor.

And here's the 2-knob version:

(https://dl.dropboxusercontent.com/u/11996927/Light_Speed_Compressor_2-knob.png)
(click for larger version)

This is untested, but it should work ok-ish. A good place to start for a value of R8 is maybe 10k.

Edit - a cap between comp pot lug 3 and R8 may be needed if the output DC offset on the first op-amp is significant. If so, this will form a HPF with R8, so choose a suitable value with this in mind.

It's now built using this vero layout, which should be a comfortable fit in a 1590B:

(https://dl.dropboxusercontent.com/u/11996927/LSC_vero.jpg)
(click for larger version)

You will note the mission creep - 12P1T rotary switch in place on on-off-on toggle for attack/release cap selection. I have caps from 22n - 100u. At 100u you get a noticeable dip as the compression kicks in when the ratio is turned up. This may be desirable, so audition your caps.
The rotary switch was more of a hassle than I anticipated as there wasn't enough depth in the enclosure to put the electrolytic caps above the switch pins - hence the squashed mess. In retrospect, I expect that an on-off-on toggle switch will give plenty of control over attack/release times, so I am not sure I would recommend the rotary switch option.

(https://dl.dropboxusercontent.com/u/11996927/LSC_top.jpg)  (https://dl.dropboxusercontent.com/u/11996927/LSC_guts.jpg)

Nathan/vigilante - sorry, could fit the '-ish' on the end of threshold.

cool build... 8)

yep that rotary looks a nightmare...

Thanks Rob.
I've built rotary switches into guitars before - Varitone style - but these use smaller value caps (no electro's), which obviously (in retrospect) aren't as tall. Live and learn.

Necro post.
I breadboareded this tonight and found it to have a good amount of compression.
Yes it is somewhat interactive with the settings of the pots. and without probing with the ozziescope.
I cant hear what the threshold pot is doing. Might be my ears, might be a breadboard mistake, miswire.
Nice neat little design Sam. Thanks and as you say it can get into near limiter settings.
I breadboarded it with nsl32 opto, TL072 op amp, bat 41 rectifier, No Feedback leds.
I didnt notice any pumping. breathing.
It seemed there was sometimes a delay in the compression but i think this may be the breadboard. or rather the way I breadboard. Compnants shorting.
I didnt rig up the 1u 10uf switch I kept it simple for the breadboard with just the one 1uf cap across the led.
I subbed the input cap to a 20nf cause thats what came out of the box first.
It's late and time for bed.
I'm gonna mess with the vactrol next and see what i can do with a 3mm red and gl5539 roll my own.
PS what are the two feedback leds doing anyway? Limiting the output voltage? Clipping?
Both?
Thanks Sam

Quote from: samhay on April 12, 2016, 03:09:18 AM
And here's the 2-knob version:

(https://dl.dropboxusercontent.com/u/11996927/Light_Speed_Compressor_2-knob.png)
(click for larger version)



I like the simple 2 knob version. It's on the breadboard now.
all parts as in my previous post except my 24k (just what was close enough to 22k, it came out of the breadboard parts box first)
I tried 10k and 47k but like the ~22k.

Suggested parts change for a 3mm red led and gl5539 and 5516 please.
Thanks

Wait shouldnt i be able to just use a red/orange led with a gl5516 as a direct drop in for the nsl32

specs
http://akizukidenshi.com/download/ds/senba/GL55%20Series%20Photoresistor.pdf

Roll my own report
(Still not ozziescoped it)
I tried a 5mm super bright with gl4437 the attack time seemed a little slower to my ears on the intial attack but once compressing it did a good squshy job.
I tried a 5mm super bright red with a waitrony ke107 ldr (500k dark) and it seemed a little closer to the nsl32
basically i'd try any combination of vactrol and see how you like it. cobber........
"You get me?"

Sounds good with a nice brass or thick glass slide too.

Hi Rich,

If you don't find the threshold control very useful, then the 2 knob version makes sense - and you seem to have found it. Thanks for verifying it works! I assume your 22k is for R8?

> It seemed there was sometimes a delay in the compression but i think this may be the breadboard. or rather the way I breadboard. Compnants shorting.
I didnt rig up the 1u 10uf switch I kept it simple for the breadboard with just the one 1uf cap across the led.

The larger the cap and/or the slower the LDR, the longer the attack will be. I am guessing that is what the delay is. Making the cap variable makes it a bit more versatile and you should try 10+ u as it lets the initial attack through nicely (so it sounds a bit more natural) while still giving some sustain.

>I'm gonna mess with the vactrol next and see what i can do with a 3mm red and gl5539 roll my own.

I never tried it will a home-rolled vactrol. This usually seems to have a slower attack, which probably isn't a big deal here. It might not quite get to light speed, but close enough for rock and roll.

>PS what are the two feedback leds doing anyway? Limiting the output voltage? Clipping?

They stop the op-amp clipping when you turn the ratio up and turn the threshold down and/or use a large C8-C10 cap. With the 2 knob version, these may not be necessarily unless you go for a large value of C10 and/or have really hot pickups.

Thanks Sam. I'll rig the electros with a switch on the breadboard later tonight and I'll put the feedback Leds in too. Is it worth scoping it?

I can't imagine a scenario where scoping a circuit would be a bad idea.
However, I added the LEDs after I noticed (heard) objectionable clipping, so if you don't hear anything you don't like, then I see no harm in leaving them out.

p.s. The impedance around the vactrol LED is really high, so if you go looking there, use x10 scope probes if you have them.

I get you, cobber, it's the right "sam hay", but the wrong "optical compressor". thanks for the vid (I prolly should listen to it agin, listen for the kazoo). and if I have to plug swap for audio, you have to osc-ope, cause sam said.

I can answer to cobber too if it helps - I have a piece of paper somewhere with a picture of a kangaroo and an emu next to my name.

In other news, I had a recollection that I ripped the envelope/compression indicator out of my build as it was contributing some noise - probably LED conduction causing spikes on the power supply. Rich - did you add the indicator?

No I left it off mate

Possibly for the best.

I see what you mean the addition of the Attack switch is good.
The feedback loop diodes too. Definitly sounds less overdriven at low ratio settings and more squished at high settings.
Again Might just be my ears.
Ozziescope time. Laters.

Try a 47k R8 cobber .....

RE: using a roll-your-own vactrol.

The ratio pot sets the gain of IC1A. Remember, this is a non-inverting op-amp, so the gain =  1 + (R(ratio pot) || LDR)/4.7k. This gain in turn determines how much compression you will get.

The dark resistance of the NSL 'vactrols' I used is > 1M* so the LDR's dark resistance is >> any possible ratio pot resistance and can be ignored giving gain from 1  to about 21.

The compression happens when the LED turns on and reduces the LDR's resistance. In this circuit the NSL's light resistance can get to < 1k.* This sets the compression ratio as R(LDR) < 4.7k and the gain can approach unity. As we are using a non-inverting op-amp, the gain can not be < 1, so we don't need a make-up gain stage afterwards (let's call this a feature).

What does this all mean? Most home-rolled LDRs have light resistances much greater than the NSL's used here. To make them work, scale up the values of the ratio pot, the 4.7k resistor (R3) and then scale back the associated caps (C2 and C3). Aim to make R3 a little larger (edit) than the LDR light resistance and the ratio pot 10-20 x larger than R3. However, the LDR's dark resistance will become limiting (to max gain) at some point, so try to pick an LDR with as wide light-dark resistance range as you can.

* datasheet: http://logosfoundation.org/instrum_gwr/tinti/datasheets/Silonex_NSL-32SR2_60721.pdf

Thanks Sam I'm far from done with messing with this.
I'll be a few days getting back to this. Cheers.

A small necro-bump to post a picture of my build from a couple of weeks ago.

(http://www.bouron.org.uk/marc/ls509.JPG)

 Very nice.
Which caps did you use for the attack/release control?

I went with the 1uF / 10uF combo in the version 1 (March 2016) schematic.  (If there was another version, I don't have it!)  TBH, I didn't get much chance to test it properly (too late, people asleep...), but my layout has plenty of room to tack on other values if I need.  Either way, cool design.  Thanks, Sam.

That should work nicely, and you're welcome.
I think my attack caps go up to 100u, but I almost never use that setting. If 10u isn't quite long enough, particularly for bass, 20 or 30u would be plenty.
Once you've had some time to play with it a bit more, I'd be curious to hear what you think.

I'm planning on putting together a version of this compressor adapted for a 15-0-15 dual power supply, but I've hit a problem.

The bottom (or top - pin 3) of the Threshold is connected through a 'minimum-squash' stopping resistor (R8) to the mid supply (4.5v) ref voltage, and a capacitor (C7) to earth. This suggests that the side chain signal is positively biased, and that in a dual supply version R8 should (theoretically) be connected to earth, and C7 should be connected to -15v.

Is this right? and if not, what should I change to adapt this circuit for a dual supply? Would appreciate any guidance.

The threshold pot and R8 provides bias to IC1B, so should go to ground on a bipolar supply.
If you use a non-polarised capacitor for C7 (as is likely to be the case for 100n), then I suggest you also connect this to ground. If that's tricky, you can alternatively connect to the negative supply.

Thanks for the reply Sam. I'll test this out on the prototype.

One more little question if I may, I'm hoping to incorporate it into a solid state practice amp, and would like the compressor to be controlled by one single pot (Threshold (with an internal preset for Ratio)) where minimum compression (maximum threshold) corresponds to no compression. This would, I suppose, mean omitting R8 altogether and connection pin 3 of the Thresh pot to earth?

Is C7 just power supply local de-coupling for the half supply ref (along with C11)?

I'm just guessing hear.

If you want a one-knob compression control that doesn't compress at minimum then the control has to be via the ratio pot. The 'Two Knob' version I designed fits this criteria and would be my recommendation (below).
Make R8 10-22k or a trimmer that you can set so that the volume change is minimised when you vary the compression.

C7 and C11 - yes, this is just to filter the bias and supply voltages, respectively. On second thought, C7 should be unnecessary if you use a bipolar supply, but then you would probably add the equivalent of C11 to filter/decouple the negative rail....

(https://samdump.files.wordpress.com/2017/07/light_speed_compressor_2-knob.png?w=1800)

Here's my first-ever vero build that came from someone else's layout - thanks to Sam Hay for both the design and the layout. Not tested just yet - it's pretty late.

(https://i.postimg.cc/2VJrFLdk/20191020-005359.jpg) (https://postimg.cc/2VJrFLdk)

(https://i.postimg.cc/NKHYNGb1/20191020-005417.jpg) (https://postimg.cc/NKHYNGb1)

Looking good.
Hope you like it.

Getting oscillation - trying it without being boxed it up. Suspecting too-long leads to the ratio and threshold pots - frequency varies with the setting of one of the pots. More later - will shield the pot leads, ground the pot bodies, and try again. Using this layout.
(https://i.postimg.cc/gwCLgvWb/light-speed-compressor-vero.jpg) (https://postimg.cc/gwCLgvWb)

Question: I can omit the filtered Va lead and reduce the number of antennas by one - that's only intended for the indicator daughterboard right?

Correct - don't need Va if you don't add the indicator (and I don't recommend you do add the indicator).

(https://i.postimg.cc/vcYkyBjn/trace-blob.jpg) (https://postimg.cc/vcYkyBjn)

Jeez, Sam, I thought you had my back! How come there isn't a warning on the Vero layout? Perhaps something along the lines of, "Make ALL the trace cuts indicated for best results." (See zoomed-in photo for The Inevitable Vero Rookie Mistake.)

Working great! This thing is very unobtrusive, a nice design indeed. I was drawn in by the full-wave envelope rectification and optical compression - seems like they are definitely a good combination.

THAT IS A VERIFIED (YET AGAIN!) VERO! Thank you for a lovely design and an easy-to follow vero layout.

Build Notes: Attack implemented with 1uF (grey box) and 10 uF (tantalum) mounted on an on/off/on switch. The 10uF caps on the board are also tantalums. Vactrol is a roll-your own, I measured my photoresistor at about ~350 ohms min and >>2M max, with a "reasonably fast" recovery, and I used a bright white 3mm LED, mated the two in some shrink tubing. A little light is getting in so the max ratio will probably be a touch bigger than what I'm hearing now with no enclosure. On the other hand it's reassuring to see the LED flashing a bit as confirmation it's working. I used Schottky diodes in the fullwave bridge, matched tightly around 234 mV drop.

My buddy Christien better like his birthday present - he was hoping for a Ross but I didn't feel like building a Ross once I saw this.

Thank you Sam!

Glad you like it.
Hope you didn't have to do too much debugging before checking the solder side of the board.

Quote from: samhay on October 22, 2019, 02:56:41 AM
Glad you like it.
Hope you didn't have to do too much debugging before checking the solder side of the board.

Thanks! Honestly, I found it  surprisingly quickly - I started verifying all the parts positions against the layout and looking at all their solder joints one by one and when I did, I was sort of looking at them relative to the board cuts. So when I got up into that corner it was plain to see.

I like Vero - someone called it "perf board's idiot cousin" but it sure goes together quickly.

UPDATE: gifted to my buddy Christien on the occasion of his 50th birthday this week. He's VERY happy with his present. As well he might be: such a nice design, Sam. I'm really amazed at how easy/quick it was to take a verified vero layout to a competed pedal, remarkable, even with the mistake it was super quick.

Excellent, and happy birthday to Christien.

Hello,
I am reviving this thread,
First, to say thank you to Sam Hay for an interesting design, a very nicely laid out veroboard layout, and sharing its knowledge on the forum  :)

Then, because I built it today, and sound passes through but no compression occurs...

I will describe my build, hoping that I can get some directions for troubleshooting / debugging :
- I followed the "3 knob version" veroboard layout done by Sam Hay in April 2016 (https://samdump.wordpress.com/welcome/projects/light-speed-compressor/).
- I used a DIY vactrol (5mm orange led + GL5516. I have seen some people mentioning that combo above in the thread).
- I know DIY vactrols can be tricky, so I also tried with VTL5C4, and a LCR0202, without any success...
- I have read that if using a DIY vactrol, we should scale up the ratio pot, the 4.7k resistor (R3) and scale back C2 and C3. So, I used 15k (instead of 4.7k) for R3, 250k for the ratio pot (instead of 100K), and 470pF (instead of 1nF) and 1µF (instead of 2.2µF) for C2 and C3.
- I do not use the daugherboard (for compression indicator);
- The ATTACK switch toggles between 100nF and 10µF.
- I have used a "sidechain" switch, by adding a 100nF cap in parallel with C6 (in one position, it is the stock circuit, with no additional cap, and in the other position, it adds 100nF).
- The diodes are 1N60 Schotty.
- I haven't used yet an Output Volume pot (10kA), so the output goes straight from the circuit to the jack socket.
- On my vero build, the vactrol is placed with the LDR at the left, and the LED at the right, with negative pin at the bottom and positive pin at the top.
- The opamp used is a NE5532.


I understand that compression occurs when the vactrol led begins to light up... and I think it doesn't light up (it's wrapped in heatshrink tube, but when i put my eye close to the little hole, I don't see any light whatsoever).
The 2 red leds do not light up either, even if i feed the circuit very loud levels.
Turning the pots or switching the switches do not produce any change in sound.
If I remove the vactrol (it is socketted), the sound is the same.

I am going to measure the voltage at the 5532 pins, but hearing the dry signal passing through the circuit, I don't think it is an opamp issue.
The veroboard layout by Sam Hay is very well done, there are few cuts and few jumpers, so i will check again, but I don't think there is a mistake there.
My guess is that the signal that should trigger the compression action is too low : either my vactrol is not sensitive enough (but i was hoping to trigger a reaction from the vtl5c4 or the lcr0202), or the tweaks i made to R3, C2 and C3 are too much (or still not enough)?
What I don't get is why the red leds do not light up when i feed the circuit some loud signal? Is there a way to check if something is wrong here?

If anyone has any suggestion, I'd be glad to try it :)

Re-
Note to self : when you already double-checked, then triple-check : the 10µF cap at the bottom right was soldered one row too low  :icon_redface:

- Good news : I can see the vactrol led light react to the signal.
- Bad news : I can't hear a change to the signal. When I turn the THRESHOLD pot, I can see the vactrol led lighting more, but I don't hear any change sound-wise. And moving the RATIO, ATTACK or SIDE-CHAIN pot/switches does nothing either.

If the ratio knob doesn't make it louder, you have another error to find. You won't get any compression if the ratio is shorted for example.

Hi, Sam
Thank you for your circuit design, and for your help.

I will check the ratio knob and around.
[edit] I changed the ratio pot for a 500K, and did nothing.
All i get is the dry signal at unity gain.

Did the modified values for R3, C2, C3 seem sensible to you? I increased R3 by 3 and decreased C2 & C3 by half, but I have no clue is the increase/decrease scale is right, or if it could cause issues...

Soooo... when you have triple-checked a veroboard... always quadruple-check it!
I think I deserve an award : on a 5 electro cap circuit, I managed to mess 3 of them  :icon_redface: :o
For the 2.2µF at the bottom left, I had read that I may reduce its value. I didn't had a 1µF, so i used 2 x 0.47µF caps, and didn't realized I had to solder them at row K (and not J).
A closer and more accurate look at the schematic made me see the mistake, and now... the compressor is spanking!  :icon_cool:

First impressions are very good : the ratio & threshold pot are very efficient, and the compressor action can go from quite subtle and transparent to really spanky, just the way I like !

So I can confirm that the circuit works fine with a home-made vactrol (using 5mm orange led + GL5516), and that the values I used for R3, C2 & C3(quoted a couple posts above) allows the circuit to behave really well (the way I think it should).

Once again, thank you for the help (and, of course, for sharing your circuit design), and sorry for the dumb troubleshooting :)

No problem, glad you got it working.

A couple more questions, if you don't mind :

Quote"C6 is a HP filter in this Lightspeed Comp."

Do you mean it is HPF in the side-chain filter (to remove some of the bass energy before triggering the compressor), or on the whole signal (affecting the whole signal)?

Quoteoptionally switch additional caps parallel to 100n

Would you mind elaborating a bit about the timing switch (attack)?
You write "There is some distortion from envelope ripple, especially with smaller values of C8-10, but if you are looking for a 'character' compressor then you might find this quite pleasant – I like it." So, I guess that smaller values for the caps means shorter timings, right?
With the 100nF alone, we get the shortest attack & decay timing, that's it?
Is it much, much longer with 10µF (I tried with 2.2µF and didn't noticed a drastic change in the attack or release)?

C6 forms a high pass filter with the threshold pot + R8. This will only affect the signal going into IC1B (i.e. not the audio). Also, the corner frequency is ~14 Hz, so it won't be filtering much unless you are playing bass.

The C8-C10 caps set the attack and release time - longer attack comes with longer release and vice versa.
You might find a noticeable difference increasing this capacitance above 2u. If you feel like it would benefit from a longer attack/decay, temporarily rig out e.g. crocodile clips and try different capacitor values.
I think I usually use values of about 1-10u, but I the largest in mine is, I think, 100u - I used a 12 pole rotary switch and neither can remember what I used, nor which position the knob gives a specific value.

That's perfectly clear, thank you.

Quote from: samhay on April 18, 2021, 05:18:28 PM
C6 forms a high pass filter with the threshold pot + R8. This will only affect the signal going into IC1B (i.e. not the audio). Also, the corner frequency is ~14 Hz, so it won't be filtering much unless you are playing bass.

So it's in the side-chain, great :)
I'll experiment a bit, and try a value setting it around 80-100Hz.
Am I right assuming that the HPF is formed by C6 on one side, and R8+the threshold pot value? And, if so, that turning the threshold pot will change the side-chain corner frequency?

Quote from: samhay on April 18, 2021, 05:18:28 PMThe C8-C10 caps set the attack and release time - longer attack comes with longer release and vice versa.
You might find a noticeable difference increasing this capacitance above 2u. If you feel like it would benefit from a longer attack/decay, temporarily rig out e.g. crocodile clips and try different capacitor values.
I think I usually use values of about 1-10u, but I the largest in mine is, I think, 100u - I used a 12 pole rotary switch and neither can remember what I used, nor which position the knob gives a specific value.

Thank you for the tip. I will try with 47µF, see what it gives and if I like it, and move down from there.
So far, I like the "stock" timings, but I am curious and willing to experiment a bit ;)

>Am I right assuming that the HPF is formed by C6 on one side, and R8+the threshold pot value?
Yes.

>And, if so, that turning the threshold pot will change the side-chain corner frequency?
No*. The frequency should not change with the threshold setting. This is like a volume control into the sidechain.

* With op-amps like this NE5532, which do not have FET inputs, the input impedance is low enough that it will have some effect on the frequency. The op-amp input impedance is parallel to R8+threshold pot total resistance.
However, this effect will be quite subtle, so I would ignore it to start with.

ok, that you for the information.
I'm really liking this compressor!

I'm willing to try the reduction indicator daughterboard, but i don't have either BS170 or 2N7000 : is there any transistor suitable for this task (bending legs is ok if required) ?

I need to revisit the indicator as I became convinced it was causing noise.
You need something high impedance. If you don't have a MOSFET, it can probably be made to work with a JFET, but this will require some changes. I'll get back to you.

Thank you, Sam.
Reduction indicator is something handy to have (at least for quick checks in an eyeblink, even if we should use our ears first).
Of course, it shouldn't affect the sound quality :)
I think that Madbean Holy Roller  (https://www.madbeanpedals.com/EP/schematics/HolyRoller.gif)features such an indicator led, and it's an opto design too, if it's of any help...?

I'm really having fun with this compressor.
I tried various cap values for the timing and the sidechain, but I found that the stock values were already great as is :)
I have built en enclosure for it, and will box it soon.

By any chance, did you look at the indicator led daughterboard design yet?

Glad your enjoying it.
Do you have any JETs you could use for the indicator?

Yes, I think I must have some BS170 or BSS138 (I need to check).

BS170 is a MOSFET and should work in the original indicator circuit as-is.
BSS138 is also a MOSFET and may work - I haven't played with them before nor liked at datasheet. I would much something up on a breadboard, or socket the FET and see if either work.

Oops, error on my side : I only have some MPF102...Do you think it would it work with your daughterboard?

Hi, sorry for the necrobump, but this is like the only place on in all the web where there are people who have built this circuit a few times. 

I've built one of these using an LCR0202 and while I'm getting sound through, I just don't hear much compression happening.  There's also a subtle fuzzed sound mixed in way in the background.

With no sound coming through, I measure ~1.6 V across the vactrol LED, and with a loud note played in it jumps to ~1.8 V.  Measuring the resistance on the other side, it goes from ~2.5k with no input to ~1.7k with a loud note (though that measurement may be suspect in a live circuit).  Do these sound like correct numbers?  Seems like not a lot of gain change.

Anyone who has a happy unit care to crack it open and see if my measurements are sane?

The vactrol specs look similar-ish to those I used.
1.6V across the LED sounds ok.
Without any input, the resitance across the LDR should be much higher.
Which version did you build?

Thanks for the reply Sam.  I built the version with ratio and threshold knobs, though not from your layout as I was combining it with an EQ circuit and had to adjust for the different polarity on the 0202. 

I think my next step is to disconnect the LDR half of the vactrol and measure its resistance out of circuit to see if maybe I just have a wonky vactrol. 

Edit - Taking the LDR out of circuit is a good idea.
The other thing you could do is to short the 10k resistor from the threshold pot to Vb and turn the threshold to 0. This should null the LED driver and you should then get dark resitance from the LDR (note that the ratio pot is in parallel, so turn this to max to measure the resistance). If that doesn't work then there is probably something wrong with the LDR driver. Perhaps one of the diodes is backwards?

So I think the Vactrol is OK, I measured the resistance out of circuit and it goes from about 3k to 1.2k when the input is silent to loud guitar.  That's obviously too low for the circuit to really work right, but it seems to be how this vactrol responds to the voltages being fed into it.

I found I can get more volume and slightly more range by putting some LEDs in parallel with the vactrol LED, sucking up some of that current and dimming all the LEDs a bit, but that seems a hacky solution.  How would I modify the envelope follower circuit to increase the spread of voltage between quiet and loud?  Does R7 control that?

The vactrol should give you 10Meg dark resitance:
https://www.datasheetq.com/datasheet-download/188542/1/ETC/LCR-0202
You can check this by measuring the resistance when the circuit is not powered - you should see the resistance of the ratio pot.

My guess is there is a problem with the LED driver. You should have negligable current through the LED with no input. If you are measuing ~3k, you have 1-10 mA through the LED.
The current is set by R7, which if 1k should limit the current to ~4.5 mA on a 9V supply.
(schematic: https://samdump.wordpress.com/welcome/projects/light-speed-compressor/)
My guess is that you have grounded one of the op-amp inputs (threshold pot?). Can you post voltages on the op-amp pins?

just a quick hi sam.... :)

good to see your still about on here.  8) 8) 8)

Hi Rob. Got a bit distracted by other things, but seems I still have notifications switched on, and it's nice to dip a toe in on occaision.

Well, smack my face and call me late to breakfast.

I had C5 backwards.  :icon_redface:

Was originally getting 5.75V on pin 7, after correcting C5 I'm getting a more normal 3.7 (which is around what the rest of the I/O pins are reading.  My Vcc is only 7.5 due to probably over-aggressive power ripple filtering, which is why it's not 4.5V).

Now the vactrol goes from ~90k down to single digits with a loud input, which is what I would expect in parallel with the 100k pot.

We have compression!  And the fizzy sound is gone AFAICT.

Just goes to prove my adage that the longer I look for the bug the dumber the mistake was.

Thanks for your help Sam, and great circuit! 

That'll do it. Glad you got it working.

Edit: Nevermind.