THE ENGINEER'S THUMB... At last, a better compressor!

[jonny.reckless]

I really love this compressor. I have made some modifications:

JFET input buffer (improved transparency with single coil pickups)
Reduced self noise (parallel up the 2 halves of the LM13700, replace TL072 with NE5534 low noise audio amplifier, reduce the base impedance of the LTP in the LM13700) in total by about 10dB w.r.t. the original design
Added a simple noise gate

[Beo]

Jonny, does the noise gate work well? Consistent for single/dual coil, clean and distortion? Is it set and forget (maybe install as trimmers)?

Thanks!
Travis

[jonny.reckless]

I have this as the first thing in my chain. I imagine it would be difficult to gate the noise once it's already been compressed or distorted. It works well for me. It's a soft gate in that it progressively applies up to 30dB of gain reduction when the signal is below the threshold. It never hard mutes. I generally don't need a gate with humbuckers. You could easily scale the range if you need to by adjusting the value of R18: smaller value makes the gate open at a lower threshold, or you could replace the pots for fixed settings if you like. If you want to make the gate switchable just install a toggle switch in series with the collector of TR3.

I use these to wire up the sockets, switch, power supply and LED:
https://www.taydaelectronics.com/pcb/diy-guitar-effects/3pdt-footswitch-diy-pcb.html

I'm planning to do a PCB when I get time.

[rankot]

I'll do a PCB if this is verified schematics. I tried to combine Fencepost noise gate with Engineer's Thumb, but didn't have enough time to complete it, and that one was really complicated. This one looks better!

[BetterOffShred]

I'll buy one

[rankot]

Few questions for jonny.reckless:

• Why have you omitted compressor's threshold potentiometer?
• Are you sure that 100n filtering caps should go from 9V to ground and 4.5V to ground, shouldn't they go from 9V to 4.5V and 4.5V to ground?
• How about Blend pot? Anyone tried to apply one?

[jonny.reckless]

I found I didn't need the threshold pot since the sustain pot has pretty much the same action. All the decoupling caps go to ground intentionally. When I do the PCB, the whole of the top copper layer will be ground plane which will keep noise to a minimum. I usually do analog audio on 2 layer boards with one layer dedicated to ground. It works really effectively for guitar stuff.

I tried to keep the gate simple. It just dumps current into the OTA control pin and therefore sets maximum gain reduction. Of course the OTA noise is then at its maximum but with 2 of them in parallel and 22 ohm base impedance it's pretty quiet.

It's not a hard mute. You may also add a cap (100n or so) across R13 to make the gate open slightly less jumpy if you prefer. I like a noise gate to open instantly so it doesn't cut off the first half cycle of the note. This is especially important with drop tunings.

I've not considered a blend pot although with the JFET buffer on the input it would be fairly easy to add one if you wanted to. I find parallel compression doesn't sound great on anything except drums.

[jonny.reckless]

I have done a quick place and trial route. This should fit into a 1590B enclosure.  Still have a bit of work to do...

[rankot]

Is it necessary to use TL071 for Vref generation, why not something like this:

Smaller footprint.

If it is necessary to use op amp for Vref, why not using NE5532 instead of NE5534, so one half of it can be used for Vref? One IC less.

[merlinb]

If it is necessary to use op amp for Vref, why not using NE5532 instead of NE5534, so one half of it can be used for Vref? One IC less.

Probably trying to keep current consumption down. TL071 needs <2mA, whereas NE5532 takes about 4mA per section.

[jamierichards1234]

Hi, Firstly I want to say thanks Merlin for sharing this excellent design. I've built mine and it's up and running, but I'm wondering if anyone has any ideas how to modify it to further reduce the release time?

I've swapped out the 470k resistor for a 500k pot which I currently have in series with a 47k resistor to stop me going too far (47k seems to be as low as I can get away with before everything goes mental). It's getting me close to where I'd like to be but it would be really nice to have the option of even shorter release times. Also I'd rather not be running it right on the edge of instability if possible! 

Anyone got any suggestions of how I might be able to do this? I don't mind losing the option of longer release times to get there..


(FYI the reason I'm after shorter release is that I'm using an external sidechain for 'ducking' to a kick drum but the signal doesn't quite come back up as fast as I'd like)

Many thanks,

Jamie

[merlinb]

t it would be really nice to have the option of even shorter release times.

Use smaller smoothing caps in the side chain, e.g. swap the 1u caps for 470nF. This will halve the attack/release times. You get less smooth control voltage though, obviously, which is the price you pay for short release.

[digi2t]

t it would be really nice to have the option of even shorter release times.

Use smaller smoothing caps in the side chain, e.g. swap the 1u caps for 470nF. This will halve the attack/release times. You get less smooth control voltage though, obviously, which is the price you pay for short release.

Just a spitballin' question; If going with smaller caps, would the type of composition play a roll in helping smooth the control voltage? Example, using polystyrene rather than polyester or metal film.

[merlinb]

would the type of composition play a roll in helping smooth the control voltage? Example, using polystyrene rather than polyester or metal film.

No way. It's not smooth as in "smooth tone, bro!" we're talking about here, it's your grandpa's good old fashioned rectified ripple, and it doesn't care what your caps are made of! 

(Actually it's only a half-wave rectifier in the side chain, but you get the idea)

[digi2t]

Alrighty-o then. Never any harm in asking. Thanks Merlin.

[jonny.reckless]

I got the PCBs for the "Engineer's Thumb redux" this evening. I will try and get one built up over the weekend and post some audio.

[jonny.reckless]

Is it necessary to use TL071 for Vref generation, why not something like this:

Smaller footprint.

If it is necessary to use op amp for Vref, why not using NE5532 instead of NE5534, so one half of it can be used for Vref? One IC less.

I generally find zeners are very noisy especially around 5V. Also without feedback the output impedance of the emitter follower will be relatively high (and dependent on current). The NPN can only source current and not sink it. The op amp based VBIAS is pretty simple and cheap and works well.  For a general purpose regulator I really like the LM317 series, they are quiet and have excellent line and load regulation. But again are unipolar only.

[jonny.reckless]

If it is necessary to use op amp for Vref, why not using NE5532 instead of NE5534, so one half of it can be used for Vref? One IC less.

Probably trying to keep current consumption down. TL071 needs <2mA, whereas NE5532 takes about 4mA per section.

I used the external compensation pins of the NE5534 to guarantee stability of the op amp under heavy dynamic gain reduction. I found this approach worked better than the feedback pole/zero of the original design. But that might have been an artifact of my vero prototype. The NE5534 is cheap, widely available in DIP, has very low noise, low output impedance (i.e. strong drive capability), low distortion and is externally compensated for noise gains less than 3. It's one of my favorite audio op amps.

[rankot]

What about gate LED indicator?

[iiimonfire]

Hey all! I've been looking for a new low-noise compressor for bass, and have been reading up on this. I breadboarded it over the weekend according to Merlin's schem: http://www.valvewizard.co.uk/engineersthumbschem2.jpg with all of the mods to make it a 5-knob pedal (volume, threshold, ratio, attack, release).

One thing I am noticing is that turning the ratio knob CW will dramatically amplify the signal, so that it distorts heavily way before ratio is at ~50%. That basically renders more than half of the pot range pretty much useless.

I've tried switching out the LM13700 with one that was working in a Keeley compressor I made - so I'm guessing that's not the culprit. I've gone over the schem and the vero layout I created as well, and I am pretty sure they are correct. FWIW, I'm using a TL074CN instead of two TL072's which I see a lot of other layouts (e.g., Tagboard, Sabrotone) using.

Any ideas what could be causing this massive distortion?