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

[samhay]

OK, fair enough.

Maybe a dumb request, but since I've found some contradictions on the net, here goes...

Would it be possible to indicate the lug 1 for the pots on the schematic? I could never get my head past volume controls where that's concerned. Just seems that the conventions for this are all over the map.

Well now you know how to decode Merlin's dialect, wherever you see CCW, you can translate to lug 1.

[digi2t]

Well now you know how to decode Merlin's dialect, wherever you see CCW, you can translate to lug 1.

Yup, got it. Thanks Sam.

[jonny.reckless]

I label the pins on a potentiometer CCW, W and CW for counterclockwise, wiper and clockwise. On the schematic symbol I show the wiper off center so you can see visually which end is clockwise. I label the pins on a SPDT switch NO, C and NC for normally open, common, and normally closed. I label the pins on a diode A and K for anode and cathode (kathodos). I label the pins on a jack socket T for tip, R for ring and S for sleeve. And so on ad infinitum.

For multiple gang parts I just append numbers 0, 1, 2 etc to this nomenclature. Pin numbers are always confusing IMHO. Some people number the pins of a TO92 transistor looking from the top, some looking from the bottom. Some people even put pin 3 in the middle. This is how it is with SOT-23 packages. It's all rather confusing and I try to do everything I can to make it hard to make a mistake. Even so I still make plenty of mistakes. 

[digi2t]

Wiper biased to the CW side. Check. Greatly appreciated.

That was

[digi2t]

jonny, can the treble boost be incorporated into the new design? It's one of the features that I really appreciate on the original design, especially with humbuckers.

[samhay]

^It's already there via C7 and R9.

[PRR]

Another convention: an arrow showing which way the slider moves for "more".





(Remember that CW and CCW are sometimes relative. Gas stove knobs are CCW for "more". The old Quad hi-fi had only part of the knob exposed, and internally turned withershins. Ham equipment sometimes have right-angle gear drives.)

[suryabeep]

Hello!
I built a 5-knob version with the treble boost switch. Everything works fine, except the LED. It's off when the effect is on (I can tell because the knobs change the sound), and on when the effect is off. How can I fix this?
I'm using a 2n7000 flipped 180 degrees in place of the BS170.

[digi2t]

Another convention: an arrow showing which way the slider moves for "more".





(Remember that CW and CCW are sometimes relative. Gas stove knobs are CCW for "more". The old Quad hi-fi had only part of the knob exposed, and internally turned withershins. Ham equipment sometimes have right-angle gear drives.)

And yet... the schematic you post also shows the pot lug numbers. For all intents and purposes, the arrows could simply be indicating the direction to Santa's workshop.

[jonny.reckless]

I finished my latest batch of changes to the gate design. I added a new peak detector, and a threshold comparator with some hysteresis. I also added separate attack and release controls for the noise gate. It's now an 8 knob monster

Here is the schematic:


Here are some photos of the revision C PCB:





Here it is in action. Sorry for the poor audio quality, it's just my phone camera, which is adding some weird compression of its own to the sound
One day I'll get a proper camera and mic for this stuff.

Overall I am fairly happy with the circuit design. It's a bit more complex than I would have liked, and at 100 x 70mm will no longer fit inside a 1590B (you need a 1590BB instead), but the added flexibility to the noise gate is nice. Let me know if you would like a PCB to build yourself.

[suryabeep]

I finished my latest batch of changes to the gate design. I added a new peak detector, and a threshold comparator with some hysteresis. I also added separate attack and release controls for the noise gate.

  WOW! I can't even begin to comprehend what's going on in this circuit. Could someone please explain how the gate works and what Hysteresis is? (or provide some links if possible)
Amazing stuff Jonny!
PS - can I replace all those 1uF caps with something smaller like a 100n? 1uF's are kinda expensive compared to 100n, especially when using so many 

[jonny.reckless]

Could someone please explain how the gate works and what Hysteresis is? (or provide some links if possible)

OK here is a brief description of how the circuit operates

IC5 is a charge pump voltage inverter. It generates a negative voltage from a positive one by moving charge between a "pump" capacitor C21 and a "reservoir" capacitor C24. There is a lot of 10kHz ripple on the output. TR3 is acting as a lowpass filter to remove the 10kHz ripple from the negative supply rail.

When you have both positive and negative supplies, the rest of the circuit design gets simpler, as you only have one ground reference for all signals.

With a single supply, it's like only having unsigned integers in your computer. Everything is still possible but you have to think a lot more about how to implement simple things. I generally prefer to generate a -8V rail rather than split the voltage and create a virtual ground at 4.5V. Not only do you get more headroom, but you can fill the top of the PCB with ground plane which improves noise performance, since real ground and signal ground are the same net.

TR1 is a n channel JFET configured as a source follower. It is basically a unity gain buffer which turns the high impedance guitar input into a low impedance signal. You could make a buffer in lots of ways. I happen to like the sound of this J112 buffer with single coil guitars, and I expect this pedal to be first in my effects chain.

IC1A is an inverting amplifier. So is IC1B. IC2 is an operational transconductance amplifier (OTA). https://en.wikipedia.org/wiki/Operational_transconductance_amplifier Its transconductance is proportional to the control current into pins 1 and 16. In this circuit, it is set up to be a virtual feedback resistor around IC1A. The higher the control current, the lower the virtual resistance, and the less overall gain (due to the lower feedback resistance). So basically gain reduction is proportional to the current flowing into pins 1 and 16. You could just use one half of the LM13700 as in Merlin's original circuit, but paralleling them up like this reduces the noise a little, and since you get 2 in a package you might as well.

IC3 is the compressor sidechain. I have not changed this from the original design, other than adapting it for a bipolar power supply. IC3A is a negative peak detector, with separate attack and decay time constants. It produces a negative voltage on C13 that represents the peak value of the input signal. R20 and C11 further smooth this voltage. IC3B and TR2 convert the voltage into a current. R12 sets the conversion ratio, in this case 1mA / volt. So overall, the gain reduction is proportional to the input signal amplitude. Bingo - compression! In my opinion it is the slightly unusual feed-forward sidechain peak detector that gives the engineer's thumb its signature sound, which I like so much. You can really experiment with the character of the compressor by trying different dual opamps in place of IC3.

IC6 and IC4 are the noise gate sidechain. The noise gate works by setting the gain reduction to a maximum (max current into IC2 control inputs).

IC6A looks familiar to anyone who has built a tube screamer. It is just a high gain clipping circuit, so that even very low level signals are amplified enough to open the gate. IC6B is an inverting amplifier with a gain of -1.  D8, D9, C26 and R28 form a peak detector for peaks of either polarity. It is important with a noise gate to detect both positive and negative peaks so the gate opens fast enough. This doesn't matter so much for a compressor hence the half wave level detector for the compressor sidechain.

IC4A is set up as an inverting threshold comparator. When its input signal is strong, the output goes low and the gate opens. When the input signal is weak, the output goes high and the gate closes. R26 provides hysteresis (positive feedback) so that the threshold to open the gate is about 4dB higher than the threshold to close the gate. This Schmitt trigger stops the gate from chattering as the note dies away. https://en.wikipedia.org/wiki/Schmitt_trigger

R23 and C17 are there to slow the edges of the transitions to avoid audible clicks when the gate opens and closes. D5, D6, RV4, RV6 and C15 make up the gate attack and decay time constants. IC4B is just a buffer to supply current into the control pin of the OTA, and hence control gain reduction. D7 is there to make sure that IC4B only sources current, and never sinks current from TR2. LED5 glows when gain reduction is being applied, either by the gate or the compressor. Its brightness is roughly proportional to the amount of gain reduction being applied.

I hope this helps 
Jonny

PS: 1uF and 10uF ceramic capacitors are just a few cents each when you buy them in bags of 50 or 100 off eBay. I tried 100n in the circuit and it lets some control clicking into the audio path which you can hear with high gain sounds. I recommend using 1uF ceramics for decoupling as shown. Why spoil the performance of the circuit for a saving of less than a dollar? These are what I used for my build. I measured them with my LCR meter and they are really good. Loss factor less than 0.1% and ESR less than 0.1 ohms. Ideal for the application. With multilayer ceramic capacitors it is always a good idea to get voltage rating much higher than the circuit voltages, as this type of capacitor decreases in value when significant bias voltage is applied. I typically use 50V parts for stompbox circuits.
https://www.taydaelectronics.com/1uf-50v-multilayer-ceramic-capacitor.html
https://www.ebay.com/itm/50pcs-Monolithic-Ceramic-Chip-Capacitor-1uF-105-50V-Pitch-5-08MM-New-M/321389847401?hash=item4ad453e369:g:-E0AAOxycD9TW82M
https://www.ebay.com/itm/50pcs-106-50V-10uf-10000nf-10000000pf-Multlayer-Ceramic-Capacitor/112250694765?hash=item1a22a9ac6d:g:GMoAAOSwux5YXhz7

[suryabeep]

I modified the original schematic to include the paralleled OTAs, however I am not sure if this is correct. Could someone take a look and tell me if this will work?
I simply took the 'parallel' configuration from jonny's schematic and applied it to the original schematic without changing anything else.
Also, at the risk of repeating myself (since my original seems to have gotten lost in the replies): the indicator LED on my ET build is on in bypass and off when the effect is engaged. How can I fix this?

[jonny.reckless]

You've got the OTA incorrectly wired. The ouput should connect to the inverting input of the opamp. R7 is 220 ohms in the original circuit. You need to move LED1 and R19 to the drain of Q2.

[suryabeep]

Thank you!
So, like this then?

[jonny.reckless]

Closer. What is R3 for? I think you'd be better off without it.

[merlinb]

Thank you!
So, like this then?

Correct. R3 is not essential. I included it in my original design as an attempt at stabilisation, but it turned out not to need it.

[jonny.reckless]

Here is my engineer's thumb redux in a case. It sounds really sweet now. All the residual noise went away when I put it in an aluminum enclosure. I connected the case to the PCB ground plane.


This is the final circuit revision I ended up with after a bit of value tweaking with different guitars.

I have a few spare PCBs. Let me know if you would like to build one.

[Ben N]

Been following this thread along. What can I say about Merlin's little compressor? It grew some! Now that it is a super-flexible uber-comp, maybe the last two things to add are a separate side-chain input (to facilitate placement anywhere in the signal chain, so long as the envelope detector and gate get fed a clean, dry signal) and MIDI control.

[EeroN]

Hi all,

I built an Engineer's thumb a few years ago and it works great. However, I'd like to get a bit more level boost from the circuit, not much but I find I almost need to max out the Level control to match the bypassed signal.

I know how to adjust the gain of a typical opamp circuit, but I'm not sure how I should go about it with the OTA in the feedback loop, so as to not mess up the compression effect.

Thanks!