PMC - Poor Man's Compressor

[jonny.reckless]

btw, where's the half-wave side chain from ? ... I've seen it before

Dynacomp, Ross compressor, Boss CS-2, DOD FX80, you name it, they all pretty much did the side chain like that with some minor variations. I just didn't bother with the phase inverter and full wave peak detector; you don't actually need it unless you are building a super fast attack limiter, and I was trying to keep component count down. As I mentioned in the original post, the whole design was heavily inspired by the circuit topology of the Dynacomp.

[Eb7+9]

i was wondering about values ... attack is still pretty fast in the original dyna, which were populated half wave, the sink device has no current limiting resistor there (yours does that with 470r) ... the release time is longer in the original dyna with 150k/10u versus your 22k/10u ... otherwise identical in structure

was just curious // well done!

[PRR]

My suspicion is that the very wide range of playing styles and expectations mean that eventually you "must" support variable time constants.

(The LA-2a is the exception to prove the rule.)

And C12 is also part of the time-action. 0.1u seems small to drive 1k and a naked bass. The first person to complain "tighter" should tack a 10uFd across C12.

Me myself, I would not be content with a half-wave, but that's my technician background.

[jonny.reckless]

Me myself, I would not be content with a half-wave, but that's my technician background.

I really don't think it's necessary for standard guitar compression where you want an attack time of 10ms to 20ms typically. The Dynacomp did it that way and everybody just seemed to copy it without trying it out any other way. I experimented with a full wave sidechain and in my opinion the difference in sound and feel was pretty negligible, and it adds to the cost and component count so went against the "Poor man's" minimalist spirit.

The Engineer's Thumb also only has a half wave sidechain and that sounds great.

[Eb7+9]

I experimented with a full wave sidechain and in my opinion the difference in sound and feel was pretty negligible ...

no doubt ...

there's been lots of talk about the dyna side-chain on this forum over the years
... it keeps coming up
 
original 70's dynacomps had boards populated for full-wave operation
but an intentionally missing jumper made them half-wave ...

see note [3] in one of my early hand scribbles

http://www.lynx.net/~jc/dynacompLayout.gif

whoever at MXR came up with the dyna thought it might be better to go FW "as the old tube designs did"
or at least have the option to use it, otherwise that stuff wouldn't have been populated in

but ...

the fact that it was left disengaged suggests a few things :

(i) obviously MXR couldn't put out a pedal design with a 600mV output clamp
(ii) had they not been perfectly pleased w HW operation you can bet your dollars that circuit would never have been manufactured ... at least as is

somebody overlooked a critical performance aspect of the design at first
realized it, had doubts ... edged bets (just in case)
and in the end were forced to ditch ... the rest is history

the only drawback some see with this side-chain architecture is its lack of potential for controlled and independent variability ... seems the best way is to use it is as was done here, with its generally fast enough attack and properly slowed down release // invariably leading to a streamline design

[11-90-an]

so i've been curious... what does this do?

wondering why this network appears twice in this schem... 

[antonis]

so i've been curious... what does this do?

wondering why this network appears twice in this schem... 

Read first - wonder after reading..
"C9, R10, R21 and C14 form a simple emphasis / de-emphasis circuit to reduce the noise (hiss) of the transconductance stage."

As for "double appearance", differential stages consist of two inputs..

[11-90-an]

so i've been curious... what does this do?

wondering why this network appears twice in this schem... 

Read first - wonder after reading..
"C9, R10, R21 and C14 form a simple emphasis / de-emphasis circuit to reduce the noise (hiss) of the transconductance stage."

As for "double appearance", differential stages consist of two inputs..

wait what... i didnt even notice that.... gah 

so basically they are filters... right?
found this...

[PRR]

> basically they are filters...

Mr Reckless' plan makes a series resistor which is 2.2k for lows and (2.2k||1k)= 0.68k for highs. Unloaded this does nothing. But you snipped-out the (important) load, so we expect highs to be 3X or 4X as strong as lows. (And we expect to find the inverse response after the processing.)

[jonny.reckless]

Basically it's a first order shelving treble boost before the long tail pair (emphasis), and a complementary first order shelving treble cut after it (de-emphasis). This keeps the overall signal frequency response of the pedal flat, but reduces the amount of hiss generated by the LTP, which can be quite a problem if you put a compressor in front of a drive pedal for example. It also makes the clipping transients you get on note attack sound a bit snappier.

[11-90-an]

Did some experimenting on this and discovered that adding a 100k pot and a 22uF cap in parallel with C12 makes a "gain reduction" control. It controls how low the gain reduction can go.

Now for the questions...
1. What is the minumum value of R17 before blowing the transistors up? (Don't wanna risk some trannies... )
2. Would adding a "ratio" control be putting a pot on pins 1 and 2 of TR13? If not, where?

[antonis]

As long as Base resistors (R16 & R18) are kept in place, neither TR12 nor TR13 could be blown up..
But there should be a limit for half-wave peak detector to be practically functional..
(by zeroing R17 value, you will turn TR12 into a crass Emitter follower.. )

[11-90-an]

Changing the circuit like this should make it a full-wave peak detector... right?
(And would 20Ω be sufficient for no blown-up trannies?)

And if anyone has an idea for a ratio mod, please share...

[jonny.reckless]

This looks OK to me, although I would question the benefit of doing a full wave sidechain. I experimented with a full wave side chain like the dynacomp but, at least to my ears, the difference was negligible.

Now for the questions...
1. What is the minumum value of R17 before blowing the transistors up? (Don't wanna risk some trannies... )
2. Would adding a "ratio" control be putting a pot on pins 1 and 2 of TR13? If not, where?

You can probably go down to 10k on R17, anything less than that and the release time will be so fast you'll get distortion on low notes due to gain modulation. You might want to go the other way, maybe a 10k resistor in series with a 100k pot for R17 to increase the release time. I always set release time pretty fast on a compressor; I like the snap it gives me. Most other guitar compressor designs have a bigger release time constant than this (200ms) for example the engineer's thumb and dynacomp are both in the 800ms range.

I don't think this would be easy to adapt to variable ratio due to the simple peak detector nature of the side chain. Maybe a small resistor in the emitter of TR13? Perhaps 100 ohms or so would soften the knee a bit. It's really a simple cheap circuit; I would probably use a THAT2181 if I was going to build a full 5 knob compressor

[jonny.reckless]

SAME GREAT TASTE, NOW ONLY HALF THE FAT
I realized I could have laid this out a bit smaller. I did a revision B PCB layout which fits in a 1590B enclosure. Still uses PCB mounted pots, sockets and switch for ease of assembly and test.
Details here: https://drive.google.com/drive/folders/1_beVxacCYdQylqaJsPzDPl9SqTk5rp4o?usp=sharing




Let me know if you would like a PCB. I also uploaded the Gerbers and NC drill file to the Google drive if you want to make your own board from this layout.

[11-90-an]

Just wondering jonny.reckless, why didn't you put the peak detector at the input side like the engineer's thumb? It would probably give adjustable threshold, attack and ratio... and sustain... So adding volume/level to that, you have a 5-knob compressor!

[11-90-an]

Finally.. a legit mod! 

For a "threshold" control, place a 100k pot in between C12 and R18...

i'm afraid that's all...

[PRR]

...Changing the circuit like this should make it a full-wave peak detector...

Except the phase splitter is practically a cripple, can't swing the lower Q into conduction. The upper Q works, but it is back to half-wave.....

[11-90-an]

...Changing the circuit like this should make it a full-wave peak detector...

Except the phase splitter is practically a cripple, can't swing the lower Q into conduction. The upper Q works, but it is back to half-wave.....

So I need to change the emmiter resistor value...right?

And also, jonny.reckless, why did tou use a transistor configured as a diode rather than a diode itself? Diodes are cheaper...

[PRR]

> change the emmiter resistor value...right?

Wrong. To preserve AC balance you change both resistors and come right back to the same near-bottomed condition.

> Diodes are cheaper...

Not always. And if you buy transistors by the scoop, a stray diode is more expensive.