Reckless compressor - an original feed-forward diode based compressor

[jonny.reckless]

@ J.R.: since you're driving with a current, you might want to arrange to drive the top of the diode modulator with a voltage equal but opposite to the voltage at the bottom of the bridge. It will take some of the load off the differential amplifier to keep control signal feedthrough. Yeah, I know, diffamps are good, but no sense making the work harder than they have to.

That's an interesting suggestion. I was considering improving the diff amp by building a 3 op amp instrumentation amplifier or maybe even using an INA103, so both legs of the diode bridge see an infinite impedance load. Do you think balanced diode bridge drive would be superior to this approach? I guess I could go the full monty and do both :-)

[PRR]

20dB better balance in the modulator is worth 20dB better CMRR in the recovery stage.

Which choice leads to better performance or lower part-count, I do not know.

[jonny.reckless]

Thanks, I am going to see what I can do to simplify the design a little and make it more DIY friendly.

[samhay]

Thought I saw a revised schematic earlier that used BJTs?

In any case, you can get a pretty closely balanced (in terms of input impedance) diff amp using a single op-amp and I have got this to work  'close enough for rock and roll' in a couple of related designs. Most recent:



Thread: https://www.diystompboxes.com/smfforum/index.php?topic=119901.0

[Ben N]

Watching this thread. Cool design, Jonny, seems like a way superior alternative to Ross/MXR types, and very Strat friendly.

[jonny.reckless]

I've prototyped using a discrete long tail pair as the differential gain stage. It's working quite nicely. I'll post more in a week or two when I've had chance to optimize and debug the design. I found I could get away without the negative supply which simplifies things a bit.

[jonny.reckless]

Here is where I have got to with the revision B circuit. It's simpler than the original with fewer components. I also simplified the side-chain and replaced the time control with fixed attack and decay of 20ms and 200ms respectively. I'll post some clips of the sound when I get chance, but it's pretty sweet.

https://drive.google.com/open?id=1y6zuyUdH2dI2hWFurxAMeCIFM8huKZCV

[jonny.reckless]

I've finished developing revision B of the diode compressor. It's simpler, cheaper, has fewer components, and is easier to build than the revision A. It still sounds pretty sweet, but only has 2 controls now: sustain and volume. It's not quite as subtle and transparent as revision A, and has more of a pronounced "quack" at higher sustain settings.

Here is the revised schematic:


Here it is in action. The video also has some photos of the PCB in the case:

Let me know if you would like a rev B PCB.

The music is from my band Saint Wry: https://open.spotify.com/album/76AYXUusxuLu0GiVGBQn6q

[Ben N]

Nice squish, no pumping or other unpleasant artifacts that I can hear on the demo, but definitely "there"--not an always-on compressor. Well done, Jonny.

Def interested in a pcb. Is Israel doable?

[jonny.reckless]

Nice squish, no pumping or other unpleasant artifacts that I can hear on the demo, but definitely "there"--not an always-on compressor.

If you would like a more transparent "always-on" compressor, I think the revision A design is slightly better in that regard. It's very transparent due to the full wave sidechain averaging.

I can post you a PCB of either revision. Message me your address and I will stick one in the mail. $5 plus postage, which I am guessing will be a few bucks to Israel.

[bartimaeus]

Thought I'd share a pic of my build of the REV C PCB! I'm really happy with how this turned out, it's exactly what I'd been wanting from the compressors I've been trying, plus anything else I could ever want. JUUUUUST barely fits into a 1590BB enclosure – the jacks are a super tight fit, but it's doable. There's even enough room in the front to cram in a 3PDT (or SPDT with a relay PCB), but I'm using it as an always-on compressor so I didn't bother. Maybe I should have gone for a 125-BB (smallbear makes them), but I'm happy. Extra switch is to toggle the treble boost option.

I really have to thank Jonny Reckless for this excellent compressor!

[diffeq]

Thought I'd share a pic of my build of the REV C PCB! I'm really happy with how this turned out, it's exactly what I'd been wanting from the compressors I've been trying, plus anything else I could ever want. JUUUUUST barely fits into a 1590BB enclosure – the jacks are a super tight fit, but it's doable. There's even enough room in the front to cram in a 3PDT (or SPDT with a relay PCB), but I'm using it as an always-on compressor so I didn't bother. Maybe I should have gone for a 125-BB (smallbear makes them), but I'm happy. Extra switch is to toggle the treble boost option.

I really have to thank Jonny Reckless for this excellent compressor!

Nice build!

[rankot]

Jonny, what's the difference between REV A (PCB I have) and REV C?

[jonny.reckless]

Jonny, what's the difference between REV A (PCB I have) and REV C?

He's referring to issue C of the engineer's thumb redux.  BTW nice build, and let's not forget to credit Merlin B for his original design. You couldn't ask for a better starting point for mods. Although this is a different thread :-)

I've only done A and B of the diode compressor. Both have videos so you can compare the tone. Basically rev A is more transparent and rev B is more squishy.

[Ben N]

Mine is awaiting some parts, but in the meantime I wish to thank you, Johnny, for keeping it (Rev B) simple in terms of parts. I think there is a total of two resistors in the whole shebang whose values are not multiples of ten, and precious few caps altogether. I think if I was wiring this up on perf, I would try to include some kind of jangle-box option, but a I'm not, I won't--unless you have a suggestion as to how this may be easily done without carving up your nice board.

[dschwartz]

What about using a rectifier bridge instead of matching diodes?
I would think that the internal diodes if the bridge should be reasonably matched.

[pinkjimiphoton]

been so caught up with work related projects, really lookin forward to the build!
thanks again mr. reckless!

[jonny.reckless]

What about using a rectifier bridge instead of matching diodes?
I would think that the internal diodes if the bridge should be reasonably matched.

You could do that, but the dynamic impedance of a typical rectifier is going to be a lot lower than 1N4148s, so the transconductance of the sidechain would need to be modified to suit. I'm not sure how well rectifier diodes in a bridge rectifier are matched when only carrying milliamps. If you've got some lying around maybe see what the forward voltages are with a DMM?

[jonny.reckless]

I think if I was wiring this up on perf, I would try to include some kind of jangle-box option, but a I'm not, I won't--unless you have a suggestion as to how this may be easily done without carving up your nice board.

What's a jangle box? I am guessing some compression / EQ hybrid? I've thought about adding pre-emphasis and de-emphasis filters around the compressor, so that the gain reduction is more balanced across the frequency spectrum rather than being defined primarily by the low end. I've experimented with a couple of graphic EQs in front and behind the compressor, but never really gotten to a sound I like. I did consider adding an insert into the side chain so you can do this sort of thing. You'd basically put a 1st order highpass filter in front of the sidechain so the compressor kind of "ignores" low frequency more. I think Joe Meek used this trick with compression in the 60s.

[Ben N]

IIRC, the Marshall ED-1 has some kind of filtering in the sidechain for this purpose--which it calls "Emphasis"--although I'm not sure if it controls threshold or the amount of compression. As for the Janglebox concept, yeah it is some kind of treble emphasized compression, as exemplified by the onboard Rickenbacker compressor as used by Roger McGuinn back in the day. Somebody put out a commercial stompbox with than name, which was probably a slightly modded Dynacomp, and there have been a few discussions around here about implementing it as a mod.