Tube compressor/limiter with single pentode

[blackcorvo]

It started as a simple preamp for bass and guitar, but after getting curious about the possibility to add a compressor, I came across this circuit, one of the first images that show up when searching for tube compressor schematics:



The concept is simple: a triode gain stage is used to isolate a rectifier circuit that provides a variable DC negative voltage whenever there's a signal, which is then fed to a (sharp cutoff, dual control) pentode's supressor grid. The louder the signal, the lower the gain of the pentode, causing the compression effect.

I liked the idea, but wanted to try and simplify it as much as possible, perhaps using a single pentode stage. At first I simply copied the circuit above, substituting the dual diode for 2x 1N4007, and taking signal from the pentode's plate. This proved to work, but introduced distortion to the signal (of course) because of the diodes. I then remembered that the screen grid can also output signal, and just tied the decoupling capacitor from the rectifier circuit to the supressor grid, and it worked!

And after playing around with values, this is the final result:



Already breadboarded and tested. Now I just need to sit down to record a few sound samples. I used a 5636 because it was the most easily accessible tube at the moment. I also have a few 5725 somwhere. I'd imagine this would work with any sharp cutoff dual control pentode out there, provided some tweaking of resistor values.

The reason I added the "limiter" is because of the way this circuit works. Since it's reducing the gain of the stage with larger signals, it's essentially working like a limiter, impeding the signal to go above a certain threshold. But sound-wise, it sounds like a compression. So I'm just calling it both.

[EDIT]
Here's a quick sound sample:

https://youtu.be/VzQeS60YYHI

(BTW it's not distorting with the guitar, it's something that's vibrating in the room that I couldn't locate at the time of recording).

I'm using a TPA3118 module board as the power amp, feeding into a 1x10 8 ohm cabinet. I adjusted the volume so the camera wouldn't compress the sound.

[mozz]

Who wants to make me a pcb layout for the top circuit so i can order some boards?

[merlinb]

I'd be interested to see a plot of gain v. input level.
Do you get much thump from the CV feedthough?

[Ben N]

Cool idea. How are you powering it?
Also, how does it compare to the original circuit? I happen to have a Russian submini rectifier tube, along with various dual triodes & pentodes, so I think that might be fun to breadboard if I can work out power.

[blackcorvo]

Who wants to make me a pcb layout for the top circuit so i can order some boards?

If I ever figure out how to use those PCB board softwares that make gerber files for manufacturing, I might try my hand at it.

I'd be interested to see a plot of gain v. input level.
Do you get much thump from the CV feedthough?

I currently don't have a way to plot curves like that, but you can adjust the "thump" by playing around with the cap values for the rectifier circuit and the cathode bypass cap. The values I used are a happy medium with what I have available here.

Cool idea. How are you powering it?
Also, how does it compare to the original circuit? I happen to have a Russian submini rectifier tube, along with various dual triodes & pentodes, so I think that might be fun to breadboard if I can work out power.

I'm using this board to power it (you can find on eBay by searching "250v 6v Nixie board"):
https://www.ebay.com/itm/394403334630

I didn't try the original one, just the rectification method. I do have a 6S6B-V single triode I could've used but I didn't feel like searching the tube box for it, so I don't know how much better the original circuit might be.

[Chillums]

Who wants to make me a pcb layout for the top circuit so i can order some boards?

I'm stuck at home today waiting on tires for my truck so I'll see what I can come up with.  I'm assuming you mean the first schematic correct?

[amptramp]

Are you getting much distortion with this circuit?

Most tube compressors use remote-cutoff tubes like the old Fairchild compressors using a 6386 remote-cutoff twin triode to avoid having the signal modulate the gain.  My Hammond Solovox uses a pair of 6SK7 remote-cutoff pentodes as a volume control.

Getting good compression from a sharp-cutoff tube is interesting and it allows smaller control voltages to do the job.  How much voltage are you getting at the upper grid?  Do you find the 6.4 Hz pole frequency for the 250 K and 0.1 µF time constant at the upper grid to be the best for the attack / decay time constant?  So many questions!

I like the circuit - at some point I may try to build it or something similar just to check it out.

[blackcorvo]

Are you getting much distortion with this circuit?

Only if I play aggressively, easily remedied by adding a Hi/Lo set of inputs like most tube amps, or turning down the volume on the instrument a notch.

Getting good compression from a sharp-cutoff tube is interesting and it allows smaller control voltages to do the job.  How much voltage are you getting at the upper grid?  Do you find the 6.4 Hz pole frequency for the 250 K and 0.1 µF time constant at the upper grid to be the best for the attack / decay time constant?  So many questions!

That I could measure with my digital multimeter, it goes down to a little over -4v at peaks IIRC. I'll measure when I'm home from work and get back to you on that.

The rest of what you said is an enigma to me, both because I never managed to learn filter calc when I got my tech degree, but also because I don't math most of the circuits I build, I go by ear + trial-and-error (Yes, sometimes things go poof, don't @ me).

[mozz]

Who wants to make me a pcb layout for the top circuit so i can order some boards?

I'm stuck at home today waiting on tires for my truck so I'll see what I can come up with.  I'm assuming you mean the first schematic correct?

Yup, only thing I see it needs is the tube filaments.  I like projects, or should I say projects like me. Tube sockets don't need a footprint, I can hardwire them. That way I can swap to other tubes if needed.  I have the tubes , they would never get used other wise.  Imay have some suitable power transformers. 

[blackcorvo]

I've updated the first post with a link to a quick sound sample.

[jonny.reckless]

Neat circuit and a great idea! What's the purpose of the 47n cap from the screen grid to the cathode? Normally you would use that to shunt away AC on the screen grid, but don't you actually want signal on the screen grid in this circuit?

[rankot]

Great! I was playing with that original schematic for some time, but since it wasn't easy to find sharp-cutoff pentode, I gave up. I will follow this.

[blackcorvo]

Neat circuit and a great idea! What's the purpose of the 47n cap from the screen grid to the cathode? Normally you would use that to shunt away AC on the screen grid, but don't you actually want signal on the screen grid in this circuit?

Tying it to cathode forms a kind of feedback circuit that increases the gain of the circuit. And I thought exactly the same thing as you, but apparently it still has enough signal left for the effect to work. Haven't tried removing that cap yet...hm. might do it tonight and see what happens.

Great! I was playing with that original schematic for some time, but since it wasn't easy to find sharp-cutoff pentode, I gave up. I will follow this.

Tubes-store has a sharp cutoff section, great way to find tube types.

https://tubes-store.com/index.php?cPath=103_105_114

Now what you gotta keep an eye out for is that it has a separate G3 (supressor grid) pin, as a lot of little pentodes have it internally tied to cathode from the factory. You can check their datasheet to verify that.

[amptramp]

The most common tubes of the dual-control pentode type are the 6AS6/5725 which appears often in military surplus equipment and the 6DT6 which was used as a quadrature FM detector for television sound.  The 5636 was an 8-pin subminiature version of the 6AS6 with slight differences in ratings.   The 6HZ6 appears to be close to the 6AS6 in performance and ratings.  The 6DT6 took over from the 6BN6 as the quadrature FM detector of choice for the television industry.

It would be interesting to see what you could do with a "digital tube" like a 6BN6 - the design kept the cathode current near constant and the action of the control grids diverted it to the screen or plate.

[merlinb]

Neat circuit and a great idea! What's the purpose of the 47n cap from the screen grid to the cathode? Normally you would use that to shunt away AC on the screen grid, but don't you actually want signal on the screen grid in this circuit?

I completely missed that! I assumed the side chain was coming from the anode not the screen grid. I'm kinda surprised it compresses at all with that screen cap in place

[Eb7+9]

I think you'd get way better (more controllable) results if you buffered the discriminator ...
an emitter follower using a high voltage NPN device would likely do the trick here

any way to plot out the control range of the tube using a 9v battery or two feeding the screen grid pot instead of signal+diodes ? ...

[blackcorvo]

I think you'd get way better (more controllable) results if you buffered the discriminator ...
an emitter follower using a high voltage NPN device would likely do the trick here

any way to plot out the control range of the tube using a 9v battery or two feeding the screen grid pot instead of signal+diodes ? ...

What makes it work is a variable negative DC voltage at G3. An AC signal is probably just gonna turn it into a tremolo that varies with how hard you hit the strings. Or an oscillator.

[Eb7+9]

What makes it work is a variable negative DC voltage at G3

yeah, we get that ...
what would be interesting to see is a rough plot of plate AC voltage against this negative voltage for a fixed input signal (ie., 100mVpp at 1khz)

[blackcorvo]

yeah, we get that ...
what would be interesting to see is a rough plot of plate AC voltage against this negative voltage for a fixed input signal (ie., 100mVpp at 1khz)

my scope is currently outta commission so IOU that one for now.