Compressor/Limiter in a bass amp

[stw307]

How to implement a tube limiter between stages in my bass amp?

I've used the forum's search, and all related topics were at least 10 yrs old with dead links.
Web search "tube limiter" and "vari-mu" gave many a good schemas with and without circuit analysis. Neglecting the optos, two main groups of balanced signal limiters seem to be PI-connected limiters inside an amp and the legendary Altec 436 and its kin. Unbalanced signal limiters are either double-triode limiters or pentode grid tweakers (attachment). Any other tried-and-true designs are welcome, I'm sure I missed some. Any experiences of limiters inside an amp would be appreciated. For example Gibson GA100 looks interesting, but how does it work in real life...?

I'd like to add a tube limiter to a 5F6A-related bass amplifier. It does not need to be aggressive, because an all tube signal chain compresses some already. But it is just rounding off corners, I'd like to have some real compressing down the peaks. Transformer has plenty of juice to drive some additional preamp tubes.

I have some 6AL5 diodes at hand and some pentodes. Dual triodes are plenty. If the goal is to tame the signal tops and get those diodes into use, which way to go? I'd like to compress/limit between 1st stage and tone stack, hence the attachment. It is not mandatory, though. Using interstage trafos = expensive = no-no.

I'm not an engineer, so... the way I see the attached picture: looks like it gives some feedback to suppressor grid when hit hard. Threshold of the diodes would let quieter signals through. Pot to zero would connect the suppressor to ground = zero effect. Is this right? Does this work at all? Or should I ditch this one?

[SEL49]

The circuit works like this...

[stw307]

The circuit works like this...

Thanks. But the sidechain 
Like I said, I'm not an engineer, but I'll try to figure this out. My analysis:
Input pot defines how much sampling 6C4 will do. Then that signal is amplified and sent to diodes. Left side of 6AL5 lets the positive lobes of signal to ground. Right side of 6AL5 lets the negative lobes of signal to C .22uF right after diodes. Cap rectifies the negative signal. Level pot mixes the rectified signal and ground. Now to the actual magic. The job of the suppressor grid is to keep electrons from bouncing off the plate, normally it is tied to ground, right? So, to get the pentode off of "beam tetrode mode", the suppressors voltage shifts when sidechain hits it with DC. This DC would be rippled, negative spikes parallel to original signals positive lobes because of inverting in the 6C4.

Is this correct or even close?

What I really can't see, is the magic happening inside the pentode. It looks like the incoming signal to suppressor is negative compared to ground. Does this block the electrons coming towards plate? Screen dissipation would not rise to melting point because it is not tied to plate but a voltage divider. If so, quieter signals would not affect the amplification factor, while louder signals would tame the pentode?

Many questions, many guesses... thank you all for educating hobbyists like me.

[Merlin]

Making the suppressor grid more negative diverts current away from the anode and towards the screen grid, so gain to the anode is reduced. In your circuit, the bigger the audio input signal, the more negative voltage is stored across the .22uF cap by the rectifier, which is applied to the suppressor grid, so the lower the gain of the pentode becomes.In this image you can see what happens to the audio signal at the anode when the suppressor grid swings about 70V positive or negative (this was in a trem circuit using an EF86 but the principle is the same as your compressor). Gain is maximum when the suppressor grid is about 0V (positive voltage doesn't increase the gain much further), whereas negative voltage reduces the gain.

[stw307]

Making the suppressor grid more negative diverts current away from the anode and towards the screen grid, so gain to the anode is reduced. In your circuit, the bigger the audio input signal, the more negative voltage is stored across the .22uF cap by the rectifier, which is applied to the suppressor grid, so the lower the gain of the pentode becomes.In this image you can see what happens to the audio signal at the anode when the suppressor grid swings about 70V positive or negative (this was in a trem circuit using an EF86 but the principle is the same as your compressor). Gain is maximum when the suppressor grid is about 0V (positive voltage doesn't increase the gain much further), whereas negative voltage reduces the gain.

Now this all makes sense, thank you. I might go and try to build this. No point of making circuits I don't understand.
Even the pros say, it is real hard to build a compressor that sounds good. Let's see what comes out of my garage... Blue smoke?

Anyone got any opinions on the PI-related compressors?

[Merlin]

Any opinions on the PI-related compressors?

They're not PIs, they're balanced. Balanced circuits cancel out a lot of the distortion and CV feedthrough that you unavoidably get in an unbalanced compressor (like your circuit). Unbalanced compression was traditionally called Automatic Gain/Volume Control, AGC or AVC.

[stw307]

Any opinions on the PI-related compressors?

They're not PIs, they're balanced. Balanced circuits cancel out a lot of the distortion and CV feedthrough that you unavoidably get in an unbalanced compressor (like your circuit). Unbalanced compression was traditionally called Automatic Gain/Volume Control, AGC or AVC.

Yes, like Altec 436 and alike. There is plenty of info of those and many good explanations how they work. I see the difference and understand the canceling principle. I'd like to build one standalone unit some day.
Searches with AGC and AVG gave me plenty of new educational material, thanks.

https://el34world.com/charts/Schematics/files/Gibson/Gibson_ga100.pdf

This Gibson is what I meant by saying PI-related. Should've attached a schematic in the first place...
Looks like a cathodyne PI connected to balanced compressor and driver pentodes. "PI-connected" would've been a more accurate choice of words...
(And the picture in the first post is not my circuit, just something that came up while searching info on this subject.)

[Lauri]

That Gibson GA100 schematic has a mistake. Here's the corrected version.

[Merlin]

Very interesting schematic, thanks for posting!