Vari-Mu compressor Sidechain/Detector explained

[StarTrucker]

Please school me on this, as there are mythical posts around here explaining the theory but I can't find the whole explanation in one place. Hopefully this will help other noobs like myself

This is my understanding of the "usual suspect" 6AL5 sidechain/detector circuit, a-la Sta-level, Rs124, Alter 436, UA 175/6 etc.

1. Blocking caps take only AC from the plates of the output tubes

2. the 6AL5 rectifier produces a DC signal that will be applied to the grids of our vari-mu tubes. More signal on the cathodes means more DC output to the grids and therefore more compression. ***Why are voltage readings on a meter actually lower with faster speeds, not enough time to charge back up? ***

3. The threshold network sets the DC operating point of the 6AL5 (220k resistors off cathodes). More DC means we need more signal from our output section to overcome the bias and begin compression. Less DC on the cathodes here means every little sound from the output tubes will be passed through to the time circuit, and we will get a higher ratio (more compression) for the louder parts.

4. Before our signal reaches the grids of the input tubes, this DC signal passes through the time circuit
4. The attack resistor controls the rate at which the capacitor charges
5. The capacitor controls the amount of voltage on the grids of the vari-mu tubes, A larger capacitor will take larger to charge, meaning slower attack AND release
6. The resistor in parallel to the cap determines the rate at which the voltage is drained out of the cap, called release
7. The signal is fed into the grids of our vari-mu tubes, the higher the DC voltage from our sidechain, the more compression (gain reduction) we will get.


What I don't understand is how the AC and DC interact in the sidechain. It's DC coming out of the 6AL5, but this DC swings from a steady 25ish volts down to 6ish in proportion to the output tube signal on the cathodes, making it AC right? Why wouldn't the SC cap dump all the AC swing right away, and the Release resistor dump the DC immediately, resulting in 0V coming out of the side-chain at any given time? With 0 volts coming out of the detector wouldn't it result in no compression?

If shrinking all the time values results in distortion, would adding a second 6al5 in parallel with the first offer more current to be able to refill the cap fast enough to lower distortion?

 

[merlin]

It's DC coming out of the 6AL5

It's rectified, but it's not 'smooth' DC.

but this DC swings, making it AC right?

It's just a rectified version of the audio signal. Mathematically this can be regarded as a DC average plus an AC component.

Why wouldn't the SC cap dump all the AC swing right away, and the Release resistor dump the DC immediately

Because the release resistor is much larger than the attack resistor. Current can flow into the cap fairly quickly, but it can only drain out slowly, so the cap fills up on average. Like turning the water fully on into into a bathtub, while it can only drain out through a much smaller hole. The bathtub will fill up as long as the input rate is greater than the output rate.

It's the same concept as a power supply rectifier and capacitor -see image. In the compressor it's rectifying negative voltages of course, but its the same fundamental process.

If shrinking all the time values results in distortion, would adding a second 6al5 in parallel with the first offer more current to be able to refill the cap fast enough to lower distortion

I assume you're talking about CV feedthrough distortion. Mainly it's the relative difference between attack time and release time that is the problem. The smaller the difference (e.g. you reduce the release time, or increase the attack time), the more 'ripple' is left on the cap, which is what causes the distortion. There are ways to improve this, but they're not as trivial as 'just use a better diode'. You need to understand the basics before wading into those waters!

 

[StarTrucker]

Never considered it as a replicate of a power supply rectifier circuit. It seemed somehow special and mysterious that the "sidechain" should be more complicated where components don't work as they normally do in something fundamental like a PSU.

Thank you for de-mystifying this!

 

[gyraf]

..never underestimate the generalizability of electronics..

 

[Murdock]

Sorry to hijack the thread but I have a question which is somewhat related. I think…
How does the sidechain amplifier output distortion in limiters with separate sc amps (670, U23, U73,…) affect the operation? Let‘s say the sc amp output just before rectification has like 5% THD. As the signal gets rectified anyway is that distortion bad? Does it add to the overall distortion of the limiter?
Or is only gross overload and thus limit in sc output bad for the compression action? And a few % distortion is negligible?

 

[StarTrucker]

Correct me if I'm wrong, but AC clipping in the sidechain amp would be the same as a lower overall voltage out of the rectifier. I.e. a 35V pk-pk signal with a clipped top, at like 32V pk-pk, would look the same to the rectifier as just getting 32v pk-pk and putting that on the grid.

Maybe this behaves like a fixed ratio after a certain point because the sidechain is fixed at a maximum signal level; more signal from the program amp results in the same amount of compression? That's assuming the distortion means the power amp isn't able to react to the full input waveform from the program amp.

Harmonic distortion from junk noise riding on your input wave might get amplified but I don't think it matters much since the sidechain reacts to pk-pk voltages at the rectifier. Maybe makes your attack and release a little non-linear?

 

[Alva Goldbook]

Hi, I think this is my first post here. I’m a tube guitar amp builder and amp tech. Not super familiar with studio circuits, but there’s a lot of over lapping concepts.

So the 6AL5 makes DC. It takes total energy from the out put and scales that down a bit through the 220k resistors as well as the natural voltage drop of the 6AL5 rectifier. This becomes DC that’s not smoothed out. It in fact varies in voltage depending on the output.

This is then sent back to the cathodes of the first preamp. The cathode biases the first gain stage.

I wouldn’t know the voltages off hand, but say the negative bias voltage is -10v, a positive voltage of +5v would effectively drop the bias to -5v, changing the bias, which in turn changes the output.

The interesting thing is that you could in principle do this with preamp pentodes by adjusting the screen voltage, but the response probably wouldn’t be as linear and who know if it would destroy the screens pretty quickly.

 

[merlin]

As the signal gets rectified anyway is that distortion bad? Or is only gross overload and thus limit in sc output bad for the compression action? And a few % distortion is negligible?

Like StarTrucker said, a few % THD is negligible. The side chain turns the AC peaks into DC; minor changes in the AC peaks due to distortion are drowned in the sea of DC.

The interesting thing is that you could in principle do this with preamp pentodes by adjusting the screen voltage, but the response probably wouldn’t be as linear and who know if it would destroy the screens pretty quickly.

Pretty much every possible way of performing AGC has been tried by somebody, somewhere. Screen control, suppressor gid control... They all do basically the same thing with different degrees of loop gain and other tradeoffs.