Vari-mu compressor question - Side chain

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Looking for some clarity in quite a specific topic. Im sure there is someone here who has delved into vari-mu compressors....?

So as I understand it, a vari-mu compressor works very basically as follows...

The signal of the final stage is decoupled through a cap and then rectified to a DC voltage. This is then sent to the input valve's grid, where it will make the grid more or less negative to change the mu of the valve, and therefore compressing the signal.

And there will be circuitry for threshold, attack etc...

My question regards a side chain bass cut.

I was playing around with a vari-mu compressor the other day and it had a bass cut switch. Where you can change the decoupling cap value of the side chain, which acts as a high pass filter. This filter cuts bass frequencies from passing to the rectifier, so audibly you dont compress these bass frequencies and compress the higher frequencies.

This was quite a sublte effect, but noticeable. I was told it was to make the mix less muddy etc etc...

So, I can not understand why changing the value of this cap would give this effect. If the signal is being rectified to DC, and sent to the grid of the input valve how can it know what frequencies to compress?

DC isnt frequency relative right?

Hope someone can clear this up.

Cheers

Charlie
 
If you have a lot of bass energy the rectification and subsequent DC level will be greater so you would get greater compression. Filtering it out would cause a lower DC voltage and therefor less compression. The compressor is not frequency specific, it just adds less compression which may make it sound less muddy to some.
 
Actually no, this doesnt clear up my confusion.

This suggests that the high pass filter on the signal when recitifed only produces a different DC voltage, so the input valve compresses less.

So all frequencies are still being compressed, just at a different rate. So it is effectively just another type of threshold control.

If you were compressing a track with a wide frequencie spectrum, this side chain bass cut would specifically allow bass frequencies to not be compressed would it?
 
Appalling English sorry, my question is..

Is this 'side chain bass cut' actually allowing bass frequencies (set by the coupling cap)to not be compressed? And if so, how is it doing it?

How does a change in grid voltage at the input valve compress certain frequencies?

Cheers
 
Sorry I may have not been clear in my question.

By side chain I mean the circuitry that decouples, and rectifies the signal, to then give a DC voltage at the input valve. Different from a side chain input where an external signal can be used to trigger compression.

Cheers
 
"How does a change in grid voltage at the input valve compress certain frequencies?"

It doesn't. Can't.

The gain element, in this case a remote cutoff vacuum tube, is not frequency selective. It compresses all frequencies equally, depending on the DC or impulse signal from the sidechain.

Filtering the sidechain determines which frequencies the compressor will act on if present in the signal, but the entire signal will be compressed as a result.

Think about a de-esser which compresses the signal when there is excess energy in the sibilance band from 5 KHz to 7 KHz. It compresses all frequencies equally but since the "ssss" is all that's likely to be present in a vocal signal at that time, it works well to remove only the "ssss".

The bass filter prevents the compressor from responding to bass energy, which I believe is more useful in the context of a mix bus compressor, when one wants to avoid the "ducking" effect where the whole mix is reduced in volume when loud bass or bass drum notes are played.

I hope this is clearer.

Cheers,

Michael
 
To save starting a new thread I thought I would post my question here.

I have built a vari-mu compressor, very similar to the Altec 436C. http://www.dvq.com/hifi/images/436c.pdf

Everything is working well, however I am having a stability issue when using it at quick release settings. The circuit goes unstable at low frequenices with quick release settings - release being the discharge time of C4 I believe. Instability is cured if I use a bigger cap here, but then I have a very slow release so that is not the solution I can use.

I was hoping someone could give any suggestions of has any experience in improving stability. I have read about curing instability when using gloal feedback and 3 LF time constants, and have followed those guidelines but this is something a bit different... as usual a little bit stumped.

Thanks

Charlie
 
There must be an overlap, at fast attack times it is very unstable. So I should address that before moving on.

It does perform very well apart from that, but it would be a much better comp if it was a bit quicker. I assume there is a way to do this.... But then a vari-mu will never be as responsive as a optical I guess.
 
It is not uncommon for the combination of fast attack/release times and low bass notes to cause instability. The reason is simply that the with fast attack/release times, the gain control voltage will actually follow the waveform of low bass notes and causes them to be flattened. It is equally common to include a high pass filter in the side chain to prevent this happening.

Cheers

Ian
 
Good to hear from you Ian.

I guess it depends on how much the instability is a problem, and if actually it is just part and parcel of the vari-mu design and sound. I will aim for a good balance between stability and fast attack/release - I thought it was worth asking as there might be another approach to get quick settings and stable bass.

I have a high pass filter at 50,75,120,250,300 which I found to be the most useful frequencies and the unit has no stability problems with that in. I am also finding that the release is a little bit sluggish in mid settings, so I imagine some experimentation with the discharge time of that cap will give better results.

Thanks, learning lots.

Charlie
 
You can get the best response by ensuring that the loop design is excellent, with a dominant pole and other poles much higher in frequency. The temptation is always to add extra smoothing so there is no audio on the control voltage. This adds phase shift and so destabilises the loop.

You could try reading up on 'hang AGC', as used in some communications receivers. Receiver AGC and audio compressors are doing almost the same thing, so techniques used in one can sometimes be applied to the other.
 
P1 is gain
P2 release
P3 is threshold
R12 is attack, but I have a different arrangement here.
M1 is gain reduction shown as current through input valve.

Thanks DF96, I will look into hang AGC. Better get the calculator out and make sure the poles are where they should be.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.