Transformerless Vari-Mu Compressor

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

Heikki

Well-known member
Joined
Jul 31, 2008
Messages
401
Location
Finland
Here's what I'm thinking about building next. There is one cheap audio transformer in the sidechain but it would not be necessary to use one there. Control voltage is brought to tube cathodes so no need for input transformer. The stage after the tubes should be able to handle approximately 25V of common mode voltage and using 0.1% resistors should guarantee good common-mode rejection. If each triode has quiescent current of 5 mA, plate resistors up to 4.7k can be used and 25V common-mode voltage won't be exceeded when going to deep gain reduction. Since the audio signal will be much smaller than the common-mode voltage, the combination of differential and common-mode voltage won't be much larger than the common-mode voltage alone.

This is probably the cheapest way to build high quality tube compressor since there are no expensive audio transformers needed.

Edit:

Attachment removed

Edit: Here is the original schematic https://tube.ghr.fi/varimu-t.pdf
Schematics to what the thing evolved to can be found here Variable Mutual Contuctance Compressor
 
Last edited:
Try it and let us know?

I wouldn't get my hopes up too high on how it sounds though, having tried many many variations over this theme. But who knows - maybe this is the ticket there?

If you build it, I'd suggest that you also try the ECC86 tube (at properly reduced plate voltages) - that is the most interesting direction Ive seen/heard so far for this sort of idea [Behaves tubish within reach of 5534 inputs].

Make sure to report back!

/Jakob E.

 
That looks very interesting! how high is B+? How is the power supply implemented? A special transformer, or two separate units? Voltage multiplier or charge pump?

ECC86 seems to be a good idea, if the gain is high enough for this application?

New design ideas are always welcome! 👍
 
Heikki said:
Here's what I'm thinking about building next.  There is one cheap audio transformer in the sidechain but it would not be necessary to use one there.  Control voltage is brought to tube cathodes so no need for input transformer.  The stage after the tubes should be able to handle approximately 25V of common mode voltage and using 0.1% resistors should guarantee good common-mode rejection. If each triode has quiescent current of 5 mA, plate resistors up to 4.7k can be used and 25V common-mode voltage won't be exceeded when going to deep gain reduction.  Since the audio signal will be much smaller than the common-mode voltage, the combination of differential and common-mode voltage won't be much larger than the common-mode voltage alone.

This is probably the cheapest way to build high quality tube compressor since there are no expensive audio transformers needed.
Since you already have a lot of solid-state things, I suggest you use a synthesized common-mode inductor to reduce thumps. See attachment.
I'm dubious about 1k as plate resistors; indeed it reduces thumps, but it also reduces the level of signal.
 

Attachments

  • synthesized CM inductor.jpg
    synthesized CM inductor.jpg
    35.5 KB · Views: 191
abbey road d enfer said:
Since you already have a lot of solid-state things, I suggest you use a synthesized common-mode inductor to reduce thumps. See attachment.

I can't see how those two gyrators would emulate a center tapped choke, I probably fail to understand the circuit. I need to simulate it  feeding it some common mode voltage and see if it cancels any of it out.
 
Heikki said:
I can't see how those two gyrators would emulate a center tapped choke, I probably fail to understand the circuit. I need to simulate it  feeding it some common mode voltage and see if it cancels any of it out.
Please do it. You will see that the input impedance for common-mode signals is about 1 kohm with the actual values and the impedance to diff-mode signals is very high, several Megohms.
If you feed it it with a perfect source (zero-ohm impedance), you won't see a change, but you must consider the source impedance of the tube/plate resistor. In your case it's dominated by teh plater resistor.
Since you're considering using only 1kohm for the plate resistors, some of the proposed circuit's values must be scaled down considerably. By making R1=10ohms, you get 30dB attenuation of all thumps below 10 Hz.
 
Spice must be lying to someone. I don't understand how that Bob Pease floating synthetic inductor (https://www.youtube.com/watch?v=AEJtajaRj_s&feature=youtu.be&t=285) could ever work like a common-mode choke and even spice seems to agree with me. It's just a floating synthetic inductor.
 
Heikki said:
Spice must be lying to someone. I don't understand how that Bob Pease floating synthetic inductor (https://www.youtube.com/watch?v=AEJtajaRj_s&feature=youtu.be&t=285) could ever work like a common-mode choke and even spice seems to agree with me. It's just a floating synthetic inductor.
Agreed. Now, what do you think about this?
 

Attachments

  • common mode choke.jpg
    common mode choke.jpg
    75.5 KB · Views: 136
Hasn't Ian 'Bluebird' been tinkering with this idea a long time ?
Might be worth a look back to previous posts for inspiration .
 
abbey road d enfer said:
Agreed. Now, what do you think about this?

That looks like something that should work. It should greatly reduce common-mode voltages but I am worried how it will handle high audio frequency differential signals.
 
gyraf said:
Dosen't it need some time-constant capacitors to get chokis'h?

No because it's not a gyrator circuit.  Common-mode voltages appear in opposite phase at the other ends of R5 which will cause them to cancel and same happens at R10. For differential signals no current flows through R5 or R10.  But I think that at high frequencies opamps will cause some phase shift which will cause current to flow through R5 and R10. Because the resistor are only 10 ohm even small phase shift will cause significant current to flow. We could increase the resistors but that would make the circuit less effective canceling common-mode signals.
 
aah, it's a common-mode short, not a choke.. :)

I'd like to try this, have a hard time wrapping my head around how it would behave in real life

/Jakob E.
 
Heikki said:
No because it's not a gyrator circuit.  Common-mode voltages appear in opposite phase at the other ends of R5 which will cause them to cancel and same happens at R10. For differential signals no current flows through R5 or R10.  But I think that at high frequencies opamps will cause some phase shift which will cause current to flow through R5 and R10. Because the resistor are only 10 ohm even small phase shift will cause significant current to flow. We could increase the resistors but that would make the circuit less effective canceling common-mode signals.
Simulation shows that even with pedestrian opamps (GBW=1MHz), the -3dB HF point would be about 50kHz.
I know that simulation is not real life, but I thinks it justifies breadboarding.
 
abbey road d enfer said:
I know that simulation is not real life, but I thinks it justifies breadboarding.

I'm going to try it out when I get to building my compressor.
 
Yesterday I got the compressor to the point where I could do some measurements and listening tests and it works very well. I used 5.5k plate resistors and 4.5mA plate current for each plate so the maximum common-mode voltage is 25V. As long as the common-mode voltages are kept smaller than the maximum the op amp stage can handle, there's no need to use a transformer. When I got the time I'm going to do some tests with the abbey road d enfer common-mode strangler circuit and see how well it works.

But for somebody who want's a cheap tube compressor this is a good way to go. No need for expensive transformers and arguably much better quality because you don't have to deal with the flaws of transformers. I used 12AU7 but any tube appropriate tube for compressors can be used. Bunch of the resistors, U1 and U4 could be replaced with THAT 1280 series Line Receiver IC's and U5 with THAT 1606.  Big advantage with THAT chips would be the laser trimmed precision resistors.

I'll post some measurement data and sound samples when I finish up the compressor.
 

Attachments

  • compt.jpg
    compt.jpg
    238.1 KB · Views: 173
No need for expensive transformers and arguably much better quality because you don't have to deal with the flaws of transformers.

..but remember that the reason we like the vari-mu compressors in the first place might very well have to do with exactly the parts you so avoid..
 
Back
Top