Transformerless Vari-Mu Compressor

More to study for me : )

Another idea, similar to above, but op amp land: We have differential amps to reject common mode in the output and in the sidechain. How about a summing amp (THAT e.g.) to isolate 'undesired' differential mode signals for injection in antiphase into sidechain CV ? Could that work to further reduce 'over-compression' ?

Script said:

How about a summing amp (THAT e.g.) to isolate 'undesired' differential mode signals for injection in antiphase into sidechain CV ?

Click to expand...

In order to have only thump left at 1240 output: Ref pin is connected to the output from the tubes and In+ to somewhere where the signal is in opposite phase, exactly same level but without thump. Where is In+ connected to?

I like to limit the maximum amount of CV that the sidechain can make. (D1 in attachment). With 12AU7 I limited the maximum amount of gain reduction to 12 dB. 12AU7 was way too distorted with more than 12 dB GR and I didn't want the possibility of thump whacking the tube to complete cut-off. Remote cut-off tubes handle big thumping voltages better but I still like to limit the maximum GR to some point. With 6SK7 tubes I can get 30 dB of gain reduction but distortion gets very unpleasant. I'm probably going to limit the max GR to 20 dB.

Easiest way to get rid of overcompression in backward acting compressor is to put a steep high pass filter in the sidechain. If your fastest attack time can be slow as 10ms, just put a steep 20Hz high pass before the sidechain.


Attachment removed



When trying to improve tube compressors there is probably some point where one should realize what's the point. You could just get a THAT VCA, wire it up as soft knee compressor, add some circuit to make a little bit of 2nd and 3rd harmonic distortion and nobody would know if you got tubes there or not.

Heikki said:

... add some circuit to make a little bit of 2nd and 3rd harmonic distortion and nobody would know if you got tubes there or not.

Click to expand...

I don't believe that. Read Designing Value Preamps by Blencowe (aka "merlin" here) about the varied ways tubes clip. Designing a semiconductor-only based circuit to make a little bit of 2nd and 3rd harmonic distortion over a wide range of signal amplitudes and with more or less at different frequencies (LF distortion can easily sound bad) is not at all trivial. I burned a lot of hours trying to make a semiconductor-only based circuit distort like a tube and failed. Even the digital recreations don't sound right IMO. Tubes are inherently more linear than transistors and have much more graceful clipping / starvation characteristics. These characteristics are probably what makes them good for dynamics devices but I say "probably" because I have never made a tube compressor.

squarewave said:

I don't believe that. Read Designing Value Preamps by Blencowe (aka "merlin" here) about the varied ways tubes clip. Designing a semiconductor-only based circuit to make a little bit of 2nd and 3rd harmonic distortion over a wide range of signal amplitudes and with more or less at different frequencies (LF distortion can easily sound bad) is not at all trivial. I burned a lot of hours trying to make a semiconductor-only based circuit distort like a tube and failed. Even the digital recreations don't sound right IMO. Tubes are inherently more linear than transistors and have much more graceful clipping / starvation characteristics. These characteristics are probably what makes them good for dynamics devices but I say "probably" because I have never made a tube compressor.

Click to expand...

There has been a lot of work invested in solid state mimics of tube overload, involving multiple patents.

The approach developed by Peavey marketed as "transtube" was imo pretty successful (but the transtube technology was used in guitar amps, not efx units).

JR

I'm not sure if graceful clipping is why some people like old tube compressors. Lets take a look at RCA BA-6A for example https://groupdiy.com/index.php?topic=44740.msg580840#msg580840. Grid current in the gain reduction stage would cause C1 to charge and clipping would be similar to RC coupled stage. I'm pretty sure the next amplifier stage will clip before the gain reduction stage. I'm not sure if transformer coupled amplifier stage with decent amount of negative feedback will clip any more pleasantly than solid state amplifier.  Assuming of course that the solid state amplifier behaves well when clipping. Since the threshold of limiting is fixed in BA-6A and the attack is fairly fast, I wouldn't expect audible clipping to happen often or ever.

There are of course designs like Sta Level https://groupdiy.com/index.php?topic=44868.msg561651#msg561651 which I would expect to clip much more an more audibly. But I think Sta Level won't clip very gracefully either.

Now I’m trying out a power supply that is powered by a 28VDC wall wart. I’m using DC-DC converters to make 200V for the tubes and +15 and -15V for the INA chips. The power supply is more expensive compared to big lump of iron but makes for very safe DIY.

I was wondering if you could improve the 'transformerless' designof a vari-mu by keeping the grids grounded and modulating the cathode voltage instead?

You could do this by placing a capacitor between the cathodes of the tube and ground and connecting some kind of low impedance buffer on the cathodes.

So instead of reducing bias voltage on the grids, you would increase the cathode voltage.

the supply voltage for the anodes might need to be a bit higher.

worth a try?

Heikki said:

Now I’m trying out a power supply that is powered by a 28VDC wall wart. I’m using DC-DC converters to make 200V for the tubes and +15 and -15V for the INA chips. The power supply is more expensive compared to big lump of iron but makes for very safe DIY.

Click to expand...

Mind sharing that PSU schematic? Sounds interesting.

MaxDM said:

I was wondering if you could improve the 'transformerless' designof a vari-mu by keeping the grids grounded and modulating the cathode voltage instead?

You could do this by placing a capacitor between the cathodes of the tube and ground and connecting some kind of low impedance buffer on the cathodes.

So instead of reducing bias voltage on the grids, you would increase the cathode voltage.

the supply voltage for the anodes might need to be a bit higher.

worth a try?

Click to expand...

Haven't you  looked at the original schemo, in post #1?
That is exactly what happens there.
That's why there is a power transistor at the end of teh side-chain.

Check also posts #23 to 28. Jury is still open.

abbey road d enfer said:

Haven't you  looked at the original schemo, in post #1?
That is exactly what happens there.
That's why there is a power transistor at the end of teh side-chain.

Check also posts #23 to 28. Jury is still open.

Click to expand...

oh wow...

hah, I reinvented the wheel!

Will check the rest of the posts.

dogears said:

Mind sharing that PSU schematic? Sounds interesting.

Click to expand...

There's not much to the schematic. It's just DC/DC converters preceded by a linear regulator to make sure the maximum input voltage to the DC/DC converters is not exceeded. Also heater/sidechain supply has it's own linear regulator to make 25V for the heaters

DC/DC converters
https://recom-power.com/pdf/Innoline/Rxx-B.pdf
https://www.tracopower.com/sites/default/files/products/datasheets/tdl3_datasheet.pdf

Wall wart
https://www.meanwell-web.com/en-gb/ac-dc-industrial-wall-mount-adaptor-output-28vdc-gst25e28--p1j

Thanks, I'd never come across those RECOM units before. They look nice, but a bit pricey!

Here is the output noise floor with gain at max and nothing connected to the input. (0dBFS = +20dBu)

Heikki said:

Here is the output noise floor with gain at max and nothing connected to the input. (0dBFS = +20dBu)

Click to expand...

That's a quite impressive performance. I notice the absence of mains frequency residues.

abbey road d enfer said:

I notice the absence of mains frequency residues.

Click to expand...

That's the advantage using DC/DC converter power supply.

For comparison here's the noise when I used SMPS for heaters and linear power supply for the rest.

Heikki said:

For comparison here's the noise when I used SMPS for heaters and linear power supply for the rest.

Click to expand...

There seems to be a significant difference in HF noise too...
The different scales do not allow precise comparison, though.

abbey road d enfer said:

The different scales do not allow precise comparison, though.

Click to expand...

The latter one also has couple dB more gain because I used U-pad for volume control that always attenuates 4dB for the DC/DC supply compressor. For precise comparison I would need to do the measurements again. With these at least the difference in mains frequency residues can be seen.

Here is a better noise comparison. Now the gain is set at 19dB on both compressors. On top is the compressor with AC/DC power supply and at bottom the DC/DC power supply. This time there's a tiny spike at 50Hz even with the DC/DC supply.

Heikki said:

Here is a better noise comparison. Now the gain is set at 19dB on both compressors. On top is the compressor with AC/DC power supply and at bottom the DC/DC power supply. This time there's a tiny spike at 50Hz even with the DC/DC supply.

Click to expand...

That's interesting. Actually, the mains residues are much lower but the LF noise is higher, which results in a slightly higher wide-band figure (+0.2dB).

And here is channel 2 from both compressors.