Transformerless Vari Mu Compressor build thread

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

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

Something that I'm a little confused on is how the standard version deals with gain. I see that RV3 comes before the tubes, and obviously we've got the threshold control in the sidechain, but then I don't think there's any user-adjustable output control (makeup gain, attenuation, whatever) after compression, right -- at least when built as stock? So I suppose the operator relies primarily on those two controls to achieve both the right amount of compression as well as the appropriate output level (obviously the other sidechain controls each impact this as well). I know this has been discussed/explained, and I just need to go reread some parts of the thread again to bring my understanding of it up to speed.
 
I suppose the operator relies primarily on those two controls to achieve both the right amount of compression as well as the appropriate output level
Yes.
(1) Dial in compression action
(2)i Increase input gain to almost unity (which leads to deeper [soft] compression)
(3) counter the now too deep compression by backing off on the threshold

Works remarkably well with Input Gain (and / or Threshold) as just a pot for fine-tuning.
 
If you need some (minor) unity gain calibration -- and instead of putting trimmers on the outputs (which also requires additional changes) , and also instead of offsetting the CV voltages for this -- you could install four 100k trimmers into the 'empty' parallel resistor slots as in the picture (looks a bit ugly in my case but works).

This allows matching the gain of the four individual tubes before the signals go thru the INA stages.

I matched (up to 0.4dB of difference between tubes before) with mono test signal into L and R channels and no compression dialled in. The trade-off is minor.

Mind you though, once compression is happening, it will all be off again due to non-perfect tube behaviour -- but hopefully better than without.
 

Attachments

  • IMG_20221027_115847.jpg
    IMG_20221027_115847.jpg
    1.6 MB
Wrapping my head around the wiring here (which is drawn up very nicely on the webpage, thank you Heikki). Am I correct that the outputs of the isolated DC-DC converters are at no point connected to "GND" (which I take to be chassis ground)? Those rails, as a rule, can return only to the Common or V- terminals of the converter?

Additionally, do V+ and V- refer to the +/-15VDC rails? I was a little unclear about those designators.
 
Last edited:
The markings on the boards are confusing. I thought I had fixed them on the newest versions on the boards, but I might have not.

V+ and V0 on the sidechain boards should be as marked A+ and A- whis is the 25V (U1 output from PS board).

The DCWN03B-15 common connects to the main board GND directly and R24-150B -Vout trough a resistor. Gustav Robert told me current always returns to its source. In the layout picture connection to chassis is made from the LED meter boards from a pad marked GND. It won't make big difference where you make the chassis connection, but if you follow the layout pictures everything should work fine.
 
Ah, I see. I have never worked with an isolated DC-DC converter like the DCWN03B-15 before, and I was confused because I thought its Common was not connecting to chassis -- my own error, I was not following its path all the way through to the LED board. So in general, if that Common was connected to chassis directly from the PSU instead with all return paths converging at the PSU, similar to what I often see when using simple bipolar LM317/337 PSUs to power boards, there would be no significant difference as long as no ground loops are created? (This question is not so much for this specific circuit, where I will just be using your wiring layout as-is, but rather in case I find myself using a similar component in the future).
 
Last edited:
Transistors have maximum emitter - base voltage usually around 5V. Connecting the sidechain board outputs together might lead to a situation where the emitter - base max voltage is exceeded.
What situation would that be ? Tube failure ?

Initial idea was to link for stereo after the BD139 SC output transistors (not a good idea for above reason). Nonetheless, I have been running/testing this cautiously for some time, always wary of course that I might end up blowing things up -- but I also have to say that generally I like dual mono more.
 
What situation would that be ? Tube failure ?

Initial idea was to link for stereo after the BD139 SC output transistors (not a good idea for above reason). Nonetheless, I have been running/testing this cautiously for some time, always wary of course that I might end up blowing things up -- but I also have to say that generally I like dual mono more.
It can happen just in regular use if the BD139 emitters are connected together.
 
I think I found it: thermal runaway.

When two transistors, tied directly at their emitters, 'dissipate heat at different rates due to slight differences in hfe characteristics'. The one conducting slightly more current might heat up faster until overheating in a supernova.

[If tied together, only after a small-value emitter-current-limiting resistor (i.e., after a resistor summing junction, so here tied directly at the tubes) and transistors mounted on a common heatsink -- might this be a solution ?]
 
Last edited:
If you connect the Sidechain 1 & 2 BD139 emitters together it might lead to a situation where the emitter - base junction starts acting like a zener diode. For example if sidechain 1 is doing nothing and sidechain 2 is putting out control voltage. This might damage the sidechain 1 transistor.
 
Oh yes, I remember now. Sometimes some things just slip my mind.

Diode junction then? I'm guessing...
 
I think he asked for his account to be disabled - he'll just have to contact admin to get it re-opened
Well I know he told me he was having trouble logging in and asked me to post here to let others know how to contact him so there's that.
Admin should be contacting members like him asking what they can do to keep him imo. Not being allowed to log in while still having member status and all content still here in isn't a true representation of this place's roster and certainly not one of the better benefits of membership. Regardless of whether or not he decides to "contact admin".
There will always be plenty of people like me signing up asking and answering the most basic things but losing another great contributor just feels like some people don't understand a big part of what makes this place so much more than a large number.
 
ANOTHER STEREO LINK OPTION

This is from a private mail exchange yesterday with Heikki about linking for stereo. He agreed I post it here.

Script
(14 Feb 2023, 10:10 pm)
In order to protect the SC out transistors from damage when simply hooking them together for stereo, would adding a diode in series after the transistor on each channel and then linking for stereo after those two diodes do the trick of protecting the transistors from possible damage should there be CV (can be high up to 12VDC, I think) only coming from one channel? SC control voltages would need to be upped a tiny bit to overcome the diode step.

If not this, I'll rewire SC for input from L&R but only one set of controls.

Heikki
(14 Feb 2023, 10:44 pm)
Adding a diode will work but you also need to add a resistor from the BD139 emitter to ground (sidechain board V0). The resistor value doesn't matter that much, 10k for example will work.
[schematics see attachment]

Script
(14 Feb 2023, 10:51 pm)

10k there to keep the BD139 happy ?

Replacing 1R. Good idea. I don't like dangling components.

Heikki
(14 Feb 2023, 11:05 pm)

I think the resistor needs to be there because the diode alone will basically leave the emitter floating when the other sidechain has higher output voltage.

Script
(15 Feb 2023, 1:26 pm)

I installed two UF4007 and the two 10k (used 2W ones cos of thicker lead)... and it works :) Will test a bit more, but confident... so basically unit is now queuing up for (re-)calibration.

Thank you very much (yet) again.
 

Attachments

  • protection.png
    protection.png
    26 KB
Last edited:
Wired it up as follows:

1st link at the SC input, directly bridging the two (L&R) rectifier diode junctions.

2nd link of L&R at the SC outputs, directly after the added diodes (replacing the 1R), forming a junction when linked, with higher of two signals driving both channels.

So the two sidechains are linked with two wires. They read the same summed L and R DC at the input, and they both send CV (higher of the two) to control both channels equally.

Basically it's two identical sidechains. This allows to use just one sidechain (by jjust setting the other THR to 'off'), or use both sidechains but with different settings. Examples:

(1)
L-SC is set to slow levelling, R-SC is set to fast limiting. The higher of the two signals drives the tubes. (I had added super fast attack.)

(2)
L-SC and R-SC have different settings for riding 'chorus' and 'verse'. Changing over is done with the THR knobs (as the higher SC signal drives both tubes) -- or could integrate SC soft bypass switches to flip between the two sidechains.
____

Anyway, having said and implemented it like that, I often stil prefer dual mono nonetheless (comp knee is soft after all) for that 'oceanic feel'.

Also, stereo linking like that can make differences in behavior between unmatched tube more noticeable. (I had added 100k trims for matching, see above).
 
Last edited:

Latest posts

Back
Top