Transformerless Vari Mu Compressor build thread

[Heikki]

When the unit is powered and you flip the bypass switch is the voltage exactly same no matter which end of R1 and GND you measure the voltage. Or is there any voltage drop across R1 or R2 on the main board.

 

[living sounds]

When the unit is powered and you flip the bypass switch is the voltage exactly same no matter which end of R1 and GND you measure the voltage. Or is there any voltage drop across R1 or R2 on the main board.

No voltage drop accross either resistor, with or without bypass engaged.

I tried applying +36V from the switching supply (with bypass engaged) and the relays didn't switch either. I tried it with another relay out of the circuit and it did switch...

Maybe unsoldering the diode will do the trick? I hesitate to take everything apart to unsolder the relay at the moment, it would be complicated the way my compressor is built...

 

[living sounds]

It isn't the diode...

 

[living sounds]

Those relays didn't work because I am an idiot. I finally took everything apart and found that there were 4 unsoldered connections on the main board... the relay power pins...

Maybe the moderators can delete everything from post 226 onward, so as not to pollute this thread with unnecessary stuff.

Sorry guys.

 

[scott2000]

Nah... The information is good.
I think there must have been a hundred messages with Heikki when I was getting mine together. The saved information from the correspondences leaks out when relevant hopefully. Not efficient...

 

[living sounds]

Got it all together and calibrated. This is one seriously awesome compressor! It's got all the power and heft of a vari-mu, but without the trafo smear. It sounds great on just about everything. I want 10 more...

Will post pictures of my build tomorrow.

Thank you, Heikki! You know your stuff. I will have a lot of fun with this thing.

 

[living sounds]

Here are a few images.

I built mine with an internal (Meanwell caged 36V) PSU. To avoid interference I added extra shielding and placed it as far away from the main PCB as possible. That made assembly quite a bit more tedious. But it worked out great, SNR is better than advertised and clean white noise only. The chassis connection is made at the PSU and not through the meter board, but I didn't notice any detrimental effects.

The tubes are very microphonic, not sure if or how much that varies. I bought a bigger batch but so far only used four. Those were easily matched via the testing instructions almost down to the noise floor.

Sourcing was a quite tedious thanks to the current supply shortages. I will have to shorten the axis of a couple pots and get different knobs. I matched everything meticulously and can now get the channels to perform very, very closely.

The frontpanel layouts and the insert plate (all supplied by Frank from frontpanels.de) made mechanical assembly a breeze. I had to mount the sidechain boards at an angle tilted up so as to avoid them touching the two 47uf capacitors.

If anyone wanted to build a vari-mu compressor, this would be my recommendation. I have built others, but this is by far the easiest, best documented, best sounding and also the least expensive. And board are so well-spaced that even beginners won't have any problems creating solder connections accidentally.

If I could change one thing, I would prefer to have the gain after the compression as make-up gain.

PS: This is a desert island kind of unit. It doesn't colour the source sound and works great on everything.

 

[scott2000]

The tubes are very microphonic, not sure if or how much that varies

For some reason, the metal ones I tried had the same issue. A couple of glass were microphonic but usually they were just out of whack. But it was much easier to find a nice set in my case with the glass ones.They all ended up being Sylvania. Maybe Heikki will get around to releasing some of his metal stash in the future.

You would love the mastering version.

 

[living sounds]

For some reason, the metal ones I tried had the same issue. A couple of glass were microphonic but usually they were just out of whack. But it was much easier to find a nice set in my case with the glass ones.They all ended up being Sylvania. Maybe Heikki will get around to releasing some of his metal stash in the future.

You would love the mastering version.

I have switched one highly microphonic tube per channel against spares, now it is much better. Are the Sylvania glas ones not microphonic at all?

Maybe I'll build a mastering version one day, too.

 

[scott2000]

Are the Sylvania glas ones not microphonic at all

I had a couple of them that were. Actually was specific in asking the seller about microphonics for this specific set after struggling with other ones ...
But in all fairness, the gains were blasted up to where they would never be so, who knows if they really were where it would matter. But the other ones that ended up in the units didn't have the same microphonics.

 

[Heikki]

If I could change one thing, I would prefer to have the gain after the compression as make-up gain.

There's a reason to have the gain before the compression stage. This way the signal level stays low enough at the tubes to keep distortion from getting too high. If the gain was after the gain reduction stage then users would be feeding too high signal levels to the tubes, keeping the gain pot at very low and complaining about too much distortion.

 

[manulaudic]

Hi all, I'd like to point out that from the mouser cart it would be ideal to order round shaft Lorlin for the side chain boards of the rotary version..

It's a pain to find knobs that orient well with the d shaft and the orientation from the PCB.

In case someone else is also using the cart, I would suggest that edit to be made

Manu

 

[Script]

Idea / question:

Plate resistor for maximum gain is as follows:
5k5 for 19 dB
4k99 for 17 dB
And there's a parallel resistor footprint.

How about sticking a 100K trimmer in there, pre-dialled to 100K (minus one turn), to calibrate overall max gain of unit or rather/better calibrate L and R somewhere around most often used max gain ? Should be safe even in case of trimmer failing/go open. -- Calibration overkill ? Bad idea ?

[But not needed when trimmer is on output for calibration, I guess.]

 

[paolomunarini]

Hi all, I'd like to point out that from the mouser cart it would be ideal to order round shaft Lorlin for the side chain boards of the rotary version..

It's a pain to find knobs that orient well with the d shaft and the orientation from the PCB.

In case someone else is also using the cart, I would suggest that edit to be made

Manu

I actually already had lorlins with round shafts so I didn't see that coming, 'cause the ones on the mouser cart were unavailable at the time I ordered

 

[richiyobs]

For lorlin... Banzai music and reichelt... Both have good prices

 

[paolomunarini]

For lorlin... Banzai music and reichelt... Both have good prices

Mine were from banzai in fact

 

[Heikki]

How about sticking a 100K trimmer in there

It certainly would be better way of adjusting the gain balance of the channels than adjusting the tube bias.

 

[Script]

More ideas:

IC SWAP ON POT VERSION

On the SC version with pots, I swapped U9 and U10 for INA2137 (instead of 2134) as suggested above for a mastering version. With the 2134, the "Input Gain" pot sat somewhere before the 9 o'clock position for unity gain, and I also don't need that much gain from this unit (e.g., to drive next unit in chain louder).

With the 2137, the 'Input Gain' pots now sit more or less at mid position when compressing mildly. Also, it allows me to drive gear before this unit louder and back off with the Input Gain knob before hitting the tubes.

With full THR and full RAT, but slowest ATT and REL, the unit can still yield hefty compression with -12dB LED lighting up (--more compression when ATT is set to faster, and even more when driven harder--), yet still operate around unity gain level, when "Input Gain" is fully CW (for makeup). This way, the Input Gain pot control feels quite intuitive to me.

 

[Script]

STEREO LINK (ON POT VERSION)

How about linking for stereo after the Darlington (after R1) ? Not good ?

This way, CV output of the two sidechain boards can and should be tuned for exactly the same value (of around 4V) to address component variation.

When tuned to same DC output, this link switch does not interfere with the bias voltage of the two sidechains, as they might be slightly different between left and right if used for fixing imperfection of tubes.

Also, the ATT and REL timing networks are still separated, just as they where before, and there is no halving/doubling of time constants when switching between stereo and dual mono (not a problem with this unit, but on other "vari-mu" units).

The biggest benefit I can see is that deviations in pot resistance etc do not matter much any more when linked. Actually, the settings on left and right channel only need to be set to somewhat similar. However, they could just as well be radically different (one fast, the other slow, as in two parallel sidechains).

Difference in volume between left and right due to tube variation could then be calibrated 'on the fly' by using the "Input Gain" pot on the front panel.

Plus, pairing that with matched tubes and it's seems just perfect to me -- also for stepped controls.

 

[Heikki]

How about linking for stereo after the Darlington (after R1) ? Not good ?

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. Another safe method for stereo link is described in this post.