Transformerless Vari Mu Compressor build thread

[Script]

Can anyone confirm that the BJT alternative to the sidechain transformer works ok?

I had intended to go BJT in the Diode 8 simply for cost reasons, but then never did -- also cos I had run out of real estate in the case.
Heikki's dual comp does without SC input xfrm. But THR is implemented differently there.

 

[C H]

The 107N is 1.5dB down at 300Hz. Data sheet doesn't say what happens in the three octaves below that. Should we care? Doesn't this mean there's a high pass filter in the sidechain whether we choose it or not?

 

[JMan]

There is a HPF at around 20Hz by default

 

[C H]

Indeed, but doesn't the transformer also impose a high pass at a much higher frequency?

 

[thelivingroom]

The 107N is 1.5dB down at 300Hz. Data sheet doesn't say what happens in the three octaves below that. Should we care? Doesn't this mean there's a high pass filter in the sidechain whether we choose it or not?

The 107N and 107V data sheets show the frequency plot. Somewhere in my messages Heikki recommended the 107V over the 107N.

 

[JMan]

Indeed, but doesn't the transformer also impose a high pass at a much higher frequency?

I do see your reasoning here and I can't speak to it on a technical level. Just going by ear, my unit does seem to be responding as I would expect well down into the low frequencies when I do not have the added HPF engaged. But I haven't verified by running any sweeps or the like.

 

[C H]

The 107N and 107V data sheets show the frequency plot. Somewhere in my messages Heikki recommended the 107V over the 107N.

View attachment 146213

View attachment 146214

Ah OK, that's not too bad. Still, wouldn't bjts be better (and a lot cheaper)?
Does the transformer step up, or down? The pinouts on schematic and data sheet don't match.

 

[scott2000]

wouldn't bjts be better (and a lot cheaper)?

I'm not having any issues with the Hammonds. Can use any transformer too obviously..
from Heikki when I was talking with him on the subject
"Resistance from R35 & R37 and the sidechain transformer inductance form a high pass filter for the sidechain. This is done on purpose. In my opinion in a backward acting vari-mu compressor there should always be a high pass filter on the sidechain. I think I have written more about this on groupdiy"

Better? Who knows.. Cheaper?
Would be cheaper to replace R6 on the sidechain boards with a jumper... if you wanted faster attack times... But then we're getting into maybe more subjective design choices.. Mods...
but again, Heikki's view..
"In my experience when limiting/compressing for effect 2 ms is about the fastest attack needed. With faster attack times you would want to make sure the tubes are in near perfect balance. Because when the “thump” gets to be high enough frequency it is hard to keep it out from the sidechain and from making that overcompression effect which can be seen in the Manley example, even though the “thump” itself might be inaudible. In my opinion ultra fast tube compressor should have a feedforward topology like the Gyraf vari-mu. Feedforward vari-mu compressors have their own unique problems and probably thats why feedforward topology is rarely used in vari-mu compressors. "


These units are laughingly good for what they cost. When I was in Bob Katz studio testing it along side some other very impressive units, it had zero issues being right at home and was preferable to some. And this was before the mastering version being made. Of course that's program dependent and the Leapwing plugin bested all the hardware he had on certain material...

Does the transformer step up, or down

In the compressor they step up
"With 107N the output levels where compression starts will shift few dB up"

The pinouts on schematic and data sheet don't match.

I think it was an error that was never corrected. Pin 1 on transformer goes to the square pad on the pcb.

 

[thelivingroom]

I finished mine yesterday, then listened to many genres of program material through it on numerous settings for hours while cleaning up my build mess and setting up to shoot photos of it, and share the same sentiment as @JMan with feeling like a kid on Christmas. This thing rips, is extremely quiet at idle, and I’m surprised that the thread isn’t 100+ pages by now. It’s affordable, easy to follow, easy to calibrate, original, modern, and was a pure joy for me to build. It sounds so good running it transformerless that in hindsight I wish I’d have omitted the input trans option and saved some $$ and wiring time.

It’s been a long time since I’ve hardwired an entire build without one breakaway connector, so I was very happy to fire it up and have every voltage test point measure spot on. A +4dBu sweep at unity gain was dead flat from 20Hz to 25kHz on both channels. The tube balance calibration measured less than 1mV on each channel (-60 and -62dBu) with the NOS GE 6SK7 steel tubes. I also bought extra bundles of NOS RCA (also steel), Sylvania JAN, and Hit-Ray (both glass), but didn’t feel the urge to swap out the GE’s. The AC-DC power adapter is a TDK-Lambda DTM65PW360C that I bought a couple of new on eBay. The datasheet for it listed the model of the Kycon connector, so sourcing the correct chassis power inlet was easy. I can also confirm that you can use any brightness LED that you want for the GR meters and it doesn’t affect the metering; just adjust the 4.7k resistor values to taste (R4, 7, 10, 13, 16, and 19). I used super bright and dumbed them way down, with the added benefit of drawing less current from the PSU. I sent the 25.3-ish volt power for them out to a breadboard and worked all that out beforehand. Each color change ended up with a different value for them to all match in brightness, as expected.

I’ve been wanting to design a front panel inspired by a vintage HH Scott integrated amp that I read about many years ago in a tube amp magazine, with a gold faceplate and slide switches, and this project felt like the perfect fit. I did a fair amount of CAD work to lay out the symmetry and figure out the side chain mounting panel (to both conceal the hardware and buy myself some space for where I wanted to mount High-Low switches), GR PCB locations, power switch plate, and case layout. I wanted the GR LED’s to be on the same centerline as the slide and rotary switches, so that meant bending an offset on the ones for Channel 2. I drew the LED with the offset in CAD, then printed it and taped it to a wood block with a cavity for the LED to sit in and bent them. It worked like a charm when installing the meter PCB’s. I clamped the face of the front panel to a piece of 4mm aluminum I had laying around, placed all the LED’s in the PCB’s, mounted the boards to the standoffs, used a hex driver to reach in and push each down until they hit the 4mm panel, then soldered them. To clean up the look of the gap through the power switch hole and keep any light emitted from the backside of the meters from shining through it, I cut a scrap piece of tool box drawer liner, used a leather punch to make the holes for it to slide over the standoffs, cut a slit in it, then affixed it to the front panel with E6000. In case of future surgery or potential initial troubleshooting with everything hardwired, I wired it with everything coming up from the bottom to the controls so the front panel can be unbolted and rotated down flat on the table.

Thank you for the boards @scott2000, Heikki for the wonderful project, and everyone else who contributed useful information in this thread!

 

[JMan]

Unreal. As usual, you've raised the bar on clean, attractive design work!