60 hz hum in Vari-mu compressor and..

[Svart]

So I built a PRR vari-mu recently and love the way it sounds(thanks again PRR) but had been plagued with 60 cycle hum in the system, but mainly one channel. I had a few extra minutes to poke around in the box and decided to try to kill this noise. I came to a good outcome but also came across some strangeness that I don't quite understand.

the channel that had less noise idled around -55db with the feeding faders @ unity. noise was mostly white/pink with a little AC noise mixed in. the noisy channel idled at -35db with the same fader setting, noise being mainly AC. strangely enough the noisy channel was the furthest away from the PSU and got much worse as I moved it closer to the psu. I then cut out a peice of steel from an old computer case and bolted it between the PSU and the main board. this helped greatly on the noisy channel, bringing it down about 10db to idle around -50db, but also bringing down the noise on the other channel to around -65. still a little lopsided but getting better.

now here are the strange things I noticed. I happened to touch one of the output wires and one of the output trafos at the same time. the thing squealed like a little girl! some kind of feedback I assume, but it wouldn't do it everytime i tried it. not sure what that was but it was really of no concern. The other strange thing was that if I let the output wires lay normally the noise level would go up another 10db, but if I moved both sets of output wires and bound them together in the center of the board exactly in between the tubes, I could drop the noise another 5 to 10db.

another strange occurence also happened by accident. When i installed the steel plate between the psu and the main board, the wiring from the psu was pushed into one of the tubes. later i went to move it and secure it and as soon as I touched it and started to move it the noise level shot up another 15db in the noisy channel. I moved the wiring around and found that the quietest spot for the wiring to be was touching the tube at the very top point. anywhere else and the noise was considerably worse.

so this leads me to ask, do tubes emit some kind of field? I'm not at all educated when it comes to tubes so i don't know. Judging from my experience I would say that they do but I still don't know.

all said and done, the box is now fairly quiet, idling around -60db on both channels evenly. It still sounds exactly the same as it did before but it's much quieter on quiet passages.

any ideas?

 

[bcarso]

True 60Hz usually means magnetic field pickup, as your experience seems to bear out. Minimize loop areas, both those that emit and those that receive! Then you can also get noise induced from an a.c. heater supply, which typically can be trimmed out with a strategic potentiometer balancing the filament supply relative to common. But since both channels would be using the same supply that's an unlikely mechanism.

Tubes don't emit much unless they are part of an oscillator*, in this case an unwanted one, and then the emission is more likely from the wires/traces connecting. Do all of the stages have few-hundred-ohm grid stopper R's?

Brad

*or if they are running at really high volts, enough to get X-rays...

 

[Svart]

not sure if they have the stopper resistors as I am uneducated when it comes to tubes. the tubes, as far as i know, are not in a gain arrangement and the output gain/drive comes from an IC. It's obviously a field of some kind maybe a ground loop, but the mainboard is grounded through the psu connection, not to the chassis through the mounting screws.. maybe I should try replacing the ground through the psu with one directly to the star ground?

interesting..

:thumb:

http://www.conditionedresponse.com/DIY/C5/index.html

just for reference

 

[gyraf]

..first try rotating/relocating your power transformer - this is the most likely source of hum in a circuit that depends on audio transformers..

 

[Svart]

thanks gyraf, I have about 12 inches of wires between the PSU and the mainboard, stretching the wire out completely and rotating/moving the PSU doesn't make much difference until it gets right up close and then of course you get more 60hz noise from the trafo, but even at 12 inches away you couldn't get the noise to go completely away. once I shielded the PSU, the noise was at it's least, but still uneven across the channels, so moving the other wiring evened it out. you still get small amounts of hum though..

It's totally usable though and if I can't lessen it even more, i'll use it nonetheless.

:thumb:

 

[CJ]

Is it 60 cps or 120 cps?

120 cps means pwr supply filter inadequate.

120 is an octave above 60 cps. (is that right?)

120 cps corresponds to a note on the guitar but I fogot which one.

lets see, low A string is 440, divide by half, 220 is real low A, 120 would be............ :?:

 

[bcarso]

...brown noise....heh

That's another whole topic, even without help from South Park.

120Hz is about a quarter-tone-flat B natural based on the A = 440Hz reference.

 

[PRR]

> kill this noise.

Don't say noise when you mean hum or buzz!!! It gets very confusing, fixing the wrong problem.

The hiss of this plan should be at least 80dB down from limiting level.

The hum and buzz "should" be totally layout flaws. If I did my design right, and you do the layout good, the hum/buzz should be well below the noise. The tube side is well filtered and also push-pull. The chips are filtered and have huge supply rejection.

> bound them together in the center of the board exactly in between the tubes, I could drop the noise {Buzz!} another 5 to 10db.

2 years later, I do see a bad assumption on my part. The heater DC is not well filtered, and unless you are obsessive with heater wire twisting and routing (maybe if you are) it will induce buzz into the system.

ADD a several-thousand uFd capacitor, on the heater side of the dropping resistor, but keep it on the power supply board away from the amplifier.

> happened to touch one of the output wires and one of the output trafos at the same time. the thing squealed

Sure. Stuff happens when you connect outputs back to inputs.

If you just touched the case of the transformer: the case needs to be grounded. For the general situation, to chassis ground; however strapping it to signal ground will probably be OK.

 

[Svart]

Sorry PRR, I did mean to keep the words BUZZ and HUM separate but I guess i failed. the white/pink noise is mixed with the 60hz hum, but in one channel the 60hz hum was very much louder than the noise by far. this seems to be much more equal now with the shielding.

I also strapped a ferrite(stolen from a monitor video cable) around the bundle of wires going to the mainboard from the psu with no real results jsut in case anyone wondered about that..

I'll try the extra cap on the heater line, as I am not *that* anal about routing and twisting :green:

the squeal was from just touching the output wiring to the shell of the trafo, mind you this wiring is with the insulation ON, this is not a bare wire making a connection to another piece of metal.

still not sure what that means though..

thanks for the ideas!

 

[Svart]

ok, update time!

Took a few minutes to get back into the box and take a look. Added the capacitor(s)(i added a .1uf to the mainboard psu inputs as well as the 100uf at the psu) to the 12v circuit which did quiet things down a few more db. I did however find something else that I missed entirely.. the layout i used was mirrored from one side to the other but when i pinned my connectors I didn't mirror them. turns out I didn't hookup the ground for the noisy side output.. not a big deal because I hooked it to ground with a strap, but once i did the noise on that side dropped a couple more db. I moved the PSU wiring around to the quietest spot and measured around -70db(almost the limit of the VU/PM!) at idle with the feeding fader at unity. pull the fader down and the unit goes down to beyond_my_ability_to_test_quiet..

I also found that this unit sounds really really good by itself.. meaning no other compressors in the chain. very clear and clean.

thanks!

:thumb:

 

[Svart]

So I revisited this problem after years of using it as-is.

I have 120hz hash on every rail except the 1.5v.  If I remove the tubes I only have 120hz on the 100v rail.  It's roughly 50mv.  I tried upping the cap values considerably but nothing helps.

still troubleshooting..

 

[kazper]

Not a expert by any means, but I remember someone talking about the ground planes or DC common are shared in the layout that has been floating around....

It's something I have scribbled in my documents for that project.

Only just a shot in the dark...

It would be nice to get someone like [SA] to do a layout for this thing.... maybe have Gustav pimp them...helo... heloo ..anyone out there...

Kevin