Marshall 2210 hum

[lowell]

This came in with a toasted EL34. Retubed it, and biased it. There is a low level 60hz ish hum. If I removed the PI tube the hum goes away. If I remove all tubes EXCEPT the PI, the hum remains. Tried a new tube there, didnt help. I put a BFC from the grid of the PI input to ground and the hum went away. I'm not really sure what this tells me. I THINK it's saying the source of the hum is before the PI but after the previous recovery stage. I clipped a filter cap across the filter node for the recovery stage V4 and no change. I checked for AC ripple and don't see more than 27mv. But not sure how accurate my meter is. It goes down to 2vac range. Thoughts? BTW all controls are at 0. If I touch the chassis with my meter probe I hear a slight tick sound and it makes a scratchy sound if I rub my probe along the chassis. Not sure if that clues us in. I've checked all filter cap grounds from the negative pins of the caps to AC jack earth. All seem fine, at 7ohms, and my leads shorted is 7ohms.

I even tried elevating the heater center tap to 10v! It made the hum WORSE!

http://www.drtube.com/schematics/marshall/2210prem.gif
http://www.davidsonamp.com/sm/images...TDpoweramp.gif

[lowell]

Ok so I am getting ~3ohms from V4 Rk ground to chassis ground. The rectifier ground to chassis ground is 0ohms. Can't see exactly where the ground traces run and what should be grounded where. I clipped a wire from the Rk to multiple grounds on the chassis and it didn't help. I clipped it to the ground of the filter cap for the switching IC and it made it worse. Suggestions?

[lowell]

Ok found the iffy ground, was the rectifier ground wire. Resoldered it to the cap and now get a solid reading. This didn't help at all with the hum unfortunately.

What do y'all think? I'm baffled. The hum is somewhat acceptable volume wise, but doesn't seem like it should be there, especially if only isolated to the V4/PI area.

[wizard333]

When you retube and you get hum I immediately think "heater-cathode leak on one of the new tubes" but if you changed PI tubes and removed the rest, that won't be it.

2210s are a high gain circuit. Lead dress matters. Make sure your flying wires from the board to the tube sockets aren't misbehaving. Chop stick them around a bit and see if anything changes. Also, the passive loop/insert points are directly before the PI on those amps, check that your loop jacks are clean.

[wizard333]

Oh one more thing:

These amps have reverb, and the return is not blocked by anything before the PI. I assume you have the amp out of it's cabinet and on the bench, so make sure the verb knob is off. Otherwise, the verb return will absolutely cause hum.

[lowell]

Yah ALL knobs have been at zero. Also have the reverb return grid is shorted to ground for now since there's no grid leak path without the tank.

[lowell]

Ok so here's some more test results. Not sure what to think at this point. I'm missing something, but not sure what.

AC grounding the PI input kills the hum.
AC grounding before C25 input cap does not kill the hum.
Both stages of PI are working - 130v across both Ra.
The presence knob increases the hum when turned up.
Disconnecting the 100k FB resistor makes the hum louder.
All power tube grids have identical -37 grid bias.
There is no AC on the grids of the power tubes, screen grids, or plates even.
There is 0Vac on the bias supply.
Moving wires to the socket and/or board don't have any effect on the hum.
Again, a new tube is in there, and the hum goes away only when the PI is removed.
All PI resistors checkout fine.
PI grid is -1.4v referenced to cathode.
Vk is ~70v.

I'm stumped!

[wizard333]

When you say "a new tube is in there"; how many tubes have you tested in that amp with this result?

Also, did you test that each tube is actually getting screen grid voltage? If the one that blew also blew the screen grid resistor, and you had 2 power tubes working on one side of the PI and 1 on the other, it would likely hum (the push/pull 'humbucker' effect would be unbalanced), and you'd hear the crossover distortion on the ends of the notes.

Don't just test plate/bias voltage on each tube, test the screen voltage right on pin 4 and see if you have current actually flowing in each tube.

[lowell]

I have checked, and Screen voltages are all good. I'd think the bias would not be the same on both sides of primary if one screen wasn't working.

I just tried a couple more 12ax7. JJ ecc83. Either I messed up my reading before or this tube biases a bit different. But now, I get 144v across the 100k Ra and 130v across the 82k Ra. Still hum.

BTW, is this Merlin?

[wizard333]

I have checked, and Screen voltages are all good. I'd think the bias would not be the same on both sides of primary if one screen wasn't working.

Just to prevent confusion, did you check the screen grid voltage at the pin or are you checking control grid bias voltage at the pin? A blown screen resistor won't change grid bias voltage showing up on pin 5, and you'll still get plate voltage on pin 3, but your tube wont work and if you check pin 4 there will be no voltage present.

Sorry if I'm misinterpreting what you said, but when you mention control grid bias in combination with checking screen grid voltage I'm getting a disconnect.

[Jazz P Bass]

That was a confusing reply.

I do believe that Lowell was speaking of the actual bias current through the tubes.

[g1]

If it were a bias imbalance issue, I don't think it would go away when you pull the PI tube?

BTW, is this Merlin?

Merlin's user name is merlinb

[nickb]

I would consider 27mV of ripple at R58 (V4a stage) to be significant. I'm guessing that when you clipped a cap to that node you clipped it to the leads of C47. Now if either lead of C47 had a bad connection to the board...

Edit: Uh. I just saw this that I missed "AC grounding before C25 input cap does not kill the hum". Rather says that the hum is coming from after C25. Still, if the max drive voltage at the PI for clipping is something like a few volts then i still think 27mV seems excessive. You have a cap filter, a choke /cap and two RC networks of hum attenuation - that has got be 100db or more you should be getting. And don't forget it's a chain and any bad link will spoil the result. IOW, it could be an earlier stage.

[lowell]

Sorry for the garbled reply.

Bias CURRENT (shunt method) is identical on both ends of primary (within a few ma), so all 4 tubes are conducting. Screen voltages were taken at pin 4, and are good. The grids (pin 5) all have identical negative bias voltage.

I've rerouted some grounding, mainly the filter cap for the PI and all preamp tubes, and moved it's negative terminal over to the ground at which all signal grounds lead to. Seemed to help a tiny bit. I've tried a new filter cap there as well. Didn't help.

I realized that if I left the BFC on the PI input, the hum went away for about 5 seconds while it charged up but then came back much louder.

One thing that seems to help is to tack in a 16uf after the last 10k dropping resitor for the supply. And this cap is grounded at the cathode bypass cap for V4. That does help.

Anyway, it doesn't seem right to do that as a real FIX. Nickb I'll check that cap connection.

[lowell]

It seems like a bad idea that they use ONE filter cap for 8 stages. Not sure this matters in regards to the PI though, as it has it's own dedicated filter.

Except the reverb send. Hmmm... gonna investigate that, disconnect it.