Using Oscilloscope to Track Oscillation

[surfsup]

Ok so here's something I discovered that seems to be new b/c I know I dmm'd this before and came up with nil. There is DC voltage on my power tube grids...

I put the DMM clips on the CC leads off the PI (after the cap but before the MV). Voltage readings:

MV @ 0: 22vdc and 14vdc on the two sides
MV @ 10: 8vdc and 3vdc on the two sides

I assume the DMM is probably getting some RMS AC so I set the scope to DC coupling and put the probes on the 6N1P socket pins and one shows positive DC and the other negative DC:

[surfsup]

ok I'm an idiot. on that DC voltage I had signal runnign through the amp. So I FOUND the problem (at least I think I did)...something must have happened. I see the PI plate is severely chopped. Blue positive swing in picture. The Ra are 100k/82k. The 82k side is dropping hardly any voltage (16v vs 100v on the other side). so the 100k has 1 mA going through it and this side with the 82k has about 0.2mA. This is indeed a problem. I swapped tubes, I reflowed, I ohmed the 82k it is 81.9k. I continuity tested the lead to the cap lead and also did an ohm test and got 0. Nothing is changing...

[surfsup]

For your continued enjoyment...I see if I put the scope probe on the PI grid while measuring the PI plate, the waveform gets chopped. I remove the probe from the grid and voila! the PI waveform is fine. WTF! Obviously probing the PI grid is pointless...unless you like wasting time testing other stuff for absolutely no reason.

[katopan]

PI grids are the only place where it is almost useless to probe with a CRO. The CRO input impedance although high, forms a divider to ground with the 1M or 470k resistor grid to tail resistor and loads or pulls down the grid voltage. As you've found that changes the PI bias/balance and stuffs up the output.

I say almost useless because if you keep the loading in mind it can still be useful to understand some of what's happening there. But you've got to majorly adjust what you're seeing compared to what is actually happening there without the probe attached.

[vibratoking]

I remove the probe from the grid and voila! the PI waveform is fine. WTF!

Explained by katopan. This is a good example of what I tried to warn you about in my post early in the thread. You can't just probe and assume there will be no influence due to the probe characteristics.

[surfsup]

Ok I have some relevant news I think. I resoldered in the Edcor OT and 125R cathode and started over. The buzziness is caused by the PI. I am almost 100% certain of this. Why? I put one probe on the cold clip stage and one on one side of the PI out after the coupler. The buzziness only happens when the PI signal out into the MV control has flat peaks. As the guitar strings dampen and the signal goes down, as soon as the flats disappear and go round again, the buzz cuts out into a clear note.

So I'm now wondering for an EL84 should the signal out of the PI be clean, just before flattening, at max signal?

[gingertube]

A schematic would have been good. I just read thru' the entire thread.

One thing I have found with these sort of problems is that when an oscillation happens above a certain signal level it is nearly always to do with an inadequate or missing grid stopper.

El84 Output Tubes - you have 4K7 minimum stopers - YES?

Next (fgrom something you said above) I would look at the stage (gain stage or PI) which immediately follows the tone stack, do you have grid stop there, if not fit one. Even 1k can help.

Also be aware that on an amp (Trainwreck) I built which was incredibly noisey (at any level) it turned out that I had accidently fitted a 470K grid stop (on the cold clip stage), changing that back to 1K eliminated the noise so you can go too high on grid stops as well but this shows up as general noise rather than oscillation above a certain level. On that same amp the cold clip stage and the tone stack recovery stage shared a 12Ax7 and I found that the when the couipling cap between the two was too close to the cold clip stage anode load resistor it would burst into oscillation above a cetain level. In that case it was coupling from the anode of the cold clip stage back to it's grid via the coupling cap from the preceeding (tone recovery) stage.

It can be other lead dress issues as well. The big thing with lead dress is to examine the high impedance points around the circuit which are far more susceptible to pick up inadvertently coupled signals from adjacent circuitry. That is, look at anode connections and top off tone stack, top of volume control points and grid feeds rather than low impedance points like cathode circuits.
That is a bit of general hint for building too. If you have to run a wire to "remote" some circuitry because of space constraints then the obvious choice for the "remoted" components are the cathode bias resistor and bypass cap coz it being low impedance will be much less susceptible to pick up.

The oscilloscope is not always the best tool. I have found on a number of occassions that clipping on the CRO lead actually cures the problem you are trying to find, particularly when scoping grid connections (due to the added 15pF or so of the CRO probe).

Cheers,
Ian

[surfsup]

Thanks for reading, sorry there was no schemo. It is basically a liverpool with some minor differences. The sounds good except for a little buzziness on top of the notes and what appears to be an oscillation when I hit certain double stops on the guitar. On some double stops it is fine. I believe this is accurate, except the screens are a single 1k5W with a 220R at each tube and the OT is an 8k. The supply voltages are right 325 plates and 322 screens, etc.