Negative voltage on grid of cathode biased output tube...

[alchemy]

Taking a look at an abomination of a homemade 5c8 twin fender. After cleaning up tons of horrible lead dress, bad joints, bad parts, etc it is starting to sound good, but one major issue: It has 2 6l6 cathode biased, and on one of the grids there is -6vdc to ground, while the other sits right at 0vdc. This voltage drops to about -1.5vdc when the negative feedback is removed. It's not the tubes nor the coupling cap. It sounds great until it's turned up towards overdriving , and then there is an ugly distortion riding the signal. I put it on the scope and there is indeed an oscillation (that seems to be modulating only the bottom 25% of the sinewave) at 50khz.
So while I am certain that the parasitic is due to the build quality, I am not sure if it has to do with the negative voltage on the output tube grid. Never seen this before.
Any ideas? Is this a common problem when oscillations are present?
It seems the whole amp is resonating at 50khz. I can wave the scope probe around inside and get it all over the place.

[Wilder Amplification]

Taking a look at an abomination of a homemade 5c8 twin fender. After cleaning up tons of horrible lead dress, bad joints, bad parts, etc it is starting to sound good, but one major issue: It has 2 6l6 cathode biased, and on one of the grids there is -6vac to ground, while the other sits right at 0vac. This voltage drops to about -1.5vac when the negative feedback is removed. It's not the tubes nor the coupling cap. It sounds great until it's turned up towards overdriving , and then there is an ugly distortion riding the signal. I put it on the scope and there is indeed an oscillation (that seems to be modulating only the bottom 25% of the sinewave) at 50khz.
So while I am certain that the parasitic is due to the build quality, I am not sure if it has to do with the negative voltage on the output tube grid. Never seen this before.
Any ideas? Is this a common problem when oscillations are present?
It seems the whole amp is resonating at 50khz. I can wave the scope probe around inside and get it all over the place.

First off, AC voltage has no polarity. Hopefully you meant "VDC". Can you please clarify this?

[alchemy]

doh! fixed it!been a long day (already...)

[es345]

... would start with the following
- open the NFB
- put out the driver tube 6SC7
- measure if the oscillation is still there
- if not close the NFB
- measure if the oscillation is still there
tell us the results

[alchemy]

Oscillation is gone with the tube removed, whether NFB is hooked up or not.

[Steve Conner]

If the oscillations are big enough, they can push the tube into grid current and generate a negative bias that way. I guess this is what's happening.

Cures for parasitic oscillation:
Grid stopper resistors on power tubes. If it already has them, try bigger ones?
Improved layout. Power tube grid and plate leads no longer than necessary, grid stopper and screen resistors mounted right on socket.
Tinker with the global NFB. At 50kHz the oscillations are more likely to be "Nyquist" oscillations involving the whole feedback loop, than true parasitics which tend to be in the 10s to 100s of MHz. If you use a different OT to the one the circuit was designed for, you may have to tweak the NFB for stability.

[alchemy]

The first thing I did was shorten all plate and grid leads, separate them, properly ground the shielded cable (both sides were grounded), and just generally clean up the wiring and layout. Still could be better, as this thing was really poorly laid out and built. I haven't tried higher value grid stops yet (has 1k now) as I was trying to determine the problem and correct it before I just covered it up. It definately seems to be coming from the PI, which by the way is not wired like the schematic. It's closer to the later self balancing paraphase, though not identical. And the NFB is coming back into the cathode of the PI, not the grid of the output tubes as the schematic shows.
The oscillation frequency is not audible like most that I've dealt with. But it is obviously modulating the signal. Seems to start at the voltage divider right before the PI, and then get exponentially worse as it goes through the PI stages.


If I use grid stops to stop the oscillation, since they are at the output tubes but the signal is getting modulated in the phase inverter, won't the effects of this still show up?

And, why is the negative voltage only showing up on one power tube grid and not the other?

Thanks, as always, for the insight!

[es345]

OK, some more remote diagnostic is necessary:
Q1: when the tube is removed, do you still have the negative voltage?
Q2: can you provide the schematic of the machine?
Q3: can you provide a photo of the layout/wiring?

[alchemy]

I did a bit more poking around and it looks like the primary frequency of oscillation is about 71khz. There's a few others a bit lower, but this is the one I keep finding everywhere. With no input signal to the amp this frequency shows up on the scope everywhere in the amp. At the second PI output, it is 35vpp!!!

When the tube is removed the oscillation disappears, as does the negative voltage.

I dont have a schematic I can scan drawn up. Basically, it's a 5c8, but with a few things changed. 6sl7 PI, and instead of the second PI grid getting it's signal from the 250k/6800/250k voltage divider (grid resistors for output tubes), it gets it from an added 470k/560k voltage divider (the center of which also goes through a 470k resistor to ground. Each output tube has 270k grid resistors to ground, and the PI has 100k plate resistors. Those are the major changes (and the feedback I mentioned above). There are probably a few other little things.

You don't even want to see a picture of this thing!!!!

What does the fact that it disappears when the PI tube is removed suggest?

[loudthud]

What does the fact that it disappears when the PI tube is removed suggest?

Too much feedback. Is there a presence control? Does the oscillation go away when the feedback is disconnected?

[alchemy]

Too much feedback. Is there a presence control? Does the oscillation go away when the feedback is disconnected?

Nope. Only when the PI tube is pulled.

[loudthud]

So, no Presence control? Looking at the 5C8 schematic, is there the 1M feedback resistor to the 6L6 grid from the speaker?

Try this: Minimize the capacitance to ground at the 560K/470K/470K node that goes to the grid of the 6SL7. If that doesn't fix it, try a small amount of capacitance across the 560K, (or directly on the 6SL7 grid and plate) perhaps 10pF to 27pF.

[alchemy]

So, no Presence control? Looking at the 5C8 schematic, is there the 1M feedback resistor to the 6L6 grid from the speaker?

Try this: Minimize the capacitance to ground at the 560K/470K/470K node that goes to the grid of the 6SL7. If that doesn't fix it, try a small amount of capacitance across the 560K, (or directly on the 6SL7 grid and plate) perhaps 10pF to 27pF.

Nope, no 1 meg feedback to the 6l6 grid. The feedback is to the cathode of the first PI stage. Not sure what you mean by "capacitance to ground". The only capacitor in that area is one right before the signal is taken and sent to the 2nd grid of the 6sl7, before the 470k to ground. Are you speaking of stray capacitance (wires, etc)?

[balijukka]

but this is the one I keep finding everywhere. With no input signal to the amp this frequency shows up on the scope everywhere in the amp.

This points to the oscillation starting at the input tube. That's where it usually do.
Try pulling the first tube and see if the oscillation disappears.

jukka

[alchemy]

This points to the oscillation starting at the input tube. That's where it usually do.
Try pulling the first tube and see if the oscillation disappears.

jukka

It's still there when the input tube is pulled, though the amplitude can not be controlled any longer with the volume pot. The only tube that makes it go away is pulling the PI. Also, the amplitude (and frequency to a certain extent) can be varied by turning what I think is the treble pot (the 1 meg pot that goes to ground via the .005 cap at the plate of the 6j5. The PI circuitry seems to have the most effect on the oscillation. Would this indicate that it is coming from this part of the circuit, or is it just coincidence? What is the normal procedure for pin pointing where an oscillation is coming from?