I feel like the wheels are coming off this build. Now 1 side of Tube voltages way low

[Enzo]

There should not be any voltage drop across the 1 meg plus 50k into the grid. DO NOT measure voltages to ground, your meter impedance will addle the readings. I can believe -41.9v, but measure voltage from that -41 point directly to the grid pin. You have 3.7v across the cathode resistors, and I expect about that same thing if you measure directly cathode to grid pins on the socket.


Almost 50v bias on an EL84?

[SoulFetish]

Hail, OCD-man! Wow, what awesome lead dress you have, there. I want to see more photos!

Those big black boxes are filter caps?

And what about those blue-handled tools at left in photo?

Sorry for the delayed reply, xtian and Dave. Thanks for the kind words.
Yeah, those black boxes are Polypropylene filter caps. High Voltage, High Surge current (not huge concern for this application), 105C rated and life expectancy of 100,000 hours @70C. If I never had to use an electrolytic cap again, it would be too soon. To be fair, size and cost still make them a sensible choice when necessary. I've also read that the higher ESR characteristics in electrolytics can be optimal in certain switching regulator applications.
Those blue handled tools are a set of desoldering picks kind of like this:


I use them ALL the time (some much more than others). If your doing any repair work on through hole components, I think they're a must have. They're usually pretty inexpensive as well.

[SoulFetish]

Ugh. I don't...
This is got be going in circles, I'm not sure where to begin..
Enzo, I'll start with you.

There should not be any voltage drop across the 1 meg plus 50k into the grid. DO NOT measure voltages to ground, your meter impedance will addle the readings. I can believe -41.9v, but measure voltage from that -41 point directly to the grid pin. You have 3.7v across the cathode resistors, and I expect about that same thing if you measure directly cathode to grid pins on the socket.

First, I wanted to make sure all the cathode resistances after the 18k loads were equal on both sides and measured through to the cathodes properly. I realized that I was testing the little bit of exposed lead which was soldered to the cathode pin. I was suspicious that the joint may have been bad and was adding resistance to the actually cathode that I wasn't reading. This would explain the increased negative grid bias. So, I figured I would reflow solder onto the tube pin and measure through to the other side of the chassis. Looked good. So I powered up and measured cathode to grid like you said. This seemed to do the trick and I was measuring -1.8V(ish) on the grid! seems like a win, right?
Now, of course the other wheel has to fall off. I measure the "good" side of the triode and I was getting -0.78V even with the adjustment at the highest cathode resistance! What the F••k?
But, I thought, maybe I'll put the output tubes in and test the 1ohm bias points in case the meter is causing any faulty readings.

Almost 50v bias on an EL84?

?? you mean -50V? not Ideally.

Hanging meter probes on the lower 12AU7 could be causing it to oscillate giving false readings. Remove the probes from the 12AU7 and plug in the output tubes. What do you measure across the 1 ohm resistors for their bias currents?

Now, Dave. With my amp plugged into a dummy load, I put in my EL84s to bias them and here's where it gets weirder.
Firstly, I was getting bias runaway watching the EL84s climb to 80mA and pulled the power. But, the weirder part was, even without a speaker, I could acoustically hear some oscillations??
This wasn't happening before, I was just trying to troubleshoot some feedback and the tone stack. Now, I don't even know...

[Enzo]

It is normal to hear the OT singing when you can't hear a speaker. Use a dummy load, and this happens in many amps.

Unless I am missing something, you have -41 on the grid and +8 on the cathode. That sounds like pretty close to 50v of bias to me.

BTW, -50v is a voltage with respect to some other point, usually ground. Bias is the difference between the points.

It usually doesn't get in the way, but in high impedance situations like the grids of a LTP PI, or like your circuit, it helps to consider your meter as if it were a 1 meg resistor. SO it forms a voltage divider with the circuit resistances, skewing your readings.

ANy chance your little zener deals are getting involved?

[SoulFetish]

It is normal to hear the OT singing when you can't hear a speaker. Use a dummy load, and this happens in many amps.

ANy chance your little zener deals are getting involved?

Can too high a resistance on the 12AU7 grids cause instability?
Zeners arent getting involved because I went back to the neons.

[g1]

Unless I am missing something, you have -41 on the grid and +8 on the cathode. That sounds like pretty close to 50v of bias to me.

He has a negative supply feeding the cathode circuits (-127V). Voltages he's getting are shown on his schematic in post #12.

[Enzo]

1 meg? A million Fender PIs have 1 meg grid returns.

Instability would mean oscillation, not a DC problem.

Scope the cathodes or the power tube grids, RF oscillation would be easy to see. if you lack a scope, google "RF probe", you can knock one together with a diode and a cap for your meter. It allows a basic meter to measure RF levels. If any are present.

[Enzo]

g1 - yes, post 12 is where I got those voltages. EVery way I look at it, his power tube grids appear to be almost 50v more negative than the cathode. Unless I am making some fundamental error.

[g1]

Yes you are correct, my mistake. I thought you were talking about the driver tube, but you were talking about the power tubes.

[SoulFetish]

Oh I probably should have clarified! After flowing solder over the tube socket, it's no longer at -41V. It was around like 1.5V

[Chuck H]

Oh I probably should have clarified! After flowing solder over the tube socket, it's no longer at -41V. It was around like 1.5V

1.5V is what you measured between the cathode and grid of the driver stage. That is a differential voltage and not necessarily a standing voltage. Since you are floating some circuits above (or below?) ground any readings for those circuits references to 0V wouldn't be relevant to actual working voltages. The way I read the thread the -41V you measured before was in reference to ground and the 1.5V you measured last was grid to cathode. So the voltage didn't change from -41 to 1.5. We don't have consistent testing to confirm exactly how much change there was.

As careful as your work looks to be I'm scratching my head over how many bad tube socket connections and solder joints you seem to be encountering with this build.

To be honest I don't even see just how everything is supposed to jive together for this direct coupling. It doesn't look to me like the circuit itself should work. But I have no experience with direct coupled stages.

When Enzo mentioned the -41V that should appear at the grid of the power tube you countered with how it was now 1.5V. But that is the voltage you measured from the grid of the driver triode to it's cathode, correct? So, what is the voltage from one of your power tube grids to it's cathode?

[SoulFetish]

1.5V is what you measured between the cathode and grid of the driver stage. That is a differential voltage and not necessarily a standing voltage. Since you are floating some circuits above (or below?) ground any readings for those circuits references to 0V wouldn't be relevant to actual working voltages. The way I read the thread the -41V you measured before was in reference to ground and the 1.5V you measured last was grid to cathode. So the voltage didn't change from -41 to 1.5. We don't have consistent testing to confirm exactly how much change there was.

As careful as your work looks to be I'm scratching my head over how many bad tube socket connections and solder joints you seem to be encountering with this build.

To be honest I don't even see just how everything is supposed to jive together for this direct coupling. It doesn't look to me like the circuit itself should work. But I have no experience with direct coupled stages.

When Enzo mentioned the -41V that should appear at the grid of the power tube you countered with how it was now 1.5V. But that is the voltage you measured from the grid of the driver triode to it's cathode, correct? So, what is the voltage from one of your power tube grids to it's cathode?

No, it does work. It's fairly simple. But I fear there is a lot of confusion when talking about voltages. So Ill post new schematic voltages. But to clear this 1 point up: There should NOT be -41(50V) bias at the EL84s. The fact that there was is a problem.
Chuck the tube sockets were the first thing to be installed. Swapping them out, one at a time, because of hardware failure, while trying not to have to redo all the wiring is a huge pain in the ass. Im sooooo close to gutting this and redoing back end. Ill also put up what the expected voltages should be.

[Enzo]

Right, the 1.5v is the bias across the driver tube, the -41v is the bias voltage on the power tube grid. They are of course separate things. I agree -41 at the grid is a problem. Would be helpful to see the other side voltages too.

Just to rule things out, measure voltage ACROSS the 50k grid stopper to the lower output tube. Should be zero volts.

Looks to me like you want more current through the driver tube. I get about 15ma with a crude calculation. That would move the -41 about 40 v more positive.

[Dave H]

No, it does work. It's fairly simple.

But not as simple as it could be. Why have the driver stage when you could eliminate the -ve supply and drive the EL84s from the PI in the usual way?

[SoulFetish]

But not as simple as it could be. Why have the driver stage when you could eliminate the -ve supply and drive the EL84s from the PI in the usual way?

True enough. EL84s easily suffer from blocking distortion when overdriven, and I didn't want to implement some of the popular remedies like the zener clipping of tubes in cutoff, etc.
I thought the solution was to decouple the capacitors feeding the grids. The design is simple, but I'm learning that maybe the implementation might not be. I'm using a 12AU7 now, but I don't really need all the drive current provided by such a higm gm tube. I'm not driving the grids into positive voltages, especially if I keep those 50k grid stoppers there, so
I'm wondering if I can tame some of the oscillation by using a lower gm/higher mu tube like a 12AT7, 12AY7, or 12AX7. But first things first, I have to track down the source of the DC bias issues and get those stable.
What's frustrating is that the driver/output tube biasing wasn't a problem at all earlier on.