Weird problem or crappy PT?

[Chuck H]

That sucks!

It occurred to me later on, while in a meeting, that you wouldn't need a "Champ" OT to tell if it's the problem. If you wire up some generic 4KΩ 40W OT and get more than a watt or two out of it, that should be evidence enough, right?

I considered that. Maybe I'll try it. Maybe not. There's nothing about the circuit that should cause the distortion rising with frequency affect. Other than the failed zener circuit, which was just an experiment and is out now, it's all testing as it should and I can't see any reason not to pin this on the OT.

[g1]

I'll take that to mean you have verified that the signal getting to the power tubes is good and consistent with the frequency increase.

[pdf64]

.. trying a 12v/5W zener between the screen filter node and the screen stopper resistor. This seemed to work fine, until I drove the circuit into clipping. Then the plate current dropped and the screen current increased dramatically turning the 1k screen grid resistor a nice golden brown. Why did this happen?

I think that it would be normal for the main HT voltage to rise when a class A / SE power tube is pushed into clipping, due to the regular bias excursion mechanism reducing the average plate current draw, thereby reducing HT sag.
I don't see why that would cause the screen grid to draw more current though?
Or why the zener diode would make things any worse?

Are you sure this doesn't happen without the zener in place?

[Chuck H]

I'll take that to mean you have verified that the signal getting to the power tubes is good and consistent with the frequency increase.

Nope. I sure didn't. But I will now.

Like Enzo sez... Never invent a reason to NOT test something.

[Chuck H]

Are you sure this doesn't happen without the zener in place?

Certain. I ran the same tests before trying the zener and then again after I removed it just to be sure.

[Enzo]

When we need a simple OT test of the go/no-go type, then it matters little what we sub in place. Who cares if the impedance matches, all we want to know is if the OT was the problem or not. Single ended OTs need a gap, but for a quick test, it doesn't matter. I have an old Fender Bassman OT I use for just everything, including using half of the primary for a quick single ended test.

[Chuck H]

Thank you Enzo and g1 for inspiring the testing and OT substitution. More on results in a sec.

And my apologies to pdf64. I was "certain" that the screen overcurrent during clipping was caused by the zener. That was wrong. The screen overcurrent during clipping does happen without the zener as well. I must not have been paying attention to the right things at the right time during testing. I'm pretty embarrassed to be "that guy" already so please be gentle.

I'm here with my hat in my hand. This problem is beyond me. Thank you to those who have been kind enough to follow along. And if you're still with me I have some test results to share. I'll try to be concise and brief.

This is a schematic of how the power supply has been modified. My intent was to reduce hum. There is no hum at this time.

*NOTE: I have changed the HV rail resistor feeding the screen and preamp from 1K (as it has been in the previous discussion) to 5k which is keeping the screen voltage well below the plate.



Performance remains poor. With a 1kHz signal @ 70mV I get roughly 5W with the wave form just barely clipping the cutoff (figure a). However, compression on saturation and then an HF peaking rise steadily as the frequency is increased. With a 5kHz signal it is quite pronounced (figure b). Actual clean output with a frequency sweep is limited to about 1W.

That is problem #1. Perhaps/likely related to problem #2. Under overdrive conditions with a 1kHz signal there is a steady and dramatic increase in the spike to easily twice the cutoff or saturation levels (figure c). This is actually much worse as clipping or frequency is increased. As clipping is increased from 'a little' to 'a lot' there is an analogous decrease in current at the plate of the el34 from 50mA to 42mA and a dramatic increase in screen current from 3mA to 25mA.



The preamp signal is clean at the power tube grid to well beyond where the problems manifest. Full preamp output shows some clipping (not a lot) on the bottom of the wave form and no indication of oscillation.

Substituting the power tube for both a 6v6 and another el34 made no difference. Substituting the OT made no difference.

The amp was pitifully thin and quiet before modification but no extensive testing was done. The modifications were to upgrade the speaker and the power supply as shown. It's fuller sounding now, but not a lot louder. Not as loud as I know it should be and suffering the dangerous operating conditions under clipping.

I'm open to run any requested tests just as if I were a total noob.

I have a hair brained theory about what might be happening. I'm putting in the next post to keep this one on point and spare anyone that doesn't want to go there.

TIA gents

[Chuck H]

This is such a simple circuit that I'm baffled. There are only a couple of things I can think of. Perhaps with the higher than average resistance on the HV winding adding the 100R resistor and additional capacitance to the power supply is slowing it down even more and choking the tube? and/or

The tube is biased cooler than 100% at about 19W. Not sure how this might relate since the wave form does clip in saturation before cutoff with an input signal above 1kHz. and/or

The dreaded conductive circuit board. I can read .5V to 3V at any point on the board by stabbing my probe into it. The preamp cathode bias looks about right though at 1.65V and 1.75V even though I can read fully 1V on the board near those eyelets. Perhaps a conductive condition is worsening or modulating when the amp is conducting signal? I haven't checked for that yet.

That's all my (perhaps useless) observations, aside from actual test results.

[g1]

This amp has no NFB, correct?
Any chance of extreme HF oscillation you are not seeing on scope?

[Chuck H]

This amp does have NFB but testing with and without showed no change in the problem. I will get a good close look at the preamp and power amp wave forms and report back.

[eschertron]

I was "certain" that the screen overcurrent during clipping was caused by the zener. That was wrong. The screen overcurrent during clipping does happen without the zener as well. [snip] Under overdrive conditions with a 1kHz signal there is a steady and dramatic increase in the spike to easily twice the cutoff or saturation levels (figure c). This is actually much worse as clipping or frequency is increased. As clipping is increased from 'a little' to 'a lot' there is an analogous decrease in current at the plate of the el34 from 50mA to 42mA and a dramatic increase in screen current from 3mA to 25mA.

The rise in screen current and the 'hook-shaped' waveform are consistent with what I've seen on champs (and any other SE amp, I presume). Part of their charm, I suppose.
I'm curious about the waveform 'B' and what signal voltages you had going in and out of the power tube when the 5kHz signal started to clip. The drawing sure looks like cutoff to me, but that can't happen without full power from the tube (or some outrageous bias shift!). There is something strange going on there at the higher frequencies. Looking forward to seeing that you found it!

[Chuck H]

As small, relaxed wiggle on the saturation half of the wave form when clipping (other than the hf spike). Nothing I wouldn't expect from any guitar amp. A close up look at the preamp and power amp shows no doubling of the trace or hair at the peaks or anything like that.

[ThermionicScott]

Wow, still a problem with other OTs... I was so sure the OT would be the culprit!

I like Enzo's heat gun/hairdryer solution if you suspect stray currents on the board -- might be worth a try.

Have you tried a ~1.5K grid-stopper on the output tube? That 330pF cap between the grid and cathode is supposed to shunt HF garbage, but I have a growing preference for stopping resistors rather than capacitors for this job. Heck, if the cap is still there, you might try removing it and seeing how the amp behaves.

[Chuck H]

Wow, still a problem with other OTs... I was so sure the OT would be the culprit!

I like Enzo's heat gun/hairdryer solution if you suspect stray currents on the board -- might be worth a try.

Have you tried a ~1.5K grid-stopper on the output tube? That 330pF cap between the grid and cathode is supposed to shunt HF garbage, but I have a growing preference for stopping resistors rather than capacitors for this job. Heck, if the cap is still there, you might try removing it and seeing how the amp behaves.

I might try removing that grid cathode cap. It's actually a grid to ground cap now as I did a little rearranging on the board. And I did add a 4.7k grid stop resistor to the power tube.

As to the possibility of a conductive circuit board... I've heard that removing moisture can fix it, but I haven't seen that. I've heated and heated boards in the past for many minutes and all the wax was glossy for a half hour during and after the work (which does make the finish look better once it cools) with no resolution. I'm very glad it works some of the time for some people though. Since the preamp signal and bias voltages look ok and I didn't read any DC on the power tube grid I'm actually wondering what the affect might be of the voltages I can read on the board.?. My next test will be to float the power tube cathode connection to help eliminate the possibility that there is something about modulating voltages on the board pissing in my Cheerios.

[Chuck H]

Oh! Hey! I just reached 10,000 posts at 8:40am on #40. Can I get a WOOT! WOOT!