Weird fault in tube amp.

[dazz]

Hi everyone.
I have this friend who's built a tube amp (schematic below). It's been working just fine for weeks, until it started developing a crackling noise, then lost most of its gain & volume and sounded tiny and distorted. Turns out apparently there were two separate faults in the amp. C202 was dead shorted, letting lots of DC voltage pass through the tone stack. That was replaced and that issue was resolved.
There's still a puzzling issue though. The cathodyne's grid seems to be drawing far too much current. I attached a picture with the relevant voltages. The R9 grid leak is dropping a whopping 13V, and that seems to be messing up the tube's bias badly. He's checked all the components in the phase inverter and they all seem to be A-OK. He also tried several different tubes and they all do the same thing.

Does anybody have a suggestion for what to try next, please? Could the PCB have turned conductive somehow hence causing the problem? I suggested assembling the cathodyne point 2 point out of the pcb for testing purposes, but a conductive PCB seems unlikely, especially considering there's not much voltage around R9 to begin with. Could this weird fault be related to the shorted C202 in the tone stack?

We're running out of ideas, so any help would be much appreciated.
Thanks.

[TimFox]

What kind of voltmeter did you use to measure the grid voltage on the cathodyne?
With the large resistor values here, this might be loading from the voltmeter when you probed the grid:  the cathode and plate voltages seem reasonable (I don't know what the plate supply voltage was) for a 12AX7 running below 1 mA.

[dazz]

What kind of voltmeter did you use to measure the grid voltage on the cathodyne?
With the large resistor values here, this might be loading from the voltmeter when you probed the grid:  the cathode and plate voltages seem reasonable (I don't know what the plate supply voltage was) for a 12AX7 running below 1 mA.

Sorry, forgot the supply voltage. That's 245V.
The multimeter used was a UNI-T UT139C. Damn, if it's simply loading the grid then we've wasted a bunch of hours tracing something that wasn't even there. Doh!

[TimFox]

I don't know that voltmeter;  most DVMs are 10 Megohm, but some are only 1 Megohm.  You can get a better measurement by (carefully) probing between grid and cathode, ratner than grid and ground.
Keep an oscilloscope on the circuit, however, in case the probing causes oscillation.  In that case, you should put a 1 Megohm between the probe tip and the grid, adjusting the measured voltage accordingly.

[dazz]

I don't know that voltmeter;  most DVMs are 10 Megohm, but some are only 1 Megohm.  You can get a better measurement by (carefully) probing between grid and cathode, ratner than grid and ground.
Keep an oscilloscope on the circuit, however, in case the probing causes oscillation.  In that case, you should put a 1 Megohm between the probe tip and the grid, adjusting the measured voltage accordingly.

A quick google search seems to indicate it is indeed a 10M ohm meter. We'll try measuring the grid-cathode voltage as per your instructions, thanks Tim.

[floobydust]

Why is your plate voltage for V2A 0V? That looks like something's wrong a stage ahead of the phase-inverter.
Shorted C202 (mica?) likely to damage the tone stack potentiometers or V2A (from grid current), depending on the pots' positions.
You could swap V1 and V2 to see if anything changes.

[dazz]

Why is your plate voltage for V2A 0V? That looks like something's wrong a stage ahead of the phase-inverter.
Shorted C202 (mica?) likely to damage the tone stack potentiometers or V2A (from grid current), depending on the pots' positions.
You could swap V1 and V2 to see if anything changes.

Oh, my bad, there are actually some 160V in the plate of V2A. I fixed the picture.
Yes, C202 was a mica cap. The thing is we tried a bunch of different tubes and they all behave the same way.
Maybe there's no problem there if the meter was simply loading the grid, maybe C202 took a pot or some resistor in the tonestack and that's why it still sounds like crap. I'll need to wait for my friend to get back to me once he's figured out the actual cathode-grid voltage.

Do you guys think he should swap those mica cap for ceramics or something?

[andy3055]

If that c202 shorted, those pots the anode voltage would have gone to ground through the pots in the tone control section. You may have a bad control.
Edit: I also like to know why there is no DC blocking from that anode to the tone controls.

[floobydust]

You didn't use cheap eBay or Aliexpress parts? The part spec is 500V and it's seeing under 200V. I've never seen a decent mica capacitor short. But the real deal Cornell Dubilier CD15/CD19 is $4-$8 each nowadays though. The type of capacitor used makes a large difference in the sound, especially in the tone stack.
Capacitor types are part of guitar amp religion and fights happen over bumblebee or sprague etc. I'm not sure which types you used but likely to start a holy war no matter where you bring it up.
Usually you'll put a say 22nF 400V on the output of V1B to block the HV from the tone stack, also it allows using lower voltage boutique capacitors to be used there.
I think you roasted your tonestack pots with C202 shorting.

[dazz]

Thanks, guys. The reason why there's no coupling cap between the second preamp stage and the tonestack, I have no idea. That's simply the way whoever designed the amp decided it should be. I attached the pdf of the project for reference. Apparently it comes from some forum in Brazil.

The guy who built it is not one to skimp on shady parts from Aliexpress or anything like that (that would be me, LOL) so I don't think those are some random el-cheapo caps. Will ask him anyway to make sure the voltage ratings are correct. We'll also focus on the rest of the tonestack and the pots in particular, as per your suggestions. 

[dazz]

Here's a picture of the infamous cap. Rated at 500V.

[andy3055]

Put a .02 uf/400 volt capacitor there to block the DC before the cap that failed-just after the anode.

[dazz]

Update on this issue.
Apparently the meter was indeed loading the grid. Measuring the bias voltage between cathode and grid produced the correct 1.6V.

Turns out the problem was the value of the R9 grid leak, which was supposed to be 56K instead of 560K. The theory being that a lower grid leak value formed a voltage divider with the 470K R10, thus taming the signal that goes in the cathodyne. Otherwise it would do weird things.

[TimFox]

C202, 203, and 204 DC-block the anode voltage of V1B before the pots of the tone stack and the grid of V2A.
Of course, should C202 fail the anode voltage goes across the "T" pot.

[dazz]

C202, 203, and 204 DC-block the anode voltage of V1B before the pots of the tone stack and the grid of V2A.
Of course, should C202 fail the anode voltage goes across the "T" pot.

Yes, I know, Tim. C202 was replaced and all the pots seem to be fine. I guess they were never moved passed the point where the resistance path through all 3 pots was low enough to cook them.

[TimFox]

I was just commenting on the lack of a DC blocking capacitor before the tone stack.

[dazz]

Just for kicks, I run the simulation with C202 shorted and plotted the power consumption at each of the tonestack pots. All three pots sweeped to get all possible combinations. The power never exceeded 100mW in any of the pots