Burning Screen Resistor

[bal704]

I'm building a JTM45 clone. I'm using 2 matched EL34 tubes, with a 470 ohm, 2 watt resistor going from pin 6 to pin 4 on each tube. When I first powered it on, doing my 'smoke test', it made noise and sounded OK, but it had smoke. I appeared that my screen resistor on my first tube, being fed directly by the B+, was smoking. The second tube has a wire running from pin 6 of tube 1 to pin 6 of tube 2. After replacing the resistor, I measured the voltage drop across the resistor (from pin 6 to pin 4), and found at least a 40 volt drop on tube 1, and a 12 volt drop on the tub 2. I say at least, because I powered down the amp as it hit 40 volts. That's almost 4 watts on a 2 watt resistor. The bias voltages are reading high, but I'd like to get the smoke problem cleared up before I work that issue.

I have 2 sets of matched EL34's, and they both do the same thing. I have the amp wired up to accept either EL34's or 6L6's, so I put a matched set of 6L6's into the amp, which resulted in 33 volts across the first screen resistor, 7 volts across the second.

I checked the socket wiring under a magnifying glass trying to find any stray wires etc that might be touching an adjacent pin. I checked the connections on both tubes and everything seems to be connected to what it's supposed to be.

Any tips on why there's such a huge difference in voltages on the different tube sockets?

[pompeiisneaks]

One thing could be a bad socket. make sure there's also no resistance between the bias input and the side that's not running as hot. You may have right general bias for one side but it has 'extra' resistance due to some connection etc. Then it causes the other tube to run hotter and conduct too much?

Not sure though, Photos may help.

~Phil

[bal704]

no resistance between the bias input and the side that's not running as hot

Not following that.....

[nworbetan]

If either of the bias feeds have a significant amount of extra resistance, then the bias voltage difference may be the cause.

[tubeswell]

Make sure you have a decent negative voltage on all of the tube socket control grid pins (Pin 5) before you plug any output tubes in

[bal704]

I'll check those things. I've already eye-balled it and didn't see any issues, but I'm going to reflow all the solder joints on the PI and power tubes to see if that fixes anything.

[nworbetan]

A leaky coupling cap from the PI is a suspect too.

[Andy Le Blanc]

I would set the bias a little too high (cold) and check static voltage conditions. First without the power tubes installed, looking for odd voltages, and then with the power tubes... could be a physical or component issue but it could also be a poor matching of the power tubes, bad tube or even a lousy OPT.

[Roe]

470r is too low with el34s

[bal704]

How do you check for a leaky coupling cap?

A leaky coupling cap from the PI is a suspect too.

[martin manning]

Power it up without the power tubes installed and measure voltage across the 220k bias feed resistors (from the junction of the 220k's to the grid ends). It should be a few millivolts or less. If it's more than that there is DC leaking through the PI coupling cap.

[pompeiisneaks]

If either of the bias feeds have a significant amount of extra resistance, then the bias voltage difference may be the cause.

Exactly what I was trying to say. Sorry if it's not clear.

[pompeiisneaks]

I'll check those things. I've already eye-balled it and didn't see any issues, but I'm going to reflow all the solder joints on the PI and power tubes to see if that fixes anything.

Just reflowing solder joints won't specifically help unless there's a known bad soldered joint. You can easily test with a continuity mode on the DMM between junctions so you see if there's low/no resistance.

~Phil

[bal704]

Power it up without the power tubes installed and measure voltage across the 220k bias feed resistors (from the junction of the 220k's to the grid ends). It should be a few millivolts or less. If it's more than that there is DC leaking through the PI coupling cap.

I get 0 volts measuring from the junction of the 220k resistors to either grid end. no tubes installed.

I did find one bad solder joint on the PI, where the 10k/0.1uf junction goes to the presence pot. Fixed it.

With no power tubes, I'm getting very little bias voltage. It ranges from -0.005 volts to -0.015 volts at each extreme of the bias adjust pot.

With no power tubes, I'm getting about -.005 to -.010 V on pin 5 of each tube. With the power tubes installed, the negative voltage on the 'bad' tube pin 5 is much more negative than the negative voltage on the 'good' tube.

I ohmed out the path from the 220k bias junction to pin 5 of each tube, and they both checked fine.

[pompeiisneaks]

That sounds like the bias tap or bias circuitry itself is bad. Test all the way back to the output of the bias right after the diode to see what level of DC negative voltage you get there, nad then trace it forwards until you see where it's going. A little googling seems to show you should have somewhere in the range of -50VDC, so I'd make sure you have that first. millivolts is a clear sign something's wrong. (I wouldn't consider that 'scripture' as I'm just googling for JTM45 negative bias and they could be wrong, but I've seen bias ranges from like -15 to -60 or so depending on tubes so it 'seems' in the ballpark for me).

~Phil