trouble with burning screen resistor

[Luthierwnc]

Hi all, It's been a while.

I pulled an amp out of the closet I hadn't messed with in about fifteen years. It started sounding bad and I never got around to investigating. As it happened, our cat at the time felt the need to personalize it. After I got it cleaned-up, I got it on the bench and found that one of the power tubes screen resistors was fried. A new one did the same.

The amp is a 15-watt EL84 head along the lines of a DC30. The power section is conventional with cathode resistors of 250R on each tube with a 47UF cap across each. Plate volts are about 360VDC. Screen resistors are 100R sand-boxes. I cleaned and checked the socket, tried different tubes, checked solder joints, checked the grids for DC, actually pulled the OT and checked both impedance and resistance between the legs. Those were all fine. The choke has about 135 ohms of resistance. Resistance from the cathodes to ground is within two ohms -- 220k grid-leakers on the PI are the same. 1k5 stoppers.

No matter what I did, the socket still incinerated the one side. Before it got too hot, that side was pulling 50ma and the other only 35. This amp has the EF86 channel feeding one side of the inverter and the TMB cathode-follower on the other so there is no negative feedback from the OT. Screen voltages are within a few volts of the plates (should be 10v lower?). A sine-wave in has a good shape all the way to either side of the power-tube grids although there is some clipping of the top half of the wave. The resistor burns from either channel. With a signal generator and dummy load, you can hear the whine for a few seconds before it gets intermittent and starts smoking. Filter caps seem good.

This has got me stumped. I might switch the transformer hots just to see what the other tube does but it is time to cast about for fresh ideas. If anything comes to mind, I hope you'll share your experience. Thanks, Skip

[sluckey]

Does the screen resistor get warm with the tube removed? Any sign of arcing on that socket? Anything connected to pins 1, 6, or 8?

[Luthierwnc]

Hi Slucky, Nope. none of the extra pins are used for tie points. I usually lick a chopstick and leave a glob on the resistor. With no tubes it doesn't evaporate quickly. With a tube it only takes about 15 seconds if they are already warm. sh

PS OT is ultralinear with the between taps taped-off. Resistance between all primary legs is uniform and consistent with about a 43% tap. Impedance tests show about 5k5 for both the 4 and 8 ohm taps.

[martin manning]

Any sign of a HF oscillation?

[Phil_S]

Look for a short to ground.
-Use a magnifying glass to look for a whisker.
-Use your meter between each leg of the resistor and the chassis.
-If you get continuity from a leg to a chassis, then you have narrowed the path to search...one side of that R or the other.
Have you tried different EL84's? Maybe an internal short?

[sluckey]

I know you said your grid resistors were good. Now check pin 2 resistance to ground. This will verify that the resistors are actually providing a path to ground for the grid.

[Luthierwnc]

Thanks gents. resistance to ground from the grid is 185k on both pins. No whiskers (been rewired several times). I've tried four tubes in that socket. No continuity to ground from either side of the screen resistors (tubes out).

However, I left the supposedly good tube in and now that resistor is burning too with very little signal. The pic is after the 220k grid resistor and PI coupling junction (tubes out). Both sides look the same. Although, there has only been one or two volts of difference between the pins 7 and 9. It should be more like 12v. Martin, I wondered about a HF parasitic. Now that it is on both sides, it's not the socket. sh

[SoulFetish]

You may not see HF oscillation at the grid. Can you get a shot of a 1kHz sine wave output into a load?
Also, are the screen resistors failing under idle conditions with no applied signal? What is the power rating of the resistors?
Do you have a way of measuring the amps idle current draw at the mains side of the PT?

[Roe]

H...
The amp is a 15-watt EL84 head along the lines of a DC30. The power section is conventional with cathode resistors of 250R ...The choke has about 135 ohms of resistance.

Too little screen resistance. With 100R individual screen resistors, you need at least 840 ohms resistance between the mains and screens supplies with normal voltages around 320-330v.

On the ac30, vox went from a 420R choke to 680R shared screen resistors (in addition to the 100R individual screen resistors)

[Luthierwnc]

SoulFetish; Here's a one-handed shot of a 1kHz signal on the 8 ohm tap with a 6.5 ohm dummy load. I had to do this fast. The resistors don't fail with no signal. The MV was dimed. It is the type that shorts the power tube grids with a 1ma pot.

Roe; I'm not married to the choke. I can put a big 1k resistor in its place (or in series). IIRC, the 2009 Clubman schematic does that. They don't fail with no signal. I have run out of 100R and 130R resistors over 3 watts so more experimenting means an order somewhere. The ones in there still work but they will have to be replaced before I'll put the chassis back in the cab.

Thanks, sh

PS I suppose I could use the UL taps too. sh

[SoulFetish]

Well... um. A couple of things..
It's helpful to adjust the time base to get a better look at the waveform. But, here's what I can see from that shot:
First, I suspect that one of your resistors may have failed open again. It looks like only one of the output tubes is conducting. I would swap the two tubes around, switching sides and see if the waveform looks the same (ie the top of the waveform in the display). If when you swap positions, the waveform switches too (where the peaks are at the bottom), then it is a bad tube.
The other thing is it looks like there is some High Frequency oscillation happening (potentially). It's tough to see in the pic, but it appears that there is some nasty distortion on the ascending and descending slopes here. It isn't super clear, but I would power down, remove the tubes, and measure the resistance from the centertap of the output transformer to chassis, and secondary. Do you get any reading?
Any chance can you put a some meter on the primary and get an measure of the amps idle current draw off the wall?
Also, a 5k5 p-p load impedance can be fine for four EL84s (you would want to double that ~ 8-11k in a two EL84 15-18W amp), and a higher load impedance is probably preferable with a higher plate voltage. However, the important thing to consider is that it allows a larger voltage swing on the plates. If you look at the characteristic curves, you'll notice that as the plate voltage swings to it's most negative point, the screen current dramatically rises. This puts a lot of stress on the screens, so I'd agree with the others that 100Ω screen resistors is really inadequate for your design and should be at least 1k to protect the screens and maintain stability.

edit: I amended the primary load impedance statement.

[martin manning]

What is the OT primary impedance? 8k would be typical for EL84's at these voltages.

[SoulFetish]

oh, yeah. martin's right. 5k5 would be for a 30W with 4 x EL84s (2 each side). Sorry, it was early in the AM for me.
5k5 for 2 EL84s could very easily cause excessive plate current. I'm going to correct my previous post...

[Luthierwnc]

There is no continuity to ground from the center tap. I've tried 5 different tubes in both slots. It isn't stable enough to get a good scope display before the resistors start smoking. The first bad side was pulling 50ma versus 35 at idle. For a few seconds it looked like my first shot.

You ought to know that this was a really good sounding little amp before its demise. I can't rule-out the OT was damaged. I did repaint it. It seems to test fine but I can't rule-out any of the suspects. Next I'll try subbing a stout 1k resistor for the choke and see what that gets me. Then I'll try the UL. Might be time for another OT but at that point, it might also be time to bail. sh

[Phil_S]

It may be time for the super secret transformer test. See article by RG:
https://www.premierguitar.com/articles/ ... r-tester-1