Orange OR120

[Ted]

Just had a 70s Orange OR120 returned to me after I replaced the valves.
The guy told me it was blowing mains fuses.

I replaced the fuse, plugged it in and waited. Nothing. It seemed to work fine.

On prodding around a little I found the quiescent current on one side had dropped to around half what I originally set it to. (They're JJ E34Ls - good for 30W) I'd biased them to around 20W per valve - Oranges like to be biased quite hot. Swapped the pairs over and the low current followed the valves...

Its sat there now - no fuses blowing, just two valves which appeared to have failed very early. A single I'd write off as a dud. A pair on one side suggests something else.

Any ideas?

[R ski]

When you removed the blown fuse, was it a glass style, if so, then was the glass extremely blacken, that would indicate a hard interrupt of that fuse. On the other hand if the blown fuse was opened up mildly, then the interrupt wasn't so drastic.

Waiting for the DC to drain from the capacitors, measure the resistance of the output transformers primaries between center tap, this value has about 10% tolerance, no need to be alarmed.

Your biased value can also be measured across the center tap to either primary leg, they should correspond with the measured resistance values, if not, the primaries of the output transformer is breaking down from the plate potential.

Sometimes tubes just crap out, mostly the power tubes, if one failed, that is why the fuse interrupted, buy a single one if that one has crapped out.

[Ted]

Got it.

Sat it on the bench left it running for a couple of hours. Nothing. All output stage voltages normal.
Ran it up to a measured 120W into a dummy load. Nothing.

Put it onto a cabinet, plugged a bass in - after two minutes playing both valves on one side red-plated.

Measured some voltages - bias on the hot side was -6V on the cold side -37V.
All resistors and coupling caps good. I poked around a bit more and the solder on one of the 1K screen grid droppers crumbled...

Soldered it back. Problem solved.

Can anyone suggest what was happening here? Looks as though a floating screen grid connection to the EL34 was causing it to start drawing current from the control grid, collapsing the bias supply on that side.
But why? I've seen enough amps in the past whose screen grid droppers have failed without any effect other than a drop in power.

[Steve Conner]

This makes no sense, a bad connection on a screen resistor should just cut the power. It shouldn't make the tube draw grid current.

Possible explanations:

#1 The 1k resistors you were poking were grid stoppers, not screen droppers.

#2 The Orange uses two tubes in parallel on each side, and only one was working because the other had an open screen connection. The working tube would be overloaded, which would cause it to redplate. JJs are known to be vulnerable to thermal runaway, so it might then run away and drag the bias down.

This doesn't hold much water, as then only one tube would be redplating, not two, unless the screen connection magically restored itself about the same time as you looked in the back.

#3 The tubes just run away because they're JJs, and none of the hypothetical bad connections had anything to do with it. (What value of bias feed resistors does the amp use: 47k, 100k or 220k? 220k is marginal even for one pair of NOS EL34s, let alone two pairs of hot biased JJs. Have they drifted up with age?)

[Ted]

This makes no sense, a bad connection on a screen resistor should just cut the power. It shouldn't make the tube draw grid current.

Which is why i'm posting here.

Possible explanations:

#1 The 1k resistors you were poking were grid stoppers, not screen droppers.

Senile and short-sighted though I am, I can tell the difference between a 1K 5W resistor and a 22K 0.5W

#2 The Orange uses two tubes in parallel on each side, and only one was working because the other had an open screen connection. The working tube would be overloaded, which would cause it to redplate. JJs are known to be vulnerable to thermal runaway, so it might then run away and drag the bias down.

This doesn't hold much water, as then only one tube would be redplating, not two, unless the screen connection magically restored itself about the same time as you looked in the back.

So if they were going to redplate why didn't they do it when I ran the amp at 120W into a dummy load? A few watts into a speaker which caused a bit of vibration seemed to cause the effect.

#3 The tubes just run away because they're JJs, and none of the hypothetical bad connections had anything to do with it.

This was a real bad connection. Resoldering it seems to have fixed things.

I must admit that its possible that one valve ran away, taking down the bias supply and which then casued the other one to do the same...

(What value of bias feed resistors does the amp use: 47k, 100k or 220k? 220k is marginal even for one pair of NOS EL34s, let alone two pairs of hot biased JJs. Have they drifted up with age?)

They are 220K - measure around 230K. Could you explain what difference changing their value might make?

[Steve Conner]

The higher the value of the bias feed resistors, the more susceptible the tubes will be to thermal runaway.

(Because the higher resistance makes it easier for them to "drag the bias down" with their grid leakage current, which is the cause of runaway.)

[pdf64]

I agree with Steve, probably loose elements within the tube, vibration caused it to short to something it shouldn't have, red plating the tube and causing the screen grid resistor to get v hot and the solder joint to it go bad.
I've had a TAD 6L6WGC do something similar.
Max permissible grid feed resistances are specced in the data sheets, usually 220k for grid bias, 470k for cathode bias. If the grid supply voltage is from too high an impedance source, the grid current from a marginal tube can overcome it and proper bias to it is lost, as your measurements show. Peter.

[Ted]

I agree with Steve, probably loose elements within the tube, vibration caused it to short to something it shouldn't have, red plating the tube and causing the screen grid resistor to get v hot and the solder joint to it go bad.

Sounds convincing. But why won't it redplate now I've soldered that screen grid resistor back into place?

[Steve Conner]

Probably just coincidence.

[Ted]

Probably just coincidence.

Hmm. Like Voltaire said:

"What we call chance can be no other than the unknown cause of a known effect"

[Enzo]

And there is also the logical fallacy post hoc, ergo propter hoc. WHich points us back towards coincidence.

[Ted]

I'm just always cautious about writing anything off as chance. I've written off as coincidence what later turned out to be vital clues so many times.

Anyway thanks for your help everyone.

[Steve Conner]

Well, so have I. But I've also chased "vital clues" that turned out to be coincidence. Cause and effect is so 19th century.

It's human nature to forget those occasions, though, and only remember the lines of investigation that actually came up with something.

[Ted]

Cause and effect is so 19th century.

Indeed. I think Sherlock Holmes had the last word on the repair of old amps:

"When you have eliminated the impossible, whatever remains, however improbable, must be the truth"