One socket causing tubes to "run away"

[MitchK]

Hey guys and gals and everyone in between!

I've got my 5150 here, I'm currently running two kt88 tube in the amp and I'm experiencing an odd problem.

One of my tubes is running away no me, I can run the amp for about 20 minutes or so but then just one of the tubes seem to overheat and crap out and the plates glow red.

Now at first I figured I just figured the tubes life was up kinda thing and I replaced the set. It is still happening though. It didn't start happening right away with the new tubes, but after about 3 months or so.

A couple things, I do have the tubes biased hotter than stock but they are still fairly low about half way between stock and the max for kt88 tubes (I can't remember the specific numbers I just remember from when I was biasing it initially). I am just running the 2 tubes not a full quad.

I have a feeling this might be caused due to the tubes over heating because I had re-engineered the inside of the chassis and I don't think the circulation of air was very good for the tubes, so I've now installed a fan.

If heat isn't the issue though does anyone have any ideas as to what would cause this to happen?

[Enzo]

OK, two tubes. The first time this happened, new tubes cured it for three months. Now it is doing it again, why can't it be a bad tube again?

Swap places with the two tubes. Does the other tube now overheat in the same socket, or does the heat move with the tube? That will tell you if it is tube or socket.

When this red-plating occurs, monitor the bias VOLTAGE at pin 5 on the power tube. If it drops, then that is why the tube is heating. If it stays up where it ought to be, then the tube is suspect.

If the bias voltage is low, then either there is a leaky coupling cap from the PI tube, or one of the resistors that feed the bias to the tubes has an issue.

[MitchK]

Hey sorry I knew I was forgetting something when I wrote that up. I have swapped tubes and it is the socket I swap the tube and the tube that was red plating is fine in the new socket and the tube that was ok is now red plating.

If I just let the amp sit idle the red plating will not happen it only happens when I'm playing and it happens faster the louder I'm playing.

So should I just hoop my meter up to the amp at next band practice and wait for it to start red plating and see what happens to the bias voltage?

As for the leaky cap or resistor issues, can you recommend how I'd go about trouble shooting which it is, or would be?

[Enzo]

If the cap is leaky, it allows +DC from the PI onto the output tube grids, thus trying to cancel the negative bias. I check the resistors for value wuth the amp cold, no power.

I am talking about the voltge on the grid right at the power tube.

And make sure pin 4 of each power tube has B+ voltage.

[Steve Conner]

If it drops, then that is why the tube is heating. If it stays up where it ought to be, then the tube is suspect.

In my experience, it doesn't work that way. If a bad tube is leaking too much grid current, it will drag the bias voltage down just the same as a leaky coupling cap would. That's how thermal runaway gets started, and Mitch's problem sounds like thermal runaway. The tube draws more grid current as it gets hotter, and the extra grid current drags the bias voltage down, making the tube draw more plate current and heat up all of its guts (including the control grid) yet more, which increases the grid current leakage and drags the bias further down... etc until the tube is almost fully turned on with a red glowing plate.

I've watched it happen in an amp head that would work fine when stood the right way up, but redplate whichever tube was uppermost when stood on its end. Both tubes were on the verge of runaway in this amp, and the rising hot air made the upper tube hotter than the lower one and touched it off first. The problem appeared to stay with the uppermost socket, so I was looking for bad grid leak resistors and coupling caps and finding nothing, till I stood the amp on its other end. :P (Does the socket that's running away in your amp have worse air circulation?)

Tubes tend to leak more grid current as they get older, and KT88s probably leak a lot more than 6L6s to begin with. Changing the grid leak resistors from 220k to 100k might be an option worth trying.

[Amp Kat]

Which socket is it ? That last socket I think all the way to the left from the OT connector is not grounded on pin one. Although unlike an EL-34 which uses this pin 1 as it's suppresor grid there could be some correlation to the runnaway. Also check your screen resistor and make sure you got good continuity there. KT-88's don't like high screen resistors and will go into runnaway if it's to high. If the coupling cap is leaking DC both tubes on that side would go into runnaway as they are both on that leg. You can monitor the bias by putting your red meter lead in the empty pin 5 socket on the same side as the runnaway tube and the black lead to chassis ground and monitor the negative grid voltage when it starts redplating. Just be careful around pins 3 & 4 as there is about 450 on those pins.

[MitchK]

Hey steve, I'd actually say that the tube that is running away has the better circulation, but I've changed things now and they should both be good for air. Can you tell me a little more changing the grid leak resistors? I can do it but they are in an awkward spot on the pcb, what will this do, whats the idea behind it?

ampkat, its actually the socket that sits closest to the OT thats acting up. As for the screen resistors, they are stock 100 ohms, I was told I should try replacing them with 1k which is what I've done. But you say they should be lower for kt88s? why is this?

[Rob Mercure]

Hey KB,

I've been monitoring this thread for the heck of it and I also am curious as to why you find too high a screen resistor a problem with the KT-88. While I've heard lotsa tales about the effect on the sound I would actually suspect that the higher resistance further prevents oscillation and generally makes the tube run cooler due to greater voltage drop across the resistor - in fact I like "slightly high" g2 resistors for the "safety" effect (but no KT-88 experiences). Not being contentious here - just wonder about your thoughts/experience with these.

Rob

[Steve Conner]

Changing the grid leak resistors to a lower value lets them soak up more grid leakage current before the bias voltage gets affected too badly. Designers always make these as big as they think they can get away with, because they soak up precious signal from the PI.

I think even 100k is borderline for KT88s, once they get even somewhat worn out. (The leakage current increases with age as the grid gets contaminated with cathode material and the tube gets gassier.) Some premium tubes have a gold-plated grid which might help.

By grid leak resistors, I mean the ones that go from the tube grid to the bias supply. I don't know if grid leak is the correct term.

I don't believe changing the screen resistors will make a bit of difference. I bought a bunch of used Sovtek 6550s off fleabay (from some hi-fi weenie who was "upgrading" to Svetlanas) to use as cannon fodder for testing my 300PS clone project. While testing them all, I could get one of the worn-out ones well on the way to thermal runaway even with a 1k screen resistor, a 100k grid leak, and fan cooling.

[MitchK]

Just to clarify which ones are we talking about?

[Amp Kat]

Mitch I think you would be fine with 1k Screens but not sure the 88's would be bothered with the 100 ohmers as Steve mentioned.If you decide to use EL-34's then you would need the 1K's so changing them to 1K is not a bad idea. I was thinking in the upper amounts of 75k or more could cause runnaway but not that low.

Rob,I used to have a big Screen manual that had everything about a screen you ever wanted to know and more but it went with my hardrive on my old computer and I can't find it but, from what I remember is when the impedance gets to a point of swamping as far as a ratio it causes the screen voltage to become positive with respect to the Anode and the plate impedance also starts to come down. The screen gets invisible and secondary emission shoots up and causes electrons to flow,bounce or however you want to call it from the cathode to the anode in uncontrollable amounts. Tube heats up and away it goes. I've also read this several times in some Ham stuff but I can't actually say I've tried it in practise or seen it but I may tomorrow.

[Enzo]

You are right, Steve, that a bad tube can also kill the bias, but I would just yank the tube from that socket and see if the bias voltage is there. If the cap is leaky, the bias will still be low from contention. SOmetimes my entire mental process doesn't show...

[Steve Conner]

R54 and R55, but also check the two coupling caps for leaks like Enzo and others suggested.