EL34 heater short to plate - why?

[EFK]

Well I just had my first "real" problem. A new SE amp build, single EL34 with JMP front end, @ 430 plate voltage and fixed bias. Got it all fired up and running well, was testing it flat-out when pffft, sound crapped out, a little puff of smoke and horrible smell. Shut it down quick, and I can;t see anything wrong at all inside the chassis. New Micalex tube socket, the socket or tube arced between pin 2 and pin 3. Nice toasty carbon track on tube and on face of socket. Now: why would it do this? My soldering is very clean, wiring is all very well separated, nothing touching or making intermittent contact etc. I will mention this is an older tube, branded "NAtional Electronics 6CA7" and "Made in West Germany." I believe this means it is a Siemens? Anyway, I'm wondering (1) if this seems more like a situation where the tube itself arced and then burned the socket, or (2) are there operating conditions that I don't know about which may promote arcing between the heater and plate?

Any information or opinion appreciated!

[Tom Phillips]

I think your #1 is most likely what happened.

[hasserl]

Check the impedance of your speaker load. An excessively high impedance speaker load can cause high fly back voltages to occur at the tube plate. With the heater at only 6.3v there is a lot of potential between the plate and the heater, so that's where the arcing occurs. Also can occur with intermittent open in the speaker wiring.

[g1]

Was that National tube new or used? If used, could it have had some carbon on it from a previous arcing event? Or if new, maybe the base was dusty?
hasserl makes a good point, if you don't have flyback diodes on the OT, you might consider them. (edit: diode no good for single ended output, thanks Steve)

[Steve Conner]

You could well be getting excessive plate voltage spikes when you crank it up. But the flyback diode won't do anything in a SE amp, except maybe suffer reverse breakdown and die.

It would be nice to verify the existence of spikes with a scope, but they could well destroy an ordinary 10x probe.

A R-C snubber across the OT primary would be better. This is also known as a conjunctive filter, others on the forum have more experience designing them than me. The resistor should be a cement block type of a resistance roughly equal to the design plate load impedance, and the capacitor a high voltage unit as big as you can go without dulling the tone too much. Probably something like 2200pF.

[trobbins]

MOV-R would be a better design choice - the MOV is a capacitor, until it becomes low resistance.

[Pedro Vecino]

National EL34/6CA7 Made in Germany is RFT (older DDR). They are very good tubes. Sometimes this problem occurs without external reasons (a fault in the tube itself) though of course it is advisable to check the external possibilities.
This is the effect on a Marshall reissue:

[EFK]

For the moment I am going to assume that it was a tube problem, as I can not see nor find any external cause. Admittedly, the tube is one of a pair that I yanked out of an old Selmer I had bought a year or two back, and while they did function just fine at that time and sounded fine, I have no idea of their history nor age. Regarding the output impedance, I am using a higher load than is usually used with EL34 in SE, at 5K into 16 ohm (matched), however this certainly is not *too* high over the usual. I was using the somewhat higher load in order to try to keep the plate current in check, as my plate voltage is higher than usual for SE and I don't want to have to bias too terribly cold.

I've looked into conjunctive filters before, never actually implemented one but maybe I'll give one a go. Going to replace the socket first though! I have some KT77 I can bias into this amp, might try one of those first before I go slapping 'band aids' around.....

[Prairie Dawg]

If you have not already done so I'd take this as your hint to fuse the B+ and the filament circuit. It sounds like the tube shorted out. The reasons may be unclear but you need a little protection of everything else here.

[EFK]

I do have the B+ fused, although oddly enough it never blew. And, more interesting, the tube does still work although I'm sure not trusting it now and won't use it for anything.

Can someone explain *how* a conjunctive filter will protect against flyback? Doing some digging, I've read that it can, but no explanation of why. Is it because it prohibits the reflected impedance from jumping around so much, i.e. climbing so high at certain frequencies?

BTW, can this situation be helped if the filaments are floated at an elevated DC, say @ 50-70V?

[Steve Conner]

Can someone explain *how* a conjunctive filter will protect against flyback? Doing some digging, I've read that it can, but no explanation of why. Is it because it prohibits the reflected impedance from jumping around so much, i.e. climbing so high at certain frequencies?

Pretty much. The load impedance gets higher and higher with increasing frequency, because of voice coil inductance and leakage inductance of the OT. When the power tube is overdriven, it is making a square wave and functioning more like a switch. The tube tries to quickly turn off the current through the stray inductance, and that's what generates the high voltage spikes at the plate. The circuit really isn't that different to the line output stage of a tube TV.

The conjunctive filter provides an alternate path for the high frequency current, stops the impedance rising at high frequencies, snubs the flyback spikes. 3 ways of saying the same thing.

The tube might well be salvageable if you took a Dremel and ground away the burnt part of the base.

The flyback spikes could be over 1500V, so 70V of filament elevation isn't going to make a lot of difference. I'd want the filament circuit solidly grounded, to protect the other tubes in case it decided to arc again.

[Enzo]

You probably will never know why it failed. One possible scenario: the tube had a certain amopunt of dirty film accumulaqted on its base. Not enough to start an arc. SOme tiny bit of crap, some debris, even a tiny spider, manages to get in between pins 2 and 3 and that is all it takes to let a anr form. Once an arc strikes, then it can continue itself without help.

Fuses are not sentient beings, they can only blow when their current rating is exceeded for the period of time it takes to blow. A 1A fuse will run forever with 1A through it. Each fuse type has a chart, how long it takes to blow with what percentage of over-current. 1.2A might take an hour. Meanwhile on the other side of that equation, the arc doesn't necessarily draw all that much excess current. Things can burn up and charcoal without exceeding the fuse limits.

people often are amazed some little half watt resistor burned to a crisp and the fuse never blew. Well, that 3A mains fuse will allow 360 watts (at 120VAC) to pass without even considering blowing. SO unless that half watt resistor managed to draw 360 watts while burning, it was never going to take out the fuse.