Mystery Problem

Suppose you didn't actually use a bench DVM with one input connected to mains earth?

I'm using purely battery-powered DVMs for this project. I don't even have a bench DVM. Perhaps I'll try a scope instead and see how it reacts to that.

It is a mystery then. Loose connection somewhere?

If the circuit has any sort of gain, are you creating a feedback path with noise radiating cables that oscillates and overwhelms the power delivery of the supply? Try the meter switched to AC, or an oscilloscope, you may still see something.

I only wish I could. The AVO has thus declined to generate any more heater supply. This *has* happened before. I suspect there's a thermal cutout which operates when an overload condition arises. This will be some primitive mechanical device no doubt (we're talking early 1940s here). Effective, but it takes an *age* to reset and I don't have an age to F about with it. I did check the DVM though, just to ensure it wasn't faulty and drawing too much current. It was fine, however. Even on a sig gen with 600 ohm impedance it didn't detectably reduce the amplitude of the test waveform at 50Hz. So it remains a mystery I'm afraid - and likely to continue to be so - at least until someone more gifted than I in 80 year old gear unravels it. :(

Oh - by the way, the meter was switched to AC the whole time (and it read the anode voltages accurately, too). That valve tester could conceivably be repurposed as a damp detector, if it's *that* sensitive to additional loading! :-D

I'm a bit confused here. You say that you set up the panels to test *a* valve, but then you refer to the *valves* glowing like they should. Did you test more than one valve, and did these use the same or different sockets - eg only octal, or octal and noval? I'm also a bit surprised that there is space to put a DVM probe on a pin when the valve is in its socket, unless the valve is partially loose. That doesn't explain what you found, but perhaps the socket is cracked and the pin connector is shorting out somehow. What happens if you put a resistor (which would draw a similar current to the valve filament) in the socket instead of the valve, and monitor the voltage across that with the DVM?

I assume, by the way, you've seen the large number of threads here:

Thank you, Jeff. You're a long-standing and respected contributor here and I value your advice. These testers were made in a different age altogether and under wartime constraints. They would be totally illegal to sell today as retail for their lack of basic safety features. I was easily able to get prods on the valve pins by partially unplugging the valve base adaptor. I can't recall the base type for these valves but the test valves I used for this purpose were all common small-signal tubes like the ECC85 and others with the same pin-outs. The issue was particularly slippery to deal with. Voltages were intermittent (in addition to the delays due to the thermal cut-out) and I just cannot waste any more time on it, I'm sorry to say.

Of course, the _last_ thing you would suspect is the DVM.....

Seems to me it's the force of your probes on the pins of a partways inserted tube that breaks the circuit.

Jeroen Belleman

Nice theory and certainly credible. However, I get noticeable sparks on the probe tips just on contact alone with no pressure. It's just weird; I've never seen sparks when probing something for its voltage level (and yes - the meter *was* set to AC volts).

You checked the DVM was working correctly, I suppose. Did you measure the voltage from heater to GND? What's the cathode voltage? Are the valves of the indirectly heated kind?

Jeroen Belleman

Yes, it doesn't behave that way when measuring anything else. For instance, the anode voltages go up to 300VAC (yeah AC as well!) and measure as expected with not a spark in sight. Yes, they're indirectly heated double triodes each with its own filament running in series with a center tapped to ground. I've put the damn thing away now and have no intention of wasting any more time on it. Someone else can have a stab at it; maybe they'll have better luck.

Were the plugs in the 10 Amp sockets on the meter, by any chance?

That was my initial thought, too. But no, everything was where it should have been. Clearly the bloody thing's possessed.