AZ31 failure - cause?

[dryad1313]

I have a Pye 15a set which I restored some years ago. This morning, after using it for about an hour, the sound went, accompanied by a short fizz/crackle. There followed a burning smell, so obviously I switched off. At first I was concerned the mains transformer had shorted, but it still puts out the correct voltages. After further investigation it turns out a filament in the AZ31 rectifier valve had burned out and fractured into pieces. As I have never had this happen before with such a valve, and because of the strong burning smell, I am concerned that there is another problem which caused it to fail in this way. I checked the 8/16uf smoothing electrolytic with my Peak Atlas ESR and it tests fine (albeit with an ESR of slightly less than 1 ohm) - even though it is quite an old Radiospares one. All the other electrolytics have been replaced. I couldn't see any components which look like they overheated, so what was burning? Could the fractured filament have been shorting out with an anode and started to fry the transformer?!

[Radio Tech]

Hi dryad1313

What I would suggest is replace the rectifier, disconnect the HT rail from the smoothing cap, switch the set on and monitor the HT on each side of the 8/16, if the voltage is there and not drifting down then the fault is further on, but if the HT is falling off or even struggling then the fault lays with the capacitor. As the cap is an old RS spares on I would be inclined to replace it with say a 10uf and a 15uf axial type underneath the chassis or find a dual capacitor of similar size and re stuff with axial types. Hope this helps

best wishes

Ken

[Sideband]

It can happen on directly heated rectifiers. The filament is also the cathode so is at full HT (that's why it's fed from a separate winding on the mains transformer). If one filament breaks (most likely due to age or mechanical vibration) the HT shorts to anode via the broken filament (the anode is at a high AC voltage) and as the anodes of the rectifier are connected to a centre-tapped AC winding (in which the centre-tap is connected to chassis) you have a situation which causes the HT to be shorted to chassis via the transformer winding causing rather bright blue sparks and flashing inside the rectifier which further destroys the filamant.

Hopefully your transformer has survived. It's unlikely anything caused the problem other than an ageing rectifier but you would be well advised to double-check/change the smoothing can.

I've had a couple of AZ31's with offset filaments where they have almost been touching the anode. I put it down to age and mechanical damage (vibration).

[GeoffK]

The HT smoothing resistor between the reservoir and smoothing capacitor may have been overheating if the smoothing cap has gone leaky.

[dryad1313]

This is all very useful advice and very much appreciated. I'll order a NOS AZ31 and do as you suggest, then report back.

[GeoffK]

Check the resistor value, it could have changed if it did overheat enough to smell hot.

[julie_m]

If the transformer has been damaged by the failure of the rectifier, it will get hot when power is applied even without any valves fitted (so no HT or LT current flowing). You can do this test now. In the absence of an optical pyrometer gun test the transformer with the outside of your fingers -- that way, if it is hot or electrified, your hand will be pulled away as your muscles contract by reflex action -- after 1, 5, 10, 15 and 30 minutes. (You can switch off the set at the mains while checking the temperature, if you are wary of exposed live terminals; the transformer should have enough mass to retain its heat long enough for a test, and there are no valves to be damaged by such switching.) If even after half an hour, it's cool enough to keep your hand there indefinitely, the transformer probably has survived unscathed. If the transformer is getting too hot, or is electrically live, switch off at once and investigate further.

If the transformer is OK (not excessively hot) and the secondary voltages seem sensible (slightly high is fine; a modern digital meter does not steal any energy from the circuit under test like a moving-coil meter), then you could use a temporary silicon rectifier consisting of two 1N4007 diodes and two 2W, 220Ω resistors (suggested starting value; adjust up or down as necessary to get the HT within 10% of nominal), just for testing while waiting for your replacement AZ31. Even if one half of the HT secondary winding has failed open-circuit, there is a way to run the set from just one half of the secondary; using a pair of 1N4007s with the anodes commoned-up and taken to chassis, to complete a bridge with the AZ31 rectifier or silicon replacement. That half of the secondary winding will be passing current all the time (so still zero net DC magnetisation in the core, which will keep it happy) but no current will flow in the other half, so the total power dissipation will be the same. Even with a break, the copper in the other half winding should still conduct heat away from the working half well enough to keep it from overheating.

[Heatercathodeshort]

As SB has mentioned, failure of the filaments in a directly heated rectifier is very common. They are supported with very light springs that can be seen on the top of the electrode assembly. When a filament breaks it is pulled in an upward direction and often shorts to the anode/anodes causing a mains transformer burn out as few radios of this vintage have fuses.
Most of the faulty AZ1/AZ31s I have come across have failed in this manner together with Marconi U10s and U52/5U4s.
I don't like directly heated rectifiers and cannot understand why they continued to be produced into the 1950's. Hope your transformer is OK, my bet is that it is. Regards, John.

[Mike. Watterson]

They were more efficient than indirect?

[dryad1313]

Problem solved! The resistor between the HT caps is actually the field coil for the speaker, which is connected via a melamine(?) socket on the back, which is where the problem was. At the back the HT had been arcing to earth via a rivet holding the socket in place, and created a carbon track. This was the burning, as was evident when I put a new AZ31 in. Scraping all the carbon off immediately cured the problem. Presumably the arcing reached a point where the strain on to rectifier valve was too much. I'm releaved it wasn't the transformer! Before discovering where the fault was I did try the supply with the HT rail disconnected, as suggested, and the voltage was fine. However, a test of the HT caps out-of-circuit showed that the 16uf section had a very high ESR of 40R+, so I've replaced that section with an axial one under the chassis.

Thanks for your help, everyone. Here's a photo of the problem socket.

[jonnybear]

I would not be happy with just scraping the carbon track as it will re a cure at at a later date, ether drill a hole through where it was tracking and spray with printed circuit lacquer, or remove the rivet and replace with a nylon nut and bolt.
John

[Sideband]

Originally Posted by jonnybear

I would not be happy with just scraping the carbon track as it will re a cure at at a later date,

Yes agreed on that....the paxolin (most likely) will have started to break down all the way through the material. It may be OK for a while but I'd be tempted to drill the rivets out and replace the socket completely (you'll probably have to find a similar type from a scrap set).

[Herald1360]

Straight on views of the bottom and top of that connector will likely produce an ID and an offer or two.......

Or you could get a Tufnol sheet offcut and simply rebuild the metal bits into new insulation.

[paulsherwin]

Pye sets of that era do seem to suffer from various short circuit problems.

[dryad1313]

Yes, I was thinking it would be a good idea to replace it. It was only a couple of years ago I had to replace one of the paxolin valve sockets in the same set for the same reason. For that I managed to find a NOS one which dropped straight in without drilling or cutting. Not so sure it will easy to find one of these though.