This is why I have questions to SC with wax paper caps. 200Meg is the minimum resistance for them, as a coupling cap (Lafayette tester) so that would be no different. As there is no significant current draw, voltage would appear in the same way as a leaking cap. I don't care about that, rarely are they that good & if its an NP cap & leaks, its a dud: Ruthless but effective.

Some of that bitumen, I swear, is corrosive. There is a high failure rate in many transformers & coils that have been "tarred & feathered", or potted with it.

My leakage test for NP caps in valve radios & test equipment, is normally an insulation tester. It also gets a run on transformers (Tag & Test) often primary & secondary,

I would suggest 500V DC on that transformer disconnected (close to surge volts) one tester wire on primary one on secondary. Might pay to check the other one? If its moisture the voltage may help split the water in to its gaseous components? Depending on construction the coil you may be able to melt the stuff off. But someone may have spare IF that will suit. (Photo).

I am single so can get away with using the kitchen oven, as a Lab oven.

You have more experience with gear of this vintage than I do. My background is in doing post-mortems of new gear to discover failure mechanisms. Unlike old radios, this happens at the steep end of the bathtub curve.

The failure mechanism of the HMV I was referring to was simply that, when the IFT was made, a wire from the primary winding was draped across the secondary winding below it on the former before the wax was applied. Pretty common construction technique but definitely a weakness given the cotton-covered litz wire.

Think about it -the introduction of polyester coating on winding wires did one hell of a lot to improve product reliability across the board.

OK I attacked this one on the weekend.

Only IFT I had available was from 50's AWA - 20mm square clip mount, you've seen the type. So I decided I'd try to hide it inside the original can.

Managed to get the pitch crumbling away enough to get the guts out of the can. Thinks "that doesn't look too bad" so decided to have a go fixing it.

Stripped the trimmer head off, dis-assembled and cleaned the micas, metho soaked and baked out the phenolic base, new compression screws. 500v leakage test across and between trimmers - OK.

Cleaned vertigris out of inside of former where it had penetrated the gap between the former and the (newfangled for 30s) ferrite core, Baked the windings, repaired damaged former with heatshink. Nice neat job, all rigid and tight. Leakage test between windings - OK.

Assembled unit and tested - AARRGH! Leakage is back!

OK, I'll give up on that IFT and glue the AWA one into the can with hotmelt and a plastic bottle cap with a hole drilled in it so the adjustment can be got at. Attached new wires. Assembled into chassis. Fired it up.

Nothing! IFT secondary is O/C!

Pull apart, discover that it's only the CT on the secondary that's O/C. No problem, let's use the whole secondary like most other sets do.

Fired it up, works well. Touch up the slugs on the new IFT, gain increases massively. We have a hot performer!

Minus 7 volts of AGC on 702 on the Northern Beaches is pretty respectable. The AGC brings backgound noise up to full audio levels between stations.

Fired it up again, cold, this morning. No more intermittent!

There is really far too much audio gain on this set. Much more than you'd expect from a 75 and a 42. I might put some negative feedback around it.......

Before you do any mods make sure the coupling caps are what they should be and the divider resistors on the AGC (if used) are correct. I am constantly finding "bench top" repairs, where any cap, which is good, ends up in a set.

Some repair people just made it work: How it worked was an entirely debatable thing. I had one HMV recently where a 6AQ5 was in the hole that should have had a 6M5. Not only was it wired with input on pin 1, it was the wrong value coupling cap, the grid leak was out of spec and the plate resistor of the first audio feeding it was also way out of spec. And you wonder why things don't run.

Having a dig, Marc, at the kind of people you're talking about.

Everything in terms of values is correct. Actually the 75 is the high gain version of this valve, it just has more gain than I expected. About what you'd expect from a pentode in that slot, actually.

What I DID do with this set (and I did mention it in my first entry in this post - hid the evidence under the big tagboard) is to fix a basic flaw in the original design - the sort of oversight you rarely see.

Well, after a few days switched off, I fired this radio up this morning while I was packing the dishwasher.

The fault is still there!

However, because of the much greater IF gain than before, if you didn't know about it, you'd never pick it. The AGC now covers it up most effectively, there is no change in audio volume, just a ~20mS AGC time constant transient as the fault appears and again when it fixes itself.

I'm not going to bother looking any further, it happens so rarely and it's now very subtle when it happens at all. Time to move on!

Sounds very much like the IF is going into regen, and is obviously right on the edge of it.
Maybe your workings have created a feedback path, with differing temperature, line voltage or something.
This would explain the excessive gain.
JJ

No, the IF gain is now normal, it was poor previously. There is no sign of regen., you can tell when tuning, not even close.

The set is connected to my long-wire antenna on the back fence via RG59 that runs down the side fence. Only way I can get AM reception here that's out of the noise.

6D6s are prone to instability if run unshielded. Not the case here, the shield is grounded via bright metal (acid washed and solder pot re-tinned) socket surround.

It now cycles, from turn-on, OK 3 mins, fault 20 seconds, OK 30 seconds, fault 20 seconds, then OK for the rest of the day. But now it's very hard to hear the change, you've really got to be listening for it. Apart from this fault the set is rock-solid stable with frequency response better than most AM radios.

Unless I can find another 6D6 to try I'm going to leave it as is.

6D6 became 6U7 with a base change. 6U7 is a horror, & 6D6 can be as well. Left unshielded 6U7 is apt to pick up anything & use it, with great effect, to destabilise the whole show.

I have had issues with 6D6 where you can have three and none with behave, the same as another. In one Midwest one of them was outside of what was posing as a Faraday cage. The cage underneath had half of several sockets either side of it & to add to the entertainment they ran a wire carrying RF through the audio and an Audio cable to the tone past an unshielded part of the RF cage.

Putting a "Goat" shield on that tube caused even more issues & I ended up with a straight sided shield? Not surprisingly this was one of the most unstable sets I have ever dealt with. Lots of shielded wire etc. to tame it. Lead dress and some modern caps can lead to all sorts of issues.

You can also get nasty surprises when not looking. I did have a hacked STC, where I found a manufacturing issue caused by lead dress, there was a second oscillation in the mixer (autodyne) which was bad lead dress.

You may have to consider my tactic on it where I found that the 6D6 was overloading the Plate detector & they're bad enough without help. I actually upped the bias on the 6D6 to -4.5V from the recommended -3. This drama occurred as I was using an Oscilloscope for calibration & initially signal tracing to see where it got lost. It was also that which exposed several of the issues with the Midwest & helped in getting rid of them.

Yes well after running it for a couple of hours this morning I left it off while I took the dog out for a couple of hours. Fired it up, no sign of a problem. This is pretty typical behaviour for this fault.

Just to check the efficacy of the 6D6 shield can I removed it. Sure enough, as soon as you tune off the station it breaks into oscillation. Tune to the centre of the carrier and the AGC pulls the gain back and the 6D6 drops out of oscillation. All pretty much as expected.

Marc, where you need to be careful with AGC bypass caps is where neutralisation has been implemented around the IF, as in some HMVs and others, by connecting about 1pF from the anode to the cold end of the IFT winding in the grid. For this to work as intended the capacitive reactance, ESR and series inductance of the replacement bypass cap must be the same as the old one. So the type of cap you use in such a circuit does matter a great deal. Use the wrong type of capacitor and you'll never get the IF stable.

This morning, no sign of the fault.

This set still has its 1930s IRC cast-lead-ended 1 watt resistors. You know the ones, nearly as big as a modern 10 watt resistor. Every one of them checked out within 10% of its marked value so I left them all in. Just maybe one of them is responsible for this fault.....

Only job left is to clear-coat the replica Tasma badge. I had it stuck on the front with DS tape (covering an unused hole) but it creased when I was trying to pull it off so I'll print another one. Then I'm done with this set, it can go to its new home where it will be used and appreciated every day.

Sorry Ian, I should have completely read your first post.
You really need to try valves first.
I may have them if your prepared to pay for postage only.
JJ

Thanks for the offer JJ. I need to have another rustle around first. I know I have a 58 and a 2A7 and the mains transformer in this chassis has a spare 2.5v heater winding. I may also have a 6D6 somewhere.

The 6D6 is an interesting bottle. It has a predecessor in the 58 and a successor in the 6U7G. All potentially just as temperamental.

Both the 6A7 and 6D6 in this chassis look like they are originals.