Pye P75A - most unusual fault

[dave_n_t]

This is not a full success story, merely one small success which, nevertheless, some might find interest in.

As is my wont, I powered up this example, monitoring the voltage on the output valve (EL41) grid. I was rewarded with reception on all bands (L,M,S), after a fashion - weak and distorted.

I gradually moved through it, replacing any resistors more than about 20% out, and any caps that appeared to be in leak-sensitive spots, or which had lost their capacity (the electrolytics).
There were only two resistors to change: 220 ohm cathode resistor for the EL41, up to 280 ohms, and a 220k anode load for the EBC41, up to about 275K;
and that capacitor, the 'tone correction capacitor, and the electrolytics - which wouldn't reform.

For good measure I also changed the negative feedback components, and the volume control - noisy, despite dissembly, cleaning + greasing.

The reception had improved marginally, so I thought it was worth a quick re-alignment. Indeed that did produce results: the IFTs were a little out, but the S+M aerial coils, and most of the trimmer caps, needed a fair bit of movement. (The dust cores in the former seemed to be fixed in place with some fairly strong grease, which a spray of WD40 eventually loosened up). I've quite often come across the trimmer caps losing capacity - I wonder if the mica undergoes some sort of changing which reduces its dielectric constant (maybe it slowly absorbs water).

The reception was now fairly good, although the quality was still disappointing (which it shouldn't be for a set with at least two different negative feedbacks); and the volume wasn't ear-splitting.

On with the thinking-cap, and some fairly detailed voltage measurements. The EBC41 wasn't passing quite enough current, so in with my best example; the 6C9 was one of only three I had, and none did much better. Actually, a new ECH42 did best; so that stayed. The EL41 grid voltage grew to about 0.5V (positive) after a few minutes, which wasn't helping distortion - so again, in with one of my scarce new ones. Finally, the AGC line was checked just in case something was pulling down the gain - but no - a healthy 4V negative of chassis on Radio 5 live (the strongest MW signal here). It was still barely acceptable in terms of volume and quality.

It was then that I got a bit annoyed with the mains hum: audible from valve warm-up, irrespective of earthing the output valve grid. This shouldn't happen in a set with hum-cancelling transformer, and a deal of negative feedback. I checked the (new) smoothing + reservoir caps - no problem. I checked the DC resistance of the output transformer primary: 525 ohms for the main part of the primary, and 6 ohms for the 'hum cancelling part': both perfectly reasonable.

Finally, the penny dropped when I was looking at the diagram on the service sheet showing the disposition of the main components above chassis. The output transformer appeared to be rotated through 180 degrees (the 'tag board' on top pointed away from the tuning scale, instead of towards it). That was the breakthrough: the wiring had been done as if it were the right way round. (It wasn't as easy to track this down as I'd have liked, because two of the three primary wires were white - if only all 3 were differing colours it may have been easier to spot).

To cut to the chase: reversing the wires (i.e. connecting them as the designer had intended) produced all of the volume and quality I'd expected: a really good performer.

What's slightly odd about this is that the wiring and solder joints appeared original: so I suspect it was wired this way from new.

But what's absolutely amazing is that even using the very small (hum cancellation) winding as the anode load for the EL41 produced a fairly loud output!

(I'll take a photo of the set once I've done the cabinet - not my favourite bit, and nowhere as good as most of the sets I see in here).

dave

[paulsherwin]

I wonder if this was done as an experiment to reduce inductive hum between the mains and output transformers. A number of Pye sets suffer from fairly witless layout design whereby inductive coupling causes significant hum even with all the valves removed.

Does the set produce background hum with the O/P TX in the 'standard' position?

Paul

[dave_n_t]

Originally Posted by paulsherwin

I wonder if this was done as an experiment to reduce inductive hum between the mains and output transformers. A number of Pye sets suffer from fairly witless layout design whereby inductive coupling causes significant hum even with all the valves removed.

Does the set produce background hum with the O/P TX in the 'standard' position?

Paul

hmmm. Don't know. Maybe another P75A owner could comment. But I don't see that it should make any difference to magnetic field pickup - the output transformer is symmetrical (apart from the tag board). The hum now is almost completely inaudible.

dave

[Nickthedentist]

Hi Dave,

Very intersting write-up, thanks!

It would be fascinating to know the history of the set. I wonder if it was used in this state all its life.

Keep up the good work,

Nick.

[XTC]

Dave,


Well spotted.

It sounds like a Friday afternoon and Monday morning job. Someone plonked in the transformer the wrong way round and made the connections. Still, it seems surprising it passed muster in the factory and surprising it was tolerated by the buyer.

You do occasionally come across manufacturing faults in sets such as shortwave coils never connected, or joints that were never soldered.

I think if you simply used the hum buck section of the O/P transformer as a primary, you'd have very little output. One way and another, with the screen current from the EL41 modulating the HT line and the supply from the smoothing cap not being perfectly stiff, you get a fair bit of AF, of a sort, across the secondary. The perception of loudness vs power isn't linear, so a couple of hundred mW sounds like quite a lot, but say three watts, only sounds like twice as much.

Pete.

[dave_n_t]

Pete,

I think you've hit the nail on the head there (screen grid current, probably).
I'm still surprised that it works so (passably ) well, though!


dave