Transformer issue

Shorting turns _would_ change the inductance significantly -- you're basically putting a shorted turn in the transformer, so you'll be more or less seeing the leakage inductance of a (possibly much) turns-reduced inductor. Three orders of magnitude indicates a really low-leakage transformer, which is surprising, but shorted turns = low inductance, for sure.

Try shorting one side of the transformer & see what the inductance change of the other side is.

If you only short a few turns you wouldn't see a big difference in turns ratio.

Say you have 2 windings A and B with a turns ratio of 1000 : 1000 and the first turn on A is shorted to the second. It's like having 3 windings A, B, C with 999 : 1000 : 1. There isn't much difference in the A:B ratio, but you have a short circuit across winding C. The 1000:1 turns ratio from A to C makes for a 1000000 to 1 impedance ratio so it won't shunt the output of a signal generator.

If you take a good transformer, measure the primary inductance, then short the secondary and measure the primary inductance again, you get a much lower inductance measurement when the secondary is shorted.

"krakpot ratbag wanker"

Maybe the copper wire has suddenly gone superconducting ??

ROTFL !!

How amazing for it to affect both sides ....

ROTFL !!

Only in some alternative universe inhabited entirely by F*****TS like Keith.

..... Phil

I think your vendor is right in changing to high-temp wire. A single shorted turn renders a transformer useless, inductance collapses as Tim said. Once shorted a turns ratio measurement via an audio test may not be all that meaningful anymore. It's like testing the cornering performance of a car with the parking brake pulled :-)

Usually in cases like this the mfg would want some failed transformers back and do a failure analysis on those. Peel them apart. With photos and so on, photos that should be shared with you.

^^^^^^^^^

I would expect that, but didn't consider the shorted winding to be a separate winding. I imagined it as simply shorted (out of the circuit). Thank you very much for the explanation.

Ah, Another of Nymbecile's Nyms.

You read as well as Nybecile, too, Phyllis.

Oh, they have a pile back already. They haven't shared any photos, though. I just wanted to be comfortable that their fix is actually going to do something.

This is becoming a rather painful issue. The fallout has gone from 1% to maybe 10%. With 15 transformers per board it's a royal PITA for manufacturing. Hopefully we'll get an ICT screen in place in a week or so but changing that many transformers is still a big kink in the line.

"krakpot ratbag wanker"

Maybe the copper wire has suddenly gone superconducting ??

ROTFL !!

How amazing for it to affect both sides ....

ROTFL !!

Only in some alternative universe inhabited entirely by F*****TS like Keith.

..... Phil

I wonder if you haven't developed a DC leakage path which completely bypasses the windings? In effect, an R in parallel with one of the windings. This could result in a low DCR reading, wouldn't affect the winding ratio measurement (assuming you're driving the primary with a low-impedance voltage source), but might give you very misleading results on your inductance measurement if the measurement technique assumes that it's looking into a pure L-in-series-with-DCR component.

Try modeling it out as an unwanted R, bridged right across the primary-winding terminals... see if that would give you the sort of curve you're seeing:

Ztotal = Runwanted || (Lprimary + Rprimary)

I think it will.

I'd solder them on by hand until further notice, until they deliver them with high-temp wire.

DimBulb repeats himself....

That's a good suggestion. I don't know if they're going to like that answer either.

"Dave Platt"

Maybe you think the moon IS made from green cheese too.

Ever heard of "Occams Razor" ??

Obviously not.

.... Phil

The production folks? Invite them for lunch and explain the situation, works really well IME. If they are kept in the loop and feel like part of the family they'd do almost anything for you :-)

I don't know which is worse, replacing 10% (perhaps a high number) or hand placing 100%. I'll certainly suggest it tomorrow, though (I'm out next week ;).

The consensus here is that their fix will work. Sometimes better the devil you know...

But doesn't it take some time to discover which 10% are bad even before you desolder and replace them?

We were just discussing yesterday whether op-amps or transformers would make for a better hybrid (interfacing to the old two-wire "party line" intercom systems)... maybe op-amps deserve another look!

---Joel

I'm 99% sure a simple ICT screen will catch them. They've been relying on DCR but that clearly isn't going to work. A simple RL timing circuit will, though.

A lot depends on how well controlled your environment is. I found that op-amps work better in our application (we needed better than