pot core

Right, the winding looked like a wire short, so the MOSFET conducted very high currents, limited by its Rds(on). The IRFL014 is spec'd at ~ 0.4 ohms max with a hot junction, so surely the MOSFET dissipated much more power than the copper wire. Apparently the MOSFET's plastic had a higher melting temp than the bobbin. It's amazing how you can mistreat them for a short while and get away with it.

Typo - "perfect" was the intended target. Could be a Freudian slip - we had prefects at my secondary school and they did have a habit of being prescriptive about stuff they didn't actually understand.

Thanks for telling me what I've said elsewhere in this thread. For small numbers of strands twisting is enough to do the job, but once you've got enough strands to have an inside/outside distinction - at a guess more than four - you've got to get more elaborate to make sure that each and every strand in the bundle is exposed to the same magnetic flux change over the full length of the winding.

I made up some transformers a long time ago using quadrifilar wire, as it made the thing a lot easier to wind. This was to generate floating gate driver power for a full-bridge driver. But, the capacitance between adjacent wires was about 300 pF, and caused HUGE problems, blowing out the driver of the transformer. I went to random-wound turns on separate sections of the bobbin, and the driver didn't pop any more. Still had plenty of coupling.

Jon

from my less than comprehensive understanding of magnetics it seems to me that it'd push flux towoards one leg causing the core to begin saturation earlier.

Not so much an issue in full wave inductive rectifier, but a small gap would be a normal precaution.

In single-ended converters, you'd need proper compensation.

RL