[Hartley118]

I guess that the way to test all these imaginative theories would be to test the transformer with and without the shorted winding.

These early coupling transformers were far from ideal by today's standards. They needed high inductance windings because they were fed from a high source resistance (e.g. a specified 18k for the anode resistance of an HL2 valve) and the inadequacies of the core were compensated by lots of turns of thin wire, resulting in high leakage inductance and self capacitance, resulting in a self-resonant frequency typically falling in the audio band.

I can only guess that the set sounded better with the transformer damped by those shorted turns. There may have been plenty of gain available from the valves to compensate for the reduced transformer effectiveness.

As you say, those four shorted turns with their self-resistance load of say 0.1 ohm would have a damping effect similar to connecting a load resistor across the 19000 turn secondary. A rough calculation comparing the shorted turns with the 19000 turns on the secondary, suggests the impedance ratio would be (19000/4)squared, or 475,000. Assuming that the shorted turns measure around 0.1 ohm, that would give an equivalent secondary load of 47,500 ohms - a not unreasonable figure to provide significant damping.

It would be great if you could get the set working so that you can try it with and without the shorted winding. If this isn't practicable, then you could simulate the effect by taking a frequency response measurement by feeding the primary from an audio oscillator via (say) 20k ohms to simulate the driver valve equivalent anode resistance and measuring the secondary output voltage into say 1Mohm. Doing this with the four turns open and then shorted should answer the mystery.

Martin