Winding a conventional AC line frequency transformer

[Paul E. Schoen]

Here is a completely seat of the pants (no actual analysis) idea.

(deep breath)
If you connect a capacitor across the primary (perhaps as part of the
line filter) that does not get disconnected by the power switch, and the
power is switched off at the worst case (for start up surge) at the
voltage zero crossing, the magnetizing current that peaks near that point
would drive the capacitor voltage through zero and apply some
demagnetizing volt seconds to the core.

We have done something similar to that on our test sets to minimize
transients when switching to different voltage taps. We use an SCR
controller set to phase fire initially at about 70 degrees to minimize "DC
offset", which occurs when an inductive load is energized. We add a power
resistor in series with the capacitor to create a snubber. Otherwise the
instantaneous application of voltage to the capacitor draws a high current.

There is also the effect of remanent magnetism, which causes a high inrush
current when voltage is applied. I believe the high inrush occurs if the
phase of the applied voltage matches the phase at which the voltage was
previously removed. The core is already magnetized in that direction, and
saturation occurs very quickly. A gradual soft-start with phase modulation
can minimize this.

I have used some large thermistors from www.Ametherm.com, but they get
quite hot, use up extra power, and must cool off before becoming effective
again.

Paul