Most of the time there is no problem whatsoever. The lights dim a bit, the chassis hums for a split second, and the amp comes up without fanfare.
Your proposition of the mechanism that produced this effect seems plausible, and is the first I have heard of the effect. It answers some questions I had, namely, why it happens sometimes and not others.
Yep. It’s residual flux in the core. How much is there depends on when the AC line was switched OFF (re the AC cycle), and when you happen to turn it back ON. Let’s assume you left the core sitting near Bsat of the magnetic material. (Remember on toroids, there is more residual flux left than E-I cores due to the smaller effective air gap). If you then turn things back ON as the AC is passing thru zero heading to produce more flux, you have an entire half AC cycle that the core is effectively saturated, and the current is limited only by the leakage inductance and the small amount of resistance. It takes several cycles of AC to bring the core back to steady state conditions, which is usually enough time to pop the fuse. If you were to continually cycle power on a transformer, even with no secondary load at all, you’ll notice it’ll get HOT. An inrush surge protector in the primary helps by limiting the available power during the first few AC cycles, allowing the core to recover to “steady state”.
There’s a really nasty issue of putting a BF diode in series with the primary of a transformer. You get really hellish currents flowing. The same diode, acting as a half wave rectifier in the secondary doesn’t produce nearly the same level of problem. Reason: You’ll find that THAT condition reflects some effective DC back into the primary (really), which is of such a phase as to partially cancel the effect of biasing the core in one direction. That’s the reason half wave rectifiers work at all!