Have you considered Zubbly's (RIP) "decoupling" trick?
Use some caps to block the initial startup "DC" ?
He did that with his heaters to let the mill get some speed up before the caps passed the AC.
I don't think it will do anything useful, Tom. If the wall-wart doesn't saturate at operating speed it won't saturate from the near-DC of startup, either. The voltage and frequency are in proportion. (The genny is just a half-transformer anyhow and the magnetic field in the wall wart transformer is essentially a model of the field from the whirling magnets in the genny - but falling off a bit at very low frequencies due to current decay from wiring resistance).
Zubbly's capacitor hack was to ease startup and get out of stall by unloading the genny at low speeds. It wasn't needed to protect the output transformers. (If anything it would make things harder on them at HIGH speeds.)
Maybe I was confused on "why" a transformer stalls a mill at start up.
I always assumed it was the apparent DC short as the voltage just starts to be generated since it is slowly rising DC at first...
That was my reason for the caps. I was not looking at the resonance at higher speeds but it would exist at some level.
Yes, that is the reason for the caps. The magnetization current of (a big) transformer looks enough like a short at very low speeds that it loads the genny enough on startup to apparently be an issue if you have a cogging problem. And yes, as the frequency approached the resonance of the caps and the coils the voltages and currents would go WAY up. But that wasn't my point.
My point was that the usual issue with saturating a transformer by feeding it low frequency power doesn't apply when the feed is the output of a genny (either a permanent magnet alternator or a cored-type whether permanent magnet, excited, or self-excited induction type) driven by a "wild RPM" prime-mover (like wind).
The transformer saturation on low frequency issue relates to lowering the frequency while maintaining the voltage. At lower frequencies the current has longer to build, increasing the magnetization of the core. Of course transformers aren't designed with a lot of extra core material. So if you lower the frequency sufficiently you saturate what you have, after which bad overcurrents happen.
But for a PMA the voltage drops in proportion to the frequency. So, to a first approximation, the current and magnetization remain the same. (In fact, as you lower the frequency, resistance has more time to make the current decay, so the transformer current and magnetization actually drop a tad with lower frequencies.) So if your transformer doesn't saturate at your maximum open-circuit voltage at your maximum RPM, it won't saturate at any lower RPM.
For generators with cores and excitation the situation is similar: The output of the genny is limited by the saturation of its own core. So if your transformer doesn't saturate at the open circuit voltage of the highest RPM (below RPMs where reactance limiting becomes significant) with the genny core saturated, it won't saturate at any lower RPM.
Putting a cap in series with the transformer doesn't help this. Further, even if you're NOT near the resonance (and the loaded transformer is acting more like a resistance with a parallel inductance than a pure inductor resonating with the cap), the series capacitor causes current to increase with increasing frequency. This, combined with the rising voltage, produces a double-whammy that might encourage the transformer to saturate at high RPM and will certainly exacerbate the problems for the regulator.