Hi Warpspeed!
I'm glad to see you stumbled on my little experiment.
Yes, I don't make any pretense that this is novel or that I'm about to invent something. I'm just diving into the arcane methods of AC motor control, because what I read in books and (you said it) see on Youtube, just doesn't tell me the whole story.
Thank you for the clear picture about core saturation and especially the play-by-play of the field collapse. I have literally seen that happen many times, now, and gotten used to it. If you scroll up to my schematic diagram, you'll see a 3-pole switch on the star-point of my resistance load. When the machine is running and I close the switch, the load comes on. Open the switch, and the load drops. Noticeably, it's a fraction of a second after I throw the switch. If I happen to stop the lathe with the load still connected, and I restart it, the machine will run unloaded, just like you say. The implications of that on a wind turbine are not pleasant to think about!
I once had a rectifier fail, shorted to ground. When it happened (years ago) to a PM generator, everything slowed down. If this happens to a capacitor-excited field, the opposite would happen.
Your tip about the battery load is fascinating and thank you for explaining it so concisely. The load from a battery bank is indeed disconnected by the rectifiers - exactly what this type of system needs. High battery bank voltage, you say...
Scrolling back to the schematic that Adriaan posted on the Part I thread, now I realize that any time the generator's field fails, the voltage driving the control circuit will remove the loads from the output. Same thing, it needs to build the field before loading it.
In various test runs, when I add the load, the line voltage drops, sometimes substantially. If you work it out, it's pretty easy to see that the more you load it, or the slower it's going for a given load, the more the voltage droops. I've seen that happen, so it comes as no surprise when you say that more load can cause the field to break down completely. Voltage to zero, then.
Variable load (hence, current) and variable speed (frequency) makes for quite the moving target.