The NextGen is a capacitor-regulated head
, and a new capacitor will solve this problem (if the old one is bad).
The PO's generator head
is an AVR-regulated one. These are potted electronics
and will be difficult to replace an individual component. The capacitor in an AVR circuit may or may not be the problem with energizing the excitation field. I suspect it is not, since the AVR controls the generator
voltage just fine when the excitation field is externally flashed.
One test is to get another AVR and see if the problem goes away. There should be documentation
somewhere on measuring the excitation winding for a leak to ground. I suspect you are loosing the residual magnetism in the winding.