One possible line of inquiry . . .
If you have anything wired directly from the battery - eg horn or brake light - then when you apply them, the ammeter will show a chage as the dynamo compensates for the load on the battery. It does it in the blink of an eye. In fact it's how I know my brake lights actually work - or not!
So . . . if there is a short, or funny fault, on any item that is not wired via the ammeter, you would get sudden charges showing there's activity (maybe unintended activity) - which is making the dynamo think it needs to provide a boost.
If you do have the brake light coming straight off the battery, I'd start by having a good look at the brake light switch, and the wiring under the rear mudguard. These wires are horribly notorious for chafing: I refuse to run any wire under any mudguard EVER - and sod 'original', along with the bendy tin retaining tags that break. And then the horn (if also wired direct). If there aren't fuses in the circuits, I think I'd go there, just in case something fries itself.
If it isn't a 'load' that is causing the needle to flutter, then unless the ammeter is defective (it's always tempting, but often wrong, to blame the tools - my old dad always said warning lights were just one more possible point of failure -and it usually cost him dearly), it could be there is a short in the dynamo end cover, or or or . . . sadly, there are far too many potential 'ors'.
But you're right, it ain't normal. And I bet the ammeter is telling the truth, or at least partly.
