Problem: Bike not charging (again), will initially charge at decent voltage, then voltage will decrease.
Inspection:
I had the bike on trickle charge for a couple of days, then unplugged her and let her sit unassisted for our 3 day camping trip, just to see what the battery would do.
Trickle charger charges at 13v.
Battery tonight after sitting 3 days- 12.50.
I figured first things' first- the charging itself was doing weird things, so that was where I'd start.
Flipped the multimeter to ohms and measured the black and white spades by the R/R. By measuring there, I would not only be measuring rotor resistance but also the added resistance of the brushes, the wiring up through, and the connector.
I got a crazy number- 85.5 ohms.
WTF.
So, something was wrong there.
I took off the alternator cover/stator.
The rotor tested at 5.6, but it was dirty with grease and carbon. (I should mention that when my valve adjustment gasket leaked, oil ran down the side of the engine and some got past my gasket)
Brushes, measured with one test lead on the brush, the other lead on the corresponding R/R spade, both measured .8 ohms. Pretty normal.
Put everything back together and we got some wild fluctuations, in the 1-200s, and with some fiddling, could get it to settle around 50 ohms.
I cleaned up the sliprings with some brakekleen and 1500 sand paper, then put the cover back on and tested- Closer to 30 ohms, but not quite 'there'.
Solution?:
Cleaned up the contact parts of the brushes with brakekleen and sandpaper, and noticed something interesting.
On the rotor, there are two nubs, one on either side, to which one attaches the wires that connect to the slip rings, presumably, as resistance between these two nubs is exactly the resistance between sliprings. One slipring nubbin was resoldered a bit 'taller' than on the 'stock' rotor, and had been making intermittant contact with the edges of the stator. We touched it with a file taking care to preserve the soldered connection.
We put everything back together at that point and resistance at the RR measured 5.8 ohms, as it should.
Conclusion:
As it was getting late and I have work tomorrow, I didn't get to really test my 'solution' but removing some of the solder material, as well as cleaning up some of the carbon dust, seemed to yield definitive results in terms of resistance reduction at the regulator/rectifier. The solder problem also made sense as far as why each individual component would test out within expected tolerances, yet as an assembled unit, the resistance was way off the chart.
