I read Bruce Wilson's account with much interest, having gone through one OEM and three Electrex RRs on my '98. The bike is now on its second OEM RR with a spare aboard. After the bike failed to charge its battery with a functioning RR Bruce's article persuaded me to examine the wiring diagram carefully. Main fuse A is a basic error in the '98's design, placing a load on the battery charging circuit.
It's easy to check: Just connect a voltmeter between the positive RR output and the positive battery terminal. If the drop depends on _other_ loads, like lights and fan, the charging circuit is miswired. That was the case on my '98.
My (somewhat radical) solution is to replace the RR connectors and put the battery/main fuse on a dedicated circuit serving no other loads. Here's the resulting schematic. The manufacturer will doubtless take a very dim view of these ideas, so you're on your own henceforth.
1. Replace the RR-bike connectors with soldered ring lugs on barrier strips. Bolted together and taped up might be neater mechanically, the barrier strips take up a lot of space under the seat. Honda's quarter- inch spade connectors simply aren't big enough. I soldered in extension wires to bring the connections up under the seat for easy access. Using threaded connections makes step 2 very easy.
2. Disconnect the wire going from the harness to main fuse B and connect the fuse directly to the RR output (easy now that the terminals are accessible). A soldered splice to an extension and a ring lug helps the reach, only a few inches.
3. Pull the heavy red wire out of the four-pin connector at the starter relay and plug it into the harness connection vacated in step 2. This puts the battery on a separate, dedicated circuit with the RR, so it sees accurately regulated voltage. Fuse A is not necessary. The bike draws less than 12 amps with the ignition on and the engine not running. Absent a ridiculous accessory load the single 30 amp fuse is adequate. Measured alternator output into a dead short is 34 amps, so it can't blow the fuse either in the presence of a normal load (lights, etc).
The battery draws negligible current at full charge, so perfect connections between the RR and the battery are _not_ required.
Perfect connections from RR to the rest of the bike are optional: All the engine control electronics have to work at cranking voltage, down near 10 volts, so a significant drop from the wiring does not matter. Dim lights will become objectionable long before the bike won't run. My bike has about 13.5 volts in the main harness with the RR output at 14-14.2 volts. The drop appears to be somewhere in the red/white wires from RR to main harness.
Any accessories should be connected directly to the RR's output via a fuse and relay controlled by the ignition switch. Don't wire them to the battery, that would simply repeat Honda's mistake.
It happens that a fan on the RR was the first and so far most persistent modification to my VFR. I still think it's a good idea.
The state of the battery and wiring _does_ seem to influence RR behavior. I don't understand why, but the changes described above _raised_ the regulating voltage by nearly half a volt. This suggested the RR does not have good internal filtering and would benefit from a capacitor across the output, but adding several microfarads across the RR terminals had no effect on regulating voltage. Faulty wiring might put the RR into some sort of "self-destruct" mode, but how remains a mystery.
[note added in postscript, January 2005]
The original RR failures happened in the low 20k-mile range. The modifications were completed by about 25k, roughly in the summer of 2000. The bike ran to ~65k miles in December of 2004 when another battery failed. Careful measurements revealed one dead cell in the battery and only a 5 millivolt drop in the RR+ to battery + terminals, but there was a 50 millivolt drop from RR- to battery -. Extra wires were added from the negative RR output direct to the battery, reducing the voltage discrepancy to less than 10 millivolts. The change had no discernible effect with the old RR and a used battery, settling at 13.66 volts at 5k rpm. 1500 rpm gave about 14.0 volts.
In the course of this work the spare RR was wired in briefly to see if it would raise the charging voltage. It did, by about 200 millivolts, suggesting the RR's do age in some fashion. The old one was put back and after 90 minutes of riding the regulating voltage seemed higher. It will be checked after a few days' running.
This represents about all that can be done to give the RR an accurate measure of battery voltage while maintaining fuse protection between the bike and battery. It diminishes (but does not eliminate) the disturbing effect of shared and imperfect grounds: The only way to do that is add a second relay in the heavy cable from battery negative to engine ground. At this writing it does not seem worth the trouble.
Since the RR was initially rewired the bike has gone through three batteries; the first failed rather quickly after a hasty dealer prep, the second was placed in reserve after a couple of years and the third (a Champion, made by Exide) failed after meticulous preparation in two years. The reserve battery (of Taiwanese make) is now back in service.
Many people seem to think stators "go bad". I'd urge them to reconsider, and measure carefully. I'm very skeptical stators are the problem.
[note added in postscript, May of 2005]
Ok, it looks like stators do fail. The original stator developed an open winding at about 67k miles. On removal about 2/3rds were obviously cooked, the remaining third looked ok.
When the new stator arrived it was apparent that the windings were mostly unsupported where they join to the output wiring. Possibly that contributes to a vibration-induced failure. A dab of epoxy solidified things, but does not explain the rest of the problem.
A careful dissection of the old stator is in order.
[added in September of 2005]
The stator was placed in a sealed can of paint remover in about June, opened it in August. There was no sign of vibration induced failure. Indeed, the stator windings seemed in very good shape apart from the obvious signs of overheating. The epoxy used to hold the windings is very tough stuff. The unsupported part of the winding showed no signs of fatigue failure.
For now I'm stumped.
[December, 2011 at about 75k miles] The charging voltage became somewhat low, so the spare RR that had been stowed away in the tail fairing was swapped in. It improved output enough to justify keeping it in place. Evidently RRs do wear out.