A brief update and a puzzle.
The update - The conversion described
above uses a Kato 17632 mechanism and light board. The configuration shown there works well with the GEVO and SD70M models, but it turns out to be too flabby for the Dash-9 and AC4400CW models. The problem is that the exhaust stack protrudes into the shell to the point where it almost touches the light board, and the bundle of wires there is too thick to allow the shell to fully seat. Rather than filing the shell (and risk daylighting the base of the exhaust stack), I opted to trim the light board and run the decoder wires across an air gap. Here is the story in photos:
The first shot shows the region that needs attention. The next two are relatively self explanatory:
This configuration has a flexible, but not floppy, connection between the board and the decoder, but it's still easy to handle, and the decoder assembly sits comfortably within the shell with the rear light nicely aligned to the lens insert. I had to add a jumper wire between the board and the positive LED lead to power the rear LED. The yellow wire is run directly to the negative rear LED lead.
To assemble this I made a simple jig from a wood block and some double-sided tape to hold the pieces in place while connecting the wires.
The puzzle - I have now done 9 of these conversions (mostly of the GEVOs and SD70M variety) and 3 of them have now suffered a head light and/or rear light failure, and I'm not sure why. Here is what I
do know:
* All 3 of these units worked fine initially, but I can no longer power the LEDs with F0 on the affected units (front and/or rear light, depending on the unit).
* The motor controls all continue to work fine (beautifully, in fact).
* All of the LEDs still work. I can power them separately with a transformer and they light up as expected.
* In doing a bit of reading, my understanding is that enabling F0 F(R) closes a microswitch between the white (yellow) wire terminal and decoder ground, completing a circuit through the positive blue wire. If this is correct, the symptoms I am seeing are consistent with one or both of these switches failing open.
I have been handling these units quite a bit and the only failure mechanism I can surmise is some kind voltage spike on the white or yellow wire, but that's just a guess. This could happen if one of these wires shorted to a live chassis. I have also been experiencing some very minor shocks when I've touched the rails over the last few days (we have had some unusually cool dry weather recently), but I would hope these decoders are not
that sensitive to spikes. I think shorting to a live chassis is unlikely since I have been aware of that possibility all along and have tried to guard against it.
Anyone have a similar experience with these tiny decoders? As is, I can relegate these units to trailing locos in a consist and/or re-map some aux outputs (assuming they still work). But I'd sure like to avoid more such incidents.