(ET) Ground fault from hell

[Christopher Zach <czach computer org>]

SO I have been chasing down the most nasty ground fault in the universe. 
Every time I touch battery positive while leaning on the frame I get 
zapped. As in "*OUCH DA*M$IT I AM GOING TO DIE" sort of pain.

Not what you usually get from 36 volts.

Anyway, after the billionth time of this I decided to try and find the 
thing. So I connect battery negative and check voltage to ground: 40 
volts. Yup. Note that the lift wire and the light wires are *NOT* connected.

I disconnect battery neg, check voltage. 0. Ok, that makes sense.

I reconnect neg, turn off main disconnect: 40 volts.

Ooohkay. So it's somewhere in the - side. Wiring diagram shows wires 
going to the shunt, then all over the place. So I pull the control fuse:

40 volts.

Not there. Then a brainstorm strikes. I can just short the positive to 
ground and see what catches fire. Just for a sec.... So I do this. Run a 
wire from the positive to frame ground.

*click*

Click. That's not what I expected. So I check under the hood at the 
relays. I'm pulling 1-2 amps, not too much, but when I pull the wire off 
 the frame I get a nice hot little arc. Then there it is:

The FW relay is clicking on and off. Oh *GREAT*. This however does 
answer the massively painful *SPARK* I get; it's because the coil acts 
as an inductor which likes to generate big currents as it collapses. Of 
course.

The question is what could cause the FW relay (or one of the myriad 
wires down there on card 4) to go to ground? Why would something going 
to ground trigger the FW relay coil even when there is no + in the 
system at ALL (other than frame)?

Any thoughts on tracing this further?

Chris