At 03:16 AM 06/23/98 -0400, you wrote:

>  This seems a little complex to me. I would sample the voltage off both
>pos and neg to earth and if the voltage is not equal (or roughly so) you
>have an earth or partial earth. You would have to determine how much
>leakage is acceptable. If you are in saltwater any leaky insulation will
>show up real quick!
>
>        Barry
Barry: I think you have a much more fundimental(?) problem, and that is the
idea of leakage to ground period. The reason the AC line voltage GFCI's
were developed is because one of the the current carrying path is the
ground. The AC system ties the return lead to ground. As a result, if a
grounded person should touch a conductor at line potential he/she will
receive a shock shince their body is across the line potential.  Some
(small) amount of current flows out of the line thru the person and then to
ground. This is refered to as a ground fault, and is the whole reason for
GFCI circuits. The GFCI measures the current goint from one conductor
(line) to the load and compares it to the current comming back from the
load into the neutral line.

If a system did not have a grounded line then there would be no problem
with a person (grounded) souching the line. This is one technique used to
prevent shock hazards: isolate the system from ground by using a
transformer to remove the ground connection. Of course a person still
cannot grab one conductor in one hand and touch the other. Your DC system
seems to fit into this category: there is no connection to ground (or water
in this case) so a GFCI won't work.


Mark Walter