On 19/06/2013 23:36, Windmill wrote:

But treating live and earth as a transmission line and using a TDR to
examine its characteristics might just possibly reveal something.

Or not... With the 'fault' being a fairly high resistance leak to earth
the reflection coefficient probably won't be high enough to show up.

OTOH one might have more luck with a couple of bent hazel sticks !

Sort of. My suggestion is as follows:

Disconnect the feed-point from the RCD. Strap the L & N outgoing
conductors together and energise them at mains voltage via a
precautionary low value fuse (1 or 2 A, say). The point(s) of leakage
will now be injecting some current into the soil. If you're worried
about safety then reduce the applied voltage using a suitable
transformer or auto-transformer. 50 V might be enough.

Now use a highish impedance AC voltmeter (an ordinary DMM will do) and
probe for ground potential gradients - i.e. earth one side of the meter
to a reference, the mains earth should do, via a long flying lead. On
the other side, rig up a probe a few inches long to stick in the ground
- this ought to a have an insulated handle for obvious reasons. Probe
around, following the cable routes, if known. Look for the highest
voltage and dig there.

If water has got into a joint box it will only be a matter of time
before the situation develops into a low impedance L-N fault. I fear
that feeding via an isolating transformer would only give a short-term fix.

--
Andy