David Farber wrote:
David Platt wrote:
While trying to troubleshoot why some kitchen electrical outlets in
this junction box weren't working, I did a bit of research.
Apparently, there is supposed to be a ground connection (other than
the screw going through the frame of the outlet to the box) from the
GFCI outlet to the junction box. (see photo:
http://webpages.charter.net/mrfixite...nction_box.jpg )
Can I just run some wire from the green screw terminal on the GFCI
outlet to the box? Where would I attach it to the box? By the way,
the problem with the outlet was that the breaker tripped in the
GFCI. There are three pairs of wires in the photo. One is the
incoming AC. One feeds another GFCI outlet down the line. The last
pair goes to a switch in the same junction box which turns on some
lights in the kitchen.

Well, let's see.

Having a ground-wire connection to the GFCI itself is a good idea, if
a good ground is available. It's not (I believe) strictly required -
a GFCI can be used on a two-wire circuit.

In your case, it does appear that the outlet box is probably grounded;
I see what appear to be bare ground wires entering at the back. These
wires *should* be securely bonded to the junction box - this would
probably be done right at the back, where the screws and wire clamps
are located. I can't tell if those ground wires are in fact securely
bonded to the box, or whether they're just sitting there.

If you're going to ground the GFCI, you'd run a wire from the green
screw terminal, to a connection point with one of those ground
wires... e.g. if the existing bare ground wire is screwed to the box
at some point, you'd want to connect the ground wire to that screw or
to another box-mounted screw not too far from it.

Now, the lack of a ground wire from the GFCI shouldn't tend to cause
the GFCI to "trip" mistakenly. A proper GFCI operates by measuring
"current imbalance" between the hot and neutral wires. If more flows
through one than through the other, then it's almost certain that some
current is flowing to ground (somewhere) rather than back through
neutral... and this is a Bad Thing (shock hazard, or shock occurring).

A GFCI's protection does *not* depend on measuring current flowing
through its own ground wire. That's why they can be used to protect
two-wire circuits, where there isn't even a good ground connection
available... this makes them useful for retrofitting older house
circuits for improved safety.

Having a good ground on a GFCI circuit is actually likely to make the
GFCI trip sooner, rather than later, if an electrical fault exists.
For example, if you have an appliance with a grounded case, and a leak
or short develops between "hot" and the case, the GFCI will trip
instantly. If the case isn't grounded through the third prong
(e.g. if there is no ground prong, or if it's cut off or defeated with
a "cheater", or if it's plugged into a 3-wire outlet with no actual
ground) then the fault won't result in current flow to ground... until
somebody touches the case and also touches something grounded. At
that point they'll get a shock, and (knock on wood) the GFCI will trip
fast enough to protect them from anything more than a scare.

So, if your GFCI is tripping, then I don't expect that grounding the
GFCI more securely is likely to change this. You may actually have a
current-leak-to-ground on one of the circuits you are protecting. Or,
possibly, interference from a strong local radio transmitter is
"falsing" the GFCI - it does happen.

Hi David,

I should have been clearer in my description. After noticing that the power
was off to the outlet, I pushed the GFCI reset button but I didn't hear a
click. At that point I figured either the GFCI hadn't tripped and there was
an internal problem, or it had tripped and there was something wrong with
the resetting mechanism. I removed the outlet from the wall and that's when
I noticed the missing ground connection. Regarding the GFCI, there was
nothing wrong with it. I read on a message board that you may have to apply
quite a bit of force to get it to reset. I tried again and it finally reset.
This was the first time this GFCI has tripped.

I looked again at the ground wiring you see in the picture. It exits the box
through a hole in the top of the box. If it's secured to anything outside
the box I can't tell. Is there some official test to perform to check if a
junction box is grounded?

sort of. My tester is a 200 watt bulb, but larger would be better. Connect
one lead to hot and the other to the box of your ground. With the load on,
measure the voltage across the bulb. If it's not the same as the bulb
connected to hot and neutral, you have a faulty ground. The connectivity
and resistance should be as good as your neutral itself.

A plain volt meter test from hot to ground is not valid, and neither are
those little pocket testers. They don't run a real current through ground,
which is something that has to happen if your equipment faults. For old
installations like in that photo, you have no idea what's really going on
or how many times it's been messed with in the past, and how secure the
connections are now.