GFCI and Shared Neutrals

[sacstinkytiger]

GFCI and Shared Neutrals

Hi,

I thought I would write about my experience on wiring a GFCI. I am not an electrician but a normal DIY amateur. I do have some electrical engineering education way back when, but it was way back when though :).

Anyway I looked onto the Internet and found a pretty good explanation on GFCI's at Leviton's web site (e-z Learn). One thing to note is that on their web presentation and on their instructions for GFCI if your receptacle box has 2 or 4 wires coming into it, not counting the ground wires then it is a DIY job. If you do not have that number of wires call an electrician.

OK I thought how hard can this be? It looks so simple every where you look on the Internet. But as with most of these things the devil is in the details. So just to be safe I went overboard on the test equipment. Mainly because I am a techie and sometimes you just never know and a piece of test equipment will bail you out of a tight spot. Any way it turns out it was a good call. Note that the list below is what I got, the best choice for you will depend on what you feel most comfortable with design and budget wise.

Equipment list:

1) Volt Tick -- Fluke 1AC II AC Voltage detector A pen looking like tester. It will find on contact through plastic insulation wires with voltage. Using electro static fields it will find hot wires.

2) Volt meter -- Fluke 117, a digital volt meter

3) Wire stripper / Cutter (Klien tools make good but expensive ones)

4) Electrical tape

5) Wire nuts -- Red and Yellow denote the most often used sizes

6) GFCI -- I got the Leviton Smart Lock from Home Depot because it was the most handily available. Slightly cheaper in sets of three.

7) Socket tester -- One with three lights usually. Tells you if you have your socket/GFCI wired correctly. Some will even test GFCI's by inducing a small Ground fault current to "pop" the GFCI breaker.

8) 6 feet of 14 Gauge wire in each color Black, White, Green, Red

9) Screw driver, flat blade or Philips

OK armed to the teeth ready to do battle. One of the key safety tips I think is not to work alone. Make sure you have someone there to keep an eye on you. They do not have to be technically competent, but they should know where the main breaker is and know how to call 911 :)

Put in the receptacle tester and it showed that the receptacle is wired correctly. So at least that is a good start I thought. In the other outlett on the receptacle I plug in a light. The test and the light are on. Go to the basement and hit the breaker. Come back and lights are off. I test the receptacle with volt meter and volt tick. Dead no voltage.

Remove the face plate and then check again for any voltage. None found. Carefully loosen the holding wires of the receptacle and pull it out. Uh Oh ... I see seven wires not including grounds coming into this junction box. And the receptacle has one white wire handing off it, two black wires and no ground. Not the standard game plan at all.

Hmmm .... I think. What is this ... I carefully again test for hot wires with the volt meter on any exposed wire surfaces dead nothing. Then again with the volt tick. It lights I have a hot wire in there ... Yikes!

What I have is three cables coming into the junction box:

1) 4 Wire connection to the main breaker panel (Edison Feed). This cable has a Red and Black hot wires, a white neutral wire and a bare earth wire. The Red and Black wires carry 120 V on a 180 degree phase differential. The circuit is known as an Edison Circuit or also known as a shared neutral. The red wire on this cable that is still hot in this case. 2) A NM Romex cable (Bedroom A feed) with a Black, White and Ground wire. The red hot cable is connected to the black cable on this and feeds a Bedroom room A. 3) A NM Romex cable (Bedroom B feed) with a Black, White and Ground Wire. This feeds bedroom B. The Black wire from the 4 wire connection feeds this Bedroom B.

All the grounds are connected together and are connected to the metal in wall receptacle box. All the neutrals are connected together. The receptacle is earthed to the junction box by its attachment screws, hence there is no earth lead to the receptacle. Note: By removing the receptacle out from the junction box, I have effectively removed its earth.

The Hot wire is the red wire. It is hot because it is the other phase of the Edison circuit that shares the neutral. Go down to the breaker panel and hit a few more breakers. Eventually the red wire is no longer hot. Note if I did not have the Volt Tick, I may not have noticed it being hot. (I therefore really recommend getting one of these, cost about $US

20). The Edison circuit was controlled by two single breakers, one for each phase of the Edison circuit. Also the volt tick allows you to test for hot wires without having to expose the end of the copper wire which can cause electrical shocks. You can test for hot wires while everything is still insulated with wire nuts and or electrical tape. I always test again with a volt meter once the bare copper is revealed.

I also check the neutrals for hot voltage with the volt tick. Note a neutral can become hot because it may not be connected to earth. For example in my junction box. From Bed Room A and Bedroom B and the Edison circuit all the neutrals where joined together via a wire nut. It I had disconnected the wire nut, and not hit the breaker for the red wire, one of these neutrals would then become hot if bedroom A had any switch turned on.

OK. I finally have a breaker box which is really powered down. I also know what the wiring is. Note half the battle is realizing exactly what you are dealing with when you have zero documentation.

I wiring in the GFCI to protect Bedroom B's receptacles. It is correct that you cannot use a standard GFCI on a shared neutral circuit. So how did I do it? A Edison shared neutral circuit can be split into two separate non-shared neutral circuits. I do this and one split goes to bedroom A and the other to Bedroom B. On the bedroom B split I place the GFCI. Note as far as the GFCI is concerned it is no longer sitting on a Edison Shared Neutral circuit. It is sitting on its own branch with it's own neutral.

I wire it up, and also connect a earth lead to the GFCI. Note that upon installation this is not required because I have a metal receptacle box that is earthed. Attaching the receptacle via metal screws to the box earths it. Note I usually put a piece of electrical tape around the GFCI to cover up the terminals.

After carefully pushing the wires around the receptacle box, I have made enough room to put the GFCI in, tighten up the screws and replace the receptacle face plate.

Now for the moment of truth. Turn the power back on and reset the GFCI. O.K. test the outlets on the GFCI, and the downstream receptacles with the three light tester. O.K. Also test the sockets of the other receptacles in Bedroom A. In each case re check the voltages of the sockets. They read

123.5 volts for ground and neutral to hot, and close to zero neutral to ground in both bedrooms A and B.

Note if you get a reading of 220 volts anywhere, that means you have an error and have connected across the two phases of the Edison Circuit. This is bad. So turn off at the breaker panel.

I check the GFCI, it resets normally and when tripped shutsdown the power. Note on the Leviton GFCI, if you have Line and Load wires reversed the GFCI will not have power at the socket, even though the green LED on the front of the GFCI will light.

OK after all that I am done. Somewhat more that I had bargin for, but at least everything is working and voltages are in order. I guess sometimes what you get on the Internet is the glossed over version "this is easy DIY" version of things.

Best, Mike.

[RBM]

I like the part about not working alone, so when you accidentally get yourself stuck between the two legs of your Edison circuit, he can run down and try to find the breakers

[Jeff Wisnia]

I've always thought it was darned clever of Edison to come up with the idea of saving 25% of the copper needed for his distribution network by using a shared neutral with plus and minus 100 volt DC conductors.

Since the power was generated at the 100 volt level the conductors had to be quite thick to handle the current coming out of the generators, so saving that much copper meant a lot.

The nice thing about a shared neutral system is that the neutral doesn't have to be any larger than either of the two hot leads, regardless of the amount of current imbalance of the loads.

Jeff

[Beachcomber]

There is no one "best" electrical system with regard to neutrals and grounds. Tradeoffs exist between safety, economy, the desire for a multi or single voltage system, and whether you choose to spend more money on conductors or insulation.

In the US North American system, having the one ground lead (the neutral) was an economic compromise put into the electric codes during WWII to save copper. We are now back to using 4 conductors (hot, hot, neutral, and safety ground) for dryers and ranges because it was determined that having any regular current flowing in the neutral, when it is also a safety ground, is not 100% safe.

Beachcomber

[M Q]

...

... This may be a little late to post a followup, but... The above statement about the neutral never having to be larger than either hot lead is not always true!! It is true for pure resistive loads (power factor=1). It is not true for reactive loads, as the current on the two phases may not cancel out. This includes gas discharge lighting (including flourescents), older computer power supplies, and some situations with motors. I once measured 38 amps RMS on the neutral of a 20-amp 3-phase common-neutral circuit that was powering an array of cubicles. Each cubicle had a flourescent light and a computer. While the breaker never tripped (and it should not have), the 12ga neutral was getting warm.

[Dave Martindale]

Now, I see how that can be true for 3-phase systems. Three sine waves of the same voltage spaced 120 degrees apart in phase always sum to zero, so in the ideal case of three equal resistive loads there's no current in the neutral. But adding non-linear loads like discharge tubes and power supplies tends to make the current waveform more like a square wave than a sine wave, and square waves have lots of 3rd harmonic content. A 120 degree phase shift for the fundamental frequency becomes a 360 degree phase shift for the 3rd harmonic, so the 3rd harmonic currents in all 3 phases add instead of cancelling. So you get current in the neutral even when the loads are evenly balanced.

But the same argument doesn't hold for 120/240 V home circuits. The phase shift between the two 120 V phases is 180 degrees. Any

3rd-harmonic content sees this as a 540 degree phase shift, which is still equivalent to 180 degrees. So the 3rd harmonic currents cancel, just like the fundamental frequencies. This would happen for any odd harmonic, actually. To get current waveforms that don't cancel, you'd have to have some even harmonic content, such as a waveform where the positive and negative half-cycles aren't mirror images of each other.

Is this ever a problem in practice in 120/240 V single phase systems?

Dave

[M Q]

...

You have a good point. With identical loads on each leg in a single phase system, the currents cancel, whereas they they may not in a 3 phase system.

However, if the loads are not identical, you may still have a problem. I will illustrate this with some theoretical, non-real-world examples. Let's say that we put an inductor on one leg (current is 90 degree phase lag), and a capacitor on the other leg (current is 90 degree phase lead). The currents in the neutral would add rather than cancel. In the real world, you might not have this extreme, but a motor on one leg and an older computer power supply on the other would result in a higher current on the neutral than either leg.

I admit that I don't know whether this is ever a problem in practice on single phase circuits.

[Chris Lewis]

I would suggest that the safety margins designed into the NEC are sufficiently large that even the worst-case real-world scenario wouldn't present a problem.

Otherwise, the electrical code would have done something about it.