It seems that all too often I get called by someone who wants a ground for their computer where the house wiring has no ground. They've had the idea that I could just take a separate ground wire to the outside or to the panel.

I thought that was a code violation because it would be creating more than one path to ground, or something like that. The best that I could do, I would tell them, is to install a GFCI for personnel protection (per NEC 406.3(D)(3))and a surge suppressor for equipment protection.

But today I was reading an old article by Rex Caudwell in which he says that "neither a surge arrestor at the panel nor a point-of-use arrester at the appliance will work properly without a high quality grounding system to dissipate the surge."

So I decided to investigate further and found me 250.130(c), in which I was surprised to learn that I can, indeed, run a separate EGC from an ungrounded recept right out to the GE or any point along the GE system.

So am I getting this right? And if so, how could I have missed this all these years?

--So I decided to investigate further and found me 250.130(c), in which I was surprised shocked to learn that I can, indeed, run a separate EGC from an ungrounded recept right out to the GE or any point along the GE system.

So am I getting this right? confused And if so, how could I have missed this all these years? blush crazy--

many miss this,not the preferred practice but compliant.
unless in Massachusetts. Not allowed here.

Wow. Well, I'll be darned.

BTW, would you tell me then, why this is not the "preferred practice"? Is it for the reason I stated? (or tried to state, in a rather bumbling fashion -- something about multiple paths to ground?)

Most like to keep all conductors for a ckt together.Or a new ckt for the device.

Electrically it's fine.

Thanks again.

The real problem is how you actually connect this. If you read 300.3(A) it says you can't just string THHN back to a ground point. (not a chapter 3 wiring method)
On the other hand I think if it is suitably protected that is better than having ungrounded outlets.
I have heard of running a new grounded circuit in the accessible areas and tapping off that with fished grounding conductors down/up to the ungrounded outlets through the wall where they are somewhat protected.
I would talk to the AHJ and see what they think.

Personally I would just run new cables for grounded outlets in the locations that actually use the ground. (most consumer products have 2 prong plugs anyway)

Okay, Greg. Lts more code to check out. And you're right: the AHJ is a very good idea.

As for "I have heard of running a new grounded circuit in the accessible area and tapping off that with fished grounding conductors...to the ungrounded outlets through the wall where they are somewhat protected" -- if I am able to fish a ground, usually I can fish the entire NM cable (as you suggest in your next paragraph).

"Most consumer products have 2 prong plugs anyway": I am looking at the adapter for this laptop rigth now and it is 3-prong!

As was pointed out, if your plug has a ground pin, the equipment expects a real ground. You are taking your chances with an adapter. I certainly would not connect an ungrounded PC to a grounded phone line. Your modem becomes the fault path.

300.3(2) says we can run the bond/ground alone. 300.4, protection must be considered as well.

That's what I meant by 'Preferred' method Tripp.
If you can snake in the ground,(chances are you need to gt all the way back to the panel anyway) A new ckt is just as easy/hard.

I'm sorry, but I was under the impression that stringing in a new ground was a practice no longer allowed. That is, all conductors had to be in the same cable / conduit.

I remeber this being depicted when grounds were introduced, and I have encountered houses built with a separate ground strung along with the 2-wire romex. Yet, I have heard many assertions that you can't do this any more ... nor can you just tie it to any handy piece of plumbing.

I do see the need for a real ground path, though. Electronic appliances often use the ground path ... either to always have power ("instant on," timers, motion sensors, etc.), or in order for the surge suppressor to work.

Prior to 1993 NEC all the interior metal water piping was considered part of the GEC system. The NEC 1993 allowed only within five feet of entry to the building as the point of ground because of all the non metallic piping.
Many pre-'93 homes that were upgraded will have three wire receptacles on two wire circuits with the ground going to the closest water pipe. This was especially common for the GFI outlet in the bathroom (required in 1978) until the exception to 210.7(D) in the 1984 NEC allowed a GFI without a ground.
The information that a surge protector will not work without a ground is absolutely correct. Without a ground it has no place for the surge to go. Also a non-ground GFI will offer no protection to the computer even when used with a surge protector.

Along these same lines, I wrote a violation to a contractor who sold the home owner on the IG receptacle and then wired it with 12/2w/ground NM cable and used a non-metallic box. "What's wrong with that?" he said.

Alan - I didn't think even a grounded GFCI was there to protect equipment. I've always thought of GCFIs as personnel protection.

Tripp, you're technically correct. However, since GFCI's have that electronic component, they manage to provide some surge protection ... by letting their smoke out!

Actually I imagine the trip through the little toroid transformer might filter out some small transients. I know an engineer who said a simple overhand knot in the cord would filter out some very high frequency noise. I don't know how true it is but I am never in a hurry to untie a knot that develops in the cord of my electronics.

GEORGE: "...an IG receptacle...wired...with 12/2w/ground NM and used a nonmetallic box."

Well, what IS wrong with that? As long as that EGC goes directly back to the service. (I love these opportunities to expose my ignorance.)

-Tripp

GREG - Now you're just hurting my head!

- Tripp

12/2 w grd is not isolated.

Now Isolated ground may require a new thread.

(Wink,Wink,Nudge)

But first, Break out the code book, You'll need it to follow along.

The Violation was..........?

Sounds like someone is worried that a metal faceplate might be used.

Okay gang- I was baiting you as is the habit with me.
Reno has covered part of it in that the plate would not be grounded and even if one were to use a non-metallic plate and nylon screws the yoke would not be grounded. There is an exception for a work around it in 406.2(D) so that's covered but the contractor would not be permitted to use the uninsulated conductor of the 12/2 w ground on the "Isolated Ground" terminal per 250.146(D)

What really gripes me tho is that the contractor had sold him a bill of goods on how much better system this was and I couldn't say anything to anyone but the contractor about his unprofessional deed.

Baiting is good!

If this one is so sly,would they not have grounded the yoke with the same conductor?

You guys are killin me. This is better than a Hitchcock movie. So don't quit me now!

- Tripp

Tying an overhand knot or putting a loop in a piece of wire increases the inductance and can, indeed, attenuate high-frequency noise and dampen some transients. Wrapping the cord around a ferrite choke will be more effective since the pereability of iron is so much higher than air, but just a loop in air will help, too.

A single loop has negligible impedance vs 60Hz, but don't go overboard or you could end up with a DIY power factor "correction" inductor and increase circulating current in your circuit.

Any intentional use of power should never use the ground path. The grounding conductor is used for grounding of any metal parts (safety ground) and surge suppression.

Most surge suppressors have 3 separate surge suppression devices in them. One protects against line-to-neutral (L-N) surges, one protects against line-to-ground (L-G) surges, and one protects against neutral-to-ground (N-G) surges. Without the ground, the L-N protection should still work, but the L-G and N-G will not.

"Should" is a fine concept, yet it is a fact that the ground path is utilized by a variety of products as a conductor for minor amounts of power.

This is, perhaps, why the NEC only bans "objectionable" current on the ground.

I've lost count of the sundry light fixtures and replacement ballasts that come with the caution "will not work without a good ground" marked right on them.

I know that some ballasts rely on the grounded metal close to the lamp for starting. Your original statement was about electronic equipment, not lamp ballasts. There is no reason a computer, TV, stereo, etc. should use the ground for power. Modern electronics all run on low DC voltages (12V, 5V, 3V, etc.), and these DC voltages are produced by a power supply. The power supply only needs the line and neutral. The ground is there only for safety.

Again, fluorescent lamps (and perhaps other ballasted lamps) are a special case, and some of those do need a ground for proper operation. Computers, stereos, etc. do not. (An earth ground may have an impact on system noise on a stereo.)

Wee, I'm not about to get into word play; nor is it my ambition to dissect everthing I encounter.

The ballasts with such lables have all also touted themselves as "electronic." Whatever is inside the mysterious black box, they seem to have left out the heavy, gooey, tar.

Likewise, we have documented, on this very site, other electronic devices that most definitely had current to ground in normal operation. Motion sensors, photocells, and even SOME illuminated switches utilize the ground as a conductor.

I can also point to the 'wall switch' type intermatic timers as something that includes the 'we need a ground' statement.

At least one type of surge supppressor relies upon the ground path to rid itself of excess voltage. That this may not be the 'latest' type is immaterial; for the purposes of this discussion, it's enough to know it exists.

It's getting pretty hard to find things around the home that don't use electronics .... I'm pretty sure my toilet doesn't - but I do have an old one!

I can understand that devices that function as switches (photocells, timers, etc.) may use the ground as a current path, since there is often no neutral in the box.

Yes, surge supressors will clamp surges between line or neutral and ground, and require a ground for that to work. The line to neutral protection should still work, but that is only 1/3 of the total function. Some of the more elaborate surge suppressors may use the ground for the line-neutral protection.

The original post was referring to a computer with an ungrounded outlet. Therefore, I was thinking mainly of electronics that plug in, not hard-wired devices.

In any case, having an actual ground is preferred even if the device still functions without it.

Sorry for any misunderstandings.

(I think you can find toilets with electronics in Japan, among other places!)

Brian- The "Ground" is not a normal current path for the items you mentioned. When a switch for example is turned on, the two wires (single pole switch) conduct the only current in the switch circuit. If one of these two wires shorts out to a grounded surface such as a metal conduit, then you are are correct the conduit now is a current carrying conductor. Quite often a switch, photo cell or mechanical timer are not fed with a neutral. If a switch has a pilot light on it or a timer has a motor then we would run a neutral to it.

I don't mean to talk below your level of training but I was confused by your statement about the switch, timer etc. If I miss understood your post I apologize.

George,

It is renosteinke that is claiming these devices use the ground as a conductor, not me.

I agree that those devices should be using a neutral, not the ground, if they need a current path other than through the switched load. Perhaps renosteinke has found devices that "cheat" and use the ground instead. (In my opinion, such devices should not have a UL listing.) My experience is with electronics that have a neutral connection (cord-connected or hard-wired), so I am not very familiar with the workings of devices such as photocells, timers, etc. that use various "power stealing" methods. I agree that they should not use ground as a current path, but I've also seen enough poorly-made products to believe that there are some out there that do.

When this issue of 'the ground as a make-do neutral' comes up, I think we run into two issues.

The first is the natural outrage of any properly trained sparky. "WE don't do that" and "that can't be right" are proper reactions.

Countering this is are the practices, even necessities, of electronic design. A 'dirty little secret' among the engineers and testing labs is that 'off' doesn't always really mean completely off, and that ground gets used for 'insignificant' amounts of current.

Which, of course, catches us on the horns of a dilemma: do we admit this, or even talk about it?
We have to know about this in order to do our jobs.
Yet, we don't want some uneducated noob to skim over a thread, and walk away thinking 'I just learned a neat new trick - I can't wait to use it - I don't really need that 4th wire to my new range after all!'

Getting back on track of this thread .... sometimes you really need a ground wire back to the panel; a GFCI might address safety concerns, but it will not create a ground path for you.

Perhaps we also need to recognize that nothing lasts forever. For example, my house was built in 1940, using the original version of Romex. That is, no ground wire. Maybe - just maybe - it's unrealistic to expect to continue limping into the 22nd century with an electrical system that is no longer adequate. Maybe there IS a time to knock it down, and start over!

I did see an article in the IAEI rag many years ago about using the ground for some very small loads (Microamps) but I thought the U/L standard changed and they couldn't do that anymore. Virtually every "2 wire" timer or illuminated switch I have seen in recent years uses the load as the return path.

George I had you the first time. You are right, without 4 wires you can't properly ground an IG receptacle unless you just loop the EGC through the yoke to the IG terminal and at that point you could have saved $5 with a regular duplex.
Of course regular readers here know that I think IG is snake oil, an opinion shared by the IBM physical planning manual.

Greg, what changed "a few years ago" was that a switch ould only say "off" IF it truly cut off the power. Illuminated switches - an example of a switch that doesn't completely cut off the power - can't SAY "OFF" if they power the lamp by letting a minor amount of current through.

(By way of contrast, if an illuminated switch did disconnect the power, using a neutral or ground path to allow the bulb to light, then they could say "OFF").

That is the "off" and no current in the circuit rule.
I thought there was another one about circuit current on the EGC.

Here's an example of an occupancy sensor that didn't have the EGC connected properly, and didn't operate properly at all. It depended on the mounting screws for grounding, instead of the EGC pigtail on the unit.




The same model occupancy sensor had a neutral connection in the past, previous models to handle the load of the electronics.