AFCI and UPS?

[Nate Nagel]

Hi all,

got a weird issue. In an upstairs bedroom, I have a computer that is running off a UPS. After investigating the wiring in the house, I have found a significant number of grounds "bootlegged." Little by little I am picking away at them, but until spring, the upstairs will remain as is because it's darn cold in the attic at the moment.

So here's the deal. I have determined that there is one properly grounded outlet in the room where the computer is, because that outlet is fed directly from the breaker panel, and that homerun is in BX not the cloth covered Romex that the rest of the concealed wiring is. So as a stopgap until I get around to fixing everything correctly I wanted to make sure that the computer's UPS was plugged into that particular outlet for proper surge protection. I bought a heavy extension cord, unplugged the UPS from the outlet that it was plugged into, plugged the extension cord into the correct outlet. Then I plugged the UPS into the extension cord, and the lights went out. Reset the breaker, figuring that I just made an arc, if I plugged it in faster it would be OK. Same effect. Reset the breaker, plugged the UPS back into the outlet that it was using before, and everything is fine. This is a less than year old APC unit.

Is there some fundamental incompatibility between this UPS and the AFCI? I'm wondering if the UPS tests for the presence of an earth ground (there is a "building wiring fault" light on it) and if so does that cause enough current to trip the AFCI? (I wouldn't have found this before, because the outlet into which it's currently plugged is not grounded but has the ground bootlegged to the neutral.) I've heard that some AFCIs are also GFCIs is why I'm asking. The AFCI is a Siemens Q115AF breaker in the breaker panel.

I'll replace the receptacle in the morning (once it's light out) and see if that solves the problem, but I'm not sure what kind of fault could be in a receptacle that would cause a breaker, AFCI or not, to trip only when a load is connected to it but not when a plug is inserted.

Does this mean also that I run the risk of having the AFCI trip if the surge protection of the UPS kicks in?

Should I just give up the idea of having everything "to code" and ditch the AFCI and/or investigate the possibility of running a dedicated circuit to that bedroom to feed an outlet solely for the computer, not protected by the AFCI?

any thoughts greatly appreciated...

nate

[Charles Schuler]

AFCIs are rather new. Any installation that new should not have faked grounds. Did you mean GFCIs?

[Nate Nagel]

No, I put the AFCI in myself after purchasing the house. However, a lot of the wiring is late-40's vintage. That was a good enough reason in my mind to think that an AFCI would be if anything more important here than in new construction.

nate

[John Grabowski]

Is that circuit part of a three wire cable (Black, red, white) used for two circuits or is it just a two wire bx with the armor as the grounding conductor?

[w_tom]

First, break the problem into parts; then diagnose those parts. For example, do you have other GFCIs in the bathroom or kitchen? If so, then good. Run that computer extension cord to that GFCI; see computer or extension cord causes the other GFCI to trip.

Have you run anything significant on that tripping AGFI circuit? For example, if the AFGI circuit is somehow wired so that its neutral (white) wire is started with another circuit, then AGFI trip will occur. For example, an iron on that newly AGFIed circuit would be a good test.

If any newly modified safety ground has somehow shorted to a neutral wire, then that also will trip the AGFI. But long before taking anything apart, first verify which suspect (computer and extension cord, or AGFI circuit) causes GFCI trip.

Second, that plug-in protector is too far from earth AND is hoping to earth via a wire bundled with other wires. Just two in a long list of reasons why a power strip protector does not even claim (in numerical specifications) protection from surges that would cause damage. The length of that earthing wire (a short distance to earth and a significantly long distance between computer and surge protector) has long been necessary for effective protection. That also means the 1940s mains box must be upgraded so that earthing both meets and exceeds post 1990 National Electrical Code requirements.

[Nate Nagel]

It's two conductor BX with the armor grounded.

nate

[John Grabowski]

AFCI circuit breakers are not required here in New Jersey, so I don't have much experience diagnosing problems with them. My thoughts are that your AFCI does not like the BX armor ground, there is something else going on with the BX cable, or it does not like your UPS. Have you tried plugging in another three prong appliance into the outlet?

[Nate Nagel]

No, that outlet has been unused since we've moved in. It is however the first one in the circuit so the only possibilities I can see are that the AFCI is actually a GFCI as well (anyone know?) and there's enough current flowing through the real ground due to the UPS's fault sensing circuit, OR that there is a physical problem with that receptacle, which I'll be attempting to rule out shortly, now that the sun is up. (I've been on a program of replacing all the receptacles as soon as wiring "issues" are cleared, because a lot of the old receptacles were loose as well.)

nate

[RBM]

AFCI's do have GFCI functionality, but they trip at 60 ma instead of the 5 ma of a GFCI

[Nate Nagel]

Update:

I replaced the receptacle. when I turned the breaker back on it tripped immediately. I popped the cover on the panel and ohmed everything out, seemed OK although the resistance hot to ground was lower than I thought (~1 meg; nothing plugged in) although granted my meter is very old and possibly inaccurate. Everything seemed OK so I tried pulling the wires

*leaving* the receptacle I'd just replaced. Then reset the breaker. Fine. Plugged in UPS. Fine. Reattached wires and replaced breaker with regular 15A breaker. Fine. Tried another AFCI breaker (I just so happened to have one laying around because I wanted to split the upstairs into two circuits eventually.) Tripped immediately. Removed the wires leaving the receptacle again, new AFCI did not trip.

So what I apparently have is something is causing the AFCI to electronically trip but not because of overcurrent. It's got to be a wiring fault because I have everything unplugged. and all I know is that it is somewhere upstairs - one circuit serves the whole second floor, except for one outlet in the hallway (apparently intended for an air conditioner.)

I'm not sure why it tripped only when I plugged the UPS in before, coincidence, or just reached some kind of threshold? who knows?

F'ing great. Of course it's about 10 degrees outside, and I assume most of this wiring is in the attic. I'm a little too paranoid to just leave the regular 15A breaker in... or are the Siemens AFCI breakers known to be problematic?

nate

[Nate Nagel]

Heh... all right, I feel dumb now.

When I started to replace the receptacle, I plugged the UPS and the associated extension cord into a receptacle in the hallway which was on a different circuit.

I just went upstairs to start pulling receptacles out of the wall to inspect the wiring/try to track down the presumed ground fault. Despite the fact that I thought I'd unplugged every appliance upstairs, what did I see at the first receptacle I came to, but a printer plugged in to a wall receptacle. It, of course, was connected (by a USB cable) to the computer that was connected to the UPS. I feel obligated to point out at this point that I did *not* connect it this way, and in fact, the reason that I feel particularly stupid is I was just explaining yesterday why it was important to connect all peripherals of a computer to either the same power strip or the same UPS so there's no issues with floating grounds etc. and voltage imbalances that end up going through USB or other ports. (was actually on my list of things to rectify.)

I dug out another extension cord (the reason the printer was plugged into a different receptacle is that it is on the other side of the room from the computer desk) and plugged the printer into the back of the UPS on one of the "surge only" receptacles. All appears to be good now.

I'm still not sure why the breaker tripped when I simply tried to move the UPS' power feed from one receptacle to another, unless the fact that the UPS now had a proper ground while the printer still had a bootlegged ground was enough to cause a problem.

I ASSume that this little exercise does indeed confirm that the Siemens AFCI is also a GFCI...

nate

[w_tom]

Yes the AFCI is a GFCI (plus). What the USB printer is plugged into should make no difference. Implied is that something is leaking via the printer. Printer on another circuit is not desireable only because backup protection is missing. But if working properly, powering from both circuits should not have contributed to AGFI tripping.

Meanwhile, temporarily disconnect neutral wire and safety ground wire of the AGFI circuit in mains box. Infinite ohms should exist between those two wires. Also infinite ohms should exist between that disconnected neutral and mains box. All tests best performed with AC electric off so that leakage and other problems do not distort that ohm meter reading nor harm meter.

Disconnected safety ground might or might not conduct to mains box - that is acceptable. But disconnected neutral must not measure conductivity to disconnected safety ground nor to mains box (test obviously done with appliances unplugged).

One final test. Ohm meter test (use extension cord to measure this) should measure conductivity from receptacle safety ground prong to disconnected safety ground wire in mains box. If this conductivity does not exist, then that test explains everything seen previously and suggests problem still exists.

Testing appliances (computer and printer) on another GFCI circuit, and tested with computer and printer split between GFCI and non-GFCI circuits should still be performed as posted earlier. Information from that test is necessary - and would provide more useful information for many reasons including because GFCI trips at only 5 ma.

AGFIs have been required for years now in New Jersey on bedroom circuits.

[Nate Nagel]

all I can say is that it apparently did, because moving the power source to the UPS solved the problem. I printed a test page after hooking everything back up, so there appears to be no damage (thank you AFCI)

I can't disconnect the "safety ground" because it grounds to the breaker box through the cable clamp... there's no separate ground conductor in the BX. (the wiring is cloth covered as well, and appears to be tinned, to give you an idea of the age of the stuff I'm dealing with.) I will test hot and neutral just to satisfy my curiosity however now that I have located *all* the appliances that were plugged in and can disconnect them all... When I get around to rewiring the second floor, I may see if I can drop a new homerun to the box, but I'm guessing that the BX has at least one hidden staple somewhere along the way. It'd be nice if I could do that though, because then I'd just pull some 12/2WG Romex and switch to a 20A breaker (assuming that everything else on that circuit is accessable from the attic.)

I could try that... I'd have to actually wire in a GFCI receptacle upstairs though, would be easier than moving all the equipment to the basement or kitchen :) (I actually do have a couple laying around though for future use in garage etc. so it's not a huge deal.)

They've probably been required here for years as well, I don't think this house has had any electrical upgrades in the last 20 years however (previous owners were not really DIY types, and were here 18 years - while the owner before that was apparently a contractor, so I'm guessing he is responsible for a lot of the work) and I'm guessing that the work wasn't inspected then, ref: bootlegged grounds mentioned above.

nate

[John Gilmer]

Unless the authorities "know" about what you are doing, I suggest you replace the AFCI with either a normal breaker or jus a GFCI breaker.

The AFCI is quickly developing a reputation among the "pros" as something that will cause the homeowner to make a lot of trips between the bedroom and the basement.

If the NEC really, really considered arc faults to be a special danger in bedrooms, it should have considered making arc detection set off an alarm rather than cutting off the power.

This scheme (AFCI for bedroom circuits) is the equivalent of having a smoke detector turn off the lights rather than just make a loud noise.

[Mark Lloyd]

you're supposed to light a candle :-)

[Bud--]

Complete bullcrap.

Manufacturer specs for protection are readily available for plug-in suppressors.

Both the IEEE guide on surges and surge protection at:

the NIST guide at: that plug-in suppressors are effective.

Plug-in suppressors, as explained in the IEEE guide, work primarily by clamping the voltage on all wires (power and signal) to the common ground at the suppressor. They do not work primarily by earthing.

-- bud--

[John Gilmer]

I was a little too hard on the NEC. The AFCI will cut off a circuit well before a loose connection gets hot enough to actually be a safety problem.

It does happen. I found a loose wire when I realized that a plug for a lamp was WARM. Years ago the fire department responded to a small fire from a "hot" outlet.

I hope that withing a few years the AFCI manufacturers/NEC will sort things out. In the past the NEC as paid attention to the problems of false tripping.

For the time being, I would suggest that folks don't volunteer to get them installed.

[Bud--]

The AFCIs out there now detect parallel arcs - line to neutral or ground (a fault). They don't detect a loose connection - a series arc. Series arc protection is required starting Jan 2008. Far as I know these devices are not yet on the market.

Parallel arcs can draw high current and I have read AFCIs detect pulses around 60A. Series arcs are limited by the load (will usually be lower than the breaker amp rating) and should be much harder to detect and separate from "normal" arcs.

Starting in the 2008 code AFCIs will be required on all 15 & 20A 120V residential branch circuits. Coincidentally these will be the new AFCIs that include series arc detection that aren't yet on the market - large numbers of devices with inadequate field experience will be installed. In this instance perhaps you aren't hard enough on the NEC. It will be interesting if all the panel manufacturers even have the new AFCIs by 2008.

At least the AFCIs out there now have a track record. Advice about not volunteering sounds much better for the new AFCIs.

-- bud--

[John Gilmer]

I guess I don't understand WTF is going on.

A "series" fault is, exactly, what?

My hope is that the gadgets detect a loose connection that might result in something overheating.

What, exactly, DOES the gadget detect?

[w_tom]

Often bedrooms have devices plugged in using a two wire (zip cord) extension wire. The bed continuously rolls over that wire until eventually the wire arcs - causes a fire. AFCI will detect that short and cut off power before an arc can cause a fire. This is why AFCIs are required for all bedroom circuits.

However, if you use a real Christmas tree, then the outlet that powers lights on that tree should also have an AFCI so that a short and resulting arc in Christmas tree lights do not create a fire. A Christmas tree fire created when his wife only turned on lights took out the entire house - killed almost all pets - in but 5 minutes. These are arcs that kill and that an AFCI quashes.