Poltergeist?

[Gustavus]

Well, not really, but that is as good an explanation as any.

In my home flourescent lamps "flash" at about ten second intervals when turned off. It is especially noticeable at night after ones eyes have become dark adapted. The mains are well grounded with a heavy copper strap, but it is not a DC static build up anyway which is why I am posing the question here. An oscilloscope placed on what would be the hot lead of the turned off circuit shows an AC -- 60 Hz of course -- voltage that increases to about 85-90 volts at which point the flourescent lamp ionizes and discharges the voltage back to around 40 volts. The AC voltage then increases exactly as though a capacitor were being charged through a resistor -- but this is AC not DC -- until the voltage reaches the breakdown voltage for the lamp again. There is no question I am seeing an RC exponentially increasing amplitude to the AC buildup as confirmed by measurement with both the oscilloscope and a high impedance multimeter.

I recently had as house guest an electrical engineer who is chairman of the EE dept at his university who said what I have just described was impossible. We spent half a day and made all sorts of measurements -- duplicating measurements I had made many times -- which only confirmed the AC voltage level on the turned off circuits was increasing as though it were a capacitor charging up. If it were induction from one lead lying near another the voltage under no load should be constant -- besides which it is happened in multiple circuits. We did confirm it is only happened on circuits on one branch of the 220 mains coming into the house.

This is not a new problem. We have to put incandescent lamps in the guest bedrooms, since the flashing of the compact flourescent lamps keeps guests awake. In the main bathroom we have had to disconnect the fixture over the sink which had a wireless control for the flourescent lamps, since eventually the flashing would turn the fixture on and stay on. It was this that made me describe the problem as a poltergeist -- who turned on the lights after we had gone to bed.

Anyone have any idea of what could be going on?

 

[Micrornd]

What country is the house in and what is it's age?
And are the outlets in the home equipped with a ground. (2 pin plug or 3 pin plug)?

 

[sm625]

I've seen it before myself. It leads me to believe it is common. If you want to get to the bottom of it, start unplugging everything except the lights, until you find the device that is causing this. I bet it wouldnt happen if one light bulb was the only load in your entire house.

 

[SecurityTheatre]

I'm going to also postulate that it's some sort of induction from a nearby wire somewhere. Perhaps given an interesting type of load on the line that causes the linear rise in amplitude.

I hated power systems so much, I've forgotten most of it, unfortunately. :-D

 

[Gustavus]

Micrornd
United States. 25 years old. No other electrical problems or unusual occurences. The loads on the two branches are well balanced.

sm625
That is one of the things the EE professor did that I had not done before -- lift loads on the circuit. The breaker panel for the house is a Square D with 40 single breaker positions -- 20 on each side of the mains. 18 of these are used to switch 220v for things like hot water heaters, range tops etc. We did not remove all loads, only everything but a flourescent lamp on the circuit we were investigating at any time. Didn't have any effect. Incidentally the house is wired with heavier wire than required by code; 10 gauge where 14 would have satisfied code, 8 where 10 would have satisfied code etc.

The baffling thing is that the amplitude of the AC is increasing exactly as you would expect the voltage to increase in a DC circuit in which a capacitor was charging through a resistor. What I am measuring is impossible -- which is what the EE professor said when I told him about the problem, and still said at the end of our effort to find an explanation. His final comment was; Beats the Hell out of me.

 

[Gustavus]

SecurityTheatre
If it was induction, which of course was one of the things we thought it might be, with no current drain, ie. no load, the amplitude of the AC would not be increasing exponentially. It would be like applying AC to the primary on a transformer and measuring the voltage level on the secondary with a very high impedance multimeter. You would see a steady voltage. If you loaded the secondary -- depending on the load of course -- you would expect the voltage to drop. Of course when the flourescent lamp ionizes -- which they do at around 90v -- that puts a low impedance load on the open circuit (the ionized gas in the lamp) which pulls down the voltage to a level below what is required to keep the gas ionized -- and the lamp lit. So after the brief flash of light, the lamp goes off. Once the lamp goes off, the unionized gas present a very high impedance and the voltage starts to build up until it reaches a level sufficient to ionize the gas again.

 

[Micrornd]

Have you verified that the neutral and ground are tied together at the point of entry of the electrical service AND that the ground is intact AND that the ground has less than 5 ohm resistance?

Check for compact fluorescents and incandescents in the same fixture as they can sometimes oscillate and induce a feedback also.

Once you have done that, turn off or disconnect all capacitor start and/or capacitor run devices or appliances (ceiling fans, refrig, well pump, etc.), both 120v and 240v.
That should stop the feedback.
Then you can turn the items back on, one at a time, checking as you do, and when the problem returns, you've found the culprit.
(which is basically what sm625 pointed out)

 

[Jeff7]

I've seen this flash from some CFLs, sometimes up to a minute after they've been turned off.
Someone tell me if I'm way off base here....
I think that it's likely just something resulting from capacitor discharge - as various caps in the power circuitry discharge, parts of the circuit may behave unpredictably, and you might end up with a brief arc inside the tube. That's my guess.
The output voltage of a capacitor is depressed while there's a load on the circuit. Suddenly, the main feed is removed, and the capacitor's output voltage is no longer able to power the load, so the load suddenly vanishes. Now with no load, the capacitor slowly accumulates some charge again from what was left strewn around its guts, until it's once again able to power the driver circuit for one final, brief hurrah. (For example, if you discharge a CRT just one quick time, there may still be some residual charge that's still capable of delivering a nasty zap, once it's had a little bit of time to sit.)
And so, you get a very short flash of light, which has the effect of draining the offending capacitor even more, leaving it unable to start up the driver circuit again, until it receives power from the line.
(Or you might get multiple flashes, depending on the CFL. I've never seen more than one flash after power-off though.)

I seem to recall having seen that effect with high-value capacitors, after the power feed had been removed (I got to play with some 82F caps at work ) - with a load, the voltage continually declines, as you'd expect. Remove the load, and the voltage begins to climb back up, and resume its normal sluggish drop due to internal leakage.

So I don't think it's induction or anything else, as the affected lamps all have the power switch very near to the socket, and they're switching the hot wire off. The lamps are metals or composites, that doesn't seem to matter. And the socket is only hot and neutral; I don't think I've ever seen an earth ground connection on one. It seems that it's mainly the brand of CFL that matters - some do it, some don't.



I've seen a more rapid flickering effect in my alarm clock, which is probably related: It has a dim green electroluminescent strip as its backlight, and an internal inverter circuit to generate the voltage necessary to drive the EL strip. My guess is that it's similar to what's going on in the CFL, except there are several flashes in rapid succession. As it happens, I can hear the frequency of the inverter circuit decreasing, so it's probably becoming less energetic as the capacitor discharges.

 

[Gustavus]

Thanks to those that have responded. Serendipitiously I have a new bit of data. I am in New Mexico and we are in the midst of a long running severe drought. This afternoon we had a heavy rain -- the first in months -- and some of the CFL's that were flashing before are not flashing tonight. Others still are though. The system is grounded with a heavy copper strap to the waterline, but rain will have decreased the impedance of the return path since the surface ground is now wet. I can't see any reason for some of the flourescents to be affected and not others. But the rain has made a difference.

Jeff7

This is not a flash or two right after the lights are turned off. As I said they flash all night and keep my guests awake, so I have had to replace CFLs in the guest bedrooms with incandescent lamps. It takes about ten seconds for the AC voltage on the turned off hot lead to rise to the point where the lamp flashes, then it repeats over and over -- all night.

 

[Jeff7]

Thanks to those that have responded. Serendipitiously I have a new bit of data. I am in New Mexico and we are in the midst of a long running severe drought. This afternoon we had a heavy rain -- the first in months -- and some of the CFL's that were flashing before are not flashing tonight. Others still are though. The system is grounded with a heavy copper strap to the waterline, but rain will have decreased the impedance of the return path since the surface ground is now wet. I can't see any reason for some of the flourescents to be affected and not others. But the rain has made a difference.

Jeff7

This is not a flash or two right after the lights are turned off. As I said they flash all night and keep my guests awake, so I have had to replace CFLs in the guest bedrooms with incandescent lamps. It takes about ten seconds for the AC voltage on the turned off hot lead to rise to the point where the lamp flashes, then it repeats over and over -- all night.

Ah, ok. I hadn't seen anything about it continuing all night. Oops.
(The flash from my CFLs, sometimes 1-5 minutes after the lights are out, is enough to startle me to fully-awake - I'm not a fan of loud noises, which thunder certainly qualifies as, so the flash snaps me to alert-status. I had assigned that same sort of thing to your guests.)



Hm...so, static then? Or some kind of high-frequency electrical noise that's momentarily ionizing the gas? Odd indeed.
I know that a Tesla Coil can light up fluorescent lights from a distance, with no direct physical contact.

And a long-running drought? So your house may be floating electrically, rather than grounded. Hm...
Saint-Elmo's Fire from your CFLs?



Edit: So, I went and hit up Google. It appears that this is not an isolated incident.

Have you got an illuminated lightswitch?
One result I found suggested a bleeder resistor of 240k in the socket.

Or, maybe it is indeed induction from a long wire run somewhere. I know that we had to do something special at work for our 277VAC power supply - long runs of 277VAC would end up inducing enough voltage in the neighboring wires that they'd be capable of turning on LEDs that were meant to be unpowered. That design was done before I started at the company, but I'm told that it was quite interesting to troubleshoot that problem. "The LEDs are turned off, but they're still lighting up, and yes I've checked my wiring."

 

[Matt1970]

I have one that does that but it's because it's on an illuminated wall switch.

http://www.instructables.com/id/Stop-CFL-lights-from-flashing/

 

[Gustavus]

To clarify -- well I hope it clarifies -- it is not just CFL lamps that flash on momentarily when turned off. Two of the lamp assemblies in the house have the 4 foot long 40 watt flourescent tubes in them. One of those is built into a bar station and is switched on and off with a non-illuminated switch. The other is the unit mounted above the counter in the main bathroom that was controlled with a wireless remote. That one had to be disconnected from the power since it was turning on in the wee hours of the morning -- all on it's own. The CFL lamps in the guest bedrooms are large table lamps using 100 watt bulbs. Those are the ones we had to replace with incandescent bulbs since as I said the brief flash every ten seconds was keeping our guests awake. The CFL lamps in the stairwell are controlled by a pair of illuminated switches at the top and bottom of the stairs. Those are the ones that stopped flashing after the heavy rains yesterday.

What one see on an oscilloscope attached to the hot wire going to a lamp as soon as the lamp is switched off is a normal 60 Hz signal which starts from zero amplitude and slowly -- maybe 20 seconds -- increases to around 90 volt amplitude. At that point the lamp flashes, the AC seen on the oscilloscope drops back to around 40 volts amplitude which then rises back exponentially to the breakdown voltage of around 90 volts in ten seconds or so. Then the whole thing happens again. It is not static buildup -- or static discharge. The oscilloscope shows clearly that it is a 60 Hz AC signal building in amplitude. If it were an AC component imposed on top of a static buildup, the oscilloscope trace would show the AC signal shifting up or down depending on the polarity of the accumulating static voltage. It is the amplitude of the AC that is increasing exponentially. That is the thing that is impossible. If it were simply an accumulating static charge, an explanation would be that the grounding is at fault.

The power to the linear flourescents comes from a transformer in the fixture, not directly from the line. The CFL bulbs are on the line. It is not easy to get an oscilloscope probe on the terminals of the linear flourescents -- but I may try jury rigging a wire into the socket on one of them to see what those bulbs are seeing. For lamps controlled by wall switches it is easy to simply open the box and clip the oscilloscope probe on the terminal going to the lamp.

 

[DrPizza]

Just a thought - could the bulb itself be behaving like a capacitor or leyden jar? Toss a little dust on the outside, and you already have charged particles inside. That might also point to why weather is apparently having an effect on this phenomenon.

 

[Micrornd]

Have you verified that the neutral and ground are tied together at the point of entry of the electrical service AND that the ground is intact AND that the ground has less than 5 ohm resistance?

Check for compact fluorescents and incandescents in the same fixture as they can sometimes oscillate and induce a feedback also.

Once you have done that, turn off or disconnect all capacitor start and/or capacitor run devices or appliances (ceiling fans, refrig, well pump, etc.), both 120v and 240v.
That should stop the feedback.
Then you can turn the items back on, one at a time, checking as you do, and when the problem returns, you've found the culprit.
(which is basically what sm625 pointed out)

Gustavus, have you checked this?

 

[Gustavus]

Micrornd

Sorry to be slow in responding. Yes the neutral and ground are strapped together at the point of entry. The ground is a heavy copper -- either 1 or 2 -- which is connected to a 3/4" galvanized water pipe with a saddle connector on the pipe. I have verified that the saddle is firmly clamped to the pipe and that the ground bus is firmly gripped by the screw on the saddle. Not sure how to measure that the ground has less than 5 ohm resistance. The galvanized pipe exits through the wall of the basement and is then underground for perhaps 50 feet before it is coupled to a much larger PVC supply pipe.

 

[Gustavus]

Micrornd

It is only peripherally related, but a bare copper ground wire runs to every outlet and device -- connected firmly to the green screw as it should be. There isn't a floating box or device in the house. All wall plugs are code two blade with round ground pin with the ground connection made to this ground wire, or in a few cases code three blade for 220. The main breaker in the Square D service box is 200 amp and no circuit is loaded anywhere near rated capacity. For example there is no momentary dimming of any light when motors -- air conditioner, refrigerator, washer etc. cut on.

 

[Micrornd]

Micrornd

Sorry to be slow in responding. Yes the neutral and ground are strapped together at the point of entry. The ground is a heavy copper -- either 1 or 2 -- which is connected to a 3/4" galvanized water pipe with a saddle connector on the pipe. I have verified that the saddle is firmly clamped to the pipe and that the ground bus is firmly gripped by the screw on the saddle. Not sure how to measure that the ground has less than 5 ohm resistance. The galvanized pipe exits through the wall of the basement and is then underground for perhaps 50 feet before it is coupled to a much larger PVC supply pipe.

This brings up a problem.
Your system is not properly grounded.

When the house was built, it is safe to assume it was grounded properly to the cold water piping per code.
You indicate the incoming water supply is a galvanized pipe (if so the house is older than 25yrs) connected to a PVC feed.
Originally the galvanized line would have been connected to a larger galvanized line or to a ductile or cast iron water main (or steel casing in the case of a well.
At some time the feed piping was updated to PVC, but in doing so this decreased the actual length of what was your grounding system from what may have been miles of relatively shallow pipe, to feet.
(That was one of the reasons I asked the age of the home.
Many homes in the US have this same problem in that when the water mains in their areas were upgraded or replaced, their electrical grounding system was/is no longer adequate.)

Galvanized pipe in the ground corrodes over time and as it does the rust and corrosion on it's surface acts as an insulator so it's resistance to conducting electricity increases.
As you know electricity takes the path of least resistance (hence the reason for the 5 ohm check).
This is why when it rained you saw part of your problem disappear, as wet ground around the galvanized pipe (and the pipe itself being damper than usual) decreased it's resistance, and it became a better ground.
When the rains stop and the ground around the galv. pipe dries out again, those problems will most likely return.

A proper ground to meet code these days consists of 1 (min) 1/2" to 5/8" copper plated steel rod(s) driven from 10' - 60' (or multiple rods driven to shallower depths forming a grid) that achieves a maximum resistance to ground of 5 ohms or less. (all this subject to local codes)

An inadequate grounding system is not only dangerous, but can also produce strange effects such as you describe since the neutral wire is bonded to the ground at the service entrance, and without an adequate grounding system can carry some voltage.

I highly recommend you have a licensed and bonded electrician check out you grounding system, at the very least, actually check with his meter, not just visually.
(Please only use licensed and bonded contractors, they will follow all std. industry practices and abide by all natl. and local codes)

Whether it will eliminate the "poltergeist" or not, (while I can't absolutely be sure without being there) I think so, but from what you describe, it will make things better and for sure safer.

BTW- You may also want to consider replace that galvanized water line with either copper or a plastic product (any local plumber should be able to tell you what is longest lasting for the soil type in your area).
The reason galvanized is no longer used in water supply lines is because it not only rusts through, but is also most prone to build-up of deposits on the inside (restricting flow) than copper or plastic products.

Hope that helps

 

[Gustavus]

Micrornd

Thanks for the very clear analysis and explanation. The house was built in 1979 -- so closer to 35 years old than the 25 I said. The ground here in New Mexico, even before the severe drought of the last several years is very dry. Caliche is a major component of the soil -- especially overlying layers of rock.

I will see about getting the house wiring properly grounded. Never occurred to me that was the problem since I am very familar with the in-house wiring and knew it to be done to code.

Thanks again for the help.

 

[mannox]

no idea. good movie though