I have what seems to be something wrong with a 3 way switch on the
light at the top of my stairs. This light can be turned off/on from
the bottom of the stair as well as from the top.
I was replacing the fixture at the top of the stair and turned the
power off at the switch thinking that I would be safe to replace the
fixture. After removing the screws holding the fixture to the ceiling
I checked the light with my greenlee gt-11 voltage detector and was
surprised to hear the unit ringing. I broke out the DVM and sure
enough 92 volts between the white and black wires to the fixture.
Turning on the light the voltage goes to 120vac, off back to 92vac.
So something seems very wrong, I went to the panel and turned off the
breaker and changed out the fixture. I am now glad I was lazy and did
it the wrong way first.
I don't understand why the incandecent bulb did not light at 92
volts. I have replaced the bulb with a CF as that is what we are
doing throughtout the house.
About the house:
1957 cape 2 wire hot/neutral with no ground for all parts of the
house. Any new work has been done with romex in proper gauges for
5 years ago we had the 100Amp fuse panel replaced with a 200 amp
Any common problems that might cause this?
Am I using electricity even though the light is off?
Think about how a 3 way switch works and the answer should be clear.
There's no "off" per se -- only the selection of which of two lines
is energized, and (on the other end) which of the two is selected
to be connected to the fixture.
The e-mail address in our reply-to line is reversed in an attempt to
minimize spam. Our true address is of the form email@example.com.
1. There should not have been any voltage between the wires with
either switch off, ie, the light went off.
2. It's likely that one of your switches has a high resistance
connection in the contacts even when the switch is off. 92 volts
between the wires doesn't mean the light should have come on. At no
load you may read 92 but if the hot wire, ie, switchleg has voltage on
it from a high resistance connection to a 120 volt source, the voltage
will drop to zero when you put a load across those wires because of
the high resistance source.
If you want to troubleshoot further, try turning off the other switch
and see if the voltage is still there. If it's not, replace the first
switch you turned off when you replaced the light.
Remove both switches from their wiring and see if the voltage is still
there. Use resistance scale on your dvm and check the switches.
Measure between the one screw that is different color then the other
two, and each of the other screws, one at a time and flip the switch
on and off and confirm 0 ohms in one direction and OL-infinite in the
other. If you get anything besides infinite in that direction, replace
You have some kind of short circuit connection in the switch or wires
that is still feeding some voltage to that hot wire when the switch is
off. Could be one of the switches or the wiring.
On 28 Feb 2007 18:07:40 -0800, " firstname.lastname@example.org"
I don't know but if so, it would be minute, not 92/110 times normal.
Apparently you have an induced voltage in that line, just like
voltages in the secondary of a transformer are induced by the current
in the primary. But the maximum amperage will be minute.
Go back and measure with a voltmeter that has about 50,000 ohms per
volt, instead of the 11 megohms per volt that iirc digital meters
have. This is normally the difference between an analog meter and a
digital meter, except for FET VOM's which are analog but have high
resistance. Are there many of those around still? They were only
popular for 10 years iirc. If you have an old one, it will say FET on
With 50,000 ohms per volt, the voltage will drop to below one volt,
but I don't know how much below. REport back to us.
It's easy to have voltage when the resistance is too high to let any
current through. Something like when you hold your finger over most
of the garden hose and the water shoots out 30 feet, but when you let
your finger off, it comes out 5 inches.
Thanks everyone for all the great responses, as for my meter it is
old, [made in early 90's] but does not say FET that I can see it is a
Fluke 8020A, I do have an old analog VM from radio shack tucked away
somewhere I will dig that up tonight and test again. I will measure
at the lamp socket this time as I won't have time to pull the fixture
until the weekend.
Measuring between the screw thread and the center point at the bottom
of the socket with the light off then on should be the same as what I
I also want to test the switches as they have all been replaced over
the time that I have been in the house. My wife did not like the
older beige switches with the heavy click action so they were all
replaced by me with leviton white 115v switches.
thanks to everyone again, running off to work now will report back
I would love to see this 92v. Everyone talks about phantom voltages, but I
have never seen one, nor has my voltage detector ever gone off on anything
but an authentic hot wire.
But on the other hand, if you really had 92v, the bulb would be glowing.
So it probably is phantom, but I would like to see it.
Did you test the other wire? Maybe it is 92v also, so the bulb sees not
voltage difference and doesn't light up. Highly unlikely, but not
On 28 Feb 2007 18:07:40 -0800, " email@example.com"
This could happen if you have the neutral switched instead of the hot.
The 120 - 90 drop could be because you are measuring with the lamps in
the circuit. If you take the lamps out and you do have a switched
neutral then tester would then read 0.
I thought you might be onto something, but couldn't actually see a circuit
where this would happen.
The unswitched hot would be at 120v. The switched neutral would be at zero
because of the switch. So I can't see how switching the neutral would
Not sure which situation you are referring to, but if you mean the hot
wire is connected to the bulb, and you are measuring voltage on the
other, the cold side of the bulb when the neutral is switched and the
neutral is open, there cannot be 90 volts there for that reason. There
would be a full 117 volts or whatever is measured at the receptacle or
at the hot side of the lightbulb.
E=IR. When I (the current) = zero, then E, the voltage drop accross
the resistor (R) is also zero. It doesn't matter how narrow the
filament in the light bulb is, with the neutral open and no current**
flowing through the lightbulb, there will be no voltage drop accros
the lightbulb or accross the wire that feeds it from the receptacle.
**But there will be some current when you attach a meter***. How much
current depends a lot on what kind of meter you use.
***I think this is a good example of the Heisenberg (sp?) Uncertainty
Principle, which says iirc that when one attempts to measure
something, he changes it. Not sure why that would always be true, but
physics was a long time ago.
But this is a good time to talk about the difference between a meter
that has 20,000 ohms per volt and one that has 11 million ohms (per
volt?, I don't know) For house current, 20,000 is plenty, but the
meters don't have the advantages of auto-polarity, and even auto
ranging might be useful sometimss. The change to digital meters might
have been started by the space program, but is driven by the fact that
20,000 ohms is not so good for electronics.
If you have a transistor that is connected by a 10 meg ohm resistor
to ground, and you want mesure the voltage on the non-grounded end of
the transistor, and you connect a 20,000 ohm meter across the
resistor, you now have two resistors in parallel, one the original 10
megs, and the other the 20,000 ohm meter. So the total resistance is
now about 20,000 ohms and current that might be flowing through the
transistor has, barring other factors, increased tremendously, up to a
factor of 500 since the resistance is now 1/500th of what it used to
be. IF otoh, you used a 11 meg meter, the resistance (impedance might
be a better term but they are both measured in ohms) of the two
resistances in parallel is about 5 megs, not such a big change.
So people in electronics, especially with transistors and not tubes
(but tubes too), needed higher impedance meters.
What is an advantage in electronics can be a disadvantage in
electricity. If you have a pretty high voltage for a home (like 90
volts) but one with a very low maximum current, and you connect a
digital 11 meg meter across it, the currrent draw will be very low,
and the voltage will not drop much because the voltage drop is
determined by E=IR and the resistance of the circuit is low compared
to the resistance of the meter. So it's like a big pipe feeds a very
teeny pipe, adn there is almost no current through the big pipe and no
If on the other hand one uses a 20,000 ohm meter, that will still be
quite a restriction to current flow, but only 1/500th and there will
be 500 times as much current, so the voltage drop will be 500 times as
While your description of what is hapning is ok, much of what you are saying
is not accurate. The 20,000 ohm meter is really that per volt of the full
scale setting. That is if you have the standard Simpson 260 meter
(concidered the best analog meter for electricians for years) and set it for
250 VAC there is about 5 meg of resistance in the circuit which is only
about half the resistance of the digital meter. In the problem above if it
is leakage or induced voltage it will read about half of what the digital
meter will. Also if you click to the next lowest scale you will see the
relative position of the meter will not change much if it is very low
Also many of the transistor circuits are high current low voltage circuits
and the tube circuits are high voltage low current circuits. The analog
meter will usually be more accurate in reading the voltages there than they
will be in the tube circuits. A 10 meg resistor in the circuit in parallel
with the 11 meg meter will register a very large change, not a small one
like you state. Anything over about 10 % is usually noticable and you have
a 50 % differance .
Well I have to say that you guys are well over my head, but as the OP
I have more data that seems to validate what some have said concerning
the DVM vs Analog meter. With my little micronta on 150V scale and
the switch off, I barley get the needle to move perhaps 10V or so.
With switch on it goes up to 119V as expected.
What is disconcerting to the guy like me who knows enough to be
dangerous, is that in the off position my greenlee voltage detector
Some observations on my part:
There are at least 2 if not 3 more wires in the box that the fixture
is hung from, this house was built in 1957 and the breaker that I turn
off to kill power to this light also powers the bathroom lights and
the lights of one of the two upstairs bedrooms. [all two wire no gound
Another thing that may or may not mean anything, but I have also
changed the breaker that the electrician installed with a GFCI breaker
as there is an outlet in the bathroom near the sink that I felt needed
So there you have it, I am going to test the switches as dreamchaser
suggested tomorrow and post the results.
I am not sure what the greenlee is but if it is similar to the voltage
sticks I have seen, somethink like a big ink pen where the thing glows if
near an electric wire.
Just try rubbing it on your cloths and watch it light up from just the small
ammount of static electricity.
Those things require a lot of expierance to tell when you actually have a
voltage that is really connected to the power lines or just picking up some
static or induced voltage.
HomeOwnersHub.com is a website for homeowners and building and maintenance pros. It is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.