Interesting problem

[Ed Angell]

Two of my friends have a very nice gift shop which requires a lot of light.
They have installed a large number, 50 or so, Lightolier mini lights with
an attached transformers. Due to the large number of lights the system is
turned on and off with a large magnetic disconnect. The problem is that
other lamps plugged into their shop receptacles have a very short bulb life.
Lamps that use solid state dimmers seem to fail with regularity also. My
thoughts are that shutting down all those transformers instantly is causing
a line spike. I would appreciate any thoughts from the group. Also, is
there a simple way to measure and quantify a line spike?

TIA
Ed Angell

 

[Victor Roberts]

My two cents - I've not heard of transformers causing spiking across a line,
let alone what should be another circuit to the receptacles. Could it be
that they simply have a large amount of incandescent lamps operating, and
they are simply burning out randomly, giving the effect of "constant"
failures? Incandescent lamp lifes are based upon 50% of the lamps lasting
that long, so "premature" failure in up to 50% can be expected. (What other
industry uses such a % of failure standard?)

Actually, almost every industry. While lamp life is specified as the
median life of a large group of lamps, the life of almost every other
product is based on mean life, which in most cases is close to the
median. It is true that many products have a guaranteed minimum life,
but that is usually much less than the mean life of the same product.

 

[TKM]

Actually, almost every industry. While lamp life is specified as the
median life of a large group of lamps, the life of almost every other
product is based on mean life, which in most cases is close to the
median. It is true that many products have a guaranteed minimum life,
but that is usually much less than the mean life of the same product.

If you plot lamp failures vs. time for a given installation of lamps
starting with new lamps and replacing lamps individually as they burn out,
there is a spike of maximum failures around the rated life of the lamp, but
just for the first time that the installation reaches that point in time.
Then the height of that spike damps down as the installation becomes a
mixture of older and newer lamps. Finally, there is a steady rate of
failures based upon burning hours and rated lamp life.

Incandescent installations are notorious for scaring the wits out of the
maintenance staff when that first spike hits; but it happens with
fluorescent and HID systems too. It just takes longer and the maintenance
folks have probably either changed or forgotten when the lamps were
installed. Clever lighting maintenance companies have been known to predict
the spike and then have capitalized on the panic calls which result.

The statistics are not unusual as Vic has said. They apply to most things
which have a life rating due to some kind of wear-out mechanism and which
are made in quantity. The curves have the same shape as the mortality
curves for human beings, for example. What's specific to lamps is the
choice of the 50% failure point as the lamp life rating. That's a
convention that was set years ago during the days of incandescent lamps when
the first lamp statistics were gathered. It's worked O.K. and has the
advantage of being a simple, single number, but those concerned about such
things will ask for the specific lamp mortality curve or supporting data to
find out what is really going on. Such data will alert you to the rate of
early failures, for instance.

Line spikes usually don't affect incandescent lamps very much because of the
thermal mass of the filament; but a higher-than-rated average lamp voltage
is
certain to result in premature failures.

Terry McGowan

 

[Andrew Gabriel]

Line spikes usually don't affect incandescent lamps very much because of the
thermal mass of the filament; but a higher-than-rated average lamp voltage
is certain to result in premature failures.

Line spikes have a really bad effect on filament lamp life in the UK.
It's one of the first things you look for if a whole building is
suffering short lamp life. It may be that the different filament
geometry at 240V makes this a more significant factor.

 

[TKM]

"TKM" <[email protected]> wrote in message

This is very interesting. I wonder if it explains something I've
observed with emergency lighting. It seems to be common to have a
large number of emergency lights that don't work when you get a real
power failure. Then in reaction sometimes you swap them all out. But
the problem recurs. It could be because you started over new, and ran
into that spike again?

Yes, if you start over with a batch of new lamps the maximum failure rate
spike will repeat. This is the basis of group relamping maintenance
planning, of course. Since you know statistically that a large group of
lamps is about to fail, but not which ones; you can replace them all just
before the spike and take advantage of labor cost savings assuming that it
costs less per lamp to replace all of the lamps rather than replace them
one-at-a-time as they fail.

I wouldn't think, however, that emergency lights -- assuming that they only
come on during power failures -- would accumulate enough hours to require
replacement. Or, are they wired such that they operate with the other lamps
in the system, but switch to emergency power when the main power fails?

Terry McGowan

 

[TKM]

Line spikes have a really bad effect on filament lamp life in the UK.
It's one of the first things you look for if a whole building is
suffering short lamp life. It may be that the different filament
geometry at 240V makes this a more significant factor.

That's interesting information. The 240 volt filaments are certainly longer
and maybe thinner. So, coil-to-coil welds would be more likely in the case
of a voltage spike; but I wonder what else might be happening.

Terry McGowan

 

[Ed Angell]

Two of my friends have a very nice gift shop which requires a lot of light.
They have installed a large number, 50 or so, Lightolier mini lights with
an attached transformers. Due to the large number of lights the system is
turned on and off with a large magnetic disconnect. The problem is that
other lamps plugged into their shop receptacles have a very short bulb life.
Lamps that use solid state dimmers seem to fail with regularity also. My
thoughts are that shutting down all those transformers instantly is causing
a line spike. I would appreciate any thoughts from the group. Also, is
there a simple way to measure and quantify a line spike?

TIA
Ed Angell

The fact that solid state dimmers plugged into the shops branch circuits
fail with great regularity, leads me to believe that this is a spike problem
and not a lamp life situation. I know that there is an inductive kick when
you rapidly kill the power to any device that has a coil and core.

 

[ddwyer]

The fact that solid state dimmers plugged into the shops branch circuits
fail with great regularity, leads me to believe that this is a spike problem
and not a lamp life situation. I know that there is an inductive kick when
you rapidly kill the power to any device that has a coil and core.

Seems like the opportunity for a new product here.
A fixed dimmer/regulator ( protected itself and the load against spikes
) to reduce voltage to the filament lams by say 15%. This should give a
much longer lamp life and the reduction in light o/p compensated by the
reduced maintenance cost.
I know such bulbs are available .

 

[Don Klipstein]

That's interesting information. The 240 volt filaments are certainly longer
and maybe thinner. So, coil-to-coil welds would be more likely in the case
of a voltage spike; but I wonder what else might be happening.

I suspect the higher line voltage in combination with the surges may be
causing arcs.

I am under the impression that an argon-nitrogen mixture rather than
pure argon is used in 120V lamps because arcs might form too easily in
these lamps if pure argon is used.

- Don Klipstein ([email protected])

 

[Andrew Gabriel]

That's interesting information. The 240 volt filaments are certainly longer
and maybe thinner. So, coil-to-coil welds would be more likely in the case
of a voltage spike; but I wonder what else might be happening.

It may simply be that the thermal interia of the filament is less,
so it doesn't self-protect itself to the same extent.