light switch blowing bulbs ??

---------

Remove YOURPANTS before E-mailing Me

This is fairly normal failure mode for the lamps. They always seem
to go at switch on and there usually seems to be a hefty current when
they do. In any case, there is quite a surge if you power on from
cold.

I've fitted an MK dimmer (4500 series) of 400W rating to mine.

This does not prevent the bulbs from failing of course, and it's stil
normally at on-time, However these dimmers have a soft start which
seems to lengthen bulb life a lot. Also, when a lamp does go, it
tends not to ping the breaker.

Andy Hall

Email by removing the word after the two letter,
top level Nordic country domain.

I first encountered this with the hall light in an elderly neighbours
house. I was routinely asked to come and replace the bulb every 4 to 6
weeks. So after several trips I replaced the switch as well. That fixed it.

A switch that exhibits lots of "bounce" will reduce bulb life, somtimes
by a large amount. This is because at switch on, you don't get a clean
transition from off to on. Instead you get a rapid sequence of on - off
- on cycles lasting possibly lasting 10s of miliseconds before the
switch contacts settle at a constant "on". This puts lots of stress on
the bulb.

(It's a common software engineering problem when reading switches under
program control. If there is no de-bouncing implemented in the hardware,
the program has to sample the switch position several times with a
delay between samples. Otherwise one simple switch on by a user, can end
up being read as half a dozen on off cycles by the program becuase it
runs so fast as to be able to see each individual bounce)

--
Cheers,

John.

/=================================================================\
| Internode Ltd - http://www.internode.co.uk |
|-----------------------------------------------------------------|
| John Rumm - john(at)internode(dot)co(dot)uk |
\=================================================================/

Debouncing DC switches is important when used for data - but I would have
thought in the case of an AC lighting circuit, this would actually just
cause the lamp to come on a tiny bit slower. Due to the thermal inertia of a
light bulb, I wouldnt have though the very fast bouncing would cause much
trouble - it wouldnt have time to significantly change temperature when it
bounces before being reconnected (certainly not to reduce lifetime to a few
weeks)?

Part of this problem could be down to the GZ 10 bulbs sold by Screwfix.
They have some at almost £6 each and economy at nearly £3 .They are
guarenteed at 200hrs and 1500 hrs repectively.
As i bought them from Screwfix i phoned them today and they are sending me
replacments for two -wouldnt agree to four as the record of me buying the
original fitting with the 4 bulbs fitted wasn't shwn on their system.
If you estimate a kitchen light ( well mine anyway) as only being on for a
couple of hours a day average 1500/2000hrs is a pretty long life .
Stuart

The assumption is that power cycling damages bulbs. Wrong.
Bulb life expectancy is determined by hours of operation and
applied voltage. Once a bulb gets down to its last hours of
operation, then something as trivial as a power on cycle may
damage the bulb. Another has cited numbers for this event
from manufacturers. A bulb will fail on poweron when it is
within its last 14 hours of operation. None of this explains
the OP's original problem.

I can only really go on the empirical results I have observed - i.e.
that "dirty" switches seem to accelerate bulb mortality.

I don`t know how long a 240V bulb takes to reach operational filament
temperature - but I would guess someware between 100 to 200 ms. If the
bounce duration is say 50ms then that could represents up to 25% of
final teperature reached during the bounce (assuming a average 50% duty
cycle of bounces).

There are possibly other factors at work here: the temperature
coefficent of resitance of the filament, and the fact that bulb filament
tends to be tightly super coiled.

The lower than nominal resistance of the cold filament at switch on will
result in a high inrush current, and the coiling will create a slightly
inductive load, which again will tend to raise inrush current.

The unsupported coil will also be subjected to not only thermal shock,
but also mechanical shock as a result of the magnetic forces in the
coil. Finally the switch bounce can occur anywhere in the 50Hz cycle -
so you could get a step rise to peek mains voltage several times in the
space of a switch on.

So in summary: switching a bulb on is bad for it. Switching it on tens
of times more often as a result of a dodgy switch seems worse - how much
worse is hard to quantify.

Question:

Do dimmer circuits run their PWM with zero crossing detection?

Also what frequency of switching does a typical dimmer use?

Most that I've seen do, to reduce heat dissipation in the triac and
reduce RFI if nothing else.

The simple mains ones switch on each zero crossing.

There are "transformers" for LV halogens which are effectively
switched mode power supplies. These switch in the 20-100kHz range
and provide DC to the load. They can be adjusted by an external
potentiometer or some by a 0-5 or 0-10v DC signal and are typically
intended to be used as part of an automation system.

There are similar products which will fill the role of a ballast in a
fluorescent lamp.

From experience with using soft starting mains dimmers with mains
halogen lamps, I am pretty certain that gentle treatment during the
lamp's warm up from cold extends the life a lot.
The MK ones that I have can be switched on at dimmed or full
brightness settings and still come up to full set brightness over
about 3 seconds. I am pretty sure that the switch is electronically
handled and possibly debounced as well.

da wrote:

I think the issue is that pulsing the switch on and off will result in
lots of very high current pulses rather than a long lower one. I
personally think this would induce premature failure, but someone else
pouinted to a site that claimed that its total hours, not how many times
you switch em on, that burns bulbs, last time we did this issue to death,..

>
>

w_tom wrote:

No, they don't. They 'bounce' synchronously from the zero crossing
point. Peak currents are *MUCH* lower.

See above. The cases are not comparable.

John Rumm wrote:

Its usual to switch on at some more or less stable part of the cycle,
and off at the zero crossing point. Also you never switch OFF at a high
current - this can'tr be do0ne with a triac.

100hz. Its done with teh mains freq. ALL they are is an RC circuit that
puts a phase delayed mains voltage on teh triac gate. It switches on at
some more or less later part of the cycle, and off again at the zero
crossing. To reduce peak currnts and more mportantly RFI, they use
chokes to slow the switch-on down. Many have 'soft start' so they can't
be turned on to give full crack into a cold lamp.

>
>
>

just out of interest, do light bulbs have a high inductance?

Thermal cycling puts almost no stress on the filament. Bulb
life expectancy is a function of things like hours of
operation, mechanical vibration when hot, and voltage. Power
cycling is irrelevant to bulb life expectancy. Unfortunately
too many cite trends without numbers. Yes, power cycling does
put a stress on the filament. Then we add this thing that
myth purveyors fear - numbers. Once those numbers are
applied, then power cycling a light bulb is irrelevant -
trivial.

One must first do research and learn basic theory before
inventing a such a theory. Bulbs damaged by power cycling is
a classic example of junk science. One indication that it
comes from junk science - no manufacturer specification
numbers are provided.

> ...

[[ Top post moved to bottom in line with news etiquette ]]

OK, Mr. Science, consider the following numbers.

About three years ago I bought a pack of 25 branded (Sylvania) 50W
mains halogen lamps. They were in manufacturer's outer packaging and
would appear to be from the same batch.

I have an application where four are used as a group from the same
switch. Because of the cost of the lamps, I keep a notebook of the
dates they are changed. Initially they were switched with a
standard MK Logic wall switch.

The pattern of use is the same every day - 8 hours - the area
concerned is otherwise too dark.

After 9 months, the first lamp failed - this is a little over the
rated 2000 hours. The remainder failed over the next two weeks.

The same pattern happened at around 18 months, although of course with
a slightly wider spread.

At that point I installed the dimmer and checked carefully that the
lamps could be set to full brightness, using both a light meter and a
changeover switch. There was no reduction.

The same set of lamps are running 15 months later, suggesting that
there is quite an improvement.

I can't identify any mechanisms other than the soft starting that
could make a difference. The bulbs should be similar since they were
from the same batch; there is no mechanical vibration and no change to
the pattern of mains supply voltage. I log the latter from one of
my UPSs and store in a database - no change in pattern.

If you can think of some other phenomenon then fine, but I have been
quite methodical in my recording.

That 1.5% voltage reduction would only increase life
expectancy from 9 months to 11 months. But do we really know
the dimmer switch was at full intensity every time? Even a
slightly off maximum setting would cause an even greater
increase in bulb life expectancy. Do we really know utility
RMS voltage has not changed ever so slightly? Very little
voltage change causes significant change in bulb life
expectancy. A multimeter may read about the same voltage
because AC sine wave is distorted - and actually has a
different RMS voltage.

Obviously I am only providing one possible reason for the
discrepancy. We really don't have good controls on the data.
Theory demands that we also measure line voltage and ceiling
vibrations. But we do know what manufacturers have been
saying for generations. They provide numbers with their
conclusions. Power cycling according, to those manufacturers,
has little affect on bulb life expectancy - as demonstrated by
traffic light whose bulbs still conform to expected hours of
operation even though they power cycle excessively. Minor
voltage variations, such as created by a dimmer switch in the
line, significantly distort those life expectancy numbers.

a wrote:

No. Pretty low. The filament coils are too small.

>
>

w_tom wrote:

I don't remember cigarette manufacturers providing data on cancer death
rates and smoking either.

Not to say its right or wrong, merely that I consider your evidence
unsatisfactory.

w_tom wrote:

No, they don't.

High peak currents will snap any area of the filament that is a bit
thinner - that has higher resistance and lower thermal inertia.

Despite published 'evidence' to the contrary, I still think that
switching full 400V peak on an ice cold bulb does more damage than
leaving it on...

Temperature of which bit?

I guess thats why my light bulbs all survive being thrown against a
wall. Its only the temperature of the filament that gets them.

Said with complete conviction. What about differential expansion, and
thermal shock. Oh, I forgot. It may kill expensively the seals on
expensively sealed microchips, but never a 20p poorly sealed lightbulb.
and ingress of air doesn't matter because you have assured us thet
filament failure is determined by temperature.

What about potential snappuing due to self induced magnetic effects?

No, can't bve relevant because we have your word that fiulament
failure is entirely due to temperature effects. Shaking a bulb hard or
subjecting it to peak ratings of over a kilowatt won't damage the
filament BECAUSE ITS ONLY TEMPERATURE THAT KILLS YOUR FILAMENT.

Do you realise, that your very insistence, leads me to disbelieve you?

I wonder how many of the so called cycle tests were actually performed
with a bulb in a cold room being switched on once per day, rather than
switched on and off once every few seconds - i.e. starting with a warm
filament, and more to the point, whether the switching mechanism in
those tests was a nasty arcing light switch, or even a moderately clean
relay contact. Or was it in fact a zero crossing triac?

I seem to recall m,any experiments carried out by tobacco companies
showing how rabbits smoking fags didn't get lung cancer.

I have a fairly good quality Fluke meter which is regularly calibrated
and is a true RMS type.

On a check, the voltage is actually reduced by 0.7% - not enough to
explain what is being observed on lamp life.

Yes

I do.

If the voltage theory is true, then it's worth putting in a dimmer
anyway and reducing the voltage by a few % considering the relative
prices of lamps and dimmers.

The initial support for this came from measuring the
life of traffic signal lamps (not in the UK -- we've
used special lamps in traffic signals for years, but
many other countries have used mains GLS lamps -- same
as used in your house).

It's easy to work out exactly how long a traffic signal
lamp has been on and how many times it has been switched
and to check relamping records for deteriming lamp life.
On the same traffic signal, you get a very wide variation
in the figures. E.g. in some cases amber is alight a
lot less time than red or green. In some countries,
red or amber are flashed all night, and green never is,
etc. So you can easily find signals where, for example,
two lights have been on the same length of time, but
one has been through 100 times more switching cycles.
The measured lamp life only depends on the total time
on, and has no bearing on the number of times the
lamp is switched.

This research has been repeated many times, but always
with the same result. This has been used to decide which
lamps in a traffic signal to replace with LEDs. It
depends on the country's signalling sequences and on the
particular timings for a junction, but the highest power
consumption and relamping costs typically come from the
red lamps, followed by the green lamps, and lastly, the
amber lamps. In many cases, the payback period for
replacing the amber lamps with LED units is too long,
and filament lamps are retained (even though they are
often used in flashing modes).

One cavet I would add to all this is that I am only
aware of such research having been done on standard
filament lamps. Much of this thread is talking about
tungsten halogen lamps, and I don't know that this
follows for such lamps -- I can think of some reasons
it might not. Ironically, UK traffic signals have been
using halogen lamps for years (they tend to be years
ahead of what you find in the rest of the world).
However, they are driven through much more intelligent
power control circuitry than a plain light switch, so
even if I had any figures for UK traffic signals, they
could not be extrapolated to household use of halogen
lamps.

--
Andrew Gabriel
Consultant Software Engineer

It would seem that the general concensus (w_tom excluded obviously) is
that bulbs are affected by switching to some extent.

I found this article the other day:-

http://freespace.virgin.net/tom.baldwin/bulbguide.html

This seems to agree with the concept that soft starting can make a
significant difference in halogen bulb life - and a smaller difference
in conventional bulb life.

Also the description of filament degredation due to thin spots would
also support the hypothosis that hard starting from cold is detrimental.

Andrew Gabriel wrote:

AND a 40W filament behaves a lot differently from a 100W filament as well.

AND I suppose the circuits that are used in some spot switches that
catually never ever switch the bulb OFF - just down to infra red
temperatures - are only there because of pseudo science as well.

>

Same reason UK traffic signals (some at least) preheat
halogen lamps: to give a more instant 'on', and to help
ensure quartz envelope is up to min temperature (250ºC) for
the tungsten-halogen cycle to operate from the moment the
filament is hot enough to be evaporating material.

As I said, I can see some reasons why frequent switching
of tungsten-halogen lamps could effect life (the more times
you switch the lamp, the longer total time it will be
operating before the tungsten-halogen cycle starts operating),
but I don't know if this is the case.

One thing that puzzles me about this, though, is that if power-on and
thermal shock is not a factor in reduction of bulb life then I would have
expected an even distribution of bulb failures against time-from-switch-on.

Common experience suggests that this is just not so - just about all of the
bulb failures that I have witnessed have been at power-on. In fact I
struggle to remember the last time that a bulb failed after the initial
power on.

As I'm the one who ends up replacing bulbs in the house I'm fairly certain
that I am not just missing the failures that happen post switch-on, they're
just not happening!

Another indication is that the MCBs trip here every time a bulb fails, and
they don't trip spontaneously IME, they can always be attributed to someone
switching a light on.

Cheers

Richard

---------------------------------
Richard Sampson

can be found at olifant with the co uk thing.

When a lamp gets to the point where it's got only a few hours
left, then it likely won't survive a switch-on surge. If you
normally operate your lamps for only a few hours at a time,
then you will only ever see failures at switch on.
Yes, you lost a few hours of lamp life, but in the lifetime
of a lamp, that is insignificant.

I suspect a lamp failing whilst operating is probably less likely
to start an arc which trips MCBs (there's less current flowing
at the moment the filament breaks, so the spark is more likely to
extinguish than to grow into a large arc). However, I have had it
happen, so whilst less likely, it's certainly not impossible.

Regards

Capitol
Andrew Gabriel wrote in message ...

IOW this article even provides numbers that are consistent
with what I had posted earlier. With no power cycling, a bulb
may last 750 hours, With power cycling, the bulb may last
anywhere from 736 to 750 hours. There is only 0 to 1.9%
improvement! 0.95% improvement is not significant. My post and
this cited article are in complete agreement. Power cycling
is not significant to light bulb life expectancy as
demonstrated by John Rumm's article, by my posts, and by
numbers also provided by Andrew Gabriel :

Hours of operation, voltage, and external vibrations when
hot determine a bulb's life expectancy. Power cycling is
irrelevant as even demonstrated by John Rumm's article. The
general consensus from those who post numbers, including the
article, is that life expectancy is determined mostly by hours
of operation and not due to power cycling. Those in
disagreement post no numbers or misrepresent what this article
says.

If it makes you happy - then I am sure we all agree with you ;-)

Did you read the section on halogen bulb failure at the end?

Extract:

"However, do not expect a major life extension unless soft-starting is
used. Even with soft-starting, do not expect to more than double or
possibly triple the life of any halogen bulb already rated to last 2,000
hours or more. "

Double or triple the life? Not sure I would class the effect of soft
starting as "irrelevant" with those figures....

Without numbers, urban myth purveyors can declare that
power cycling decreases bulb life expectancy. Without
numbers, then all kinds of fiction is possible. Power cycling
destroys incandescent bulbs only because of speculation using
concepts also known as junk science. Those ideas are made
irrelevant once confronted by numbers. Presented were numbers
and basic theory along with experimental evidence that
demonstrates power cycling does not significantly reduce
incandescent bulb life expectancy. And yet those without
numbers would still argue that an industry proven conclusion
is wrong. Instead they would argue their experience is more
valid than fundamental science.

You may not agree with the industry. Fine. That is
irrelevant. But others are offered fundamental theory, field
experience (ie traffic lights), and 'numbers' in direct
contradiction with your experience. Even your own cited
article says power cycling is insignificant - in direct
contradiction to what you claimed it said.

The question is not what you will do. The question is what
will THEY conclude. Will they use one's conjecture on why
power cycling destroys light bulbs OR will they use
conclusions repeatedly cited from industry science - that
includes numbers? Fundamental difference. Some will still
believe urban myths. Others will appreciate facts
demonstrated by numbers are more honest. Numbers are
fundamental to so many discussions about urban myths. Urban
myths are promoted without numbers.

Yes it was. Look at the original posting. My comments were
specifically related to my own experience of mains halogen lamps and
soft starting them. I even provided some data illustrating the
effect.

Also, please follow the requests that several people have made and
follow the postiing convention. This is at the bottom of the thread
or with short clips in the middle, but please never at the top.
It makes it extremely difficult to follow the flow if you top post.

Light bulb technology in the US is incandescent - almost
exclusively. Are bulbs in UK mostly Halogen technology? At
$0.25 per bulb for conventional incandescent vs $4.00 per for
equivalent halogen, then halogen bulbs are rare. Are these
prices consistent with UK? Are halogens more standard in the
UK?

John Rumm's article suggests possible reason for halogen
failure but does not provide numbers. For example, halogen
bulb power cycles 100 times every second. Those cycles are
(obviously) irrelevant - too short. So how long much an
outage be to affect halogen life expectancy as explained? Is
switch bounce with tens of millisecond interruption sufficient
to cause failure? Or does that bouncing simply create the
equivalent of a slow start? Numbers are not provided to make
workable conclusions.

However, best evidence is always in the body. Where are
Andy's bulbs failing? At the filament ends as cited in John
Rumm's article?

At best, we can only speculate from Andy's experience that
such interruptions are relevant. But is sparking during
switch make or switch break? Bouncing would only be during
make. Making and breaking would create different electrical
characteristics.

Data on soft starting makes the whole problem more
interesting. Many ways of soft starting exist. Interrupting
the 50 Hz cycle as dimmer switched do. Limiting current with
an inrush current limiter (the positive temperature
coefficient thermistor). Slow power up by permitting one AC
peak, then none for ten peaks, then one cycle, then none for
nine peaks, etc. Which kind of slow start is used in Andy's
dimmer? And how many milliseconds is that powerup cycle
(obviously more than 10 milliseconds)? Each slow startup
procedure would have different results for halogen bulb power
up and life expectancy, as defined in John Rumm's article.

Lot of questions. Anyone have trends to fill in the
blanks? What kind of soft start is in Andy's dimmer?

One of the requirements of real science is that theories are not only
exposed to peer review but also that experimental and empirical evidence
support the theory.

One of the observations that I, and it would seem many others have made,
is that when a lighting circuit is experiencing premature bulb failure
for no other obvious reason, switch replacement can correct this problem
with all other factors remaining unchanged. What do you suggest is the
explanation of this?

A quick google search of +"bulb life" +"soft start" would suggest that
there are a large number of switch gear makers who promote soft start
technology as a way of increasing bulb life. Which would indicate soft
start is an industry wide recognised way to increase bulb life. It does
not seem to far fetched to extrapolate that a "nasty" hard start is
perhaps a way to reduce it.

not really - the OP discussed the issue of failure resulting from cold
switch on. The issue of "cycling" is in many cases different because you
are dealing with a "warm" bulb - e.g. the traffic light examples.

I suspect the "urban myth" to which you allude, is that switching a bulb
on and off frequently will cause it to fail prematurely. I don't believe
however that anyone in this thread has suggested anything supporting
that argument. What we have said is that "noisy" switching of a cold
bulb (even infrequently) does seem (in our experiance) to reduce bulb life.

As a result we have explored various physical characteristics of bulbs
and switching that may help explain this.

w_tom's (rather confrontational) dismissal of these discussions as
irrelevant/junk science do not actually help progress the discussion
further toward a better understanding of the realities of what we are
observing.

It would seem that many on this fora have provided numbers and
observations to support their discussion.

It would seem this thread has probably reached the point of generating
more heat than light (like a light bulbs!) but is perhaps an interesting
exercise in seeing how long a group of otherwise normal adults can spend
discussing light bulbs!

I too would promote soft start if I thought it would
increase sales. I would not put out application notes that
claims soft start is effective. I would leave the many who
use junk science techniques to promote the advantage of my
'soft start' product. I would claim it contains soft start -
and let those urban myth purveyors promote the product. I
just offer the technology. Let others promote the myth.

I would imagine (with no evidence) that the most popular bulbs in the UK are
incandescent, followed by compact fluorescent, followed by halogen. Halogen
bulbs are likely to be used in specific cases where high lighting levels are
required, particularly in kitchens. Over the last decade, there has been a
huge shift towards the more environmentally friendly compact fluorescent
bulbs.

Christian.

[top posting fixed]

As others have demonstrated on this group, soft starting is effective at
increasing bulb life, the increase being most noticeable with halogen
bulbs rather than conventional incandecents.

However remember the question is not if soft starting will increase life
- it is, will noisy switching reduce it? The observations of this group
are that it does - I have yet to hear your explanation of why this is so.

That you dismiss this observed behaviour as being a figment of our
collective imaginations because it does not fit your personal belief
does not alter the reality.

Judging by a google search of some of your other "contributions" to
usenet, I think you might more appropriatly say "I too will troll every
newsgroup I can find, and top post bollocks as if it were fact, in an
attempt to see how many people I can get to argue with me!"

We suspect soft starting has extended halogen light bulb
life expectancy. But we don't know what kind of soft
starting. A switch that bounces on startup is soft starting.
And yet that type of startup is suspected to decrease bulb
light expectancy.

Dismissed are conclusions not based on explicit facts -
especially those without numbers. Only facts are relevant
here. Numbers are missing to define soft starting for bulb
life expectancy. One type of soft starting in a dimmer switch
appears to extend bulb life expectancy. Another created by a
bouncing switch (assuming it was sparking when powered on)
appears to shorten bulb life expectancy. Numbers would go a
long way to defining the difference. No numbers (IOW how the
soft start occurs) have yet been provided.

If that bothers you, then good. It may be enough to provide
numbers. Your article did provide good numbers on
incandescent bulbs. But it provides no facts on soft starting
halogen lights - just some ideas. Original post demonstrates
a form of soft starting that causes increased bulb failure.
How do you reconcile that discrepancy? Numbers.

Stop trying to make this personal. Stick to the facts,
man. We don't even know what kind of soft starting works or
how long such a cycle must operate - even from personal
testimonies. Numbers were not provided.