2
\$\begingroup\$

Does lifetime of tungsten incandescent lamp shorten by every turning on, directly connecting to electric grid, ie not using special device to turn it on smoothly, and how much? It was said to us in school that it shortens its lifetime.

How much it (lifetime shortening) is compared to lifetime shortening by stable ligthting? If a lamp user goes away from a lamp, planning for some time of absence, how much time of absence worth turning the lamp off and then on, for the time of absence?

When it is too long, it worths, because saved electricity is more expensive than the increase of price by having to buy new lamps more frequently (because of such behaviour). When the time is very short, it also worth, because lamp does not get cooled and do not get shock.

Do not take in account that frequent flickering causes decreased efficiency of getting light per amount of used power. (Ie if user for some reason makes frequent flickering, calculate only price of electicity and price of lamp, do not calculate amount of light produced; and it anyway produces heat efficiently and heat+light may be goal of user, anyway, consider it is ok for him since he does so).

I have found also confirmation of that the lifetime shortens in an answer on this site: https://electronics.stackexchange.com/a/214397 :

During rapid switching the filament would not cool completely before the next switch-on so it would not be stressed so much. The longer the 'off' periods the harder it is for the lamp.

My domestic experience was that lamps on dimmer switches lasted 'forever' relative to the other lamps in the house. This was due to the rotary dimmer being turned up over, say, half a second and gradually increasing the current while the element warmed up. There was no sudden inrush current.

And this question is similar to that question, but in content part of the question it is asked about 120 hz flickering. And an answer there is already accepted. And the 2 answers there do not answer about how much the lifetime shortens, ie there is no formula nor links to investigation result papers are given.

\$\endgroup\$
  • \$\begingroup\$ @Bimpelrekkie just interesting, i thought about that when led lamps were not widely used. \$\endgroup\$ – qdinar Feb 5 at 16:53
2
\$\begingroup\$

The tungsten filament thins with use and it will tend to fail when switched on because the current surge causes mechanical stress. This gives the user the impression that the lamp would have lasted much longer were it not for the turn-on surge.

The life of an incandescent bulb is strongly (exponentially) related to RMS voltage so even a small decrease in voltage results in much longer life. Unfortunately, the efficiency also drops with the voltage so you have a choice of getting more light per watt that is whiter or getting much less light that is redder and having longer bulb life. Consumer bulbs were designed to have a certain trade-off like 1500 hours average life. It's no problem making a bulb that will last 5000 hours, it just will be very inefficient and will use a lot more $$ in electricity to get the same amount of redder light.

I don't think there is much evidence that soft-start materially prolongs the life of an incandescent bulb. Many years ago I did a feasibility study on prolonging lamp life in a high-reliability application and we concluded it was not worthwhile.

If you still have any incandescent bulbs, I suggest turning them off the moment they're no longer needed, even if you're just leaving the room for a few minutes.

| improve this answer | |
\$\endgroup\$
  • 1
    \$\begingroup\$ Hi, Spehro. My own experience with domestic rotary dimmers on incandescent reflector spotlights was that the lamps "lasted forever" which I put down to the soft start. I was also aware that the dimmer would reduce the voltage somewhat. It was probably a combination of both. \$\endgroup\$ – Transistor Feb 5 at 17:29
  • \$\begingroup\$ I have heard claims (including from one competent engineer personally known to me) that a small series resistor significantly improves lamp life. I've not ever tried it. \$\endgroup\$ – Russell McMahon Feb 6 at 0:04
  • \$\begingroup\$ @RussellMcMahon They used to sell coin-sized thermistors to do soft-start. I don't think they sold very well. Some combination of soft start and reduced voltage (and some heat in the back of the socket). \$\endgroup\$ – Spehro Pefhany Feb 6 at 0:53
2
\$\begingroup\$

There are actually two mechanisms involved. An incandescent tungtsten filament in a vacuum will not melt - but it does lose surface material due to evaporation. As a practical matter, the thickness of the filament will vary slightly from place to place on the filament, and because the filament is thinner (less cross-sectional area) its local resistance will be slightly higher, and its temperature will be higher. This means that the evaporation rate at that point will be slightly greater than the rest of the filament, which means that, while the filament as a whole will grow thinner, the thin spot will lose mass even faster. Eventually, the thin spot will get very thin, and usually a turn-on event will provide enough mechanical stress to break the wire and cause the bulb to fail.

This is not necessary, and sooner or later a bulb which is run continuously will fail due to the hot spot getting so thin that it can no longer maintain structural integrity. But in general failure will occur when the bulb is turned on. In this respect, turning on a bulb does indeed shorten the lifetime of the bulb, but especially in the early stages of the bulb's life it has no noticeable effect.

The reason that halogen bulbs have longer lifetimes than normal bulbs is that the halogen scavenges tungsten from the inside of the bulb, then dissociates when it contacts the high-temperature filament and deposits the tungsten back on the filament. You might think that you'd get more tungsten plated on hot spots (because they're hotter, naturally), but the difference is not significant. So halogen bulbs are just as likely to fail on turn-on as regular bulbs, rather than failing during steady-state operation.

| improve this answer | |
\$\endgroup\$
1
\$\begingroup\$

Arrhenius Law prevails for predicting reliability due to thermal stress alone. The way to accelerate life testing is to raise the ambient temperature around the bulb (glass bowl). Old porch light globes can reduce the life span to 3~6 months.

The other factor is the surge force of 10x cold start current when in sync with the 141% peak of an RMS voltage during switch On. Perhaps you recall the times when the bulbs go poof, yet worked perfectly the fine last time. (Yes we know this is a joke like saying the match was working last time I tried it.) A bigger joke is that Trump announced the cancelling of the ban on incandecent bulbs yet said nothing about climate change issues in last nite's state of the union.

So you want to reduce the ambient temp. A good way to extend bulb life span is to use use a dimmer or a soft start or get warm white dimmable LED ceiling spot lamps or 4500'K ~5000'K 4ft FL tubes with high efficacy in tri-phosphor and 50kh life span.

| improve this answer | |
\$\endgroup\$
  • \$\begingroup\$ Tony - Store your incandescent bulbs. It's at least reasonably possible that their CC contributions will be appreciated 20 or 30 years from now. If not, their antique value may offset storage costs :-). \$\endgroup\$ – Russell McMahon Feb 6 at 0:06
  • \$\begingroup\$ Colour temperature is a personal preference. 4500-5000 °K is about "daylight white". I agree heartily! :-). I'm also happy enough with 6000 °K which is getting a bit "bluish" for some. I severely dislike attempt-to-emulate-tungsten CT's under 3000 °K. || Tri-phosphor costs more but gives a higher CRI (colour rendering index) - to get CRIs close to 100 (colours match tungsten) you generally also need yukky sub 3000 °K CT. Sadly. \$\endgroup\$ – Russell McMahon Feb 6 at 2:04

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.