Re: Light bulb lifetime

Hiya!

It also depends on startup temperature. I recently baught a batch of 60W normal household bulbs. I replaced all the ones in the house in one go, as we had different values and ages. The ones in the house are still working 3 months later, while the ones in the outside lights have blown about 3 times, and they only go on from about 10pm to 6am, and they ALWAYS fail at turn on. It has to be the cold that's making them go pop.

Yours, Mark.

hans>

We have lightbulb equivalent screw-in fluorescents outside. Absolutely protect the vents from rain! Other than that, they last for years and years. Outdoor-rated ones are more expensive, and for what?

Probably empirical. I got the info from

The formula has also been published in the GE miniature lamp catalog.

I believe that relationship is grossly incorrect. The lifetime VS percentage of supply voltage is a nasty power law; a small percentage change in excitation voltage makes rather large changes in the lifetime. Use M=1.145^(100-D) where D is the percentage of design voltage that is applied, and M is the lamp life multiplier.

Also, one must consider vibration from all sources, *including* that induced by the filaments own magnetic field acting on itself - meaning an AC supply will cause a much shorter life as compared to a DC supply.

Larger inrush current due to lower filament resistance (due to temperature).

Is this also an empirical equation from tabulated xps, or is there a theoretical derivation to this?

Why? It's probably an approximation that's valid over a limited range.

Where did you get that formula?

I believe that most of the vibration of an AC driven filament is due to thermal effects rather than magnetic effects. Also, what you say about DC is valid only for filtered DC; those devices for extending lamp life are just diodes; the unfiltered DC still causes vibration of the filiment.

Paul Cardinale

As far as i know, nobody bothered to do a theoretical derivation. If the filament was a straight line wire, perhaps it could be done, but since most are coiled or double-coiled, then simplifications would make any derivations incorrect. A starter to try a derivation would be the filament temperature as a function of power input; one would have to model the cooling by the internal gas that was added, along with the envelope cooling (horizontal or vertical, shape, size, external temperature). Just that part ain't symple. Then add in the mass loss due to the boiling off of the tungsten as a function of temperature; this is part of the degradation of the filament which shortens its life. That is where the filament geometry (straight, coiled, etc) comes in. If AC is used, one then must know the frequency and current; add in the filament shape for structural analysis along with magnetic interactions to derive vibrational stresses - the other part of the degradation of the filament which shortens its life. That ain't symple either. And i may have left out other important parts. So you see why empirical equations are used.

Use of a diode increases the vibrational stresses, but the average power applied is less, so the temperature goes down, giving a longer net life. If one compares that life with the same lamp driven by pure DC at the same total average power, then the diode driven lamp would always die first, and the lifetime might be considered unacceptable in light of the life achieved from the DC driven lamp. The vibration is due to both magnetic and thermal effects, but i think the magnetic effects dominate.

Thanks a lot for your thoughts, Robert. Very nice. I am inclined to believe that the reason that nobody has tackled with this from the theoretical aspect, and the fact that the two given empirical equations show such a different picture as in Life (hrs) = 1000 * (230V/220V)^(-12), and M=1.145^(100-D) may point towards a very different issue and cause that governs and determines the filament's life time.

Tungsten is a very brittle metal. So, when it gets extruded into the familiar spiral filament form it will not be a smooth and uniform wire shape as when you were do the same from/with Copper or Silver. The W-filaments will have *micro* cracks. These micro cracks present a different cross-section in the filament(wire), giving different resistances, giving different temperatures, giving an avalanche effect.... and a single one of these crack/imperfections is sufficient for the filament to burn & blow at its largest crack and smallest cross section. So its use-life time may primarily be governed by the production quality and smoothness of the W-wire. Given such inherent uncertainties in the manufacturing it is understandable that a theoretical treatment may be impossible. hanson