Maximum Lamp Life

[MNicolai]

The topic has been discussed before, but I'm looking for a conclusion to what has been a large science to many of you.

It was concluded that preheating = bad, primarily because it wastes electricity, and does not extend lamp life because the tungsten halogen cycle either isn't activated at such low intensities, or it is but the lamp cools back down before a show or rehearsal starts anyways.

BUT, it's known that running a lamp under the recommended voltage can extend lamp life exponentially.

With that known, what can be done to extend lamp life?

I believe the obvious answers, if feasible for a specific application, are to program dimmers to that the 100% on the console, is actually a lower load percentage on the dimmers. Another possibility is to create longer default rise-times.

So in addition to that question, what is the most practical, yet functional means of extending lamp life. I know people will come up with different possibilities, but one of the questions I'd like to be delved into further is which dimmer load is the best as the max, and for which lamps?

For starters, I'd like this to be based around HPL575X and HPL750X lamps, with potentional to be extended to other lamp types.

 

[David Ashton]

A key point is that as you reduce voltage to increase lamp life you also reduce the efficiency at a greater rate, and you reduce the colour temperature so that lights with blue gels will not reproduce that end of the spectrum.
Because there are changes in efficiency, colour and lamp life all interacting in non linear ways there can be no simple answer to your question.
In brief, reducing voltage to increase lamp life will cost more because the saving in lamps will be more than offset by the extra power used.

 

[JD]

The simplest way to extend the life of a lamp without sacrificing color temperature or breaking the halogen cycle is to reduce thermal and electrical stress on the filament. The easiest way to reduce the stress is to slow the turn-on rate. If a slow fade can be used great. If not, then a quick turn-on ramp as compared to a straight bump would be helpful. Tungsten, like any other metal, accumulates stress damage over the life of the lamp. Even in a quartz lamp, the tungsten also thins out with age. (Usually due to uneven re-depositing in the halogen cycle.) At some point near the end of life, part of the filament is so thinned or stressed that the inrush of one last turn on causes it to rupture. Although lamps may fail during steady state, most fail during the inrush. Another (yet to be proven) extender of life may be the use of sign wave dimmers. Like autotransformers, there is very little "lamp sing." This is the vibration of the filament that is audible at certain dimmer settings. It is after all, the filament in motion, which cannot be good. Even the extent of filtering on phase chopped dimmers probably varies lamp life. Lastly, I have noticed in the past that lamps on a very long cable run tend to last longer. It is not so much the voltage drop on the cable when the lamp is lit as it is the voltage drop on the cable during inrush, which some estimate may be 5 to 10 times the normal operating current. Once again, the lamp gets a softer start.

 

[MNicolai]

I don't know about anybody else, but I know I can afford to lose 5-10% intensity on most of my dimmers.

 

[Casey]

A couple of choices that I would look at:

Profile you outputs (dimmers) on you board. Make a profile that outputs 95% when the channel is set to 100%. That should effectively double lamp life. That said, I've seen situations where a designer has asked that the voltage at the lamp be tested because the color looked muddy, so if your working with a pro designer this might not always work for you.


If you have any influence on how your shows are cues, I have found it really helpful to preheat in cues rather than all at once before the show starts. So if you have a cue where many lamps bump from 0% to 100%, preheat them at 2% or 5% in the previous cue. As a lighting designer this is a trick used to get lights to bump on faster, but it also should (in theory) extend lamp life as the filament is actually preheated for when it is used rather than preheated then allowed to cool down only to be bumped on later.

And the longer rise-time solution might be a good answer for you, but most dimmer installations tend to me somewhere in the middle road of 350microseconds and anything longer than that can be phohibitive if you are doing really quick cue sequences, not to mention that to change the rise time on most dimmers you have to change out the Triacs, SCRs/SSRs, or even replace the entire dimmer pack (depending on the dimmer/manufacturer).

And if cost is an issue at all, get a quote for an installation of sinewave dimmers just for a laugh. They are several time the cost of SCR dimmers.

Another thing that is more likely the problem if you burn through lamps much quicker than you think you should is because of voltage spikes in your system, however to fix this problem you need what basically amounts to an industrial sized UPS (uninterrupted power supply) to be installed upstream of your dimmers. That is obviously not a cheap solution.