Just found this on a BMW forum...
Recently had a low beam light error message, replaced the RHS bulb (easy DIY), put a new Phillips D2S bulb in.
The remarkable thing, or not, was how much brighter the new bulb appeared than the existing 'old' one.
So, I replaced the LHS bulb too, and struck a perfect match.
Clearly, the 'old' bulb was not as bright as a new replacement bulb, of the same type.
I am not a bulb guru (someone must be).
Do Xenon bulbs age gracefully, or do they burnt bright then die?
Not only do they fade, but they colour shift, getting bluer as they age. The Philips D2S 85122+ are supposed to maintain brightness and colour temperature longer than the D2S 85122's.
What's not often understood is that by the time the lamp nears end of life, and undergoes the color shift dance, 50% of its projection strength has been lost for quite some time, although brightness may continue to remain relatively the same as before. This projection strength factor (aka, lumen depreciation, or decrease in light output) is not often viewed/known by most drivers to be a problem as long as the lamp still functions and burns, but it's this gradual loss in strength over time that takes away from the sharp contrast and cutoff around the fringe distribution @20-40' forward of the car.
I've heard from Osram Sylvania engineers that their 35w 4100k D2S lamps are rated at 2700-3000 hrs life, and w/in the lighting industry, 12v metal halide lamps in general carries a 50% lumen depreciation. w/in the first few hours of lamp life, lumen output would be 100%, or ~ 3000 lumens. At ~ 3000 hrs of use, or end of rated lamp life, we can expect ~1500 lumens, or 50% loss. IF the trend was linear, then midway of lamp life, we'd see 25% loss or 2250 lumens. But the trend isn't linear, rather in published data from lamp mfrs, they show the trend to be digressive. Thus, midway in lamp life, the lamp is closer to 35% depreciation, or 1950 lumens. Thus, larger loss occurs w/in the first 1/2 of a lamp's life vs. the last 1/2.
IN addition, the 2700-3000 hr lamp life is based on a sample of lamps being allowed to burn continuously. Let's say this sample qty=100 lamps. When it takes 50% of that sample to burn out, that's the published rated lamp life. Thus, you can see how unrealistic this rated life is in real world applications, where there's numerous restrikes (turn on/turn off) plus extreme ambient temperature and humidity variations and vibrations all working towards reducing that rated life hours. There's really no data to show how far off real world lamp life is vs. rated lamp life, but it's safe to assume that no one's seeing 2700 hrs of use. If I had to guess, I'd say we'd be lucky to get 1/2 , or 1500 hrs, or on average 1 hr of use every day for ~ 4 yrs. So let's assume that we can expect 4 yrs out of the lamps; thus by yr 2, the lamps will depreciate 35% of initial lumen output, and by yr 4, 50% depreciation. It's sound practice in the maintenance sector of the roadways & parking lot lighting industry to incorporate a change out plan before lamps burn out mainly to get back the lost lumen depreciation for maintaining public safety. Personally, I feel lamps are cheap parts nowadays, relative to gas prices and all other things BMW, and so if you've not seen the light, as it were, then replacements should be part of your routine maintenance.
Anybody buy any fo the xenon replacement lamps off Ebay??