We’re very glad to have you here, theoGreece!
2 Thanks
I was also curious how the OP noticed the loss of output, but they stated that the SST40 experienced a change in the color emitted, which is often the result of phosphor degradation.
The luminous flux of a source (measured in the unit lumen) is essentially the product of spectral power density and luminosity function (relative visibility of each wavelength), integrated over all visible wavelengths. Nothing here suggests logarithmic scaling. If “amount of light” is defined as the luminous flux, then doubling the lumens exactly doubles the amount of light.
1 Thank
I’m not understanding your issue, so I apologize in advance. Are you saying that after you notice the drop in brightness, you immediately switch freshly charged batteries, or are you using the same ones?
Yeh, of course LEDs (especially white) dim over time. That’s how they usually “fail”.
Used to be the number of hours of “lifetime” was the time at spec to degrade to 50% their original brightness, but now because of complaints that was increased to 70% original brightness.
I got a Feit circular LED replacement “tube” for a fluorescent tube, and it was initially so bright that I just used the over-the-sink light vs the overhead. Now, it’s just comfortable, so I know the LEDs were/are driven pretty hard and already degraded some.
There are 2 main ways white LEDs degrade: emitter and phosphor.
Getting driven hard degrades the (blue) emitter just like any other monochrome LED. The curve is an “exponential”. Ie, it drops fastest and hardest initially, then “levels off” over time. I put that in Scare Quotes because I’m not positive if it’s an actual exponential (eg, ϕ = ϕ0e-kt blah blah blah) or other, but that’s close enough.
Then there’s degradation of the phosphor. The harder the emitter’s driven, the more the phosphor gets “burned off” and less effective. Anyone remember “burn-in” on monochrome CRTs? Like on a POS terminal with sales-screen, you could actually read the burnt-in characters, borders, etc.
Same happens with colored phosphors. Maybe the phosphor that reemits at red is more sensitive, burns off faster, and the tint shifts towards blue as a result. Etc.
Anyhoo, that’s why I couldn’t understand why lumen-hounds would push emitters to the top of their parabolic curves, dumping twice the current/power through the emitter to squeeze an extra 10% more lumens out the pipe. I never pushed mine beyond where the curve just starts to gently flatten out. I practically got larfed out of the room.
And of course, thermal paths have a lot to do with it, too. Stick an emitter on a cryo plate to cool it, and you can push more current through it and get more light from the emitter, but maybe still burn off the phosphor? Who knows. Have an emitter dangling on a shelf that’s resting only on a ledge with a crappy thermal path, and it’ll cook at even low current levels.
And PWMing LEDs keeps the same color/tint for the most part, vs color/tint shifting from lower to higher current levels. So while the LED emitter won’t be damaged by overcurrent because the duty-cycle is so low, what’s happening to the phosphor? It’s maintaining the same color/tint because it’s getting hit so hard, albeit at lower overall power.
The love for PWMing LEDs goes way way way back to LED-driver ICs (Maxim, etc.), and the “preference” in consumer items to keep the color the same when being driven at only, say, 10%, via a 10% duty-cycle vs decreasing the current to 10% of maximum. It’s like driving a screw via an impact driver vs a torquey drill.
So… yeah. Expect LEDs to permanently and irreversibly dim over time.
5 Thanks
Firstly, welcome theoGreece to BLF!
The thread is getting confusing because people aren’t reading. The original claim is 15+ flashlights have a visible permanent drop in output from new.
One or two I can understand as being poor QC resulting in bad reflows or bad thermal path, but 15 across different brands seems unlikely.
A big issue I have is these claims are based on visual comparison, with “paper burning test” given as evidence. We know the human eye is not good at detecting small changes in output, it is even more challenging as this is over time so you don’t have a side-by-side comparison.
To understand better, you would have to build an integrating “sphere” (box, tube, etc), there are lots of guides here. Even something simple like a lux meter will give you objective measurements.
Do first also check the threads and contacts are clean, if you’re experiencing drop in output.
1 Thank
The situation you are talking about is called “Luminance decrease/Lumens depreciation/Light decay”. In theory, this is an inevitable process, but few people can really perceive or measure a significant decrease.
For well-designed flashlights, there are usually temperature control and heat dissipation to prevent this from happening. For poorly designed flashlights, failure to light up is a more common problem.
In addition to LEDs, the driver components themselves may also age and cause output to decrease.
But for some kind of flashlights/drivers, the battery voltage dropping causing max output to drop, that’s really an “elephant in the room”. 1Lumen has some reviews that measure “turbo at 3.6V”, but not every review has it, I don’t know why.
Just looked through three datasheets from Nichia, Luminus and Cree and couldn’t find any mentions of expected lifetime.
Nichia includes results of the reliability tests in datasheet and Cree has a separate document that explains their testing aproach.
If I understood this document correctly, Cree claims that their emitters after 1008 hours of operating at max current (see datasheet) have no more than 15%/25% (depending on LED technology) luminous flux degradation.
So even if those dozens of thousands of hours are official data, I think they are given for some “typical forward current”. For instance 519A is binned at 1400 mA, SFT40 - at 1500 mA.
2 Thanks
Oof. Only ~1000hrs is a far cry from the much-touted “25,000hrs” claims.
But yeah, spec-sheet “maximum current” might be, say, 3A for an emitter, but in flashlights that gets pushed to twice that or more for “turbo” this or that.
I can’t get arsed enough to actually look up the spex and do this, but for anyone interested enough, pick any 2000lm flashlight and find its emitter. Look up its spec-sheet max current. Find out how much current is actually pushed through it in the light at turbo or even high. Betcha the latter is about twice the former.
And that’s in a dinky flashlight with a hand being figured into the thermal path in wicking away heat through conduction and blood-cooling (sounds pretty amazing, but that’s another story…). Spec-sheet maximums are usually done on a fixed but big honkin’ heat-sink.
And pushing these emitters to get those nice fat numbers is for just that, the numbers game. Appealing to The Muggle™ by claiming huge numbers is what sells, not longevity or efficacy preservation.
We’re (usually) not talking about enough current to fry an emitter, but to burn off phosphor and weaken the emitter enough to cause that accelerated drop in output. I’d dare say burnt spots on the dome or a poofed emitter are fairly rare. Output dropping like a rock after a few hours of cooking? Yeah, I’d say that’s pretty common.
Don’t forget that going from 1000lm to 2000lm is maybe a 30% visual increase, so the reverse is about the same, too. Halved output might not seem like that much, visually. That’s why some lights have a closely-spaced 50% mode and full 100% mode, despite the visual difference not being that much. Twice/Half the current, though, makes a huge difference in runtime.
5 Thanks
Is it possible that using a light to burn paper is causing it to significantly overheat and that’s damaging the emitter?
1 Thank
Can’t say to remember any normal review with a “Time till paper burned” column.
If you have that many lights that degraded that quick, then you should probably play the lottery because you have gotten the random draw of bad products at a very abnormal rate.
1 Thank
Might be
1 Thank
No. Lumens are a linear scale.
Human perception of brightness is roughly proportional to log2 of the actual brightness, meaning that it takes about four times the lumens to appear twice as bright.
4 Thanks
Very true, but that does not mean that we cannot see changes in light output amounting to much less than 4 times the output. I think , for me, the lower limit seems to be about 20% brighter.
1 Thank
So you are saying doubling the lumens doubles the brightness?
I don’t think so. This has been mentioned many times on BLF.
It doubles actual brightness exactly.
Visual brightness, though is a log response, more or less. That’s how we can see in everything from candlelight to daytime sun.
5 Thanks
This is an example of what I was referring to. Perhaps I explained it incorrectly.
1 Thank
The quoted thread is itself a mis-application of the inverse-square law, which describes decay of intensity over distance, which has nothing to do with perceived brightness.
I’ve seen folks who claim that perceived brightness is logarithmic in luminous flux, and I’ve also heard claims citing the inverse-square law, which suggests that perceived brightness is proportional to square root of luminous flux.
I find the former more convincing because a perceptible difference is usually multiplicative: you can’t always see a difference of +50 lumens with a side-by-side comparison, but you surely can see a difference of +50% (i.e., x1.5) lumens. That is, the minimum perceivable difference is proportional to existing brightness. This multiplicative nature of perception suggests log-scaling.
4 Thanks
My first decent light was a Fenix L2D. I used it for cross country running, and ran it on max for an hour. Eventually the LED wouldn’t produce anything but a feeble beam. Less than 100 hours of use. Since then my lights have been fine.
I’m surprised that no one has mentioned that the percieved drop in lumens might not be an LED burn in issue, but rather just that, an issue with perception, i.e. retinal burn. Also, if you wear glasses, are they transitions? It’s possible that maybe the lights are emitting enough radiation to slightly darken them without you realizing it.
Don’t those darken on exposure to UV? The glass in front of the LED will filter out any UV light.