This is my last post regarding this 'off topic' not brought here by me.
Being in the EE field all my life trying to wet my foot in the matter started reading the manufacturers' datasheets. My first purchases around a year ago beside a nice Romisen included some emitters and drivers. With very basic equipment what I found concur 100% with the specs provided . Being this tests over XR-E or XM-L the beneffict of running the emitter beyond specs is minimal, around 15% or nill if much pushed. The gain is noticeable with instruments, at naked eye there are not difference between .
What we have here. The OP of the thread linked found the same, other posters with noticeable knowledge agreed with him. Match in his MAD 6 XM-L notice the same.
Well , some people in the other site are saying thats his flashlight reach 1000L at full specs and a 'turbo' mode of 1500L at 5A , well beyond specs...50% gain of a lab number...
How would be the reaction of us if a Ebay's Chinese seller tried to convince us that thermal drop don't influx his torchs therefore his 2A 400L XR-E Q5 aren't a lie. You get the picture, pun intended.
And , no, Cree do not understated his products, if any they overvalue them stating 1000L emitter when that number is in lab condition, his datasheets are very accurate.
One doesn't need to visit CPF for this - I could point out Foy's UF980L review, where the XM-L is clearly overdriven with good results: https://budgetlightforum.com/t/-/2284
Overdriving an LED might produce more light, or it might not. There were at least 3 controlled tests on that page which show diminishing and then negative returns over 3-4A. The problem with accepting CPF claims without verification is that many of the people there are petty braggarts who are clueless when it comes to anything technical and thus confirmation bias is very common.
For example, they "expect" that more juice equals more light, and thus hold up instances where this is true to validate their opinion, whereas if the opposite is observed to be true, they consider that to be either a problem with the setup or just some bizarre exception. You can readily see this in the thread. Also consider their stance on budget lights, which they apparently feel lack the special sauce that expensive manufacturers inject, and go through all sorts of mental contortions to "prove" that a metal tube with a bulb at one end and a switch at the other needs to cost $50-100 or more.
In Foy's case, his main claim doesn't take into consideration hotspot size which is a primary function of larger reflector. We don't know what current the led is driven at, and we don't know emittance as this varies, whereas the tests above do show this very clearly.
To recap, it's most plausible to me that 3A is simply the manufacturing baseline for CREE, not unlike a 2ghz processor. Some can go to 2.5ghz or prize samples from the middle of the wafer to 3ghz. Now granted, a relation of voltage to gate speed is not necessarily linear nevermind comparing 1 junction to a gazillion, but I think the basic principle still applies.
Agentex post above remind me one of the many fallacies used. First, Foy's torch don't prob nothing, we don't know the output well drived, surely more than now at 5A.
Regarding Cree: There's nothing implausible about the hyptothesis that some or even most of their emitters will produce a bit more light when overdriven. Cree has an interest in providing a reliable product and writing their specs to encourage use in a manner that ensures their reliability guarantees.
Regarding Foy: The statement that his XM-L is overdriven was not based upon the light, but upon the power consumption. We know how much power the battery is providing - around 15W, possibly more. Unless he has an incrediby inefficient driver, the LED must be overdriven. The assessment that the results were good was based upon the amount of light we're seeing. Yes, that's somewhat subjective but it's clearly not the dramatic loss in brightness predicted above - that would be obvious. If he let the temperature get really out of control, then sure - eventually things will go down hill in a noticeable way.
Sorry, I make the mistake of exaggerate a little, couldn't resist the vibe.
Given my measures posted some time ago the XM-L T6 at 3A give 840 L at 5A 820 L ( the 2 first seconds , after that you can guess )...at 4A 900L sagging fast.
After that test never could reach 840L at 3A the emitter was permanently damaged.. still get a lot of light of it. 750L at 3A...
Again, Ron, I don't know where the confusion is. The only people providing numbers show a consistent picture of reality that the xml seems spec'ed closer to its limits than previous emitters, at least for now.
This is like people who make claims about 4ghz from 2ghz processors, just don't expect anything close when you buy yours, and least with computers the numbers are on the surface.
I don't know what the confusion is about either. I mentioned that overdriving the XP-G would probably provide more light until the emitter was damaged. I provided a link to a thread with evidence of this for both XM-L and XP-G emitters. I also mentioned that some people didn't have such good results with XM-Ls. Everything has been consistent with what I said in post #22. I don't see what there is to argue about here.
I would expect an XP-G to be brighter at 2A than at 1.2A - until the magic black smoke escapes. That's fairly typical, though not guaranteed. With some googling you can find examples from people who have done this and posted the results. See, e.g., this thread on CPF: http://www.candlepowerforums.com/vb/showthread.php?308569-No-real-benefits-to-overdriving-XML
The people who posted their result show a definitely decline from somewhere in between 3-4A (there are 4 graphs from 3 emitters). The people making wild ass claims in the thread don't count.
The OP claims no benefit from overdriving his XM-L, but others have photos showing the benefit of overdriving both XM-Ls and XP-Gs.
Again, some people took pictures of arrangements that aren't comparable anyway. Showing that one emitter in one light looking about twice as bright as another emitter in another light at 20% more power isn't just a bad comparison, it doesn't even make any sense.
The graphs that fit your assumptions count, but photos that don't fit your assumptions don't count? The experimental setup those photos isn't as bad as you suggest.
The graphs are done in a verifiable and controlled mattered, the photos are just random pictures of various lights. Plus they don't make sense even assuming a linear rampup, unless you want to change the claim to output scales exponentially with power.
The images look mostly plausible. Some look a little better than one might expect. Consumer digital cameras aren't scientific instruments. They'll be brigher when there's more light. We can count on that, but unless we know more about the image processing done by the camera, but it's iffy to make quantitative assessments based upon jpeg output.
The graphs are just graphs. We can only hope that they were done in a verifiable and controlled manner. We don't have anything to fall back on other than word of the people who made the graphs.
Because one is done on a single emitter with bench supply and lux meter, the other is just a random photo of several lights.
No, digital cameras scale reasonably linearly. Tone curve redefinition is generally only done on SLR's or in post.
Yours is exactly the kind of crazy confirmation bias commonly shown on CPF. Next thing we know, you're going to be claiming a XML in a $300 light is much brighter than a XML in a $30 light, and it'll be shown by putting a 4A driven Fenix with giant reflector next to a 1.5A p60 drop-in with low batteries.
They're done by people. The difference is that you think the people who say they did their measurments with these instruments are more credible than the people who posted pictures. I think they're probably all truthful and credible and that we're just seeing small sample size effects.
Digital camera sensors are linear. Digital SLRs tend to be a bit more conservative in the jpeg processing, while P&S cameras tend to be less faithful, but in neither case should you assume that jpeg output is faithful. (Keep in mind that the values in the file certainly won't be linear because sRGB is a nonlinear color space.)
Believing that brighter photos are the result of an actual situation with more light = crazy confirmation bias? Your comments above are way out of line.
Yes, instruments on one emitter is more credible than pictures of random lights. Confirmation bias has nothing to do being truthful, but rather about being competent.
P&S cameras do not recurve, they at best perform minor chroma shifts to balance white and be more aesthetically pleasing. sRGB is remapped to your monitor anyway so it doesn't matter how it's stored.
New DSLRs can recurve much more agressively, eg Nikon D'Lighting.
More light can come from any number of factors, which is why it's important to isolate them to the ones we're concerned with, like current, which is what those instruments do. And yes that's what confirmation bias means, interpreting ambiguous cases to mean what you prefer. For example, I'm not saying the photographed 5A lights are not brighter than his 4A lights, only that information is insufficient to make such a determination. Whereas information is sufficient in a case with one emitter, a bench supply, and a light meter.
yeah...Beckett picked up first regular-season win in Cleveland to lead the Red Sox to a 4-2 win Tuesday night over the Indians, who couldn’t quite muster another late-inning rally. Indians have an incredible home field winning percentage ..I wonder what the odds were on this game ?
Regarding random lights: It's kind of funny that you're huge variation in output from LEDs in the same bin, but so unwilling to accept differences in tolerance to overdriving. One would hope that the people taking photos used LEDs that were from the same bin and were close in output at lower current. In both cases, trust is required believe that the experiment was done correctly. You are just choosing to believe people who posted a graph over people who posted a photo. That's a leap of faith on your part.
they at best perform minor chroma shifts to balance white and be more aesthetically pleasing. sRGB is remapped to your monitor anyway so it doesn't matter how it's stored.
New DSLRs can recurve much more agressively, eg Nikon D'Lighting.
Nearly all cameras take some liberties in how they transform the linear, raw data coming off the A/D converter into sRGB. First, I'll point out that there's no reason to think they wouldn't - the goal of the manufacturer is often to produce pleasing photos and rarely to produce scientifically accurate ones. Most cameras have a contrast setting. Guess what that does? Second, it's easy to test for yourself that they aren't completely faithful.
D-lighting and similar tricks (which are also available on P&S cameras, BTW), can do some massive adjustments to the tone curve. I wasn't suggesting that any of the users in the CPF had had such features enabled and I didn't check the exif headers to see what features were available on the cameras they had. I was just pointing out the fact that we shouldn't be looking for a linear relationship between amps and values in the sRGB file (after linearization) because because the camera (esp. when used to produce jpegs) is not a scientific instrument. It will shift things around a bit. Failure to see a linear relationship doesn't necessarily mean that the user botched the experiment; it could just mean that we're looking at jpeg photos and not raw files.
I will point out, again, that you are convincing yourself that the people who produced the graphs did the experiments correctly and honestly while choosing to raise questions about the photos. The bias here is in where you have placed your faith.
Perhaps I have a too optimistic view of the people taking photos. While I find it hard to believe that they could mess things up so badly that we are seeing increased output from overdriving when, in fact, there is a decrease or no improvement in their samples, I will admit that it is possible. Lots of things are possible, including sample variation in LEDs.