Thanks for doing these tests! I have wondered how the EA4W performs over long time. Are there any simple mods one can do to make it more efficient? Could this make it brighter?
Very interesting; thanks for doing and sharing this.
very informative… thanks! 
UPDATE: More thorough testing has proven this comment to be inaccurate. Please ignore until we can finalize some testing. Accurate results will be published. To give you an idea, things are going in a good direction.
Recently there has been a debate over who’s data is right when it comes to thermal sag in flashlights. The discussion didn’t always stay nice, unfortunately. A link to the thread is in the OP.
It turns out, both sets of data may be accurate. How? Different types of measurement.
I just did a 30 second throw sag test (all my tests before this were lumen tests).
Now, up until I did this test, my rational was that for measuring thermal sag, throw or total light output would be fine. Apparently not.
The results of a thirty second test. These are relative, no quantification for now. Just enough to confirm what I started to suspect.
EA4W (lux at ~2m) I didn’t care about actual candela numbers, I’m just looking for relative here.
0s 3200 lux-
2s 3040 lux-
4s 3010 lux-
6s 3010 lux-
.
.
.
30s 3010 lux-
I’m serious; no throw sag after three or four seconds. Total output sags a lot in the same period, but throw does not. I cannot yet explain these findings, but now it appears both of our data sets do match; they are just measured differently. I will not jump to conclusions or even begin to suggest possible causes. More data is needed.
I know this raises more questions than it answers, and I intend to do more testing on this.
Edit: added ‘lux’ after the measurements to make it clear those are lux readings , not candelas.
Sorry, but that makes no sense. The beam profile doesn’t change so surely throw and lumen output must remain directly proportional to one another?
That’s what I thought too. Until I get to the botton this, my results in this thread are to be considered suspect. I’m going to work this out with offline with another member’s help. I’ll update the thread with the outcome.
Up to this point, my results have been sort of following the thermal design of the light. This may be a coincidence.
Also, my lux readings of the EA4W contradict what I had assumed about the relationship between total output and throw. This needs to be confirmed.
Finally, the issue of equipment. I do this hobby on a relatively shoestring budget. My equipment is far from professional. I have been assuming that the equipment would at least be good enough for relative tests. So far, everything I measure one day will give the same result days later. In most cases, my relative results match expected values.
My time for this hobby is extremely limited. I take on tasks like this thinking I can provide some useful information. Unfortunately, due to the various potential issues with my measurements I may be doing more harm than good.
If the outcome of the investigation shows a way for me to get consistent reliable results, I’ll continue the tests. If not, I probably cannot afford the time or expense to get set up properly and provide quality results. I’ll leave those types of tests for members that are better equipped and more experienced in doing them.
Did you use fully charged batteries for both tests?
If the cells were getting a little low for this latest one the light may have dropped down from the turbo a little which could mean a more steady output at a slightly lower level for a while.
Tests that measured lumens were on a power supply. My quick lux test was using recently charged NiMH. The light appeared to stay in Turbo. During my tests of the EA4 on a power supply, it is very well regulated. Adjustments to the power supply voltage were easily compensated for by the driver with an increase in current. I d not suspect a supply issue.