Well, here we go. Some rather surprising results along with some not so surprising results. I have not quite finished interpreting the data, but it's interesting to say the least. I would like to have data on emitter temperature, but that's not going to happen. Anyway, here it is.
EA4W XM-L U2 NW and HD2010 XM-L2 T6 1C on a copper SinkPad
Time |
EA4W Temperature (Deg C) |
EA4W Output (Lumens) |
EA4W Current (A) |
HD2010 Temperature (Deg C) |
HD2010 Output (Lumens) |
HD2010 Current (A) |
0:00 |
24.0 |
772 |
2.21 |
23.8 |
1146 |
4.03 |
0:30 |
27.0 |
685 |
2.19 |
25.8 |
1104 |
4.03 |
1:00 |
28.8 |
677 |
2.19 |
27.4 |
1099 |
4.03 |
1:30 |
30.6 |
671 |
2.19 |
29.2 |
1092 |
4.03 |
2:00 |
32.2 |
667 |
2.19 |
30.8 |
1087 |
4.03 |
2:30 |
33.6 |
664 |
2.18 |
31.6 |
1082 |
4.03 |
3:00 |
36.0 |
659 |
2.18 |
32.8 |
1080 |
4.03 |
3:30 |
37.2 |
657 |
2.18 |
34.0 |
1076 |
4.03 |
4:00 |
38.0 |
652 |
2.18 |
35.4 |
1074 |
4.03 |
4:30 |
39.2 |
649 |
2.18 |
36.4 |
1071 |
4.03 |
5:00 |
40.6 |
644 |
2.18 |
37.2 |
1069 |
4.03 |
5:30 |
41.8 |
642 |
2.18 |
38.4 |
1066 |
4.03 |
6:00 |
42.6 |
639 |
2.18 |
39.2 |
1063 |
4.03 |
6:30 |
43.4 |
635 |
2.18 |
40.0 |
1060 |
4.03 |
7:00 |
45.0 |
633 |
2.18 |
40.6 |
1057 |
4.03 |
7:30 |
45.6 |
630 |
2.18 |
41.4 |
1055 |
4.03 |
8:00 |
46.4 |
626 |
2.18 |
42.6 |
1052 |
4.03 |
8:30 |
47.6 |
624 |
2.18 |
43.0 |
1050 |
4.03 |
9:00 |
48.6 |
622 |
2.18 |
43.6 |
1048 |
4.03 |
9:30 |
49.2 |
621 |
2.18 |
44.2 |
1046 |
4.03 |
10:00 |
49.8 |
618 |
2.18 |
44.8 |
1043 |
4.03 |
And because we all love pictures, here's a graph.
The major initial drop occurs in about 7-10 seconds on both lights. The EA4W continues a less dramatic drop for about 15 seconds. After about 30 seconds, both lights seem to heat up at about the same rate, the EA4W warming up quicker. A quick glance and you might conclude that thermal transfer in these lights is similar because after the initial drop they warm up at about the same rate. You might even conclude that the EA4W has better contact because it warmed up quicker.
This is incorrect. The much larger drop exhibited by the EA4W indicates that the emitter got hot quickly. Once the thermal differential between emitter and heatsink rose enough to move the ~10W of heat, things stabilize. I attribute a good portion of this to the dielectric layer on the star, the rest to the lack of good heatsink coupling to the outside world.
On the HD2010, the direct copper bond requires a lower temperature differential to move the ~14W of heat away from the emitter. After things stabilized, the rate of increase is the same.
I intentionally chose more power for the HD2010 because it has a larger heatsink and I wanted the temperature rise to be about the same in both lights. I almost got it right on. I underestimated the power output of the EA4W and overestimated its heat capacity.
Now let's look at some percentages.
Lumen drop from 0 to 30s: EA4W - 11.3%, HD2010 - 3.58%
Lumen drop from 30s to 10m: EA4W - 9.8%, HD2010 - 5.5%
Total Lumen drop: EA4W - 20.0%, HD2010 - 8.9%
First, anyone concerned about burning up their EA4 (and probably the EA8) emitter has nothing to worry about. The emitter is probably getting hot, but no where near 150C. In my 10 minute test, I'd guess it got to 80C or 90C. That puts the body-to-emitter differential around 40C. In order to get to 150C, the body would have to get to 110C. There's no way 10w of heat could get this much material (surface area) up to 110C in open air.
Working in the other direction, with a body temperature of 50C, the body-to-emitter differential would have had to be 100C in order for a 10 minute turbo run to approach the emitter max temperature. A 100C differential would be difficult to obtain, without trying to do so.
I think the surface area and contact points of that EA4 heatsink is adequate to move 10W of heat with 40C of thermal 'pressure'.
Whoever has the capability, I'd like to see a similar test run on the EA8. I really cannot imagine how NiteCore could have designed it to overcome thermal dropout. Especially after seeing inside and these results from its similarly designed EA4 cousin.