Probly none that you will find all that useful. I used up all my XHP-35s so all I have left is the one installed in the light. I went ahead and ran that and the XML2 de domed again to get the VF, but none of the watt intervals really intersect.
(sphere recalib for .34 multip.
(bench supply used)
Cree XM-L2 U4 de-dome in Shadow JM26 head
Volts
watts
Lumens per watt
1000ma
380
3
3
127
2000ma
683
3.3
6.6
103
3000ma
969
3.5
10.5
92
4000ma
1179
3.7
14.8
80
4500ma
1285
3.9
17.55
73
Lux tested @ 1 min 238,500 on 26650 cell
Cree XHP-35 HI C4 in Shadow TC500 ( bench supply used)
Volts
watts
lumens per watt
1000ma
1030
12.9
12.9
79
2000ma
1526
14
28
55
2500ma
1564
14.6
36.5
42.8
I would imagine if I had data at low ma like I do for the XML2 it would show the XHP35 out performing the XML2 by a good bit. However with its smaller footprint driven very hard ( like we all do) in a flashlight it really doesn't do any better in lumen per watt.
The head of the XHP-35 light heats up a lot faster too. If I decide to get more of these I may run some more side by side tests in bigger lights for a slightly more stable top end.
I think the heat generated is almost exactly equal to the power consumed minus the power that comes out as visible light. Some infrared is radiated because of the heat and some maybe due to impurities or defects, but those are very small corrections. Since the efficiency is about the same, the heat is just proportional to the power in.
Ya with more than double the wattage at maximum it takes about 5 seconds before you start to feel it getting gently warm. By the 2 minute mark it would feel good on cold hands in the winter. Since my new batteries came, I may see if it can handle some 5-10 minutes like that.