You're right about that. However, based on my testing of these diodes with the Aixiz modules, I have found that even with modest wavelength shifts on the order of 4nm to 5nm they have gone LED after a few tens of hours. This implies that there is some accumulating facet/coating damage at the higher ma (and higher mw) settings that occurs at junction temperatures ( based on wavelength) that are no higher than those that occur when the diodes are operated within spec. and only passively cooled. Furthermore, my original goal in testing these diodes was to try to fill a gap in the red wavelength when constructing large multi-diode projectors. The alternative to the LOC-815 diodes are very expensive 200mw 642 nm diodes that look to be about $70 each. The LOC-815 diodes that I've tested operate at 657 when passively cooled on a large lab style heat sink and cost about $10 each. When cooled to -30C with a single stage TEC (in an Aixiz module and like you point out this is important) the LOC-815 puts out at least 260mw at 649 nm. But, when mounted in a custom (not fancy) copper heat sink the wavelength drops to 647nm. A second stage TEC is still not complicated and at -50C 645nm is reasonable. I'll be testing this next and let you know.