Considering that the LHP531 which got the 6500K, 5000K, 4000K, 3000K and 1800K in 70 CRI rating, the LHP73B will probably follow with the same CRI, if these two emitters are made out of the same LED dies. The same thing will probably happen with the upcoming 9x9 LMP emitter.
1 Thank
So out of something like the convoy m21f this thing should be putting out 5,000 lumens in 4,000 k tint whereas the xhp 70.2 4000k puts out 3000 lumens?
No, LHP73B outputs same or slightly smaller amount of lumens as XHP70.2 (at the same power).
The dies are for the most part irrelevant for CRI. Pretty much any regular LED uses the same monochromatic blue light for the die, regardless of CRI. CRI is determined by the phosphor mixture.
4 Thanks
This would surprise meāthe output difference between 70.3HI and 70.2 seems negligible, so I would expect the LHP73B to beat XHP70.2 in output.
6 Thanks
Thank you for this review!
Nightwatch does not have similar light bulbs. At the very beginning, when we were customizing the 1148D light bulbs, we developed the 43mil 16-core 7070 light bulb 7021, specifically for modifying the MS32. This is similar to the LHP73B, but its performance in all aspects is much better than the LHP73B. Nightwatch might iterate the 3v 7275 light bulb based on the latest customized 9v 7018I, but due to the lack of product planning and funding issues, I think the probability of mass production is very low.
3 Thanks
I miss my M21b xhp70.3 4000k light I lost. This might go in its replacement
2 Thanks
Why is that? I was about to suggest Simon puts this in the 3X21D, but now maybe not.
2 Thanks
If I didnāt care about throw, Iād give this emitter a try.
That said, Iām loving my new flashlight with the LHP531 emitter, which is reasonably throwy. 
1 Thank
Itās simple: at more than 22 A, you hardly get any more reasonable luminous flux (which is almost not visible anyway, especially without any comparison), but all the more heat and, above all, a massively reduced service life because the chips have to work close to Tj.
Is 10-15% more luminous flux at almost double the power (!) really worth it? Iāve now reached the point where Iām no longer interested in high-performance lights that squeeze every last lumen out of the LEDs. Durability, ease of use, and design (material) are now much more important to me. My EDC has only 450 lm and I am absolutely happy with that.
15 Thanks
Iām not happy with 450 . Every time I turn on my m21f I always bump it up to the ~1000 lumen mode. Iāve noticed that seems to be preference
Any light that cannot sustain 800-1000 lumens for over an hour Iām not really interested in.
My21f has been holding up pretty fine but Iāve also noticed that whether or not it can maintain a thousand lumens without getting too hot, is highly dependent on the ambient temperature
When itās slightly chilly outside, I have no problems with it, but when itās 90Ā° at night humidity it gets pretty hot
1 Thank
Thing is - 3x21D uses 25A driver and LHP73B gains much more between 22 and 25A than SBT90.2.
4 Thanks
I agree that pushing LEDs to their limit is generally not worth it in flashlight applications, but are we talking about the same emitter here? There seems to be a discrepancy between your quoted statement and the graphs you posted.
Your graphs show that at 22 A, you get around 7450 lm @ 70 W (106 lm/W). To add another 10% flux we need to go to 25.6 A or 82W (100 lm/W), only a 17% increase in electrical power - far from double. 100 lm/W is still excellent. For comparison, in your test of the SFT-40 6500K, you recommend running it at 85 lm/W.
At 30 A, the flux curve looks like it is only just beginning to flatten - assuming the typical curve shape it looks like the maximum in your lab test should be beyond 35 A. When I first saw your graphs, I assumed you simply stopped the test at 30 A because you used a 30 A power supply.
I asked Simon about running the LHP73B in a 3x21A with the 25 A buck driver. I assumed 25 A would be just perfect for this LED. He however replied that it would be too much for the LED - which confused me.
Is there some overheating behavior going on at around 30 A that is not obvious from the posted graphs?
I can understand Simonās point of view. My tests are not directly comparable to the use of an LED in a normal flashlight. In my test, the LED is screwed onto a large heat sink with ideal heat transfer (this is what you call lab test) whereas in a flashlight this is not the case; there are tiny irregularities in the lamp head, and above all, the mass of the lamp head is many times smaller than that of my heat sink.
What still works in my test (with a corresponding reduction in service life and massive loss of efficacy) can lead to overheating or associated damage/destruction of the LED in a lamp, especially when operating close to the maximum possible operating current.
I donāt know whether Simon built and tested this configuration (25 A driver + LHP73B) himself, but if I were in his position, I wouldnāt take any risks either, especially since the probability of a defect and thus a possible customer complaint (with the associated costs if a replacement lamp or LED has to be sent) is significantly higher.
5 Thanks
Iād agree with that except for one thing, thatās from yuor own test:
SFT-90 maxxes out at 20A on your perfectly cooled test station, in a less perfect scenario itās going to maxx out earlier. Yet Simon is driving them at 20 amps. I think itās less about whan an individual LED can actually do but Simons runs a company, he has to more or less stick to the manufacturers claims.
The SFT-90 in the test was likely a pre-production sample, since I bought it when it first become available . This may mean that the maximum current is only just being achieved or that other deviations from the datasheet are occurring, that mass production of these samples is still in its beginning and that production errors can still occur due to fine-tuning of the production line or something like that.
However, the pre-production sample is also stated in the first few paragraphs of the SFT-90 test.
I donāt know whether the currently available SFT-90s are better, but I canāt test this either, as I donāt have any current samples.
LEDs from mass production can also perform unexpectedly poorly (Iāve had this with an old XP-G2 and another LED, could be from Lumileds so far I remember, these had much lower max current due to insufficient heat dissipation I assume) because there was a problem in production and they slipped through QC. It happens.
I donāt know whether Simon conducts his own tests. Given the quantities he sells, however, it would make sense to do so in order to avoid any surprises and the associated increase in complaints.
Someone tested a newer batch SFT-90 recently and iirc it did ~10% more flux than yours, but obviously that could also be a less accurate test setup (I think some simpler integrating devices overestimate the brightness of throwy emitters?).
That might put the SFT-90 over the SBT-90.2 in terms of brightness even despite the worse heatsinking and less bond wires if true.
Could also be a better flux bin, or another geometry of integrating device, or different luxmeter, or heatsink, or thermal paste used etc.
10 % is not a huge difference tbh, flux bins can make 7 % difference alone.
1 Thank
I assume Simon does his own tests.
Even if 20 A is not optimal for the SFT90 (say it could have more output at 17 A), maybe he found it to at least not damage the LED.
Maybe heās selling a non-ideal config in this one case because people want the SFT90 but he doesnāt have a suitable ~15 A driver available. I donāt think LED specs come into it much, flashlight manufacturers are notorious for completely ignoring LED specs (and its what we all want).