There are some things I don’t understand. I have modified two CULPM1/CSLPM1 flashlights. I tried to keep them at the same focal point, and the SFT25R brought a measurable improvement at the same current.
Although I used a cell phone to measure, I used the same setup as in many previous tests, all measured at 5m. The SFT25R consistently produced higher numbers than ever before.
Why is this? The luminance of the SFT25R doesn’t seem to be as high as the CSLPM1. Could it be that the round die provides better optics, or is there some other variable at play?
The CULPM1 has 4040 footprint, which allows for much better heat dissipation from the LED chip in comparison to CSLPM1. But from experience Osram emitters are mostly on the lower flux binning side since these are not binned like with Cree or Luminus emitters (they just ship one huge flux bin group which has everything from bad to extremely good in them). I think this is an important factor we should consider here.
Warning: The first sample tested of my SFT25R 65000K F8 from Kaidomain produced a ‘green egg’ hotspot with 0.0175 center duv in a T3 SMO @8A. OP tamed it to 0.0115.
The SFT-25R from Kaidomain are all listed as being ‘B’ binned (BB, CB and DB), which is above BBL.
The SFT-25R from Convoy, tested above, is listed as being ‘G’ binned (AG) which is below BBL.
This BBL difference may also explain the Convoy F6 flux bin (as @14500 mentions above) vs the Kaidomain F8 flux bin.
Not sure how useful this info is but the Luminus datasheet claims F8 as the minimum flux bin for 6500K SFT-25R yet the Convoy samples are F6 and listed as engineering samples ( SFT-25R-WS65-A2-ENG(F6-VJ-AG) ) so they may be the only ones of their kind.
Also, the Convoy samples are ‘VJ’ voltage bin which is the lowest bin listed in the datasheet, the voltage bins aren’t listed by Kaidomain but the Convoy sample tested by @koef3 may be binned differently enough to other available samples to justify going against the ‘statements in advance that the forward voltage of the SFT-25R was very high and that the tint was often very green, as with Luminus.’
I find the component tests performed by @koef3 and others very valuable indeed, and without them my torches would simply not be as functional, capable, safe, and as understood as thay are.
But this test is a very good example of why it is important to understand the context of the data provided, and that the specific data provided can only really be used to form a general opinion of a component’s capabilites rather than used as definitive data for a whole model range.
Thanks for the quote. Actually, when I and others call it "high Vf’, it is actually according to the datasheet, which says that the VJ bin is in the range of about 3.7-3.9V at 8A, which is consistent with the results of OP’s test. This Vf is not essentially high, but it is not low compared to the SFT40 and CULPM1 at about 3.6V at 8A. Most of the samples I have seen so far are VJ forward voltage bin.
This is quite significant for use with 14500. Even an H10 can only supply 3.9V8A at full charge. At 90% charge it will sag to 3.6-3.7V and at 50% charge 3.4V meaning that the current will drop rapidly, and already be below 4.5A by mid-charge. Essentially the current will decrease linearly as the cell discharges. Not exactly desirable.
I hope that they eventually release a 5050 version that has much lower Vf and that can handle 10A - a hypothetical 5050 ‘SFT30R’?
Don’t think that this depends on the footprint. The size of LED chip and footprint is more crucial for maximum current and efficiency. It is more depending on design (lateral with bonding wires is always worse in regard of Vf) and LED chip quality. Look at Nichia, their chips have really low Vf. A hypothetical SFT-30R would be in same design with bonding wires and these Sanan LED chips, so higher Vf is still likely.
It would be interesting if Nichia is releasing true round die emitters some day.
It would also be very nice if Nichia released 5050 emitters using a normal 3-pad footprint with an isolated thermal pad too.
B35AM and 144A are noticeably hindered by this, with the MCPCB quality having a notable effect on the maximum safe current.
The B35AM is a wonderful LED for single LED boost driver lights, with great light quality and efficiency - but the footprint and the resulting thermal resistance keeps it from being the probably best all-round/floody high CRI flashlight emitter we have right now.
Sadly I think Nichia will keep their secrets, for the same incomprehensible reasons they’re not available from normal component stores like Mouser and Digikey.
A 4040 or 5050 B35AM would be a dream emitter, as well as a domeless 144AM in an XM footprint.
This is crazy! I’ve seen LEDs with this issue due to unevenly distributed phosphor (Nichia 144A having an extremely bad batch), but not from Luminus. What does the image of the die look like when projected by an aspheric lens?
