Today (well, yesterday) I modded my brand new Sofirn IF25, twice.
I hesitated for years but now I finally bought a Sofirn IF25, the one with tint ramping from 6500K to 2700K. And it is much fun, really happy with it: it is small for a 21700 light, the chubby light feels good in hand, the UI is well done, the output is good and the tint ramping is magic.
What bugged me though is that the 6500K leds are only 70CRI, but what’s worse: the tint is miles above the BBL, at medium output level the duv of the 6500K leds was +0.0145. This also caused the intermediate tints between 6500K and 2700K to go far above the BBL. Also, at 4000K the CRI was only 80CRI: as well as a tint ramp this light, as you can expect, has a steady CRI ramp.
The first thing I tried was swapping the 6500K leds to the same leds from kaidomain (had them in my stash for a while but never tried them out sofar). KD claims that their 6500K SST20 leds are BA tint which is on top of the BBL. That would not fix the CRI, but it should fix the tint ramp which would shift down to around the BBL now.
It went not as expected, at the same medium level I measured duv +0.0175 now, even worse than stock . I lost the KD tint lottery once again!
The only option now was an entirely different led, and because all other decent 3535 leds have a larger die than the SST20, the hotspot of the warm leds would become smaller than the hotpot size of the cool leds, and thus during the tint ramp the mixed hotspot would be cooler tinted at the edge than in the middle. Question is: how noticable is that and does it matter a lot?
The two options I had in stock were 1) Nichia 219b V1 5700K R9050, and 2) Samsung LH351C 5000K 90CRI (the smaller die version of the well-known LH351D that has a too large die for this optic). I went for the Samsung because the leds receive almost 4A at max setting, and the Nichia can hardly handle that while the LH351C is still under its max at 4A. That the tint ramp will only go as high 5000K does not bother me the least, I dislike really cool tints and to stay above 90CRI at all tints is much more attractive.
Here you can see the size difference of the dies after reflow.
And here is an impression of the resulting tint difference between the leds, although photographing tints is not very accurate.
It so appeared that the LH351C leds used were even a bit under 5000K, at max output I measure 4750K in the hotspot, at medium just 4550K. Still the tint difference over the tint ramp remains large and useful despite the cut-off at 4500K: from cosy warm to a no-nonsense-true-colour-revealing neutral.
In the mixed beam you can indeed see the somewhat cooler edge if you look for it but it is not very noticable even on a white wall. The different die size appeared not really a problem.
I was wondering how to show how the tint changes during ramping without transferring picures from my (standalone) spectrometer to computer and then do tedious photo-editing, so I came up with an animation (with a gifmaker phone-app) of screenshots of the spectrometer. This was fast and easy and it show it well.
The measurements were at high-medium output setting of the stepped output mode-set of the IF25. At 4550K the ramp starts at duv 0.071 which is for my taste not too far above the BBL, no green or anything, and from 3500K and down it stays almost right on top of the BBL. It ends at a nice warm 2580K.
For fun here is the spectrum change for the same set of tint screenshots:
As for the CRI, it is 90+CRI at all tints, going from 90CRI with R9=46 at the coolest tint, to 96.5CRI with R9=81 at the warmest tint.
I had fun modding and measuring this light, this alone was already worth buying it . But I think it will get some use too, I can see me grabbing this one on trips from now on.