the mcpcb is available from Zwerglein02 (he is in Germany). He does outstanding work! jlhawaii is using another version of Zwerglein02 boards for 5050 footprint LEDs in D3AA mods to FFL 505 LEDs, and I think they also work for SFT-40
(hopefully I got those footprint variations correct)
Thank you kindly for the pointer. Indeed, I didn’t think to check for PRs, and following the commit link from your PR, I can see the expected changes: defining a new flashlight model “ts10-rgbaux-lowfet”, and the associated model-specific files, so everything makes sense now.
I now understand that someone getting one of these “new LED batch” TS10s, with the added confusion of identifying themselves as model “0000” as you pointed a few posts ago, and naively (but justifiably) assuming they could simply flash them with the previous “ts10-rgbaux” firmware, would fry its LEDs as soon as they try and use turbo.
I think this needs to be posted more prominently than just buried deep in this thread; I’m going ahead and making a specific post about it and pointing it here.
Thanks, Jon! Now I know at least how to find these numbers in a “photometric CRI report” like the one @sp5it posted.
What I still don’t know is what these values mean Saying goodbye to my naive old days when just a single number for CRI (eg “CRI >=95”) was all I needed to know
To avoid getting this thread even more OT than I already did, if you could point me to something I could read to educate myself about that, I would be doubly grateful.
I’m a real fan of R9 as it makes people look more like actual living humans than Simpsons or zombies. It’s utterly essential for medical work and practically mandatory for a lot of camera work because blood is red, and thus has a huge effect on skin tone that is relevant for both medical diagnosis and general aesthetics.
96/91 is a lot better than 90/52. It averages out to about 70, but actually hits an uncanny valley as it does most colors fairly naturally while screwing up what is arguably the most important one. I can handle CRI 79 lighting better than 9050 as it’s at least blatantly unnatural.
Next to it is the old “Dogfarts” LH351D, which looks greener than it’s high duv suggests because in addition to tint, it has no red to counter the green. The Cree (center-left) is straight-up low-CRI, and the 219c next to it is also 9050. The old V1 TS10’s could hang with the 9080 Nichia’s in the bottom row and right column, but it seems that this new batch can’t.
As for the CRI number itself, It doesn’t help that many parts of the world use Re (R1-14) for CRI while North America tends to use Ra (R1-8). So I take all CRI numbers with a grain of salt unless they either specify which range they are using, append an R9 value, or simply show a graph like what you get from a Colormunki.
TL:DR - The CSP2323 isn’t what it used to be. It used to be a small-footprint 519a with neutral tint and 9080 color rendering, and now it’s a small Dogfart.
What a great post! The differences of LEDs on Caucasian skin really show of the R9 values. Did you set the WB to 5000k or something different on the photos?
yes, the V1 tint duv 0.0009 is also much better than the new V2 tint duv of 0.0064
this supports comments suggesting that early CSP LEDs had more neutral tint, the new CSP have rather high Duv by comparison… and apparently also considerably lower R9…
iow, the trend on the CSP is that it is now brighter but greener tinted and worse Red rendering, but more lumens
in fact, too many lumens for the FET in the TS10, more likely to release the magic smoke
unless we Disable Turbo or load Sammys new 50% FET hex file
LOL! Well Said!
thanks for digging up the test of TS10 V1 from zeroair and doing the comparison analysis to the test from sp5it
What a fantastic post. Thank you @jerv for taking the time to illuminate us!
Now I think I’m starting to understand what all of that Rxxyy stuff means: the Rxx part indicates what “component” (ie, red, green, blue?) of the white light it’s talking about, and the yy part indicates how much “fidelity” (compared to solar light?) is has. Please tell me I didn’t get it totally wrong…
Thanks to sp5it for measurement
Maybe a silly question but is possible mix E21A and CSP leds on TS10 board?
Are them electrically compatible?
If yes manufacture should replace them with E21A that have much better tint
Not my beamshots, but I assume that WB was locked by whoever took them.
The LH351D is quite similar there as it has more lumens, both absolute and (especially) per watt than the 519a. The LH351D got it’s nickname from it’s part number, but I’m thinking that the current-gen CSP2323 should be called “puppy farts” as it’s similar in sacrificing R9/duv for lumens but has a smaller footprint.
Mostly right. The issue here is how CRI is calculated. As stated in my previous comment, the way many calculate CRI is strictly R1-8 and excludes R9 (Deep Red), R11 (Deep Green) and R12 (Deep Blue) in addition to the other colors in the R1-15 Extended CRI (Re) test.
The yy part is R9, and pretty much any light that can get an Ra 90+ (the xx values) will do quite well with R10, R13, and R15 and pretty well with R11. The blue spike inherent in LED’s causes issues with R12, often oversaturation especially at CCTs above 4000K, but red is truly the hardest part for LED’s to get right. Warm-tint junkies often over saturate R9, and while it may look good to their eyes, it often looks mediocre to meters.
As an aside, I do think that my day job contributes to me being the sort of CRI baby that I am. Between spending a few years in the optometric/optical world and seeing some interesting things about the subjectivity of human vision, as well as having sensory issues and unusual thought processes that differ greatly from neurotypicals, I have a different perspective (on many levels) when it comes to CRI. If nothing else, I’m more keenly aware of certain things.
Define “same”. After my experiences with my TiTS10’s, I no longer believe that CSP2323!=CSP2323. And the emitter-frying of the HD10’s doubles down on that. And I think that next time LatticePower has an idea of how to make the CSP2323 “better”, Terry and Lucille should have a “discussion” with them.
I have not bought an HD10 and have not seen a spectrum test to determine if the LED is 9050 instead of 9080… (9050 is in the CuTS10, 9080 is in earlier TS10 Ive seen tested.)
the latest HD10 do not ship with Sammys new 2024-04-20 firmware that has only been used on the Copper TS10 so far. If someone buys an HD10 and it comes w 0000-2024-04-20, that would tell me that Wurkkos is using the new Easy Burn CSP…
The latest HD10 (w flashing pads) ships w 2023-07-29… IF those lights had the same CSP as on the CuTS10, they would be burning up the LEDs… No one has reported that their HD10 LEDs have burned up, yet.
someone on Reddit recently bought Three Black HD10… Two of them arrived w No Flashing pads and had firmware 2022-07-25…
iow, the Black HD10 appear to be older inventory… Im waiting to hear reports of the latest Orange HD10… I want confirmation they have flashing pads, threaded bezels, what firmware is on them, and I would really like to see a spectrum test with R9 data.
Im only basing my comments on reports from other buyers, that lack any mention of burned LEDs. It is also possible that no one uses Turbo on a headlamp?
IF you buy an HD10 and are worried about Turbo burning the LEDs you could install the same hex that comes w the CuTS10…
I thought it was but Im getting old and got so many different lights.
Fwiw…In the included manual it says you can unscrew the tailcap half turn to prevent accidental turn-on… … …they should change that.
Saying goodbye to my naive old days when just a single number for CRI (eg “CRI >=95”) was all I needed to know 
