LED test / short review

koef3 · May 24, 2024, 11:44am

Luminus SFT-70 3000 K 95 CRI

2024-05-24

Link to original test: Klick

Bin: unknown
Color group: unknown (3000 K)
CRI: 95

Attention: Data sheet not available as of 24.05.2024.

The emitter was purchased from Kaidomain in May 2024.

The most striking feature compared to the already tested cool white SFT-70 is the significantly more orange luminous surface, which is typical of warm white LEDs (especially those with a high CRI). The design and footprint are otherwise identical to the cool white variant and require no further explanation.

at 7000 mA (official maximum current): 2183 lm @ 6.99 V
Power at official maximum: 49.0 W
Efficiency at 7000 mA: 44.6 lm/W
Maximum reached at 9.6 A, at this point 2389 lm @ 7.48 V
Power at maximum 71.8 W
Efficiency at maximum 33.3 lm/W

Data for 25 °C Tsp (at 85 °C the luminance values are around 13 % lower).

As with the cool white version, the achievable increase in luminous flux above the official maximum flux is low. In the absence of a data sheet for the warm white SFT-70, the official maximum luminous flux of the cool white version is assumed here. As with the warm white SFT-40, the efficiency of the maximum possible luminous flux is very low and corresponds to that of an overcurrent halogen lamp.

Nevertheless, this LED offers a higher output than the SFT-40 can deliver, and the luminance is high enough for throwers. The LED tested here was not damaged after running at maximum current for over 5 minutes.

The beam corresponds to that of the cool white SFT-70 and there are no disturbing artefacts or rings. If an SFT-70 in cold white offers a good beam in a given optic (reflector or lens), a 1:1 emitter swap is easily possible.

The light quality is simply perfect. There is no annoying tint, apart from a slight reddish tint (‘rosy’), which is hardly noticeable due to the warm CCT and makes this light even better. The colour quality is very good, the high colour rendering value is particularly evident when compared to other 70 CRI LEDs with the same CCT.

Ra: 94
R9: 89
CCT: 3114 K
duv: -0.0032

Luminus can do warm white. The SFT-70 3000 K 95 CRI provides a very beautiful light, can be replaced directly with 6 V-XHP50 LEDs or other SFT-70 emitters thanks to the 5050 footprint and has a higher output than the SFT-40 in the same CCT.

Thank you for reading this short test.

Greetings, Dominik

v1.0.0

carsknivesbeer · May 24, 2024, 4:48pm

Thank you for all of your hard work. It is really valuable to get hard numbers for all of these emitters!

ControlTheController · May 24, 2024, 5:56pm

Thanks for the review!

This and the SFT-40 is great for straight drop-in LED replacements on 5050 emitter lights - no need for the hassle to change MCPCB footprint.

This just feels like a 3000K SFT-40 but with more dies. (Im)patiently waiting for 4000K…

elemental · May 25, 2024, 3:06am

Does anybody know if there will in fact be a SFT-70 4000k released this year? Based on the recently released SFT-40 4000k, and the correspondence in tint between SFT-40 and SFT-70 3000k, I might guess that the 4000k might be both an lumen and throw increase over the 3000k, while still being high CRI and neutral tint? Could be a awesome 6v XHP50.3 HI replacement.

James_C · May 25, 2024, 3:53am

Is there much tint shift throughout the beam profile? I have some XHP50.3 hi 3000K, I don’t recall the bin right now, that looks nice but has the “Cree rainbow” to some extent.

Thanks for the review!

elemental · May 25, 2024, 4:16am

Very little tint shift in the SFT-70 3000k. Nothing comparable to the Cree rainbow of an XHP50.3 HI 3000k, and a much nicer 3000k tint to begin with.

koef3 · May 25, 2024, 7:25am

Tbf, for XHP50.3/70.3 HI I never saw this “Cree rainbow” as extreme as with older Cree (XP-G3, -L2, XHP50.2). This effect depends also on the reflector used.

Light_Veteran · May 25, 2024, 7:37am

Thank you Dominik!

Pavlo · July 10, 2024, 2:15pm

Does anybody know if this SFT70 shifts to green at lower current?

billy_s · July 17, 2024, 3:03pm

I have some G1-HA3 bin and it’s not green at all level. It looks better than my unknown bin sft40 3000k.

koef3 · July 17, 2024, 6:25pm

Overlooked this post

My samples are perfectly warm white, without any disturbing tint or weird tint-shift. No green or yellow.

This supports the theory that these ugly tint-shifts for cool white CCT for these ‘Luminus’ LEDs is due to the phosphor mixture Sanan is using.

14500 · July 31, 2024, 6:51am

Quick PSA: do not run SFT70 5000K at 8A.
It will be less bright than the 6500K at 5A (both in Convoy M21B hosts).
I did not observe any damage, but something is off.

koef3 · July 31, 2024, 7:31am

Thank you for the warning. It seems something seems different with the phosphor mix of 5000 K Luminus LEDs, the SFT-40 5000 K had the same issue, but was also heavily burned in a light flux test.

14500 · July 31, 2024, 5:14pm

Yes, same goes for SFT40 5000K: do not drive at more than 8A.
I tried a Convoy 12A driver, but it turned angry blue (>7500K) so I swapped it for a 6500K which seems just fine.

14500 · July 31, 2024, 5:17pm

It’s a bummer, because I much prefer a 5000K-5700K thrower.
I wonder whether this is also the same for the 5700K variants, and if SFT25R will have the same issue.

QReciprocity42 · February 21, 2026, 7:20am

Randomly noticed that my SFT70 3000K seems visibly worse at rendering deep blue objects compared to the SFT40 3000K, next to a halogen light for standardization. Checked a bunch of colored papers and textiles and the results were incredibly consistent, so I decided to return to the tests of the 2 emitters to see what’s going on.

It appears that while the SFT40 3000K uses a 450nm pump, the SFT70 uses 460nm! The same trend is also observed for the cool white variants. This 10nm shift explains why deep blues look worse under the SFT70, and makes the SFT70 not just a larger SFT40 version, but a fundamentally different emitter.

On the bright side, the SFT70 3000K renders greens in a more faithful way–dark greens especially appear more saturated and less pale than with SFT40. This is probably explained by the less pronounced cyan dip. Also, it seems to make wood/skin tones extra rosy, even though the tint on white wall is slightly greener than SFT40, for my samples.

Spectrum overlay below–it is very easy to see the offset in the blue region between the SFT40 and 70.

lndger · February 21, 2026, 10:22am

I’ve been curious about this since reading the review. I couldn’t understand why it’d be so different to the sft-40. Thank you.

On a related note, I got a KR1AA recently with a sft-70 3000K (first time I’ve actually tried this emitter in person for some reason) and was genuinely surprised by how nice it was – subjectively much nicer than the sft-40 3000K emitters I’ve tried. Perhaps this difference here is related to why.

QReciprocity42 · February 21, 2026, 4:05pm

I also find the SFT70’s light to feel nicer/softer in practice despite the lower CRI, and am tempted to attribute that difference to two factors:

The SFT70 has much better power handling, more than the difference in LES would suggest. As a result, the SFT40’s spectrum quickly saturates with blue even at medium drive power, while the SFT70’s spectrum remains un-distorted even at higher modes.
As the spectra suggest, the SFT70 has a much less pronounced cyan dip. In my experience with various blue- and violet-pumped emitters, I find that cyan dip is responsible for making the light look pale (especially on white surfaces and on skin), and making reds look worse due to lack of contrast. The SFT70 is thus nicer in this regard.

I still haven’t been able to fully explain how the SFT70 makes skin/wood tones extra rosy despite a slightly less rosy tint on white wall, but perhaps this is due to the white-wall rosiness of the SFT40 coming from too much blue.

crbnfbr · February 26, 2026, 10:02pm

I think the difference in blue spike has to do with the lower current the four dies of the SFT70 get compared to the SFT40 single die.

First pic is an SFT40 with Convoy 5A buck at 0.1%. The spike isn’t as high as the SFT70 at low current. Unfortunately, Convoy’s 0.1% isn’t really that low, but I’m guessing at even lower output the spike would be closer to 460nm. Second pic is an SFT70 with LumeX1 at mid ramp, you can see it’s spike ~450nm, same as SFT40.

I have no idea why the spike changes so much with current though.

QReciprocity42 · February 26, 2026, 10:13pm

That’s an interesting hypothesis, and certainly does not contradict your observations. I find it strange because the peak wavelength of LEDs is known to shift up (i.e., in the red direction) when subjected to high temperature, so I would expect the same to hold in high current density. But if this is true, then the SFT70’s lower power density should put the blue spike to the left of the SFT40’s, which is the opposite of what we observe.

To check this, could you run some color reports for the same LED (say SFT70) across different power levels?

Your SFT70 spectrum here looks very atypical with a pretty spiky blue and correspondingly pronounced cyan dip (despite the high R12!). I wonder if Luminus changed the production process/materials between batches, which they tend to do frequently.