LED test / review - Sanan SFN40 3000 K 90 CRI - unpleasant yellow light, but very low Vf and good beam

Flashlight Modding and DIY Parts
1

LED test / review EN


Sanan SFN40 3000 K 90 CRI


2024-07-06


  • Type: multi die (9x), domeless
  • Bin: unknown
  • Color group: unknown (3000 K)
  • CRI: 90
  • Rated voltage: 3 V
  • Max. Forward current: 25,000 mA
  • Max. Peak current: unknown
  • Viewing angle: unknown
  • Thermal resistance: unknown
  • Max. Temperature Tj: unknown

Note: there is no official datasheet available!


The LED tested here was purchased from Kaidomain in June 2024. It is listed there as “SFN40 9x core 3 V”.

The SFN40 is an LED without a dome (called domeless). Its special feature is the 9 LED chips, which are arranged in a 3x3 grid with small gaps between them. There are no bonding wires or visible ESD elements. The phosphor has been applied on and slightly next to the LED chips. The individual LED chips are barely visible due to the thick phosphor layer.

The footprint corresponds to the familiar 5050 format. The SFN40 is compatible with normal XM or 5050 boards. The symmetrical housing means that standard centering gaskets for 5050 LEDs can be used.

The illuminated area is 12.1 mm² in size. Due to the light emitted from the side and the gaps between the individual LED chips, it is almost impossible to determine the exact illuminated area. It is noticeable that the phosphor has not been applied evenly to the illuminated surface; the light emitted at the outer corners has a higher color temperature. Depending on the optics used, this could have an effect on the beam.







  • Maximum reached at 24 A, at this point 3295 lm @ 3.53 V
  • Power at maximum 84.7 W
  • Efficiency at maximum 38.9 lm/W




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


In the absence of an official data sheet and operating parameters, it is difficult to evaluate the performance. The maximum current of 25 A (!) specified by Kaidomain is not achieved even with excellent cooling at lab conditions. In a flashlight these maximum values would be even lower. The 5050 footprint is the primary limiting factor here due to the limited heat dissipation surface. It is likely that the stated 25 A does not represent the official maximum current. The thermal resistance is extremely low and should be somewhere in the range of 0.8 - 1.5 K/W, which is a very good value.

The Vf is extremely low due to the design. 3 V at around 7 A is the lowest forward voltage I have ever tested in a 3V class LED. Even the Luxeon MZ achieves around 3.1 V at 7 A. This is due to the 9 LED chips connected in parallel, each of which is only supplied with around 0.8 A at 7 A and therefore has a very low Vf.

The very low forward voltage means that the use of classic AMC7135-based drivers is no problem, and constant current regulation is guaranteed over a wide range of battery voltages.

The SFT-40 is very inefficient due to the very high CRI of 95. At its core, the SFN40 is quite powerful and offers quite high efficiency for warm white light.



Despite the unevenly applied phosphor and the light emitted by the sides of the LED chips, the beam is perfectly fine. (The yellowish tint may vary slightly depending on the screen, but on a calibrated OLED the color corresponds pretty much exactly to the yellow tint seen in real life).

There are no noticeable artifacts such as rings or donut holes. Using an OP reflector results in an absolutely harmonious and uniform beam with a large corona.



The tint is very yellow, which is to be expected due to the high duv of over 0.008. Despite the nominally high color rendering, the light is not pleasant at all. Colors appear washed out and less saturated, like if they are covered with some gray haze. If good light quality is required at 3000 K, emitters such as the Nichia 519A / 719A / B35AM or Luminus SFT-40 / SFT-70 are clearly preferable.


  • Ra: 92
  • R9: 61
  • CCT: 3208 K
  • duv: 0.0084

Briefly summarized: The SFN40 is a powerful LED with very low Vf, and quite high efficiency. It is compatible with 5050 footprints and provides a good light image in reflectors despite the 9 LED chips.

However, the light appears very yellow and the nominally high color rendering value is subjectively not very pleasant. In addition to a lack of sources of supply in the West, no official data sheets are available, which is why the use of this LED must be viewed with caution.


Thank you for reading this test. :slight_smile:

Greetings, Dominik


v1.0.2
15 Thanks
2

Interesting review, thanks for this. One small suggestion: would you consider displaying the output graphs with some kind of normalised values for the voltage & current? i.e. halving the voltage & doubling the current of the 6V emitters, for example. This would make it easier to compare the 3V & 6V emitters directly in terms of efficiency.

1 Thank
3

Power on X, lumen on Y axis would be interesting - since he already measures current and voltage, that should be no problem.

With linear drivers, only current is relevant for efficiency, but with buck/boost, power is.

4

I don’t want to change the values I measured. I also don’t think the values can be divided by half just to get values similar to 3 V emitters.

I will consider this for future testing.

1 Thank
5

I agree, I wouldn’t do that either. It’s a 6V LED, so it should be listed with the proper forward voltage in the diagram.

If anything, to adjust, you’d have to half V_f and double current for 6V LEDs so they would fit (basically pretend it is a 4P instead of a 2S2P chip arrangement), but I feel like that would cause too much confusion (“3V B35AM? Where???”). Also, current over bond wires and drop would probably be affected by such a “physical” change, so just changing it in the results would skew the results a bit. I’m pretty sure with identical chips a 6V 50.3 HI would perform (a tiny bit) better than a 3V 50.3 HI, for example.

If you want to fully normalize the values, swapping X axis from current to input power would be the way to go, but that’s an extra step of work…

6

That’s what I did for the 719A vs 519A from Djozz data and it’s very useful for comparing similar LEDs of different voltage, actually it’s also good for LEDs with the same nominal voltage since the Vf might be different enough to noticeably affect the efficacy.

image
image1292×994 57.3 KB

2 Thanks
7

This looks useful, yes! Now in the day of Bucks/Boosts in most higher end lights imo power is the interesting metric, while current is pretty much irrelevant.

8

Ok, I will edit my templates and put a chart like this in. I will post these in upcoming tests.

6 Thanks
9

@thefreeman @ebastler which metrics for charts would you also see in future tests?

10

For me, personally, the lumen over power is the most interesting. But I might be a bit of an exception, I’m a sucker for efficiency. Highly efficient drivers, highly efficient LEDs - I just love maxing lm/W (without sacrificing tint, I want no green efficiency monsters).

If you also determine cd/mm² for every step (not sure if you do), luminance over power might be cool too. Gives a better feeling how throwy or floody an emitter is than just looking at the table.

11

For now I am not measuring cd/mm^2 for every step. This is way too time-consuming. The tables show measurements taken at fixed points. They are not enough to show them as a chart.

I am currently updating my test program for the lumen over power charts.

1 Thank
12

Figured as much, since you only show very specific steps. So no automated measurement. I can relate with not wanting to do a lot of steps manually… Even just measuring voltage/current curves for Anduril lights with 7ish steps in default stepped ramp is a major pain, and that is probably a lot less work than candela :smiley:

13

This effort is the reason why I currently have no plans to make charts for luminance. Maybe at a later point. The problem is, I have to reduce the time per every test so I am able to test more LEDs instead of only a few.

1 Thank
14

First power comparison:

Quite impressive, for a 3000 K 90 CRI LED. The highest-binned XHP50.3 HI is more efficient, but not as much as you would expect. The ugly yellow tint of the SFN40 remains, of course.

6 Thanks
15

Thank you for the power comparison! Very helpful!

16

Thank you for making this, it is useful to compare.
I have it’s cool-white cousin (SFN-43) in a Pioneman K75 and it is impressively bright.

Have you tested the SFN-60 (from Hank) before?

17

It’s really great to see this review @koef3!
I’m happy that I bought the 6500K SFN40. There is a bit of yellow in the center, indoors, but it’s hardly offensive and unnoticeable outside. This LED is brilliant bright white with a very nice tint.
This pic is with the WB set at 6500K:

20240831_124733
20240831_1247331920×2280 99.2 KB

This pic is with auto settings:
20240831_124555
20240831_1245551920×2126 135 KB

1 Thank