LED test / review - ChangDa SFH55 6500 K 70 CRI - Just pure insanity!

LED test / review EN


ChangDa SFH55 6500 K 70 CRI


2024-09-22


  • Type: multi-die, 16-chip
  • Bin: —
  • Color group: — (6500 K)
  • CRI: 70
  • Rated voltage: 3 V V
  • Max. Forward current: 50,000 mA
  • Max. Peak current: — mA
  • Viewing angle: — °
  • Thermal resistance: — K/W
  • Max. Temperature Tj: max. —°C

Note: there is no official datasheet available!


The emitter tested here was purchased from Kaidomain at the end of August (reflowed onto a 20 mm board).

The SFH55 is an impressively large LED. With dimensions of 9.9 x 10.9 mm and a 9090 footprint, it just about fits on the 20 mm board. The size of the illuminated area is also impressive and atypical for conventional high-power LEDs.

Due to the asymmetrical shape, the use of classic ring-shaped centering gaskets is hardly possible; on the other hand, a slight de-centering should hardly be an issue with the huge illuminated area. If precise centering is required, a special centering aid must be designed and manufactured.

The 16 individual LED chips are surrounded by a white silicone potting compound and covered by a transparent silicone layer. It is noticeable that the LED chips have been mounted offset from each other in a regular pattern, possibly to enable 16P (16x parallel) wiring. This wiring enables the low forward voltage of around 3 V.

The illuminated area is 31.9 mm² in size. However, 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; the real active area is larger, at 38-40 mm². The LED chips are - to the best of our knowledge - manufactured by Sanan.






  • Maximum reached at 55 A, at this point 13027 lm @ 3.35 V
  • Power at maximum 188 W
  • Efficiency at maximum 69.3 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 50 A (!) specified by Kaidomain is slightly exceeded here - in contrast to other ChangDa emitters, where the given maximum current is hardly reached at all. It appears (if Kaidomain’s information is correct) that ChangDa does not specify an official maximum current in the presumably existing data sheets, but only the absolute possible operating current with ideal cooling. How this was determined is unknown, but it is possible that ChangDa calculates it using data from the LED chip supplier.

The performance of this LED is simply extreme. With 13000 lm (!) from one LED package, it achieves by far the highest value in my tests to date. The XHP70.2 HD and XHP70.3 HI are easily beaten. And the efficiency is extremely high. Due to the 16 chips connected in parallel, in each of which one sixteenth of the maximum current flows (55 A / 16 = 3.44 A), the Vf is extremely low, although it is still relatively high when viewed on the individual chip. The Vf of the SFH43 is significantly higher, probably also because the chips and their heat dissipation surface are smaller and there is possibly no voltage binning on the part of ChangDa or the supplier (huge bandwidth in binning?). However, this is difficult to verify due to the lack of a data sheet.

Direct drive (operation of the LED without driver directly on the battery) is possible. With a cell with a very good voltage range (Samsung 40T, for example), 40-50 A can easily be possible! The official discharge current of the cell used must be observed!

Operating the SFH55 at 55 A for 20 minutes did not result in any damage to package and/or LES or changes to the electrical parameters/characteristics.

Although the luminance appears to be quite high, this only applies to operation at the maximum possible current. In practical operating conditions (30-40 A), the luminance is less high. In comparison with other measured values at the same operating current, it is important to note that the SFH55 is a 3 V LED, whereas XHP70.3 HI and others are operated at 6 V.

There is no sufficiently large reflector for a beamshot. Normal-sized reflectors with a 9 mm emitter opening are too small and could cover part of the illuminated area. Therefore, the use of such an LED with such a small reflector is not recommended and can lead to damage to the LED due to heat build-up/too high radiation density as well as to a beam with artifacts.

At first glance (with a reflector that is too small), the beam corresponds to that of an XHP70.3 HI: hardly any color distortions, hardly any artifacts. An OP reflector delivers the best results.



The tint of the sample tested here is very good. It is a cold white at around 7400 K, with a duv close to or just below zero. There is virtually no disturbing tint shift with increasing current - as with cool white Luminus LEDs.

Otherwise, the spectrum offers no surprises; it corresponds to that of a standard cool white LED with a CRI of around 70, although the color rendering of this sample is slightly higher - presumably due to the slightly higher red component.

It should be noted that series variation can occur; other specimens (especially from other batches) do not necessarily have to have a tint close to or below duv.


  • Ra: 73
  • R9: -17
  • CCT: 7413 K
  • duv: -0.0004

This LED is simply brute. 13000 lm from one emitter is unprecedented, apart from COB LEDs with a completely different design and LES size. The efficiency is very high thanks to the 16P design and the Vf is extremely low. Thanks to the 9090 footprint, it is also interchangeable with other LEDs, although the reflector must be large enough.

Thanks to the good beam quality (especially with OP reflectors) and the very low tintshift over angle, it is suitable for flashlights where just pure power is required and where color rendering is of little relevance.


Thank you for reading this test. :slight_smile:

Greetings, Dominik


v1.0.2
14 Thanks

Thank you Dominik for the testing, it is much appreciated

This is one of my favorite emitters for larger extreme output lights (multi-21700). Glad to see the new higher-powered power supply getting use, 55A maximum brightness is just incredible.

2 Thanks

Yeah, finally a proper opponent for the new high-power PSU :smiley:

2 Thanks

It would also be interesting if you could test the 11x10mm version of the SFH43, it is used in the Haikelite HK08 (one of the brightest lights for less than $200)

1 Thank

According to this post it is just the same as the 7070 I already tested but only with bigger package, I don’t think that these are more powerful as the 7070 version as this does not exceed the maximum power possible on 7070 footprint.

3 Thanks

Judging by pictures, the SFP55(.2) which appears in the Lumintop GT4695 for example, is probably also made by ChangDa, which also explains the nomenclature. Based on this, the SFP55 should be even more extreme, as it is equipped with 25 55-mil chips instead of 16 as in the SFH55 tested here.

3 Thanks

It would be very interesting if you could test it, there is also the 25-die SFH43P and the even-larger SFY55 8x8 (64 die), but I’m not sure where they might be available other than Changda’s Taobao store

1 Thank

So far I know there is no way go get the SFP55(.2), apart from salvaging it from one of these flashlights.

Since I don’t want to buy one of these lights only to get theLED I cannot say when I am able to test the SFP55(.2) :neutral_face:

It might be possible to obtain the SFP55 by asking Lumintop, IIRC a BLF member did that in the past. I’ve gotten other hard to source parts by asking nicely.

If shipping from my country to Germany wasn’t problematic (multiple returned packages without a good explanation) I would gladly help out.

You can get them from Lumintop (with a MCPCB that weighs like 20kg)

Or you buy them from the ChangDa Taobao store, but they aren’t declared as “.2” so maybe they are a slightly lower bin

1 Thank

What is the best way to contact them?

I sent them a message on Aliexpress, to the “Lumintop Official Store”

They might give you an inflated price if buying only 1, but that is just a “make it worth my time” price, so they can usually include other small cheap things like springs or o-rings for free.

1 Thank

What do you think how much will 2 LEDs would cost?

No idea, maybe $10 each? You will probably need to ask, especially if you only need 1 MCPCB, as they’re a gigantic chunk of copper (probably 3-4mm thick) with very heavy gauge traces.

Ok, I will ask them.
I don’t need this chunk of copper since it would not fit on my testing rig lol

Hopefully they sell these LEDs without copper MCPCB chunk.

It uses a proprietary footprint, does not go on a regular SBT90 board. You might have to ask if they have a smaller MCPCB available instead.

The thermal pad is wider, and might short circuit to one of the pads on the sides.

That could be a problem since I can only use 20mm MCPCB on the heatsink. We will see… I asked them for two emitters. Maybe I am able to modify the MCPCB chunk to fit on my heatsink.

1 Thank

You could try to contact Nightwatch
Although I had no success with that in the past

Their SFP55.2L Atleast has a standard 25mm MCPCB