Sofirn C8F 21700 Teardown and Review

Thanks for the review and also the analysis on why the Sofirn C8F 21700 may sometimes be underperforming… Hope Sofirn will solve the “underperforming” issue soon.

They already know the old FET is not the best, so it hurts performance a little. They have already switched to a better FET.

I don’t know if all the lights have a weak contact on the driver retaining ring. Some may.

I also recommended they swap the led wires to 18ga. It’s a cheap way for them to boost the output even higher.

Here’s an update. I swapped out the factory wires (I think 22ga) with some heavier 18ga. I ended up shortening them about an inch, but that’s barely long enough. I recommend 1/2” shorter at most.

Once I got a decent connection with the driver retaining ring, output jumped up quite a bit. Look at the bottom of the page. I was in a rush so I only did Turn On and 15 seconds for both tail amps and lumens with each battery.

The Sofirn battery gained about 2 amps and 400 lumen. 3,450 lumen, wow!

The 30T battery gained about 1.5 amps and 300 lumen.

The 50E battery gained about 1.4 amps and 300 lumen.

The little Samsung 30Q gained about 1 amp and 300 lumen.

So some pretty nice gains just by swapping to 18ga wire. :beer:

But I thought your lumen sphere is calibrated with Maukka lights so it should be reading about the same as mine. So if Dale’s measurements are similar to mine, then shouldn’t it be similar to yours?

Very nice info. So the 30T measured 3,400 lumens, which is less than the Sofirn 4000mah because of bad connection? Also the performance of the Sofirn 4000mah is very impressive. Seems like a rewrapped Samsung 40T.

I don’t know why, but my lumen tube seems to read a bit lower than others. Not so much with other TA Tubes, but with some other different designed measuring tubes.

Like when it comes to the xhp70.2. Dale can measure lights with that emitter at 6000-7500 lumen.

With my L6 I can only get 5300 lumen at turn on. This is with 2 different color temp leds and Liitokala cells drawing 17.6A. Even with Aspire 4300 cells drawing 19.6A, I only see 5500 lumen at turn on.

Keep in mind that Dale does not use a TA lumen tube. He uses a lumen tube design with no diffusers and calibrated by a wide variety of factory lights. (Unless he has recently bought a Maukka calibrated light). I think he always uses turn on numbers and never anything lower.

So I always take his, and everyones, lumen results with a grain of salt. People should not take my lumen numbers as exact, but just in general.

I am really surprised your lumen numbers and Dale’s are really close. Have you compared the output of several factory or modded lights between him and you or just a few?

Besides the C8F, I haven’t compared my numbers with any of his. I guess the diffuser on the TA tube evens out the beam for more accurate result whereas other lumen measuring devices might favor concentrated beams.

I would say the 30T and Sofirn 4000 are pretty much similar in this particular test. I think the emitters and circuit are starting to max out. We might see the 30T gain an advantage again if we swap out the skinny driver bypass wire with an 18ga. I’m guessing here.

If we go back to the skinnier stock led wire test, we see the 30T has a definite advantage with about 0.4A and 200 lumen more. So it is the better cell when it comes to high amp loads and having lower internal resistance. Dale has used a pair of 30T on a triple xhp70.2 light and they drew about 48A-50A. I don’t think the Sofirn 4000 would do as well in this extreme test.

Also, I did this last set of tests pretty fast and the light only cooled down for maybe 2 minutes between each pull. It’s not ideal, but still gives a pretty good idea of the differences.

We also know the Sofirn 4000 is not a 40T. It’s made in a different factory (in China) for the electric vehicle market. It’s definitely a good cell. Maybe it’s a Lishen LR2170SA 4000mah? The Liitokala 4000mah 21700 is a rewrapped Lishen and this cell performs really well also. I don’t know, but they do seem very similar.

Yes, very throwy lights do measure even higher with his tube. I think I found his tube sometimes measures between 8% to maybe 16% higher than mine on average.

He may know a few other tricks to boost output as well. Like using warm batteries fresh off the charger. I’ve heard warming up the batteries gives them some extra oomph. He might replace the lenses with UCP which have lower losses, etc… Dale just gets all the highest output. I can’t compete with him. He’s the king! Lol

Didn’t know China makes batteries for electric vehicles too. Would love to see a HKJ comparison Sofirn or Lishen LR2170SA vs a 40T and VTC6A soon.

Probably should send Sofirn this info so that they can improve the wire gauge being used and increase the lumens output to the correctly stated lumens (C8F 21700 stated to be 3500 lumens, and so far the stock form varies from around 2300-3000 lumens based on these user reports in the forums, which is way short of the “3500 lumens” claimed output.

I mentioned that I sent Sofirn this info in post 56.

They probably should just reclassify the light as 3,000 lumen. That would be more accurate.

The only real change between the 18650 version and the 21700 version is the battery size. It doesn’t make sense that a battery change alone would give you an extra 1000 lumen. An increase of about 500 lumen is much more realistic.

I hope they change the wire size, then they could rate it above 3,000. A real hotrod then.

I’ve been working on and off on trying to figure out how the switch light circuitry works, but so far I’m at a dead end. It’s certainly not a traditional switch light setup that uses two different color LEDs. On those you typically have one wire that acts as a common ground or common positive and then three more wires for the switch and two colored LEDs. The C8F has a switch, two colored LEDs and uses four wires, but the design is very different. Maybe someone here that knows more about electronic circuits can figure it out. Here is what I have so far.

A couple of the circuits.

Wire #3 does not go to ground to make the red led light up.

Both red leds are wired in parallel. Both green leds are wired in parallel.

Here is the basic led circuit minus the switch stuff and the duplicate leds. This is the most up to date.

If anyone can make sense of it, please let me know.

anyone if they will be doing the c8t in 21700??

Store Sofirn kindly sent me a driver for C8F 21700
In my opinion it is different from the original version.



I’d kind of like to see a 26650 C8F or something even floodier.

Did you measure the output?

Sofirn Samsung is very kind. They had a brand new C8F 21700 shipped to me from Amazon for free. I was shocked to see the label titled to SKV89 and was wondering what the heck could be in the package. The output, unfortunately, measures the same as the output of the one I bought originally from Aliexpress and is also the same as the output after swapping in the free replacement driver into the Aliexpress C8F 21700.

Regardless, I sincerely appreciate Sofirn’s generosity and excellent customer service.

You should try the new Sofirn SP33 with xhp50.2. It has a little bit larger hotspot. Nice 26650 size. The boost driver kind of spoils me due to it’s consistent output and low heat production. It has replaced my C8F.

I think I actually have two of those coming to give to the inlaws as a Christmas gift. I also got a single C8F to play with.

Received my new order and it is disappointing to see no change in output. The plus side is the thermal sensor making it safer and more user friendly for my customers.

I get no higher than 2650 at 30 sec

It’s puzzling as to how they got 3500 lumens. Their ratings are usually not that far off. I see d_t_a mentions a wire gauge change boosts lumens a fare bit. I’m not going to bother as that would be a big hassle to perform on all of them