Sofirn C8F's Barkuti review

Disclaimer: my Sofirn C8F sample was provided free of charge for me to express my humble opinion on it. :-)

Well, this time it's been a very looong time since I got the torch in the mail. Many issues these last months in my life have somehow lead me to put off with this again and again but I knew I had to make it and finally there it is. I sincerely apologize to Sofirn for this but, on the other hand, pretty sure this will serve to awaken a renewed interest in this pretty solid flashlight.

Prologue notes.-

Here you will find detailed tailcap current measurements, which again proved hasslesome as in my previous C8T review though this time I ended up low temp soldering a wire over the cell cathode connected to my custom 10mΩ current shunt to easen my life and get more accurate figures. I also had to get the figures in a less steadily fashion because after each few on the trot measurements it became obvious the increased cell temperature was causing a performance increase which was twisting the validity of the results. After all this logging mess I ended up noticing the driver spring bypass had broken but @#$%, I wasn't going to repeat this stuff so deal with it please. The discharge graph is plotted with the raw measured values, thus you can observe the influence of the wire atop cathode contact resistance variation.

Begin!

AliExpress' Sofirn C8F product page.

Product features and specifications: sorry, can't be arsed, doesn't helps not being able to select text in the product page }( or God knows where's the link to the embedded javascript spawn page thing with it. Check the above link, thanks.


In brief, the Sofirn C8F is a redesigned C8 style flashlight with a shorter triple reflector (mm height) embracing 3x XP-L 5700K emitters set in parallel over a ∅30mm DTP/thermoelectric separation copper board, with a nice chunk of aluminium beneath for improved thermal transfer and cooling.
Flashlight weighed 198.8 grams in my kitchen scale, 244.8 grams with a 30Q cell inside it.

Underwater ability test.-

The underwater ability ;) test for this C8F ended up in shipwreck, but of course I have to admit I kept the flashlight ≈2 days submerged in a wash :-D tub. After removal from the wash tub the flashlight wasn't turning on and I noticed a drop of water or two over the driver retaining ring, but everything worked out after a drying break without disassembly. I later noticed in the product specifications :facepalm: “it can be used underwater 2 meters for 30 minutes”. I believe you should specify “for up to 30 minutes” my Sofirn fellow dears.

Tailcap current graph.-

The following current graph was done with all group 3 levels but moonlight, using a Samsung 30Q cell: low1, low2, medium, high1, high2 and bruto (funny related spanish word :-D with the same letters as turbo). Ended up taking a nice deal of values. At some point during the current sample gathering the driver spring bypass got busted, can you guess where? ;-)

zakwilson's Ceilingbounce runtime graph.-

The below graph was performed with an unknown “high discharge” chinese OEM (?) cell (rewrapped to look like a VTC5A) with the torch started up to high2 and then turbo. I held the flashlight in my hand up to minute two and then came back to feel how hot it was with my hand from time to time up to minute 6 or so (see :-D the initial graph tremors):

Disassembly and a couple driver pics.-



Beam pictures.-

EDIT.- It must be noted that beamshots were not taken with the flashlight battery fully charged, maybe at ≈3.9V or a tad below. I do not really like to keep my cells fully charged and this also serves to reflect the flashlight output in more realistic conditions, imho. I doubt this really affects much how the beamshots look anyway.

Beamshots for visual inspection of the different output levels were taken in group 2: low1, low2, medium, high2 & turbo.

Most pictures were taken in HDR mode with my old Moto G 2013. You will also find a last set of beamshots (non-HDR shots taken with a Lenovo Zuk Z2 Pro) with the flashlight modified with Samsung LH351D CRI90+ 4000K emitters for comparison purposes. The runtime graph was also obtained with the Samsung emitters and a slightly lower capacity (2700+mAh) and discharge cell (versus the Samsung 30Q for the tailcap current graph), mod and test which also included AWG18 board to driver wires (tight fit in the holes duh!).

First a couple comparative beam shots versus the C8T, both non-HDR and HDR (Moto G 2013):


Above: C8T non-HDR shot



Above: C8F non-HDR shot

Above: C8T HDR shot

Above: C8F HDR shot

Main shot set.-

Below are sets of 6 shots each for control plus all of the group 2 different modes, from control to brightest: control, low1, low2, medium, high2 and turbo.

Train station, Moto G 2013 HDR:





Field plot, Moto G 2013 HDR:





Outskirts well-off site, Moto G 2013 HDR set:






Outskirts well-off site, CRI90+ 4000K LH351D modified C8F, Moto G 2013 HDR set:

Outskirts well-off site, CRI90+ 4000K LH351D modified C8F, Zuk Z2 Pro non HDR set;

this set was taken mainly to see how the beam looked with that tint:







Well, this is all. Hope you like this, gonna sleep now (close to 5:40 AM now). :UGHH:

Cheers ^:)

P.S.: see beam pictures' EDIT.

2018/08/25 03:45: edited for a fix or two. :-)

2018/09/04 22:34: edited to add LH351D modified outskirts well-off site photo set taken with old Moto G 2013 in HDR. LoL!

Thanks for the review. I measured 2,836 turn-on lumens and 2,706 lumens at 30s with VTC6 using TA Lumen Tube calibrated with Maukka calibration lights, which is pretty good considering Sofirn rates it at 2,580 lumens. It’s a very practical light unlike the Emisar D4 that dims rapidly the moment you turn it on. However, I don’t like the hold button to turn off. I wish it was click for on/off.

Thanks for the review!

I am interested in buying some of these drivers but the ones without the side switch and when i looked up the Mosfet the data sheet said it has a max output of 6amps continuous.

Would you know any specs on this Mosfet i tried to read the number of your pic 64186a4n1r. I think i found it 30/20amp.

Last time i got the BLF drivers they where mediocre i think ill start building my own soon. I need some one to lead my way on what components to buy.

Well, nowaday's Sofirn seems to be displaying speaking truth ethics, which is the loving way to go imho.

The C8F looks sort of like a C8 from the outside. It is heavier and beefier, holding less reflector height/space and more driver height room namely for the switch.

No problem with driver interfaces for me SKV89, once you get used to them all is fine. :-D

Cheers my dears :-)

AON6418 datasheet @ kynix.com

Not bad with the AON6418 but, wanna see a driver I love? Look at this:

3000mA 3-Mode LED Flashlight Driver Circuit Board AON7520 regulated MOSFET driver @ FastTech

AON7520 datasheet Insanely good MOSFET for flashlight use. Its extremely low RDS(ON) allows it to use an in series sense resistor with super-low sense voltage (≈30mV), thus MOSFET plus sense resistor voltage drop stays below 50mV typical even at 10A!

Of course it uses PWM, measured its frequency with VISO Flicker Tester at 10.13ms period, 98.7+KHz. :THUMBS-UP: badly. :|
HKJ told me this driver uses the AON7520 in linear mode by adjusting VGS via sense resistor feedback. Completely PWM free driver. Excellent!

Cheers ^:)

Originally posted on Thu, 08/23/2018 - 11:39. Fixes. Edited once more for a further clarification with regards to driver operation.

Hello! :-)

Do any of you precisely know the PWM frequency of the C8F?

Sheesh with the VISO Flicker Tester, seriously. Any tips and tricks in this regard? :|

Cheers ^:)

Good review Barkuti, thank you. The C8F is one of my favorite lights. :beer: . :+1:

The driver Sofirn uses in the C8A & T, and also sells separately, uses a dual channel mosfet, with the channels used in parallel. The rating for a single channel is 6.5A, so, with the paralleled, it can handle 13A continuous. I bought a couple a while ago, then decided to start building my own drivers, so I haven’t had much reason to use them. The UI is nice and simple, and I’d be happy with it for a lot of lights, but moonlight mode is only accessible in a mode-group that includes non-hidden strobe.

Thanks for the review Barkuti. Lots of good info contained herein. The CF8 is one of my favorite lights. I think I have about 5 or more of them.

Added a few more pics (outskirts well-off site, LH351D modified flashlight) taken with the original Moto G 2013, for comparison purposes.

Cheers :-)

Was there no thermal paste on the MCPCB or did you clean it?


Thermal paste there was, of course.

I inadvertently decapitated one or two stock XP-L emitters upon disassembly, the ones in the picture are Samsung LH351D 4000K CRI90+ from AEDe's second batch (T6 chromaticity bin emitters, 4000 - 4125K; pic looks warm because of unadjusted daylight white balance).

Cheers

Did you measure, what top current do you get with the LH351D modified C8F with the same battery (Samsung 30Q)?

I would like to modify my C8Fs, but I would like to keep them just below 10A.

Thanks for the info lads!

The samsung leds look good. Did you get any readings on output between the two?

I cant speak for him but the ones Dale and I have built with LH351D’s will pull over 10A on FET drivers with a VTC6. I suppose you could make the driver wires longer if you needed to.

Answering you both…

Sorry, no integrating sphere/tube here. Also, :| no idea where did I placed my homemade shunt resistor. Sheesh!

Pretty sure the thing is pulling more current now, this is because I replaced the stingy stock AWG24 driver to emitter board wires with AWG18 ones. Overall output is more or less the same after modding, I believe. Higher quality output, and higher power consumption. Tint is more pleasing, this was to be expected.

Cheers ^:)