Sofirn C8 Quick Review

111 posts / 0 new
Last post
korpzgrinda
korpzgrinda's picture
Offline
Last seen: 12 hours 14 min ago
Joined: 01/09/2017 - 16:52
Posts: 319
Location: Eastern Woodlands in the Wilderness of North America

Nice! That looks pretty cool all mixed and matched! Smile

Barkuti
Barkuti's picture
Offline
Last seen: 1 hour 16 min ago
Joined: 02/19/2014 - 14:46
Posts: 3919
Location: Alhama de Murcia, Spain

I have a few 9926A dual n-channel MOSFETs lying around, not yet unsoldered off broken powerbank PCBs. These:

Banggood is a bunch of @#$%, I have been forced to view the source code of the page to be able to access the image links. Makes hot linking their stuff pictures much harder, at least if you're using Chrome browser. JavaScript related crap. 

 

Ranting aside, according to @#$% datasheet the 9926A MOSFET features… (edited all of this useless information). This means the combined dual channel resistance won't go much above 20ish mΩ even with a semi-drained cell on a hot flashlight.

 

Cheers 

Originally posted on Fri, 10/06/2017 - 03:36; useless data fix.

Innocent

Barkuti
Barkuti's picture
Offline
Last seen: 1 hour 16 min ago
Joined: 02/19/2014 - 14:46
Posts: 3919
Location: Alhama de Murcia, Spain

User AnhTran has a little Nichia 219C 5000K R9050 sale going on, just saw his last message. 

 

Cheers Party

Innocent

Jouna
Offline
Last seen: 6 months 2 weeks ago
Joined: 01/25/2015 - 13:12
Posts: 63
Location: Europe
Barkuti wrote:

stack solder another 8-pin SOIC MOSFET on the driver and cream the Emisar D4‘s output records with the help of a Sony VTC5A or the newer Vapcell INR18650 2000mAh. LoL! Big Smile


 


Cheers Party

By curiosity; how much amps/ watts does that stack solder add?

diego2017
Offline
Last seen: 1 year 10 months ago
Joined: 10/06/2017 - 03:04
Posts: 1
Location: UK

Thanks for the review. Tailcap/ switch, very good quality u-shaped tailcap and great switch

play bullet force, router login 192.168.0.1

Jouna
Offline
Last seen: 6 months 2 weeks ago
Joined: 01/25/2015 - 13:12
Posts: 63
Location: Europe
diego2017 wrote:
Thanks for the review. Tailcap/ switch, very good quality u-shaped tailcap and great switch

You’re most welcome Diego Smile

Barkuti
Barkuti's picture
Offline
Last seen: 1 hour 16 min ago
Joined: 02/19/2014 - 14:46
Posts: 3919
Location: Alhama de Murcia, Spain

Jouna, setting up more 9926A MOSFETs in parallel reduces the overall RDS(ON) and increases continuous current handling accordingly for as long as the heat dissipation also improves proportionally, yet although this is not the case it is with regards to the instantaneous figures and for the continuous to some extent. With two 9926As stacked the driver should be handling 12 - 16A and optimizing the tail switch resistance and losses would be the next in line thing, imho.

 

 

Cheers 

Innocent

Jouna
Offline
Last seen: 6 months 2 weeks ago
Joined: 01/25/2015 - 13:12
Posts: 63
Location: Europe
Barkuti wrote:

Jouna, setting up more 9926A MOSFETs in parallel reduces the overall RDS(ON) and increases continuous current handling accordingly for as long as the heat dissipation also improves proportionally, yet although this is not the case it is with regards to the instantaneous figures and for the continuous to some extent. With two 9926As stacked the driver should be handling 12 – 16A and optimizing the tail switch resistance and losses would be the next in line thing, imho.

Whoa Smile I thought some improvement in output but not quite that much Smile Thanks though Barkuti Smile If and when you add that to your C8F do tell about results.

Barkuti
Barkuti's picture
Offline
Last seen: 1 hour 16 min ago
Joined: 02/19/2014 - 14:46
Posts: 3919
Location: Alhama de Murcia, Spain

I'm smelling some sort of misunderstanding here, Jouna. Please bear in mind I said the driver should handle 12 - 16A with two 9926A chips (each chip holds two n-channel MOSFETs, thus stacking two means 4 MOSFETs), I say it just because the datasheet suggests up to 8A per chip (with a standard level of cooling we can't guarantee in such small flashlight driver PCB), not because I am predicting you'll get such a crazy amperage. If you're stacking the chips add some thermal paste over the roof of the ones below (and proper soldering, of  course).

MOSFET stacking example:

AO4410 over Si4324DY units., drain terminals can be observed at the center. That little solder peak rubbing the peg is from a wire below the  MOSFETs.

 

Cheers Party

P.S.: you may want to edit your previous message, I think it's missing a [/quote] thing, Jouna.

Innocent

Barkuti
Barkuti's picture
Offline
Last seen: 1 hour 16 min ago
Joined: 02/19/2014 - 14:46
Posts: 3919
Location: Alhama de Murcia, Spain

XXX-Man seems to be having current draw issues with his C8F. The triple emitter MCPCB is probably not adequate, no direct thermal path. Thermoelectric separation and/or thermoelectric separated (=DTP) copper boards these chinese folks like to say.

 

Cheers Smile

P.S.: leave those XP-G2s onboard for a while, kotobuki. Wink

Innocent

kotobuki
Offline
Last seen: 12 hours 15 min ago
Joined: 10/17/2016 - 09:41
Posts: 197
Location: US

Thanks for your infos, I’m charging my Samsung 30Q. I will test with C8F ver “XP-G2” tomorrow.
Let’s pray Big Smile

Jouna
Offline
Last seen: 6 months 2 weeks ago
Joined: 01/25/2015 - 13:12
Posts: 63
Location: Europe

Better? Yes, I did misunderstood. Thanks for your info Barkuti.

Barkuti
Barkuti's picture
Offline
Last seen: 1 hour 16 min ago
Joined: 02/19/2014 - 14:46
Posts: 3919
Location: Alhama de Murcia, Spain

I am getting the C8T Jouna, mentioned it before. After witnessing this manufacturer C8F screw-up, looks like the single emitter versions have a much better chance to work right. I'll be checking out the emitter board and other details in my unit. Hope the XP-L HI comes mounted over a direct thermal path board, if not I won't be fully testing the flashlight until I can fix that. I'm serious.

Maybe the thing is already calling for an emitter swap. 

 

Cheers Smile

Innocent

Jouna
Offline
Last seen: 6 months 2 weeks ago
Joined: 01/25/2015 - 13:12
Posts: 63
Location: Europe

Ok, sorry Barkuti. I’m thinking also change emitter from XP-L to XP-L HI emitter. I like throw Wink With direct thermal path of course.

Barkuti
Barkuti's picture
Offline
Last seen: 1 hour 16 min ago
Joined: 02/19/2014 - 14:46
Posts: 3919
Location: Alhama de Murcia, Spain

Took a peek over the ThorFire TA13 (aka Sofirn SF30) drivers I took out from two recently modified units. I can see the same 10-pin unknown microcontroller onboard, next to a 3-pin voltage regulator marked as “662K”. kotobuki, can you please comfirm the 3-pin thingie next to the 9926A boasts such 662K markings?

It is more or less obvious to me now what Tracy Wan (from Sofirn) meant when she said high and low power MOSFET: it's the 9926A, plain and simple, with both channels driven for maximum output and just one for a bit less rock n' roll.

Even cooler fact for those of you daring DD lovers is that the TA13 (Sofirn SF30) uses three small AO3400 MOSFETS (marked as A09T or whatever) each with a 200mΩ resistor in series! (edited) Clarification: the whole R200 stack is in parallel (66.67mΩ combined), and likewise are the AO3400s driven. The AO3400 is rated for up to 5.7A each and its resistance at a low VGS doesn't climbs much above 20mΩ; if you dare to bypass all of the R200s you may end up with a surprise. The stock XM-L2 may not survive, Texas Ace blew up one the newer ones at just ≈5.5A.

Take care. 

 

Cheers Smile

Originally posted on Sat, 10/07/2017 - 14:23; clarification added.

Innocent

kotobuki
Offline
Last seen: 12 hours 15 min ago
Joined: 10/17/2016 - 09:41
Posts: 197
Location: US
Quote:
kotobuki, can you please comfirm the 3-pin thingie next to the 9926A boasts such 662K markings?

It has no name on it! I did try to clean any residue but still can’t see any name on it.

By the way, my C8F version “XP-G2” with Samsung 30Q get hot quickly, I scared to keep it on high mode too long to test for step-down automatically.
So, I will use low drain current battery with it, may be Walmart 18650 (Westinghouse). Facepalm

Agro
Agro's picture
Online
Last seen: 2 min 40 sec ago
Joined: 05/14/2017 - 11:16
Posts: 4399
Location: Ślōnsk

Barkuti wrote:

Took a peek over the ThorFire TA13 (aka Sofirn SF30) drivers I took out from two recently modified units. I can see the same 10-pin unknown microcontroller onboard, next to a 3-pin voltage regulator marked as “662K”. kotobuki, can you please comfirm the 3-pin thingie next to the 9926A boasts such 662K markings?


It is more or less obvious to me now what Tracy Wan (from Sofirn) meant when she said high and low power MOSFET: it’s the 9926A, plain and simple, with both channels driven for maximum output and just one for a bit less rock n’ roll.


Even cooler fact for those of you daring DD lovers is that the TA13 (Sofirn SF30) uses three small AO3400 MOSFETS (marked as A09T or whatever) each with a 200mΩ resistor in series! The AO3400 is rated for up to 5.7A each and its resistance at a low VGS doesn’t climbs much above 20mΩ; if you dare to bypass all of the R200s you may end up with a surprise. The stock XM-L2 may not survive, Texas Ace blew up one the newer ones at just ≈5.5A.


Take care. 


 


Cheers Smile


$11 for a SST-40 + FET flashlight? Interesting. Smile
Tom Tom
Offline
Last seen: 4 months 3 weeks ago
Joined: 09/10/2017 - 08:30
Posts: 1163

Stacking FETs (no, they are not “MOSFets” these days) is not, In my opinion, an avenue worth pursuing.

The one being talked about here looks like a very good one. Each of the two FETs inside has under 18 mOhm on resistance, and whilst each is rated at 8A continuous, data is provided to four times that.

So the original dual package should be quite happy to at least 16A continuous. Dissipating no more than 2.3 watts at that current. Dropping 0.144 volts maximum.Which if properly laid out on a footprint with lots of copper plane, is well within spec. (Case temperature de-rating allows about 65C at that power)

Nevermind these numbers, this dual FET package will have, datasheet, passing 16A, under 9 milliohms resistance. I would suggest looking elsewhere to improve the overall resistance in the current path before worrying about the FET.

All you get by stacking on another one is a (theoretical) halving of Rds on. From maybe 9 to 4.5 milli Ohms. Vanishing returns, and very little more thermal dissipation in a stacked configuration.

Then again, I confidently expect it will carry on working at much higher current levels and temperatures, except that it is off the published datasheet.

Barkuti
Barkuti's picture
Offline
Last seen: 1 hour 16 min ago
Joined: 02/19/2014 - 14:46
Posts: 3919
Location: Alhama de Murcia, Spain

Little buggers at the factory trying to hide the component markings? 

It was curious for me to see the C8F advertised with XP-G2 emitters inside, to find these user samples here with XP-G3 units and overcurrent problems. Usually happens when non-knowledgeable people takes the decisions. 

A spring de-bypass or two would help for XP-G3 (and Nichia 219B/C) emitters.

 

Cheers Party

Innocent

Barkuti
Barkuti's picture
Offline
Last seen: 1 hour 16 min ago
Joined: 02/19/2014 - 14:46
Posts: 3919
Location: Alhama de Murcia, Spain

Tom Tom, the Si9926CD datasheet lists a maximum of 8A per package. The thermal resistance is too high and the little FET bugger will end up like a roasted burger. I'd stack another package just because of reliability. 

 

Cheers Party

Innocent

Barkuti
Barkuti's picture
Offline
Last seen: 1 hour 16 min ago
Joined: 02/19/2014 - 14:46
Posts: 3919
Location: Alhama de Murcia, Spain

Agro wrote:

$11 for a SST-40 + FET flashlight? Interesting. :)

$9.06 for a TA13 at Banggood only leaves me $1.94 to get an SST-40 on copper DTP star. 

 

Innocent

Tom Tom
Offline
Last seen: 4 months 3 weeks ago
Joined: 09/10/2017 - 08:30
Posts: 1163
Barkuti wrote:

Tom Tom, the Si9926CD datasheet lists a maximum of 8A per package. The thermal resistance is too high and the little FET bugger will end up like a roasted burger. I’d stack another package just because of reliability. 


 


Cheers Party

Stacking another one on top will make almost no difference to thermal dissipation. You would have to lay them out on separate bits of the PCB, for the heat to be conducted through the leads into separate areas the PCB where it gets dissipated.

We are agreed, this little package would be toasted without any decent thermal path into the board. Stacking another one on top won’t improve anything there.

I think the two FETs inside the one package are quite capable of handling the current, and the power. And the PCB, presumably so.

The rest is over to you, if you think e.g. reducing from 9 to 4.5 milliohms is of any consequence..

Agro
Agro's picture
Online
Last seen: 2 min 40 sec ago
Joined: 05/14/2017 - 11:16
Posts: 4399
Location: Ślōnsk

Barkuti wrote:

Agro wrote:

$11 for a SST-40 + FET flashlight? Interesting. Smile

$9.06 for a TA13 at Banggood only leaves me $1.94 to get an SST-40 on copper DTP star. 


 



$8.99 on e-bay.
$2.09 for SST-40 when buying 1, $1.6 when buying 10.
These are on stars, though I don’t know if they are DTP. Most likely not copper…but if it’s DTP I would stick with it for such cheap light. If it’s not DTP then this light won’t be nearly as cheap….
Barkuti
Barkuti's picture
Offline
Last seen: 1 hour 16 min ago
Joined: 02/19/2014 - 14:46
Posts: 3919
Location: Alhama de Murcia, Spain

Agro, these are DTP mounted emitters: “thermoelectric separated” search at FastTech

The boards these leds are mounted over:

  1. https://www.fasttech.com/products/1611/10017569/3741400-20mm-copper-base-plate-for-cree-led-emitters
  2. https://www.fasttech.com/products/1611/10017569/6222100-20mm-copper-base-plate-for-cree-xm-l2-xhp50-led

I believe both boards are DTP, only from different providers. Hope FastTech were more clear with their product descriptions.

Reflow the cheapie on one of these boards, or buy a ready-to-go SST-40 on custom KDLight copper star at Kaidomain

 

Cheers Party

Innocent

Jouna
Offline
Last seen: 6 months 2 weeks ago
Joined: 01/25/2015 - 13:12
Posts: 63
Location: Europe
Barkuti wrote:

;SST-40 on custom KDLight copper star at Kaidomain


 


Cheers Party

I can recommend those with experience. First fried two cheap SST-40 with poor star/MCPCP. Then ordered those and they handle amps at least up to 5.5A in my setup. Edit: Someone told me that he drives those at 8A.

Jouna
Offline
Last seen: 6 months 2 weeks ago
Joined: 01/25/2015 - 13:12
Posts: 63
Location: Europe

“according Cissy from Sofirn driver is direct driven.” That I updated today and I measured tailcap amps of C8A: I got with freshly charged NCR18650GA after some spread 6.10A @ High. Have anybody measured amps with different/ better batteries from any other model (or same) of those C8:s?

Edit: Peak was around 6.3A.

Edit2: Now there is in add that driver is direct drive.

Edit3: Led gasket have been chanced but only for XP 3.5mm footprint leds:

Jackie
Offline
Last seen: 5 hours 46 min ago
Joined: 10/12/2017 - 23:20
Posts: 82
Location: Singapore

C8F,Cissy told me will have a upgrade model, with 3 groups and DTP MCPCB, Love

Jouna
Offline
Last seen: 6 months 2 weeks ago
Joined: 01/25/2015 - 13:12
Posts: 63
Location: Europe

Just talked in other matter with Cissy from Sofirn. She asked me something, I answered but maybe here are also people that could say their opinions. It’s about new C8F. There are two kind of driver groups and Cissy asked which one would be better. I quote Cissys message straight here and do tell what you think.

“Switch: one side switch
1. 3 groups modes

One click enter group1 : Low- mid-high

Double click enter group2: LL-Low-mid-high

Three click enter group3: firefly-LL-Low-mid-high

In any groups when light on, double click enter hidden mode:strobe-SOS-bike -strobe. three click enter turbo mode 1950 Lumens

2. 3 groups modes:

Turn on enter group 1: low-mid- high

When light on, four times quick press switch to change groups, cycle from group1- group2-group3-group4

In any groups when light on, double click enter hidden mode:strobe-SOS-bike -strobe. three click enter turbo mode 1950 Lumens”

And like Jackie said above C8F is coming with DTP MCPCB because of problems with heat.

3-D photo of C8F to come:

Box
Offline
Last seen: 1 month 2 weeks ago
Joined: 08/26/2014 - 06:53
Posts: 635
Location: Australia

One group with high only please! Big Smile

Lightbringer
Lightbringer's picture
Offline
Last seen: 10 hours 12 min ago
Joined: 08/30/2016 - 14:12
Posts: 8856
Location: nyc

I voted for plan B if need be. No one’s going to remember to click 3× just to light it up from off. Best to just set it and forget it.

But doubleclick-to-blinky is just evil. Rather doubleclick-to-turbo or something, instead.

If the µC is flashable, maybe some other eswitch-enabled UI could be stuffed into it.

Unfortunately, it says something if a UI is unwieldy enough to have to replace it with something else.

09 F9 11 02 9D 74 E3 5B D8 41 56 C5 63 56 88 C0

Pages