Sofirn SF14 - Misleading Runtime

I bought a Sofirn SF14 flashlight and went to measure its NiMH runtime at High. It was a bit strange that nobody did it before (usually I check my results with others), now I understand why…

According to the specs I was expecting 230lumens for 1h 30min at High. Instead I got this

WTF? I re-read the manual and noticed the “Due to the protective setup, … ” line

OK. Then let’s measure the first 3 min period. Here it is

The beam starts above 200 lumen, then within the first 60 seconds drops to 82 lumens, then at 200 sec to 28 lumens (Medium) and stays there until you bring it back to High again.

I managed to bring it back to High 18 times before the battery (standard eneloop) ran out.

According to my measurements Sofirn SF14 can peak 230lum, but won’t stay there for more than a second. It can switch to High 18 times for an “accumulated” 65 minutes averaging 100lumens.

Suddenly Sofirn SF14 does not look as good as the specs would suggest…

AA Cycler

No suprises, cheap flashlights(low quality drivers) usually behave like this…
If you want good (flat regulated) runtimes, you have to pay a bit more…

Thanks for your tests AA Cycler!!
I have and like a lot the SF14 (recently modded it), but never run a test on it!
But this is very useful and I guess that you should let Sofirn know about it for a correct advertisement and driver improvement, if that is the case!
Very useful information :+1:

Well dangit I just ordered one of these for my mom. Then again she will be happy as long as it’s small and turns on for a few minutes when the button is pushed.

Very nice tests and graphs AA Cycler!
Super well done and useful. :+1: :beer: :partying_face:

Ops wish I had seen this before purchasing it!

A gift it is then.

FYI, some of the advertized SP10B runtimes are heavily exaggerated as well:

Your expectations are understandable, but that is not how Lumen specs work

not correct
all flashlights drop from maximum very quickly, within seconds. That is even true for the D4.

the key weasel clause in the Ansi Lumens runtime measurement is, “runtime until the lumens drop to 10% of initial output”

there you have it
expecting 230lumens for 1h 30min is not how lumen specs work
that spec means that there was a moment when the light was making 230 lumens, but the runtime ended when the light dropped to 23 lumens

there is NO flashlight, anywhere, that can maintain its maximum initial lumens for the entire stated Ansi runtime

this is a common consumer misunderstanding, caused by the way the Ansi lumens runtime spec is calculated. You really need to see a runtime curve to know what is actually happening.

Ran into this extreme down-regulating when I did a runtime graph on my Coast Polysteel 600, advertised at 710 lumens on high. I knew it would throttle down, but didn't expect that much.

good example
I dont buy lights for their High Mode
I focus on their medium mode because it CAN be maintained in flat regulation for the stated runtime

An AA battery simply does not have enough power to produce a steady 230 lumens, but it can certainly do several hours at 30 lumens

If you need multi hour runtime at more than 100 lumens, you need more battery capacity than AA

An 18650 light, rated for 1000 lumen maximum, cannot maintain 1000 lumens either, but it can maintain 500 lumens for an hour

When buying a flashlight, looking at runtime charts can be very helpful in determining the actual lumen levels a light can maintain for a desired number of hours.

Flat regulated output, is not the same as Ansi runtime specs at all, its good to learn the difference, and check runtime charts before buying a light based on expectations of maximum brightness runtime

newbies make this mistake all the time

they post things like
“my flashlight is defective, it wont hold turbo for more than a second but the box says it should last 72 minutes”

then someone has to explain the 10% Ansi lumen levels rule

certainly the advertising is misleading
runtime charts are the solution to this misinformation and misunderstanding

A low-ish medium mode is what I’m seeking out for a good travel headlamp. I hate lights that have “medium” modes around 80-100 lumens which is way too high and kills runtime. Give me a nice 20-30 lumens or so that I can easily tap into and know that I can get 8-9 hours out of.

here are some runtime options with 3 different battery types
aaa 4 hours 20 lumens, Lumintop Tool w Nichia (pocket clip works on a hat)
16340 Olight S1 Mini High Cri 50 lumens 6 hours (pocket clip works on a hat)
AA 21 hours 30 lumens, Zebra H53w

You have several SOFIRN torches, and I have a few of its torches to satisfy my usual needs. I'm really surprised to see your test data, thank you for sharing.

Thanks for this. The zebralight really does seem great, but the price is an ouchie. Also not a fan of how many damn modes they have! BTW any reason you recommended the H53w instead of the H53c?

The Zebralight SC52 and SC5 have been doing all that for years. It’s certainly possible, but you need to have a decent driver.

I dont own any zebras, but I hear many people cant tolerate the green tint of the C models, and that they prefer the Tint of the low cri of the W

I was just giving an example of a light with the ~30 lum mode and 8 hour runtime you wanted
afaik, the Zebra remembers 3 main modes, the others are selectable to suit your personal taste for extra modes, but you dont have to go through them all to make the light useful.

sounds good, maybe you can post a runtime chart showing flat regulated output @ ~200 lumens, so we can see the multi hour runtime

Interesting co-inky the subject of runtime graphs for single AA Zebralights has come up. Ran several runtime graphs on my SC53Fc last month. Was going to post these on CPF before things there went south. FYI, this floody version was a Zebralight custom build when I ordered it, but now it's on their regular list.


Three different batteries were used:

(1) Energizer Lithium

(2) Panasonic Eneloop Pro

(3) Energizer Max alkaline

Make your own conclusions on how well regulated this model of Zebralight is.

This first graph is with an Energizer Ultimate Lithium. You get about an hour before brightness drops to 70% (or about 200 lumens). I can't explain the wonky rise in brightness during the first 35min. Anyone have thoughts on why there's a rise in brightness that didn't occur on the Eneloop or alkaline battery?


The wonky rise in brightness above made me repeat the test again, to see if it was duplicable with another lithium. It was.


Panasonic Eneloop Pro. Here again, we get close to 200+ lumens for about an hour, assuming the Zebralight spec of 285 lumens is valid.


Finally, an Energizer Max alkaline. Even though ANSI FL-1 is only 30min, the bunny kept running for 5 days, a testament to how well the Zebralight can run at moonlight levels down to about .8V. Also, I suspect the alkaline's ability to recover at very low current draws, contributed to the 5-day runtime.

Yes, you definitely need to use an Eneloop to make the Zebralight give a regulated output. I think those SC53Fd graphs match what the SC52’s deliver, though I think the cool-white and neutral white “w” versions are more efficient than the Fd.

From my testing, the SC5w is better regulated than the SC52/SC52w, and also about 10-15% more efficient. I’m not sure if efficiency was improved with the latest MkII models of the SC5.

I think if output and runtime are your main concerns for a 1xAA light, then I don’t think you can beat the Zebralight SC5.

Very informative video, thank you!

Now I know ANSI FL1 Standard is pointless, because it can be so easily manipulated.

I will use my way of calculating runtimes - sampling every minute and waiting for the output to drop below 50% of the so-far averaged output.

This way Sofirn SF14 does 131 lumens for 3 minutes at High, and 27 lumens for 8h36min at Medium.

There actually are and it allows me to categorize manufacturers into two groups (based on the NiMH flashlights I tested so far and runtime graphs posted in this thread)

  • honest - Fenix, OLIGHT, ThruNite, JETBeam, ASTROLUX, LUMINTOP
  • cheaters - Manker, Sofirn, Coast Polysteel, MecArmy

I like to further categorize cheating into two categories:

1. Those that simply sag because of a drop in voltage under load. That is, they don’t have regulated drivers.

2. Those that intentionally abuse the FL-1 standard so they can advertise ridiculous output and runtime stats. Companies like maglite do that; they step-down not just because of an unregulated driver, but down to just over 10% initial output so it runs for a really long time.

I don’t really have an issue with #1. Any lithium-ion light that uses a FET driver for max output, does it. #2 annoys me.