Like i mentioned before, almost all modern cells according by datasheet can go down to 2.5V. Why not to use it?
It’s never ever a good idea to rely on a devices LVP. If you start discharging below 2.5v then it’s not safe. Anything below 2v and I bin them, not worth the risk when we have no idea of the internal condition.
Mooch said anything below the manufacturer’s specs and you can never tell whether they’re safe for use.
You’ll lose cycles by dropping the voltage to 2.5v continuously. HKJ said he never runs discharge tests down to 2.5 volts anymore due to damaging some cells, even though they were rated to 2.5v
Charge voltage experiments with lithium ion batteries showing how capacity varies with charge voltage and higher cycle live with lower charge voltage This experiment shows, At 3.5V you have approximately 3% of the battery’s capacity left. There’s no good reason to go below that. And by only charging to 4.1v you get 2x the life cycle. 4.0v and you get 4x the life cycles.
That’s definitely better than blinking on. It seems it blinks off on higher modes based on my runtime graph below, but blinks brighter on lower modes according to OP’s findings. That’s an interesting design decision.
They key word is “ostensible”. My sample does end up shutting off. According to Richard, there is some variance in the resolution of the 7135 chips used and some of them cannot provide power down to the lowest modes of the driver. It’s worsened with low cell voltage too. My one sample will not power on in any mode below level 4/22 with a cell below 3.4V or so, if I remember correctly. Odd behavior. I happen to like it though because I prefer my lights to have LVP.
It’s worth noting that Anduril seems to behave similarly, which I like a lot. The difference is it actually does have an advertised low voltage cutoff at 2.7V. Such a great firmware.
Battery charge percent HKJ also created some charts and measurements to show remaining capacity at various voltages. The magnetic charging skilhunt should make it very easy to measure voltage drop while running. But if people are using potatoes or turnips to power their clock than their measurements cannot be trusted. Olights make it very easy to measure voltage drop while running also.
https://www.youtube.com/watch?v=yXmnmvDl-ao 
His trying to make a huge case out of something that is inherently no big deal reeks of trolling. If I was Skilhunt I would consider this defamation of their product.
Great thread and I now have a new favorite post!xevious 
Good info Oli, thanks!
I have an 18350 H03 with Quadrupel Dark Horse driver and it blinks, but it's a nice blink.
Thanks to this thread I'll probably be picking one of these up, as much as I love my H03 18350 I'm really enjoying not messing with taking batteries out of lights any more. I have this giant tank desk that they'd get under if I mishandled when removing from charger and spring propelled them away.
Nice to have an easy charging option AND be able to swap in a replacement cell.
!https://imgur.com/a/US8Shx0!
!https://imgur.com/a/CWOoylk!
Sorry for not inserting photos correctly, just doesn't work for me any more and I'm too lazy to figure out what I've done wrong
You need to use Imgur's Direct Link (like you used to.)
share link, html embed, BBcode, markdown, copy link, grab link, share link on every last S&M platform, nothing works. Hopefully the links that I posted take you to a couple beautiful photos of my H03 mini and not a Rick whatshisname video “never gonna give you up”.
Yeah, given a choice between a flash and a pulse at low voltage I’d go pulse, but between us: at that voltage the end is near enough that you won’t have to suffer long unless you’ve drained it and are trying to scrape by with moonlight mode.
Excited about the programming option. Haven’t dug too deep but I’d like to set it up so no more than 300 lumens on highest. Got a little reading but almost ready to pull the trigger unless OP makes me a sweet deal and I buy used.
You don't see "Direct Link"?
I had the same problem with my H03R, I don’t buy Skilhunt lights anymore. Not useful when they start blinking away with 30% of the capacity left.
ETA: The blinks happen so fast many meters won’t pick them up or display correctly when charting the data. More like a quick strobe than a blink. I had to watch mine to capture the time it started. This isn’t the same as other drivers showing flashing due to low power - it’s a programmed strobe with a low-voltage activation.
My H03R on H1 has a nominal runtime of 200 minutes to 2.9v/80% brightness with a 3000mah 30Q. The low-battery blinks start at ~140 minutes. So that’s a full hour of unusable runtime.
It may be useful for some users, but it’s not adequately explained in the technical literature on the Skilhunt webpage, at least it wasn’t last I looked.
Once you upload an image to imgur, right click it and select “view image” or “open image” (might change depending on the browser).
For your first image you should get “https://i.imgur.com/Vmx7Lu8.jpeg”. If you put that between exclamations, the image should display correctly:
It’s also worthwhile to recognize that most of the people in this thread saying it’s neat or useful feature, didn’t buy the lights, the handful who have, most don’t like the feature. So jumping on OP to defend something one doesn’t own or use is silly. I’m still using my H03R until I can afford something better but as noted above, it won’t be Skilhunt.
Seems worse than I thought. The H04 got removed from my list of next buys after this thread.
I have removed the tube off my H04 and connected the battery via wires soldered onto the driver. I’m going to get some confirmation on these claims instead of looking at runtime graphs, I have my DMM to monitor the voltage on load.
I’m running it on H1 and will see when it starts flashing and whether its stepped down or not as the graph shows above
I agree, it is annoying and unnecessary, but their drivers are crap anyway, so many failed, that i do not look at h0xx, as a flashlight, it is a host imo, a good host, easy to work with. but nothing more, It needs more work to be a good light. I can not trust their electronics, and i’m not even talking about horrible UI, but thanks to custom builders, we have great drivers for those otherwise good hosts.
…but, you also don’t have an H04 RC.
I ran a test just now with leads and checking the voltage on load.
The above graph isn’t accurate for the H04 series, Skilhunt must have tweaked it if what you show above to be accurate for the H03.
Cell used Epoch 3000mAh which is close performing to the 30Q
Test ran on H1 level
Dimming started at 3.07v
3 quick flashes to off every 30 seconds starts at 3.00v
light cut out at 2.694v
time from 3.00v to 2.694v was 31 min 17 sec
After the cell was left sitting for 5 minutes the voltage climbed to 2.79v
1 Thank
As a dealer I haven’t had one return yet in over 100pcs H04 RC sold, I’d call that pretty decent if you ask me.