I do runtime tests using 15 or 30 minute alarms on my phone. At each alarm i take lumen teadings plus voltmeter readings
Thanks, Jon, I might try that.
Jerry
I like lights that have battery level indicators, which i crosscheck w a voltmeter.
Zebra does it, my Utorch S1 mini does it, although its not well calibrated, it helps. At one flash i know its at 3.7v. Without having to open the light.
my 0light S1 Mini has a low battery indicator light that turns on Red. That’s really idiot proof for me. It does that at 3.4v. I carry a spare battery, in a spare light 
lol
I hope you find stategies that fit your needs
Do you think your 18650 figures would be similar to 26650 batteries?
Imo yes, those are reasonable numbers for any 3.7v liion
I also find that my utorch drops out of medium @3.5v, so if that happens, i swap in fresh.
I’ve noticed with protected cells when they are empty show higher voltages then the same unprotected cell. I always chalked it up to the higher resistance from the protection circuit. Like a normal cell drained to 3 volts under load shows about 3.2 to 3.3 when placed on the charger. And some protected cells and worn out cells with higher internal resistance will go up to 3.5+ once the load is removed.
For a cell like the PLB 26650 5000mAh (LiitoKala INR26650-50A), at 3.6V no-load less than 2Ah of capacity are left (to infer this just look at the 0.2A discharge curve and maybe add 25 - 30mV because of the very small additional (besides I × R) loaded cell voltage drop. Other cells (different chemistries and etc) may vary.
With regards to deep cycling being detrimental for li-ion… I've heard it many times, but found absolutely no proof of it, whereas high voltage (high state of charge) is known to be detrimental and there's plenty of experiential and experimental proof. Thus, as far as I am concerned I'm gonna give that a Big Bullsh1t Award. 
Cheers ^:)
Remaining battery capacity depends on the voltage, the discharge curve of the cell voltage, the current draw at the moment of reading the voltage, and the average discharge rate.
Without these things all you will have is a rough guess.
Lol
You misunderstand, my point. I will try to clarify:
It is not bad to recharge liion at 50%.
It is not necessay to drain completely before recharging
Liion does not build recharge level memory
I agree liion can be stored half empty
If you want to disagree with something I said you should criticize the 3.6 V comment which is not accurate.
Or better yet just post your own opinion instead of taking the time to criticize mine
These voltage to remaining capacity charts you see floating around, were mostly published and used when there were very few 18650 battery types and makes around. They were always considered a close estimate even then. Now you have several different types of 18650’s coming from each manufacture of the many, including a lot of china cells. The discharge curves vary a lot, so there measured resting voltage after partial use vary’s also, making the old voltage remaining capacity chart even further off. The chart is just a estimate and isn’t as reliable as it use to be because of so many differences in cells now a days. To make a somewhat accurate chart, you would need to discharge each type of cell and make a chart for that cell. Even then there could be slight differences. HKJ’s graphs would get you a close estimate under load, but you would have to use the same load as he tested with to get a close estimate. That reminds me, HKJ always tries to test two battery’s in his test because of the differences that sometimes causes different results when testing. These voltage to remaining capacity charts are just a estimate.
So far, I've recharged the batteries when they reach around 3.6v to 3.7v but always wondered what would happen if I just let them regularly go down to 3.0V. Never did that yet except when doing a capacity test on the charger.
Some members did a voltage reading after letting cells sit for two months.
I did a voltage measurement with a small selection of cells at 50% capacity.
Yes, that’s it. Thanks.
Jerry
Don't take it too seriously jon_slider. I wasn't specifically criticizing you but that stray belief, I just added a cheap joke over the thing. Pretty sure it may be demonstrated that deep cycling li-ion could also be harmful but, how deep is deep? At the usual loaded cut-offs we often use, 3 to 2.75V, harm should be pretty small if anything. Shortly after removing a cell from one of my fully regulated torches (3V flashes) the cell rests at 3.3+V.
Cheers :-)
At the current time all my 3 battery test stations are busy, but I hope to test some batteries for remaining capacity later this year.
I have not decided exactly what batteries to test, but I will be looking for some of the never types.
Little question for HKJ and other technically :-) inclined fellows. As far as I understand, a typical consumer UPS (an old Riello iDialog Plus 60 whose Pb battery died) could be effectively overhauled with standard li-ion cells, 4S BMS and 4S LiFePO4 balance boards. Since it comes with a 6S Pb battery, the UPS should charge its battery to 14.4 - 14.6V, this means the 4S li-ion setup would get at least around 3.6 to 3.65V per cell. From what I can infer from HKJ's graphs, even with a 30A load a cell like the PLB 5000mAh would still deliver 1.5+Ah down to 2.5V, typical li-ion BMS cut-off voltage. Since we limit ourselves to around ⅓ of peak cell capacity, even for a continuous discharge cell temperature would never exceed any limits. Thus, a 4S2P INR26650-50A battery pack would for sure deliver an average of ≈2.8V at 30A, for 84W/cell and 3+ minutes of battery life at peak load. Even at cut-off, 10V × 60A = 600VA at the input. Target SAI load is a network connected terminal computer for ≈300VA at most, with no possibility of SAI communication and always on.
Makes sense? To me it does, and 8x/12x cells plus 2x 40A BMS and balance boards would cost little money. 
Cheers :-)
It’s probably cheaper just to replace the SLA battery with another one.
I don’t know how much a specific battery for your UPS would be, but on some UPS devices, they can connect to an external battery for extra run-time. A 1000 Wh deep-cycle battery is about $100. You’d need to use the full capacity of a hundred 18650 cells to equal that, which would cost several times as much.
Bang-for-the-buck, lead-acid is still the way to go, as long as you don’t have to lug it around.
Problem is deep cycling. The battery is going to be deep cycled, period. No way around this because:
- UPS communication is not possible.
- Terminal is always on: after-work hours, weekends, etc.
The tiny stock lead acid batteries get damaged with just the first deep cycle, and 2 or 3 after this they're cadaver.
Cheers ^:)
The UPS should have a cut-off at some voltage, preventing the battery from being drained flat. Yes, it will be hard on a SLA battery to drain it low, but unless you’re doing that regularly, the deep-cycle batteries are built to handle that better than regular car batteries (or perhaps the crappy stock battery?).
Even if you ruin a deep-cycle SLA battery year or two, it’s still probably cheaper than a lithium-ion battery of the same capacity.