Choose the discharge current level closest to that at which you are operating. Then follow the discharge curve plot line down to your cutoff voltage, 3.6V or 3.8V or whatever you choose. Whichever cell(s) deliver the highest capacity at that cutoff voltage will be the cells that give you the longest run time.
Odds are that the cells running for the longest down to my 2.8V cutoff will also be the ones running the longest down to a higher cutoff. You could use these tables to narrow down your choices:
If HKJ has tested any of the cells you’re considering he has a much wider range of discharge current levels you can check for highest delivered capacity at your cutoff voltage.
Ah is just amps * hours.
Convert the delivered capacity in mAh to Ah (Ah = mAh / 1000). Then divide the number of Ah by the current (amps) and you’re left with hours. Then you can convert to minutes. Since the discharge is constant-current you can do this.
You can do this directly using mAh and minutes but you need to handle the units carefully. I leave this as an exercise for the reader.
I was thinking you could get that number just from the math but I deleted my post on that since I think the difference in voltage sag at lower current levels will have a big enough effect to make the math method not accurate enough.
Nevertheless, those QB26800 cells from Aloft (the gold wrapper ones) don’t live up to my expectations. Originally was supposed to be ~10 mΩ IR (1kHz). And they cost me much with the shipping to Canada.
Looking forward to the eventual test with the green and grey wrappers. I would like to contribute but these were above my paygrade ($54 / 2)
But all this analysis gives you an insight into what we flashlight enthusiasts have as current capacity, that’s to say below ~3.6 volts isn’t worth considering for a power-hungry LED. And they have some at above 30 A !!!
Have to add, I bypass the FET driver for such high demand – hardwired direct drive, single-mode ON.
No good boost regulators to help out with that?
Hmm…actually, there would be a decent chunk of heat to deal with using that regulator and I bet you’re already dealing with a lot of that from the LED.
Part of the exploration costs. Others in that build and 100+ sank in. Put aside till I find a better cell or lower Vf emitters.
Just now Simon (Convoy) is developing a power boost driver. I’m in line for a few…
Heat isn’t a problem if there isn’t a thermal cut-out. I live in a cold climate and use work gloves with these lights. I do check on the tube heat – wouldn’t want a runaway battery.
And my builds do consider the wattage. This one is above 1 kg.
I’m thinking Vf is about as low as it will get. At least for remotely affordable ones. The cells will probably be a better option but none will give you want you’re looking for I think. Three minutes at 25A above 3.6V is about 4.5Wh of energy…that’s a LOT.
A super-efficient boost circuit might be the best solution. Anyone working on that?
Oops, you mentioned this while I was composing.
Perhaps I should be building with multi-cell hosts….
Just tonight was looking into a 3x21700 model from Simon. See if he would sell a hollow shell as a host.
As an end note, that P45B would be a good cell for the aforementioned build. Inferring from your discharge graph, could sustain 25 Amps for 2 1/2 minutes. Looking forward to the availability this autumn.
Yea, I can see how run time would simplify things…direct read, no math needed.
If there were a bunch of requests for a particular re-graphing I could do that. Too many different individual requests each day otherwise. My patrons can request re-graphing though.
While this seems expensive in comparison with P42A, it’s still cheaper then buying a P42A, 30/40T in a walk in Vape shop or electronics store, at least here in Europe.
