AA Flashlights - "Cockroach" / "Vampire" / "Drain-me-till-I'm-empty"

People have had success with LED swaps in the infinity ultra? I’ll have to see if there are any pics or write ups. With all the warm/neutral and higher cri 5mm options out there that would be pretty sweet!

Which I’m NOT going to do. Can’t stand the thought I make this one go poof.
It was a gift from my youngest brother, and even Canadian dollars don’t grow on trees.

I use that that Archer for my kids’ night light. I figure it has at least 7,000 hours run time so far. I actually gave it to my son for his birthday 2 years ago and it’s held up to all the drops and bangs. Impressive.

I also have the TH20 and didn’t realize it will over-discharge batteries. I also love that light. Thrunite hits a certain price/quality sweet spot with the right implementation.

Wow, that’s a lot of hours. Only 43,000 left until the LED dies.

The 50,000 hour rating isn’t a failure rating. I think it’s rated down to 70% brightness. Someone correct me on the percentage if I’m wrong.

There’s a write up on swapping leds over at the other forum. Just look up ” Gerber Infinity mod step-by-step”. Its been several years since I modded mine, The head was really tough to get unscrewed but I didn’t add any heat to try to help loosen the loctite. If you get the head off the mod is really easy. Polishing the reflector hole helps too, while you have it apart.
These would make a pretty sweat swap in a infinity [1900k edition is in!] WTS: 1900k-5800k 5mm LED 95+ CRI

I EDC a Thorfire TG06S as a pocket utility light that I use otherwise-spent alkaleak cells in. Unlike my "gentlemens' dress pocket light" Acebeam M10, which always starts on L, the TG06S starts on H, which is what I want for general utility convenience.

Ive noticed the TG06S will commence starting on M as the cell weakens, eventually starting on L when the cell is mostly shot. If I use the low cell long enough the TG06S will eventually fail to start at all even on low. The light gives plenty of warning that it's about out of juice. I've not tested a cell that's dropped that low, so I can't vouch for how depleted it is. Next time I wring one out that low I'll post the resulting voltage of the spent cell.

TG06S might be a reasonably good vampire.


got a few free baby c8 $1.99 aa lights.
these drivers make useful light down to about .5v
and will restart at .7v
add a n219c on a cheap star and a dime.
and a energizer lithium.
stash in emergency kit.
not bad for something that can be considered by most to be disposable.
battery dead and tossed in the recycling ewaste box.
i get this places recycling stuff and have found all sorts of nice stuff.
lots of laptops.tons of battery packs.power tools.
a few weeks back i got a new looking magcharger.
it worked fine once i replaced the bad wallwart.

Link to these “baby c8” hosts?

noname tactical

Not exactly a cheap option, but the Peak LED Solutions El Capitan would be an excellent cockroach/vampire light if you had it configured as a low output, no QTC light.

It doesn’t get much tougher than Peak’s lights and as far as being a battery scavenger, my Eiger AAA configured at level 1 beat even the mighty Fenix E01 on operating on seemingly dead cells, I’d imagine the AA El Capitan at a similar drive level would be no different.

The TG06S finally ate the last few useful electrons from it's ancient (2/2005) Kodak Photolife alkaleak. It's been starting on L for about six weeks or so. The past week L has dimmed & begun flickering after about two minutes run time, eventually becoming nearly useless unless in a completely dark space. Last night it shut down completely. The cell tested 0.807v on a DMM immediately upon removal.


Usually a good 2×AA (or 2×AAA) light will wring more juice out of the weaker cell first, unless perfectly “balanced”. Luxpro, Sofirn, got too few that actually run down too far; the light seems to go mental first, then completely boembs out at about 2V total.

you do have to watch, as lightbringer suggested, that as batteries deplete to zero, you make sure they are not leaking

because doing that, running them to zero, is the exact thing that causes the cell case to corrode through, and start leaking crud out

do that - and a flashlight can be completely killed in a week, for half a dozen different problems.

just saying - vampire usage sounds fun but it has risks

clocks are vulnerable to this too, if you have say 20 clocks on a wall with only some of them running
the problem is that if one quits, you may not even notice til a year later then the whole thing is corroded

A member here approached me about making a low output, long runtime boost driver to fit the Convoy T2 (aka Manta Ray S1, also available as a host at Kaidomain). Target output was ~1 lumen. No mode switching. Currently 17mm but could be made to be 15mm or about any size.

I went through a couple iterations and ended up creating one that ran an XM-L2 at 2.5mA for ~0.7 lumen for 98 hours on a 900 mAh AAA Eneloop Pro. So on a AA battery, it should run for 12+ days. It runs down to ~0.7 V.

Current is set with a single resistor. Current output could easily be set higher or lower as desired.

Oshpark link. BOM:

  • TLV61220 Boost IC
  • Inductor, 1210 footprint (I used a 10uH but 4.7uH should work as well; look for lower DC resistances)
  • 2x 0603 or 0805 capacitors with 10 uF (pretty much any 10uF cap should be fine)
  • 1x 0603 or 0805 current set resistor (resistance = 0.5V / desired amps, so for 3mA: 0.5 / 0.003 = 167 Ohms)
  • Spring and wires

That sounds really interesting. Depending on cost I would be interested in a couple. Maybe at your tested output and maybe something a little higher like 3-5lm. Is this a constant current driver or PWM? What is the input voltage range (ie: is it adaptable to 2AA)?

  • Constant current (though technically due to the inductor there’s a high frequency ripple… but that’s how boost drivers work).
  • 3-5 lm isn’t a problem, though I need to see which of these drivers would be most efficient at that output. While the ZXLD381 didn’t do great for super-low current, 3-5 lm might be in it’s sweet spot
  • Re: 2AA… in short: shouldn’t be a problem. Long version: the TLV61220 is rated for 0.7 - 5.5V. It has a 0.5V feedback, so if a LED is dropping 2.5V at the preset current target, the boost driver will put out 3.0V (2.5V for the LED and 0.5V for the feedback). If the input voltage exceeds what is needed to hit the current target, I think the TLV61220 will probably go into pass-through. In which case, it might be a touch brighter than anticipated until the battery voltage falls a little bit. With NiMH (2AA ~= 2.4V) this should be a non-issue. With 2 fresh alkalines and a very low current, it might start out just a touch higher than it needs. I can test this out more.
  • Cost: partially depends on how many I make. For the TLV61220 circuit (the better option for really low output) PCB + components are around $2.50 in materials (plus Mouser $3.49 shipping). I would charge actual shipping amount and next to nothing for assembly. For the ZXLD381 (good for currents in the 6.5 - 20mA range), I already have PCBs and components on hand… cost would be like $2 plus actual shipping.

Is 0.7lm the lowest stable output achievable?

Nope! I measured 0.7lm at 2.5mA to the LED. I could drop that down as low as 0.1mA (anything below that and efficiency would likely drop considerably). So even if you wanted 0.1lm, that should be well within reason.

This is great news :wink:
I may be interested 2x17mm and at least 2x15mm, in case you can ship them internationally.
Would it be “good to use” with other leds too? (ex: Luxeon V2)

Thanks for your work :wink: