Here some further pertinent info for fran's thread. My light is the iTP A3 EOS Titanium with Cree XP-G R5 (mode sequence Lo-Med-Hi). These are my own tailcap readings with fresh cells:
| iTP A3 EOS Titanium XP-G R5 | High | Med | Low |
| Alkaline abc | 0.903A | 0.339A | 0.042A |
| Alkaline xyz | |||
| Varta NiMH | 0.88A | 0.329A | 0.041A |
| Eneloop AAA | 0.906A | 0.308A | 0.040A |
| Grey UltraFire 10440, protected | 1.2A | 0.214A | 0.015A |
| Blue UltraFire 10440, unprotected | TBA | TBA | TBA |
I also measured actual runtimes, both on Protected 10440's and Eneloops:
Protected 10440's runtime tests:
As we know, the ITP A3 EOS (Cree XP-R5) head does run on 10440's flawlessly. Runtimes on 10440's depend heavily on the built quality of the particular individual cell. Of my grey Ultrafire 2pack cells, cell1 seems to deliver more capacity than cell2. Instead of measuring tailcap reading (amperage) i am measuring the runtime in real life tests. On high-mode i wrapped the torch tightly with snow (It finally snowed this winter in Germanyikes) which later on converted to ice. In either case the torch was really cold over the whole testrun time and it was safe to operate it continuously on High-mode with Protected 10440's. Let's recall that selfbuilt never tested the iTP A3 with 10440's on High-mode. So the following test results maybe a world-premiere on the inet, brought to you by kreisler the permabanned:
| HIGH | 10440, cell1 | 10440, cell2 |
| testrun1 | 5m45s | 3m27s |
| testrun2 | 5m30s | 3m17s |
| testrun3 | 14m21s | 9m27s |
Testruns 1&2: After the given time the protection of the cell trips and the light shuts off completely. Not even 6.0mins whaaahhtt??!! yepyep. At least with the brandnew cells i got from Dealextreme (Grey Protected Ultrafire 10440's from DX). Maybe the Grey Protected Trustfire 10440's from DX have a considerably longer runtime than that, but i highly doubt it. Please be reminded that this output (at least with a freshly charged 10440 and in the first few seconds, i.e. before it begins to dim) is brighter than the Quark X AA 8-mode 1x 14500 on its High-mode (i.e. brightness level 4outof5; fyi level 5outof5 is called Max-mode, 1/5 is called Moonlight-mode). And the above results teach us that, if you use the Blue Unprotected Ultrafire 10440's from DX, you should not run it longer than 6mins total on High-mode because after 6mins (or 3.5mins, see cell2!) you begin to overdischarge the cell and 
haha S)
Testrun3 was conducted after many other testruns on the same cells (the below testruns on MED-mode plus the testruns(!!) with the Tank E09) but this time i didnt freeze the entire torch with packed snow. Instead i immersed the tip/top part of the A3's Titanium head into cold water bezel down, and the rest of the head and the body were in plain environmental air at room temperature. Interestingly, the ~1cm water immersion acted very well as heat sink and the body, the LED/driver, and the cell were operating at "normal" temperatures (~25-35° Celsius, estimated) instead of ~0°C of the snow-ice-water mixture. This experiment (testrun1, testrun2) demonstrates the effect of diminishing effective cell capacity when the cell adopts freezingly low environmental temperature.
And let's measure the other modes too, tested at room temperature and with intermittent cooling by human hands:
| MED | 10440, cell1 | 10440, cell2 |
| testrun1 | 1h14m25s | 1h16m51s |
| testrun2 | 1h16m34s | 1h13m27s |
| testrun3 | 1h15m03s | 1h16m12s |
Not bad at all, huh. Cell2 doesnt seem that bad after all. On Med-mode at room temperature it performs even better than cell1 (see testrun1 and testrun3). Let's recall that selfbuilt had measured only 1h7min with the iTP A3 Q5.
And finally, for the first time on the inet, someone is measuring and publishing the iTP A3 XP-G R5 runtime data on 10440's on Low-Mode:
| LOW | 10440, cell1 | 10440, cell2 |
| testrun1 | 15h59m55s (2.774V) | 16h31m22s (2.769V) |
| testrun2 | 16h33m38s (2.685V) | 16h54m07s (2.653V) |
| testrun3 | N/A | N/A |
Here again the runtime depends heavily on the quality (capacity) of the charged cell. Taking the highest measured runtime as reference (record mark) then the iTP A3 runs for almost 17.0hrs on Low-mode with Protected 10440's until the protection circuit trips.
Eneloop runtime tests
High-mode:
| High-mode (iTP A3 w/ XP-G R5) |
testrun stopped | remaining offline voltage | recoverable to | capacity consumed |
| Eneloop cell#1 & testrun#1 | 59min | ~1.14V | 1.2379V | 706mAh |
| Eneloop cell#2 & testrun#2 | 60min | ~1.08V | 1.2205V | 743mAh |
| Eneloop cell#3 & testrun#3 | 62min | ~0.94V | 1.2149V | 737mAh |
| Eneloop cell#4 & testrun#4 | 63min | ~0.88V | 1.1982V | 773mAh |
From the above table we can determine the Eneloop runtime in High-mode as 63min (=1.0h). selfbuilt had clocked 59min, HKJ had estimated 1.0h, Manufacturer's specs are "55 Minutes". FYI i made my Eneloops come off the charger at 1.54V. Freshly charged Eneloops can have a voltage between 1.45 ~ 1.54V and a typical voltage would be 1.47V. So my measured runtimes reflect the absolute optimum, and in real life with an "Eneloop charged full with a dubious quality NiMH charger" your runtime will be lower than my record mark.
Med-mode:
| Med-mode (iTP A3 w/ XP-G R5) |
testrun stopped | remaining offline voltage | recoverable to | capacity consumed |
| Eneloop cell#1 & testrun#1 | 2h15min | ~1.17V | 1.2330V | 703mAh |
| Eneloop cell#2 & testrun#2 | 2h19min | ~1.11V | 1.2039V | 753mAh |
| Eneloop cell#3 & testrun#3 | 2h22min | ~1.01V | 1.1860V | 765mAh |
| Eneloop cell#4 & testrun#4 | 2h29min | ~0.83V | 1.1815V | 779mAh |
From the above table we can determine the Eneloop runtime in Med-mode as exactly 2h20min (=2.33h). selfbuilt had clocked 2h9min as did HKJ. Manufacturer's specs were "4 Hours", which is a total lie, shame on Olight!!
Low-mode:
| Low-mode (iTP A3 w/ XP-G R5) |
testrun stopped | remaining offline voltage | recoverable to | capacity consumed |
| Eneloop cell#1 & testrun#1 | 19h17min (dead) |
~0.82V | 1.1395V | 807mAh |
| Eneloop cell#2 & testrun#2 | 18h16min | ~0.85V | 1.1714V | 777mAh |
| Eneloop cell#3 & testrun#3 | 18h2min | ~1.09V | 1.1856V | 759mAh |
| Eneloop cell#4 & testrun#4 | 17h58min | ~1.111V | 1.1870V | 757mAh |
18h39min ~0.85V 1.1758V 786mAh (falls noticeably out of regulation)
On Low-mode HKJ estimated 10.4hrs, i had expected 20hrs, and manufacturer's specs were "50 Hours", which is a complete lie, shame on Olight! In my real life tests the absolute max runtime until the light drops 100% dead is 19h17mins. From the above table we can conclude that the average runtime on Low-mode down to 50% brightness is 18.0hrs.