They both have incredibly unsustainable turbos for their size (FT03 or EA01). Start either of them around 1000 lumens though and they will probably hold it for quite a while. Emisar lights with boost drivers start a lot lower and maintain much higher sustainable output through their runtimes. The problem isn’t Anduril, it’s pushing 80+ watts through a inefficient driver into a flashlight that can’t move the heat out fast enough. You don’t notice a ~20 minute stepdown from 2000 lumens to 1000 in an efficient light (D4V2 w/ boost driver), but you sure notice the 4 minute stepdown from 2500 lumens to 200 lumens in lights with a less efficient driver (FT03).
The FT03 might have Narsil firmware. Mine does. It in fact drops down and never recovers, but that’s how Narsil works, no?
not at all
they are all valid topics… and I think we are in complete agreement 
including on my main point that the OP has the Option to:
Raise the Thermal Limit, to Raise the floor of the stepdown 
It compares the past temperature to the current temperature to determine how fast it’s changing, and in what direction… and if it looks like it will soon overheat, it throttles down.
However, turning the light off/on clears its thermal history buffer. This resets the “slope” to zero and disables the predictive response for a while.
Let’s say it has a limit of 50 C, and the current temperature is 45 C. Should it ramp down? The answer depends on what the temperature was in the past, like 30 seconds ago or whatever time the firmware is configured for.
- Past = 35 C, Present = 45 C: Predicted temp is 55 C. Response: Throttle down.
- Past = None, Present = 45 C: Predicted temp is 45 C. Response: Stay at current level.
- Past = 55 C, Present = 45 C: Predicted temp is 35 C. Response: Throttle up.
That’s why cycling the light can make it stop adjusting the output.
1 Thank
Correlux is probably right.
Android, for example, has to be done the same way. Each device is a bit (or a lot) different and adapting the firmware and software properly is much of what defines the user experience.
Any firmware or software that wants to run on a variety of hardwares will have this issue. Either it’s bloated up with code that won’t apply to THIS hardware (just in case it had been installed on THAT hardware), or it has to be tweaked for the specific machine it is actually operating in. Your flashlight doesn’t have enough memory to cram full of unnecessary code, so…
At the end of the day the manufacturers that take responsibility on their end to adapt the base code to their specific hardware are going to provide a better user experience.
I have only two Anduril lights. Both manufacturers have a link to github/toykeeper on the product page, so that’s something. Both work well for me. Though, I have no need or desire for long-duration high-lumen output.
Edit: Goodness, I was somehow looking at a cached copy of the thread or something? 29 posts but when I click “save” and it reloads there are 61. With some great technical details. Thanks everyone!
The economy car versus hot rod car analogy is bad for so many reasons but I will explore in that direction just a little bit. Your goal is to cruise along at 70. You haven’t defined what the economy car is capable of for a top speed. In the hot rod car you want to drag race up to 80 and assume the car will automatically settle down to a cruising speed of 70 all by itself. You are building up a lot of heat drag racing to 80. Try not flooring the car and gradually increase your speed up to 70. That’s what you should do with the stepped ramping. Out of the seven preset step modes, Try level 3 or 4 and see if the output stabilizes at that setting. If it does, then move up one level every 4 minutes until it gets too hot. Then you should know at what level you can turn it on every time without issues. And again, no way on the zebra light maintaining 1900 lumens for beyond 2 minutes.
I too have difficulty with attempting to set a proper thermal temp setting on my Sofin SP36 pro and my Wurkos TS21. As much as I have grown to like Anduril, I find it rather difficult to deal with, as a newbie to the software. I really wish it had a better way to achieve sustained lumen output, without stepping down so aggressively. A more gradual step down over a few minutes, rather than a few seconds, would be more desirable. I wish I had the expertise to change the code to suit my needs. More options and a better way to calibrate the thermal setting should be the goal of Anduril.
agree! Being able to find Sustainable Output is a valuable feature, to me.
possible strategy to find a sustainable ceiling, use lower ceiling with higher thermal limit… examples:
1. set ceiling to 90/150
2. set Thermal Limit to 50C
or
- set ceiling to 110/150
- change Thermal Limit to 55C
or
- set ceiling to 120/150,
- change Thermal Limit to 60C
Nice to see you back, TK!
Im using timed stepdown in my FW. Here is sample of 10sec. automatic step down to last used level. I found it very convenient , no need to double click to return. 
else if (actual_level > 130) { // if more than HIGH level
if (arg > TICKS_PER_SECOND*10) { // 10 seconds
set_level(mem_level); // mem_level = last used level
}
return MISCHIEF_MANAGED;
}
One other thing that I don’t see mentioned here is the ability of the light to shed heat beyond a simple weight measurement. Perhaps I am wrong about this but I don’t think so. Seems to me that reflector lights are able to shed more heat than low profile TIR lights like the TS21. I can set that light to a lower level and it will get hot much faster than an SC31 Pro or SP10 pro set at a similar output level. All run Andruil so that is not part of the equation. The TS21 has by far the most mass so that also is not part of the equation. Could be a triple SST20 is just horribly inefficient compared to the single SSt20 or the High CRI lousy tint LED (I forget which one) in the SP10? I very much doubt the difference is accounted for by the efficiency of the LEDs. Pretty sure that the ability for the light to shed the heat is significantly different between the lights and in this case the light with the most mass actually sheds the least heat.
agree
a light with 3 LEDs, can make 3x more heat than a light with 1 LED, IF both lights are set to the same Lux on Target…
this is because the 3 LEDs are illuminating a larger area… like taping 3 separate lights together. Also uses up the battery 3x faster.
this has nothing to do with the common knowledge that triple LEDs are more efficient… they are, but under a different scenario… that is… IF the triple and the single are both set to the same lumens output… this is not the same as equal Lux on target…
The triple will look dimmer than the single, when shined on a wall, if both lights are set to the same lumens. Because the triple spreads the light over a larger area. In this case the heat would be similar (triple would be cooler actually), based on lumens output…
Look at this recent post from @gchart. Look specifically at the first graph. Whether he starts on turbo, changes the temperature settings or starts on the level 6 or 5 of the stepped ramping mode, at roughly the two minute mark the sustained output is all the same. The point is, just start at level 5 of the stepped ramping mode and don’t create a whole lot of heat at the beginning for nothing. You’re using up battery power and creating heat for nothing.
I agree.
thanks for linking to those charts…
Actually, taping 3 lights together will triple the candela, because the three hotspots will coincide and effectively triple the lux on target.
The reason a triple-led flashlight is less efficient (in terms of cd/W) compared to similar sized single-led flashlight is because each led has a smaller and much floodier optic (or reflector). Just nitpicking a bit…
agree IF the 3 hotspots overlap… otoh, if they are spread apart to illuminate 3 separate spots, then the size of the illuminated area will be 3x larger, but not 3x the Lux
both is possible
in the case of a triple optic, they “usually” Do illuminate a larger area than a single LED
I did some tests with a lux meter, and it clearly showed that to match lux from a triple to a single, it requires more lumens from the triple, because the hotspot is larger
===
20-clicks will set the thermal limit to 50C.
From the Anduril 2 manual:
“Click once per degree C above 30. For example, to set the limit to 50 C, click 20 times. The default is 45 C, and the highest value it will allow is 70 C.”
Thank You for catching my mistake… You are absolutely right! Much appreciated.
Or, in anything updated in the past year or so, the user can hold the button twice instead of doing 20 clicks. Menus use 1C for +1, or 1H for +10.
So 50C would be 1H, 1H.
yes, thank you
it is a Great feature
I use this feature… but did not want to overload the chicken w too much info… I dont think his learning style matches my wall of text teaching style
nor do I know what model firmware is on his lights… in any case
Im very happy with the latest Anduril 2… really glad the blinks in the ramp have been eliminated in my TS10, SC21 Pro, SP10 Pro…
Also really like the option to slow down the ramp speed, and disable Turbo…
thank you for all your support!