Set the ceiling to 110/150
UnSustainable Output is not an Anduril specific issue. It is simply a matter of Too Many Lumens for Too Little Mass. It applies to All flashlight UIs that allow UnSustainable outputs.
without reading everything posted already, it could be 4 easy things:
temp sensor not calibrated
temp limit set too low
weak battery, low battery
high ambient temperature
Nowadays the thermal regulation in Anduril is VERY good, it is much improved since a few years ago. And a gamechanger is that the user can set the ceiling and the thermal limit (and calibrate the sensor). Simply set a sensible ceiling that does not rapidly heat the flashlight, this way the thermal regulation can do it’s thing without overreacting. You must experiment to find a good ceiling for your flashlight.
There is a related problem - many budget flashlights are using inefficient driver designs that generate a lot of heat. So the heat is coming not just from the LED but the driver also. This is not the fault of Anduril. The only way things will get better is if you demand better drivers from manufacturers (and if you are willing to pay). They merely give customers what they want, right now most people want cheap FET drivers which can achieve high output for a few seconds.
1 Thank
Temperature is measured with Attiny IC and IC is on PCB. PCB components can go up to 85degrees and more. So to set 45 degrees ceiling is nonsence. Set it to whatever suite you . Like 65 degrees or max 70. Just dont overheat battery it does not like more than 45 degrees 
. Properly made flashligt should have temperature sensor fixed to host, but not on plastic PCB.
My copper Tool AA doesn’t seem to be dimming after a few minutes, but it is too hot to hold without gloves or something.
The AA Tool is not an Anduril light, and has no Thermal regulation…
the drop in output is caused simply by the battery voltage dropping:
.
chart by zeroair
@jon_slider : Yeah, that's probably true, one doesn't want to overheat the Lumintop Tool AA and some common sense must be employed.
Relating to the overheating subject, I have a Convoy S2+ x8 on the way, and Winter in Wisconsin is pretty cold, so while Turbo might not work in 20°C for very long it might work with enough Winter -5°C cooling, otherwise I can see me changing that into a x5 so I still have Turbo mode for 5+ minutes (x4 can be 100% continuous).
I also downloaded the Biscotti code as modified for Convoy, however it is known to have memory 'loss' issues so would rather not tinker with getting it working. (Convoy uses a locally rewritten closed-source version of Biscotti to cope with a sketchy driver design--the whole thing seems shoe-stringed together just enough to get it working. Chabuduo. I don't know how @ToyKeeper ((the Biscotti author)) holds it together so well, but credit to her however she does it.)
I don’t think this is so much a problem with Anduril, per se, but as mentioned above it’s a complex firmware that should be adjusted to the host by the manufacturers (that seems to have gone by the wayside save for Hank’s lights when TK assisted with various models). It can be tweaked/fooled by the user on both ends of the temp calibration…and that basically brings you close to where we were years ago with overheating toasty lights that you turned off or wore a glove to extend the time needed to turn it off. The trend has been toward compact high lumen lights for awhile now, but physics hasn’t changed a bit. Get a larger host that can dissipate heat better, or use the light at lower lumens, and the step-down will be greatly reduced as long as the cell can keep up with the voltage needs and its own heat.
That, or get lights with boost drivers for sustainable output and generally less heat depending on design.
But I really don’t think the firmware is the biggest culprit. Manufacturers adopted it (great!) but several obviously don’t fully understand it and did not take the time to apply it correctly with their designs (heck, can hardly even get them to honor the simple request of mentioning and linking to the open source root for such a generous gift as this firmware is to the whole community and that they are profiting from with no R&D expenses of their own…).
Hopefully boost drivers will be the next thing more manufacturers will jump on. Really does cure a lot of ails.
Respectfully, while we have seen that with several manufacturers, I don’t believe it applies to Simon at all, not even a little. He is cost-conscious when he believes an expense may not be worth it/may not sell well to his customer market, but he displays the opposite many times and goes ahead to deliver more expensive designs and such if enough people seem interested. He’s an enthusiast like us. The actual expenses involved in tweaking and prototyping and stocking new skus and such are far greater than what many people might think, but he has still expanded his line so greatly in the last few years, taking that risk. The driver issues are a different saga, no “good enough” cost saving implementation.
That whole thing is an interesting cultural concept, though…kind of like the issues with “sharing” and intellectual property. Big differences in perspectives and not entirely rooted in greed.
You are asking for a stepping solution when in smooth
ramping mode. This is an oxymoron. It is so simple to switch Anduril to stepped ramping if you want to land at the same spot every time on any specific light. Anduril 1 and 2 both appear to have 7 levels when in stepped mode. Level 4 or 5 will probably be about right on many lights. Edit, There is no way that any zebra light is going to maintain 1900 lumens.
Number of steps as well as floor and ceiling brightness can even be customized. 
Chicken Drumstick, if you are after sustained lumens, choose a light that has an efficient driver (less energy wasted as heat) and sufficient thermal mass to handle the heat generated. It also helps to choose a LED that is appropriate for the host (an SBT-90.2 is way overpowered for the thermal mass of a small thrower like the D1, and will always be a hotrod).
Because the light has anduril, you at least have the option to pick a memorized brightness at turn on that is sustainable, or you can as jon_slider suggested set the ramp ceiling to that level. These are both solutions to make the most out of your lights.
I have mentioned the ” The Dance ” Aduril does on many occasions.
No matter where I set the top of the ramp, most of them ramp down and then slowly ramp back up to just under the top of the ramp. Example” FW4A with SST20s ” top of the ramp set to 550 lumens. It will stay there for about 60-90 seconds then ramp all the way down to 250 lumens, then slowly work its way back up to 450-500 lumens. The crazy part is after this dance , it will stay close to the top of the ramp. Most of the time the light isn’t hot to the touch. If you use Turbo and get the light Hot it will ramp down even further.
The original FW4A with XP-L hi is the only one that I set to 800 lumens and it pretty much stays there till the battery starts to run low
All this is verified sitting on my Lumen Tube ( Yes I know it’s not a Calibrated piece of Sophisticated Equipment, but it does show the up and down of the lights perfectly )
I think it’s unfair to call it an Anduril problem. It’s a small light problem. Physics only allows a light to dissipate so much heat, the smaller the light the lower that limit will be whether it’s an Anduril light or not. Many Anduril lights have inefficient drivers which exacerbates the problem, but the problem is the same regardless of UI.
Look for lights with more efficient drivers, that’s going to be the biggest factor in determining sustainable output for a light of a given size.
Your point is well taken. Perhaps I am wrong, or at least my assumption is limited in scope. Out of the budget light manufacturers Convoy has some very good products, will customize them upon request, and the prices are very competitive, so that does say a lot for Simon. I've also rediscovered Mountain Electronics, so there are options to use more efficient parts in a Convoy host if one is still not fully satisfied.
Exactly. Anduril is actually the solution for hotrods, having a very competent thermal management, and adjustable modes (with the stepped ramp).
It’s a bit diy in terms of finding the sustained outputs, but, on the other hand, sustained output changes based on ambient temp, wind, host, emitter, driver, how the flashlight is held, etc.
If you set a ceiling of 110. Can you still get to the higher output if you still want too? If not, this isn’t a solution, just a compromise.
Also, when you are spending good money, i.e. a lot on a light. You really shouldn’t have to then go and do the R&D to set all this up. And if you ever need to reset the light, all of these setting will be lost too. It’s not as if setting them up is truly simple either, as it is so is to miss click. Or believe you have clicked correctly, but the light doesn’t respond in the expected way.
That’s fine. But why does it sometimes dim massively even when the body isn’t hot or warm? And how come cyling the light can sometimes produce a more stable output without changing anything?
read my #24post again