Toykeeper got back to me and said she was not consulted to do a version of Anduril specific for this light. So the thermal response code may not be optimal. Mateminco probably just used an Anduril build from another light and adapted it to this one. This means the thermal code might be reacting too quickly on Turbo or the FET might be transferring heat directly to the MCU.
We are back where we started. 
for the price… just too many issues etc… bleh 
too me a rushed light asking premium money but failing to deliver…
Not every product is a runaway hit.
I wonder if a FT03 can be bought, then buy a SBT-90.2 and a MT-G2 mcpcb to put it on. You’d have to enlarge the centering ring, but in the end you’d get a perfect light. Good reflector, good driver, maybe even a cheaper cost if you don’t count your labor. One downside is this led is not easy to find at the moment.
Is there anything that can be done (like a relatively simple fix) to increase the turbo runtime on this light?
We haven’t really figured out why it’s stepping down prematurely. If the FET is responsible for heating the MCU, you could add extra mass to the FET. You take a cube of copper and attach it to the FET with thermal adhesive. What it does is make the FET heat up at a slower rate. This theoretically would make it step down when it’s supposed to.
Someone needs to test it, though.
Does your FT03S step down after 20 seconds using a high drain battery even though the exterior isn’t very hot?
I still don’t know if this is a widespread problem or not.
What temp do you have it set at? Even at 75°C mine steps down in 20 seconds. This only happens when the cell is at above 80% charge. Might be the FET overheating but I’m not sure.
limit 55°
So it looks like your light steps down at about 25 seconds?
What battery did you use?
Do you know if the exterior felt hot at 25 seconds or did it seem like it stepped down too soon?
Seeing as how the output went back up I’m thinking it’s just the software reacting prematurely and not the FET prematurely heating up the MCU (at least not all the FETs fault). The software measures the temp going up fast so it reacts sooner rather than later. A lot of Anduril lights act like this and there’s not much you can do about it. Still, adding mass to the FET might help smooth out the big dip in output or at least slightly delay when it starts to ramp down.
This is some interesting science.
How do you like the NI40?
I have the SBT-90.2 led in a Wildtrail D90V2 host, it has a Lexel FET+5+1 driver that runs flintrock's bistro firmware. On a fresh VTC6 battery (high drain 18650) it draws 19A at max. At maximum setting with a fresh VTC6 battery I measured the temperature between the fins near where the ledshelf is with a IR thermometer:
| time after switch-on (s) | fin-temperature (degC) |
| 20 | 38 |
| 40 | 48 |
| 60 | 55 |
| 80 | 62 |
| 100 | 68 |
| 120 | 73 |
At 120 seconds it stepped down, I will have to look up in the UI if it is a timed stepdown or a temperature-stepdown but I like that it is at 73degC (the led will be hotter but still some distance from desoldering temperature), a good safety measure and the tube is still cool enough to hold, I need no gloves to switch it off when needed.
I would think that the FT03 should be able to handle more heat than the humble D80V2, so two minutes before stepdown sounds reasonable, not 20 seconds
I think we learn that it's time to go from measuring the driver temp to using the MCPCB temp for our high power flashlights...
it needs just one check-run with a new flashlight model to learn how the led-temperature relates to the MCU-temperature, in temperature difference and in delay, and then adjust the step-down parameters in the firmware accordingly. If this is too challenging for a company like Mateminco, they need to hire a first-year student for a couple of weeks to acquire that expertise. Tip: It works perfectly fine for the D80V2, time to look at the flintrock firmware.
The problem is you can maybe set it up so that Turbo reacts well, but then when you heat the light slowly (like on a med level) the response will react very differently and people will complain again. Toykeeper was constantly testing and rewriting the thermal code for the FW3A and still had to use a compromised version that only worked “decently” for all power levels.
BTW, Mateminco didn’t write any thermal code for the FT03S. Anduril is all Toykeeper and she wasn’t brought in to tweak the thermal response for this light.
Basically that is what I was implying, bring someone in who is capable of writing/adjusting code. And if a smooth temperature stepdown is not possible on all modes, they then can make it maybe a basic instant stepdown to a sustainable level that can be overruled with a simple click. But it starts with being able to control the whole flashlight, not just the hardware. Sofirn is a step ahead on this, they do write code and new firmware for their lights. (If their firmware decisions are the same as we would make is a different subject of course)
I think Mateminco or Astrolux are a bit too cheap to pay for outside help.
They could maybe have their software writer make a new UI which works similar to Anduril, but it won’t be as good. It never is. The knowledgeable buyers know NarsilM and Anduril are a nice functioning UI and that helps to sell the lights. What can you do, they like to maximize profit and get the lights out fast.
On a side note, if the FT03S steps down too soon and it’s not very hot, can’t you just dble clk again to bring it back to Turbo? I seem to recall it having that feature.
FT03S 75°C (room 22°C) 4 mins (Tailstanding no fan)
End of test external temp: 51°C
Cell voltage at start around 4.1V (P42A)
Temperature regulation settings tuned to a pocket size hotrod but applied in a large light? (so based on far more extreme temperature predictions in the near future than are actually happening).