Thanks for your measurements, that’s exactly what I getting on my D4v2 except it’s way worse in my case because I am using hotter HiCRI diodes(SST-20) and hight current Sony VTC6 cells.
Are you sure your VTC6 is still good? After not many cycles these cells are not what they were as when they were new when used at their intended power (lots of amps discharge and high current charge)
I repeated the measurement with Samsung INR 18650-30Q cells, once with a temperature limit of 60 °C, once with a temperature limit of 70 °C:
The simulation result of the new code for this scenario looks like this:
(… still waiting for the reflashing kit to arrive, but I can’t wait to test my new thermal regulation code on an actual flashlight)
I think they are still fine. I don’t discharge them to 2.5V(3.3V usually ) and the cell temp during discharge definitely doesn’t exceed 80 °C. Sony explicitly says in their data-sheet that cell temp shouldn’t exceed 80 °C.
Thanks for the graphs.
Looks way better than it is coded now. At least your code brings back the control when the flashlight will drop the output to the user and the results are very close to the target temp.
Hi guys, I’m new here. I’d like to get a new king for my flashlight hill and I think it will be D4V2, but I’m not sure yet do I want aluminium or brass version (Ti is out because worse thermal conductivity). I like the weight of aluminium but I also really like switch back lights of brass version. Do you know is this upgrade coming any soon for the aluminium version? I also like to have raised switch ring for safety reasons and noticed ring is stainless steel on brass version. I’ve always thought ss will cause corrosion to brass, could this be something to worry about?
Yes, stainless steel will heavily corrode brass in theory. But depends on what stainless steel is used. 301 or 304 should be the least of the problem with the common stainless steels when being tight on brass.
That cycle life test is an extreme case, and probably not representative of how VTC6s fair in more typical use. Sony themselves suggest a 3A nominal charge current rather than 4A, and they rate the cycle life of the battery at a 10A discharge, not 15A.
At least in the flashlight world, I doubt very many VTC6s are being charged at 4A and discharged at 15A continuous, over and over. I don’t think the D4V2 would even draw 15A continuously because of thermal regulation. And on the charging side…maybe I’m behind the times, but I don’t even own a charger that will do more than 2A CC…
Still 50 cyeles is not really anyting at all considering the claims of cycles on these type of cells.
Yet the VTC6 is the one that suffered the most out of all those 3 cells in terms of capacity, one thing to not by comparison since the conditions were the same.
15A is common for direct drive flashlights, but indeed continuous 15A it is quite hard for that.
For such cell that needs to be discharged in direct drive flashlights I would only charge it with high current so I can use it fast again.
However I believe using 1 cell only is not a good practice for battery life, one should should at least have 3-4cells cycling thru single cell flashlights.
A short time lapse of the heat development with XPL-HI V2 5D emitters:
If your D4v2 is consistently leveling off at a temperature well below what you’ve set the temperature limit to, you could try reconfiguring the “current temperature” to fool the light into thinking it is cooler than it actually is.
Turn on the light and let it heat up for a little bit so its noticeably above room temperature, then configure the current temperature: From OFF —> 3 clicks —> 2 clicks (to get to batt check) —> 2 clicks (sunset) —> 2 clicks (beacon) —> 2 clicks (tempcheck) —> 4 clicks (thermal config) —> then on first rapid blinking click 20 or 21 times.
The light will think it is room temperature when it is actually hotter. This should cause the output to level off at a higher temperature.
Don’t overdo it though as you could cause injury or damage the light if you set it too high. After configuring, I recommend testing it with a fresh cell and checking that it’s not getting too hot.
I’ve already tried calibrating the sensor to the temperature lower than the real temperature but it doesn’t seem to make that much difference. The flashlight still drops the output prematurely. 0-8-15 User ’s measurement clearly show what happens when you are increasing the temperature limit. Maybe lowering the sensor temperature even lower to archive the temp limit of 90°C or 100°C or even 110°C will help but I don’t think that’s an appropriate way to address this problem.
I’ve tried 92-95 °C temp limit. Ceiling bounce graph is below.
It takes about 0.2 of a minute(12 seconds) for D4v2 to start severely dropping the output, not to mention that it has already started dropping it from the very beginning.
You may want to try with a different cell and see how that goes.
Sorry, but I don’t understand what the cell have to do with the way how the flashlight manages its thermal regulation.
Don’t waste your time on this. The issue is the underlying thermal regulation mechanism. I’ll try my best to improve on it once my reflashing kit arrives.
“I’ve already tried calibrating the sensor to the temperature lower than the real temperature but it doesn’t seem to make that much difference. The flashlight still drops the output prematurely”
Is the drop in output due to thermal regulation if it droops fast no matter of the thermal settings?
The current firmware monitors the rate of change in temperature. If the temperature rises much faster than 0.375 degrees per second then it will always drop output early, no matter how high you set the temperature limit. And if the temperature rises slower than 0.375 degrees per second then it will overshoot the set temperature limit.
That’s why the output drops prematurely at first, but then overshoots the thermal limit later on.
It seems so, it’s just how thermal management works in Anduril . AFAIK it was already mentioned in this topic by others.
Thank you for your efforts, I think a lot of people appreciate them. If you’re fine with it please share the results of your work when they will be ready.
That’s the plan.
Great! :+1:
It really is a huge help. This is something which has needed improvement for a long time.
Do you think it would help to increase the temperature resolution, or only to change the algorithm? I was thinking about left-shifting the values and maybe increasing the sample rate, with the values still going through a lowpass… and then it could work with fractional degrees instead of only integer degrees. This may also make it possible to narrow the window of temperatures it will accept.
The algorithm needs to change too, of course. I’m just not sure if extra resolution is worth the bytes it would take to implement.