Your temperature result is good. I see that you have forced airflow, so it's approx 45 deg C @ 13-14 min. I started off at the same temperature as you did, and I got 49 deg C @ 12th min. Nil airflow.
After the Sky Ray 3800 incident, i have sort of stopped doing full bore runs. I must have done > 20 full bore runs on XTAR 2600, Solarforce 2400, TF 2400, NCR18650 on that light. Luckily when the driver got a dead short i was doing protected cells and not NCR18650. Definitely i'd be looking at some expensive repairs to my decor then. (area full of tempered glass!)
At the 55th minute, the output is 80% (I misread it a little to be the 60th minute). The "knee" refers to an important aspect of any graph and this represents where something is really useful. For example, Mitro refers to Sanyo 2600 cells to have the best knee in his discharge graphs. If you dabble in hifi/audio, a frequency graph has got to have a lower cut off point and it is usually the f3 or -3dB point before the graph really dives. In the above graph, we choose 80% and it is the point where the graph really dives after that.
Should have clarified that technical part for you. Hope that helps....
You must not confuse the heat test with the runtime test, they are done separately and with very different test conditions and purpose.
The runtime is with forced airflow and a normal battery load, this is a simulation of actual use conditions (Light in hand and moved around).
In the heat test the light is tied down in the test bench and supplied with power from a bench power supply (Adjusted to average battery voltage i.e. 3.7 volt for LiIon), the purpose is to see how the light handles high temperature, secondary how good the cooling is.
It does not look like the S1 has any temperature throttling and the led does get hot, but not dangerous hot (The XM-L datasheet has a curve with relative output depending on temperature).
I see...I am looking at the blue graph actually. So I am getting a bit hotter temperatures but not a lot. Usually i am not worried about the LEDs, in my experience it is usually the driver that cooks first. heh...
That is the standard deal, it is the same for chargers and batteries.
But if you check the reviews on my website, you will see that many of the light/chargers/batteries are some I bought (I always write a note, when I get the stuff for free).
2pcs of the UCL 72.7mm x 1.85mm thick gives 1.48% cut in output. (each one is 0.74%). Original glass is nearly 10%. AR tint is green in colour. No rattle, the bezel screws back on ok but there is a very slightly bigger gap between the bezel cap and body than before. Original glass is 3.5mm and this is 3.7mm, a minor difference of 0.2mm. In reality you absolutely won't notice it.