First: Kudos to the team, and Thorfire, for stopping to take stock and get this right.
A correct but difficult decision. I am all the more impressed with the leadership demonstrated here. This has encouraged me to order my second torch.
When this is resolved, which I am confident it will be, it will further strengthen my respect for Thorfire. They are giving it their attention.
Second: here is my take on what might be happening. Just a theory.
The technology for making natively coloured LEDs differs depending on the colour. Technically, the wavelength emitted is a function of the bandgap of the compound semiconductors used. Practically this means that the forward voltage needed to light up the LED is different for each colour.
As it happens, green LEDs require one of the highest forward voltages. Typically above 3V. I.e. much the same as a single lithium cell. Leaving almost no headroom for further voltage drop across the series resistor used to set the operating point (current level).
If I was asked to specify e.g. a typical green 3.3V LED, to consistently light up from a typical single cell (say voltage ranging from 3V to 4.2 V, I could not do it with just a series resistor.
I would argue for using a different colour, e.g. red, orange, yellow etc, with a Vf around 1.8V, leaving plenty of headroom for the series resistor to control the current across the full range of battery voltage.
I would only consider using a typical green LED if it could be driven from more voltage, e.g. in a 2S light, and even then, preferably from an MCU pin set up as an “open collector” or “low side drive” output capable of handling the full voltage.
The MCU in a 2S torch, would be expected to be supplied from no more than 5V, possibly as little as 3.3V, so a direct drive from an MCU pin might be marginal at best.
In the Q8 the MCU is driven from battery voltage (minus the drop cross the reverse protection diode) so tracks the battery voltage as it varies with charge level.
I read that the switch PCB was derived from that of the Thorfire S70 a 2S torch. Is it possible that the same green LED is being used in the Q8, with perhaps resistor value change to try to compensate for the 1S Q8 ? If so, maybe it is right on the edge of working, and e.g. small shifts of Vf as the LED warms up, or during initial ageing, or between the two LEDs, are enough to tip things over the edge.
If this turns out to be the case, I suggest that the green LED be changed for one with a much lower Vf (and different colour), and the series resistor value adjusted to suit.
Anyway, I’d just like to throw this speculation into the discussion. If I had my Q8 already I could do a few simple measurements to see if this has legs. I.e. anyone with a torch with functioning LEDs, and a multimeter, please measure the voltage directly across the LED whilst it is illuminated (I.e. Vf). and report back.