At full power? A few seconds. Have you used an Emisar D4? It’s a lot like that.
At a more practical power level? It should run indefinitely at several hundred lumens, depending on how you configure its thermal parameters.
Here’s an output graph from a few months ago, showing thermal ramp-down and then a relatively stable level:
That first minute is likely to ramp down faster than shown in the graph though. It was actually too hot during that first plateau, so I made it adjust down faster.
When started at the default ceiling level of 1100 lumens, the result looked more like this:
This test was 9 months ago though, and things have been updated since then. Mostly, it just shows a rough idea of what to expect when using a high mode which isn’t turbo. I think this test also had a lower temperature limit configured. The stable level depends on the user’s personal configuration and environmental conditions.
The thermal response really depends on a lot of different things… so it’s hard to give concrete answers. I think it’ll probably ramp down very quickly on turbo though, in most cases. And at lower levels the response happens later and slower.
It’s largely based on predictions, so the rate of change has a huge effect on how the light reacts. If the temperature is rising very quickly, it’ll make a hard turn downward. But if it’s rising slowly at the same temperature, it’ll only make a gentle turn.
At full turbo though, there’s sort of an override which forces it down to a more sane level as soon as the prediction goes above the ceiling. So, expect it to only do turbo for a few seconds before it drops to the maximum regulated level of ~1100 lm *.
* Lumen numbers are from my cheap uncalibrated light box, and I suspect the real numbers may be a bit lower. I really should order some reference lights from maukka sometime, and build a proper integrating sphere.
This could be interesting, just to see how it pans out.
The Q8 had a similarly massive interest list, and, remarkably, a very high uptake from those on it.
Sufficient that it continues in production today.
Of course the first batch had a few niggling problems, but these were, I think, sorted out as well as could be hoped, thanks to swift feedback from BLF, The Miller’s determination, and Thorfire’s responsiveness in swiftly doing a little re-design and re-making a part (and replacing the production manager).
Due to serendipity I missed the first-batch problems.
The Miller had to take a chance on signing off the Q8 when he did, the team could have fannied on forever over the minutiae, it was very interesting technically, but there was a market window, beyond which interest would have slowly died, credibility been lost, reputational damage to BLF, loss of an interested supplier (Thorfire/Sofirn).
My suggestion: Build no more than say 50 in the first batch and sell them at a really good unrepeatable unmissable price. First come, first served. No warranty, but as much support as can be run from China with someone coordinating. We won’t get Wim, but is anyone else prepared to step-up ?
Make it clear that these are actually pre-production prototypes, and would best be bought by experienced people who are prepared for a few bumps and squeaks in exchange for a great price, and the opportunity to improve the thing, if necessary.
Then wait for feedback here. Give it at least two months. Then reflect, make any changes that seem sensible, and build the main run.
This is how we do things in Aerospace, typically A model, B model, C model (all internal development), P model (clients first sight of the thing, and chance to evaluate and complete qualification, not to be used in a serious situation, unless needs must), then full-scale ramp-up. But this should not be necessary for a simple torch.
Simply making only two pre-production prototypes, then sending them to only two people, most heavily involved (and least likely to be able to critique their own work neutrally), and who seem to have gone off-grid, does not seem to me the brightest of ideas.
I also suspect that there ought to be some final tuning of firmware, by the Goddess, to suit whatever emitter choices have actually been made (still unknown). When (and how, without a couple of P-models) is that going to happen ?