RMM sent me a small mod to make to Bistro's temp monitoring to adjust better with higher powered lights. Been wanting to give that a go, just haven't had time. I just have to make sure it's implemented carefully for both ramping and mode set operations. Actually I'm not that familiar with exactly how Bistro works - I assume you can manually bump it back up to turbo, then temp. control will bump you down again - after a delay I assume?
I was think'n of using turbo timer as defined now, then maybe temp. monitoring as a backup, so it would have a fairly high threshold, but that may require a decent calibration.
I can only second that. I own a cheap SRK with 7 Latticebright XM-L fakes inside, on turbo mode it outputs around 1200 lumens (from the advertised 8000 in high mode. It has a very bad heat transfer but I pimped it with proper thermal paste on the ledge where the LED boards sits on the body. Made the test a few days ago and let the light run for 20 minutes on high. The whole light was VERY hot afterwards, there was not a single spot on the aluminium body that was cold enough to touch. I can only imagine how quickly a can size light will get extremely hot with PROPER heat sink and heat transfer and > 4000 real lumens - must be a matter of seconds. So much heat must go somewhere, and I start to doubt if this light will be usable “in real life” on high or turbo mode.
Well, actually I have 3 soup can lights that do over 4,000 lumens now, 1 at about 5,000, the other two over 5,000. They get warm in 30 secs on fresh good cells, but I would not call it hot. All 3 have been heavily modded pretty good thermally, all in the 15-18A range - the M6 of course is pretty good stock, just needs copper MCPCB's mainly.
DD/FET based lights certainly have some regulation built-in, just by depletion of the cells raises their resistance, therefore reducing amps. I left my 5,000 lumens M6 on in full power possibly up to 40 minutes, and it was untouchable - very hard to even quick touch the button to turn it off. Results? No damage whatsoever to the light, LED's, not even the batteries that I am aware of. Just 3 circular burn marks on the hand towel I had the light wrapped in, in my bag. Not recommended, but no flames, not even damage, bit of luck.
I can’t say I don’t have a preference for “real” thermal monitoring, because I do. Its one of my favorite features of my little Zebralights and makes them one of the first lights I grab if I’m going to be using a light for a longer period of time and don’t need multi-thousand lumens for the job. I was out on the beach the other night with my Nitecore TM16w and kinda got irritated at having to click it back to high every few minutes even though the damn light was barely warm to the touch.
That said, I do understand that its not an easy problem to solve. So I’m not judging and will certainly buy the light no matter what solution is implemented. I was just asking… and maybe expressing my preference a bit You guys are all doing awesome thing here and I’m just watching from the sideline, and trying to be patient.
Here's the BLF Q8 driver (V3) in the 4X Securitylng clone with 4 25R's @4.22v:
Held on max/turbo, I used one of those yellow IR temperature guns, nothing fancy, in a room temp. of 64F, measuring highest temp below the bezel:
1 min: 106F
2 mins: 120F
2.5 mins: 124.5F
3 mins: 131F
Stopped there - quite hot. Cool down with it OFF:
2 mins: 107F
3 mins: 101F
4 mins: 97F
5 mins: 93F
Fresh charged cells will of course make a difference, so this is about the worse case, maybe little worse on 30Q's fully charged. Think I noted 16A measured on fully charged 30Q's previously.
This isn't close to the quality host we are working on, but gives you an idea at least. This 4X has been modded of course, with a copper shelf and bonded to the outer shell with an aluminum based epoxy.
A temperature controlled step-down has thusfar been proven to be difficult to build into a driver, DrJones’ driver with PID regulation even does not work flawlessly in all builds, only L4P’s LD2 with external temperature sensor works well but is difficult to assemble.
Alternatively, a timer stepdown is easily incorporated in the firmware. When, how many step-downs and by how much can be empirically determined from prototype samples, just like Toykeeper did for the BLF-A6. As said above, it needs not be very precise, just prevents the flashlight from becoming dangerously hot.
I'm think'n for this BLF Q8, the MCU is out there in the middle of a pretty large PCB, in a fairly large open pill area, so it may be staying quite cooler than the surrounding housing. Could be tricky for sure. I couldn't recall about the BLF A6 - did it ship with the timer stepdown enabled by default? Not sure if I ever checked.
With Narsil, it's a bit more complicated - 12 different mode sets, smooth ramping...
What I did for ramping was to engage turbo step down only if over a certain level, and throttle it back to a lower level. I haven't attempted to test it much, tweak it, calibrate it at all. It's probably gonna need something more sophisticated with multiple drop-down levels, and yes - you are right -- will need to be empirically calibrate based on something close to the final shipping version.
For mode sets, it will also have to change, because I believe currently, turbo stepdown only works at max/turbo, and simply drops down one mode. For most multi-modes, I think it's ok because the 2nd highest mode is 35-40%, but might be 1 or 2 using 50%. Also if you select mode set #1 (one mode of max) and turn off moon mode, I don't think the step down will work, but haven't actually tried it ().
While I agree with everything you said, my only concern is ‘late’ changes to the firmware. As I understand it, we missed the latest version of the Bistro firmware due to Manker jumping the gun and programing the drivers with a backlevel version. I do understand how and why it happened, I’m just not sure how the Q-team can keep it from happening again, other than being hyper aware of the potential.
Well we are not in the stage where final drivers are being flashed for production so it is very nice to how much Tom is able to put in there, with plenty of time to work out bugs. (Bugs are being dealt with now.)
At a certain point there will be a “Narsil Q8 version” and that will be the one to be used, regardless of further progress of the software I imagine.
And it is so cool to see how much can and is mplemented in this stage on Narsil that was so great when it all started.
There definitely was some dis-connect goin on at the time, but this is different since we have an open direct line with the techies (entire group) and the manufacturer. You can debate what's better - open group comm. or single point of contact, etc. - pros/cons each way.
For me as a developer (for too many years to mention), I like management/sales/support to act as a buffer so I don't get bothered, but sometimes you just have to get involved directly to get something done or fixed quick.
My little Tank007 E09 gets hot on a 10440 but I can hold it though some folks couldn’t. At that point, cell depletion ceases the temp rise, making for ideal free ‘thermal regulation’
So what is ‘safe’? Cool enough for my calloused heat-resistant hands, or a baby’s delicate skin? IMHO ‘safe’ is where no damage happens to light, cells, or common surfaces one could reasonably expect this light to be set on. That’s past where one might get burned. Just as looking into the light on turbo might damage someone’s eyes, there’s no way you can protect against everything. I think 50% ipstepdown is good enough though safer could be better.
Another option might be thermal fuses but I don’t know how small these come, or how much resistance they have, or even if they come at levels this low. Would need soldered-in replacement when blown but would serve the purpose and they are pretty reliable. Plus they would prevent repeated events to keep noobs safe if still unhappy over the light they broke from not following the directions
Let’s face it: to make a light stupid-proof cannot be done at this performance level. Good to best use the available potential in it, but not if that costs too much money or takes away other features we want. Users will always have to assume some responsibility in use of everything, and the Q8 is no different in that.
If you ever try to solder on a driver which you’re holding in your hands, you’ll see that PCBs are not that bad at conducting heat through all the traces The MCU’s temperature measurement might lag somewhat but I don’t think it will be that far off, especially in a large light which heats up more slowly.
As for the complexity of all the modes, a safe and easy way is to bypass modes and act directly on the FET PWM level. It would leave the UI in a weird state though because it won’t know about the true output level. But perhaps it would be acceptable as a failsafe.
I don’t understand 1/2 of what you folks are discussing, but to my way of thinking any heat protection should be to save the light; the operator might have a good reason for running it that way (too hot to hold). I can’t imagine one though. We’re all adults, if we don’t understand that these things get too hot to touch we probably need to go back to grandads flashlight. Heck, get burned once and I bet it won’t happen twice. Just like a child around a stove. Anyone in their right mind won’t use it without tiptoeing lightly the first few uses anyway. Or at least I always do. We’ll figure out pretty quick just how hot they get, and how quickly they do it.
I’m trying to say that we should be able to operate the light to it’s capacity without hurting it. Otherwise it would be like having a formula one car but having it governed to a certain speed that Granny shouldn’t want to drive at. Yes, Phil, we assume responsibility for it’s use, I think that’s implied.
I may not understand most of what you folks are discussing about Narsil Q8 and all that, but from what little I am able to catch, it sure does sound as though the right folks are involved. Carry on, and thanks!
A better analogy might be most modern sports cars which come with all sorts of electronic stability controls but also a switch to disable them. So it’s safer out of the box but you can still use it to its full potential if you wish to.
I agree that there are plenty of good reasons for running the light too hot to touch. You might have it on a tripod, be wearing gloves, want to cook your breakfast on it, etc.
in high mode. It has a very bad heat transfer but I pimped it with proper thermal paste on the ledge where the LED boards sits on the body. Made the test a few days ago and let the light run for 20 minutes on high. The whole light was VERY hot afterwards, there was not a single spot on the aluminium body that was cold enough to touch. I can only imagine how quickly a can size light will get extremely hot with PROPER heat sink and heat transfer and > 4000 real lumens - must be a matter of seconds. So much heat must go somewhere, and I start to doubt if this light will be usable “in real life” on high or turbo mode.
You guys are all doing awesome thing here and I’m just watching from the sideline, and trying to be patient.
).