Comment in context that I never build a flashlight…
From an engineering perspective, when I saw that led, I wondered how it was going to cool down. Particularly if overdriven. Aluminum is a decent conductor but not good enough that it won’t allow a hot spot. It would have to start with a flat, finned bottom head and forced ventilation. Even then. Liquid cooling would be the only option.
Full circle to CPUs heat issues.
Obviously less issue when dispersing small LEDs.
Related to that……I love it when manufacturers claim 1000 lumens in almost keychain size flashlights. I guess they usurped thermodynamics……
What do you mean? 4S Battery or 4S LED?
HaikeLite has 4S Battery for MT07 with 6V LED config or the MT09R For 12V XHP35 HI.
4S Emitters makes no sense because you have to boost the voltage very high.
Sure, a 4S battery and 4 XHP35.2's in parallel. Well, as we found out with the BLF GT, the Vf of the XHP35 HI's are high, high enough to be a problem. The XHP35.2 datasheet indicates it's not much of an improvement on Vf, about still as high as the XHP35. Of course with 4 LED's and only 4 cells in parallel, the problem is even worse, but with 4S2P cells, it's slightly better.
I would think a BLF GT driver would work with this setup, just that I don't know how long it would last driving the 4 LED's at 2 amps per LED, for example, before it falls out of regulation. The BLF GT driver is a buck driver, a boost/buck would be preferred but more costly, and we don't have a design to use with our favored MCU, the ATtiny85 that runs Anduril and NarsilM.
You can totally have a 1000 lumen key chain light, assuming you have a battery capable of enough amps. I think DB Custom has a little light he calls the Texas Poker using a 10440 battery that does 1100 lumen.
The thing most manufacturers leave out of the specs is TIME. Of course, some manufacturers specs are total made up BS to sell their product. Still, if you had a key chain light that did a real 1000 lumen it would only be able to run at that for a very short time. Maybe 20 seconds for instance.
The MT09R that Texas_Ace built can do 20,000 lumen, but only for about 30 seconds.
The copper head Astrolux S43S (that I reset the thermal calibration on to keep it under control) only lasts 35 seconds on Turbo (drawing 18A).
So time is an important factor when talking about run times and is often missing from the specs.
The GT that Matt put the gbx3590 in is a massive amount of aluminum that takes quite a while to heat up. It would probably do 12,000 lumen for over a minute. Maybe 1.5 minutes. Something like that. You would only need a flat, finned bottom head and forced ventilation if you wanted a sustained 12,000 lumen which is totally doable. No need for liquid cooling.
I’ve done research on all the really big lights to see what is the highest sustainable output with no moving air across them. Typically this involves a really large light with a lot of surface area to dissapate the heat and stabilize the temp. The Imalent DX80 is the current leader at about 6,000 lumen.
With active cooling you can easily have a sustained 10K to 15K lumen. These lights already exist. Even 30k lumen might be sustainable, at least in theory. If your interested, I’ve got a thread on actively cooled lights here.
To do it right is probably more expensive than manufacturers want to spend. The Acebeam X65 uses 5 xhp35-HI, but costs about $400. I don’t know what driver it uses, but maybe a buck type to keep the output carefully controlled.
You want to build a cheaper version? I think the Q8 mcpcb and carrier system will work, but the big issue is the driver. You might be able to carefully adjust the resistance in the total light to be able to use a cheap FET driver, like the Haikelite MT09R, but it’s tricky.
A proper Buck driver would be better, but might be expensive. I see Lexel makes a buck driver capable of 8A (good for running 4 leds at 2A each). He has something for the Q8, but it’s for the original battery tube, not the new carrier design.
He can probably tweak his pcb design so you can transfer the ring from the original driver to his new driver and be good to go. I don’t know the details, but it certainly looks possible. Link to his Buck drivers.
Output would probably be around 6,500 to 7,000 lumen. So not too much more than stock. The xhp35 also has a larger die, so it probably won’t throw any farther than the xpl-hi version.
I’m not seeing a big step up in brightness or throw. Maybe just a small step up. Are you sure it’s worth the trouble and expense?
Same/similar throw, larger hotspot, regulated buck driver (at least to ~3.4V?), 3K more lumens. The benefits are probably even greater if you compare the two lights for a longer amount of run time and not just a snapshot with topped off batteries.
Pretty good if you ask me and it’s probably not that much “trouble” but it certainly will be expensive compared to how much the stock light costs.
4 series batteries and 4 parallel leds in the case of 4 xhp 35.2’s.
(i’m thinking along the lines of less amps drawn at the battery end, and less amps generating heat at the heatsink for the same output as the 6v versions).
With the sofirn and the extended tube/cage, you could have a 4 series battery cage that would double up. As a single the usual 4 cells positive to negative, and to convert to 2 cages for a ‘parallel’ set up, reverse 2 cells in each cage for 4S2P voltages.
which ever 12v led used, it would just be a matter of the wattage in heat to sink for either a monster few seconds lamp or less led’s for a longer use flashlight.
(having parallel batteries would up the posible amp draw a little?)
i’ll say at this point, i’m working on past experience with other non Atmel chip stuff, and finding if you up the amps the heat generated goes up very fast, verses up the volts and keeping the same amps doesn’t seem to generate much more heat at all…all sub 12v stuff though.
Your thinking the xhp35-HI does 2,025 OTF at 2A?
IDK, I usually think of them at about 1700 lm.
Looking at my notes (and digging out a calculator), I did measure my GT once using an L6 reflector and no lens at 2A on my TA Lumen Tube and got 1,860 lm @30s. So that times 4 would be 7440 lumen.
Then I also measured my Sofirn Q8 at 4,330 lumen @30s.
That’s an extra 3,110 lumen or a 72% increase in output! That’s much better than what I was thinking in my head. Maybe this is a good idea. :+1:
Correct, lumen numbers are probably high cause I used a little bit of a fudge factor but relative to each other it should be about right like you found. However, I think higher flux bins exist than what was used in the GT and like Tom hinted at the 35.2’s do have a very slightly lower vF which helps.
the trouble and exspence? thats what i’m looking into.
i thought the xhp35.2 and the xpl hi were the same size foot print…my bad, that and was just looking luminums and amp draw on a cree chart.
i’m looking to extend burn time basically… i can be out for 3 days up to 5 days a time, so if i can get 3 or 4 times the run time for a few less luminum’s i can take that, but obviously not too many.
i’m basically aiming for 3000 luminum with throw at less battery drain, with a boost to the 5/6000. I have a haikerlite MT07s with an xhp 70.2 and at ‘hi’ setting it states 2300 lums, turbo being 5000lm’s, the wide beam is useful to me, but a bit less wouldn’t hurt, which is why i was thinking of the XHP 35.2’s (asuming the 35.2’s are more throwey than the 70.2’s). From there i was thinking 2 or more xhp35.2’s would out throw and out last in burn time than 1 xhp70.2 running in 6V mode @4.8amps.
in short the 70.2 running at 12v 2800mah verses 2x 35.2’s running at a 2 amp parallel battery draw… would the two 35.2’s out throw the one single 70.2? if so staging it up to 3 or 4 xhp35.2’s verses the 4 xpl hi’s at 8 amp battery draw is it?
Wattage is amps x voltage so the heat at the heatsink will be the same. There is no 6v version of the xhp35. Are you thinking of the 35.2, that there might be a 6v version of it? Same rules apply. The heat will be the same.
There is no need for this. Just make all your carriers 4S. They are already designed to stack in parallel to keep the voltage the same. So you can add another 4S carrier and get 4S2P. Add a 3rd carrier to get 4S3P, etc… The voltage stays the same, but you gain capacity and the ability to draw more amps.
A xhp35 at 12v and 2A generates 24 watts.
At 6v and 4A it generates 24 watts for the same lumens.
If you lower the voltage to 6v and keep the original 2A, then of course it will run cooler, because it is only generating 12 watts. So there is no free meal to be found on the led side of things.
Of course, the higher voltage and lower amps is easier on the driver components, pcb’s, springs, etc…
If you are happy with the heat sinking of the current Q8 then you dont need to worry. If you change to a buck driver the light might even be cooler in the “medium” mode groups. However, like Jason explained. The power is similar and ultimately has to go somewhere.
If you are not happy with the Q8 heat sinking then you need to choose another (bigger) light cause no amount of tweaks will make any meaningful difference.
The foot print size is the same, 35x35, but the die size is what I mentioned. This is the size of the yellow square where the light comes from.
The xpl-hi die is 1.91mm x 1.91mm
The xhp35-HI die is 2.35mm x 2.35mm
It is the ratio of die size to reflector diameter that creates the hot spot size. (I’m being very general and non specific here as it’s more complex than this) Generally speaking, a big reflector and small die give you throwy lights with small hot spots. Small reflectors and big dies create floody lights and big hot spots.
So when you compare the xpl-hi to the xhp35-HI you get a bigger die which creates a bigger hot spot and this kind of reduces the gains you get from the extra lumens. So I would guess the throw distance of each light would be similar. The xhp35-HI would have a bigger hot spot which can be helpful, but probably not reach any further.
I will have to think about this part some more. Maybe someone can think of the right light for you in the mean time.
with the older stuff i was using, the amps create more heat than the volts do for the same figure of watts, but, non of the ebay stuff had atmels etc, all pwm, and current control with LM regs.(burned a few fingers with the mosfet’s version of watts heat lol)
missed your reply Tom, it was at the very top of the page as i was darting about with other posts appearing before mine lol…sy.
is there a 12v driver that would cope with 4 Series batteries, with the atmel included, i have to admit i do like the programmable aspect quite a lot.
…………………
that aside, i have just checked my MT07s with the xhp 70.2, it is the 12v version, and running on hi for 2300lumins it soon gets to 40 degrees after 8 ish minutes, at 10-15 minutes its hitting 58degree’s C, to hot to hold and the silicone switch starts feeling a bit tacky.(i havent hit the ramp down on booste yet though, i think thats set at 85C).
i have another zoom flashlight that has an xhp50 in it, i think its a chinese version, takes 2600mah at the battery end (2s1p) on hi and never gets past warm.
the heads a 50 ish mm big focusing lens with 4 or 5 short fins on the outer. no reflecter, shelved.
so maybe 4x 35.2’s would be feezable heat wise in the Q8
