Extremum - 550 lumen from keychain flashlight

Hi. Good questions. Let me explain.
First - we need about 800 ma to drive XML2 to 550 lumens cause we’ve chosen a highest bin - U3.
3.7V*0,8A = ~ 3 watt (in fact we charge battery till 4.2V). Cree Xml-l2 U3 deliver about 150 lm @ 1 watt. But we have a focusing lens. So we refocus it’s from 170 degrees to about 80 and make bright almost uniform wide beem and add aditional lumens to the center spot. Please look at photos.

Second - led is mounted on thick aluminium pcb that works as a radiator (wtith a help of case of course).
Third - we’ve tried turbo mode till the discharge of a battery. It’s about 4.5 minutes of full pwer and ~ 3 minutes of still very bright light that is starting to fade. After 8 minutes it’s about 200 lumens left. 0-4.5 minutes is a full brightness. Highest temperature on the led during that experiment is 70 C while it is outdoors 25C

We tested cell 10 times and after that it has about 90% of life. We think after 20 sessions in turbo mode it’ll have only 50% of capacitance.
We’re honest about that - if you want such light from keychain you need to find a compromise.

s. Thanks for good technical questions!

s. We’ll show real world comparison and charger soon!

But why not an U4 bin!?

Eh… actually… according to CREE, you need ~1125mA to get 550lm out of XM-L2 U3. With losses in the lens, you’ll need probably close to ~1200mA to get 550lm out the front.

Did you test your light in an integrating sphere to see what the actual OTF light output is?

Eh? When we started that project U3 was the highest we could find. U4 already exists? :wink:

But don’t forget that we re-focus 130 degree light in a much more narrow beam. Without the lens Xm-l2 is 130 degrees width , with lens and reflector - ~80. Yes, you loose output on lens, but earn on focusing peripheral light to center.

Here

U4 exists and is sold here in individual quantities. 1A is a very cool white tint around 6500K and not what most people around here prefer. I have one of these, dedomed (which warms up the color temperature considerably) in a Fenix TK61.

I suppose if you’re chasing numbers, you should probably go for this. You might get another 30 lumens out of it. I don’t find chasing numbers you need instruments to detect all that entertaining anymore and would rather have something with good tint and color rendering.

I agree.
But this would mean go for the XP footprint (3535) in stead of the XM footprint (5050) so you can reflow a nichia 219C on it, or a XP-L2 V6 for that matter (preferably 3A or 3D tint) if you want high output.

I don’t want to sound judging, really, but it seems to me that you mixed up lumens and candelas…
Lumens are lumens, it’s a quantity of light…you can refocus as much as you want, 100 lumens will not jump to 200 lumens without feeding the LED with more current.
However, when you focus the light from 130 degrees to 80 degress, yes you are bumping the intensity of your beam, so candelas…

Good project for kickstarter, not for blf.

Seems that smb dont know the difference between lumen and lux. And also there are too many funny things in description:

In fact we see two cheap fr-4 xml boards with one 7135 based driver.

Test video

Iso 800, F 3.5. All flashlights are in equal conditions.

That’s a prototype. Serial units will be on aluminium pcbs. All pcbs on photos are our custom developments.

So your developers decided to make xml footprint board for switch :person_facepalming:

That’s a prototype! We didn’t want to order additional boards for a button if xml board suits it fine.

s. Are all Hohols are so angry? :wink:

So at 800 mA XM-L2 U3 will produce about 350 lumens, correct? Not as impressive then, IMO. Still interesting

But then focused to a 80 degree almost uniform beam it’s still impressive (at least we think so) - look at video.
We can squeeze more lumens, but we want that flashlight to be usable. Not a one second toy.

Are you going to ignore my remark about lumen and lux/candela ?

[quote=X3]

[quote=ProjectOne]

You’re right. Maybe we’ve used term ‘Lumen’ incorrectly. We’ll study that.

Come on…really??