Random idea for multi-die LED flashlights

An EDC flashlight with a 9 or 21 die configuration similar to this:

-With the center die on max and everything else off it would become a long distance thrower with a tight beam

-With all the dies on low power it would have high efficiency since each die would run at low current

-With all the dies on high power it would put out a ton of lumens

Of course there is currently no good LED for this application, but imagine something like an XHP70.2 with 9 instead of 4 dies, an the center one can be independently run.
By running a single die it could get very high intensity for long throw, but you have the option of turning on the rest for high efficiency or high lumen output.


Note: this isn’t the same as a triple or quad led flashlight like you commonly see around, since all the dies are together they would use a single optic (resulting in more throw than multiple-optic lights) and would also have a single spot when all dies are on (assuming they are close enough together)

This is a very interesting idea.

You would need a massive reflector though… which is why this is an actually excellent idea! Massive reflector with a 6mmx6mm multi-die emitter with multi current path.

Having dual channel drivers with different output voltages is difficult, but doable with multi-inductor drivers:

With a 2S buck boost converter support, you could power the separate 3V emitter up to 10W, then for higher output levels, all of the emitters would be connected in series through another trace, which would result in a massive 9000 lumens output in exchange for a lot less throw.

However, that idea is very complex, and would require a complex costly MCPB, and a costly driver. Would not be too hard on the MCU side of things since we already have multichannel capable UIs.

There are 2 simple ways to make it more bulky, but less complex:

Go the Niwalker way and put 2 separate sets of emitters and reflectors. While bulky, it allows for separate flood and throw, and a real sense of power when both are on.

Multi lens aspheric. Take a dedomed XHP 70.2/Nichia 149AM then put it in either flood or throw mode. You do lose the max throw and flood power though.

By EDC, how big overall would it be?

IMO an EDC is either a 2 x AA, a 1 x AA/14500, or a single 18650 to at most a single 26650 - and weight wise the 26650 is definitely pushing it.

Point being if it’s truly light and small enough for practical EDCing to push this to a reasonable meeting of potential output you’d need a healthy (boost?) driver, no?

But I do like this interesting idear. When are you gonna make a prototype? :sunglasses:

If it was something like 9x 1mm dies having it run off of a single 18650 or 21700/26650 would work fine.
Good single cell EDC size, doesn’t need a big optic either if all the dies are close together.
There are several EDC flashlights with XHP50s and 70s.

It could be like maybe the size of 9 oslon black flats or something.
A single one could run at 4-5 amps for crazy throw, but you can run 9 of them at like half power and get 4000lm+

9 XP-L HIs would also be able to do over 1000lm at like 180lm/w, very high efficiency.

I like it. I’ve been thinking about emitter arrays lately, and the interesting opportunities they present in power/thermal limited situations.

Consider also a mix of color temps and/or CRI. Could have one type of emitters at the corners, another between the corners, and yet another in the center.

Wire it up, Endy. :student:

Easy for moi to say. :laughing:

If anybody could do it, you can. In your spare time of course. :open_mouth:

Osram actually makes LEDs like that, but only 4 dies not 9:

0 spare time :frowning:
Trying to do too many things at once and getting almost none of it done.
I need to focus on a few things at a time, so after I finish building the second syniosbeam with CFT90 I will probably take a break from flashlights and focus on my other stuff.
Next year, 2019, I’ll start the next flashlight.

We discussed that before XD16 came out. Not terribly hard to do, but XD16 is inefficient. E21A is efficient, but good only for flooders as it’s too thermally limited.

Yeah I saw what easyb (I think it was him) made with 9 of them.

The difference here would be that having all the dies on a single large chip would have really good thermal performance when only one is lit up due to the larger pad on the bottom, as can be seen from the large luminus LEDs.

Then with some high efficiency dies, like XHP70.2, lots of lumens and efficiency could also be obtained.

I’m pretty sure we will never see an LED like this from cree, but it would be cool for EDC flashlights to have both throw and high output+efficiency options (without having multiple optics or LEDs or other complex stuff).
Simply by turning on different sections of the LED.

9 XP sized leds….not with a reflecotor but with aspheric? Costly, but otherwise sounds good.

Yeah, something like that :slight_smile:
Switching between 1 and 9 dies will go between throw and output, and focusing with the lens will switch between spill or tight beam.

This way you can have maximum throw, medium throw with large high lumen spot, or high lumen flood.

It should be even possible to mix different LEDs.
CSLPM1.TG in the middle, LH351D on the rim?
Or, for those who can’t stand cool white, XP-L + LH351D.

How would it work with a wide-open collar?

It is still a prototype but Osram is working on a 1x1 mm high power led that could be used for this, it just needs a 9-up MCPCB.

Made me also think of this: Osram is working on a 1000 ‘pixel’ led that can be individually controlled. It will however not have the lumen density to make it throw very well.

Had the exactly same idea 1.5~2 years ago, but problem is "four-nut" hole with 9x,25x etc. discrete LED arrays. To eliminate hole(s) op reflector must be used, which reduces throw significantly.

With LED like 9-die XHP-35 HI this would work with SMO reflector.

Yes, it definitely needs to be a single LED, not 9 or more LEDs soldered beside eachother.
9 die XHP35 is a perfect example, since the current one is 4 and has no visible cross.
The main feature would be turning on the center die individually.

We will likely never see an LED like this produced, but I thought it would be good to write the idea down somewhere so it’s on record :slight_smile:

Interesting idea.

And the LEDs don’t have to all be the same. Put a very high intensity LED in the center for max throw. Surround it with bigger-die, lower-intensity emitters that can be activated for max lumens and flood.

Could make a great searchlight. I don’t think this would be practical for an EDC though unless you’re just making a mule with no reflector.

If EDC flashlights can fit an XHP70.2, then there wouldn’t be any problem with a 9 die XHP35 in the same light size :slight_smile:

Bare tiny chips are widely used in automotive industry last 5 or 6 years. Just few problems:

- Osram wont sell them if you are not working for BMW

  • You need a small factory to mount this chips on your own substrate
    And yes you wont find many information about this products on osram website.

Some further thoughts on doing it with aspheric and collar….
Combining multi-led idea with collared zoomie enables making a light of the size of Brinyte B158 which delivers 500+ kcd throw and 5+ klm flood. The former sustained, the latter being turbo.

But aside from the regular challenges with collared light (patents….), there are also new ones.

If we want wide beam, we need wide collar opening. Also, collar won’t focus on the flood leds, reflecting their light partially on other LEDs (which is good as it enables recycling) and partially on the PCB.

If we want wide opening there are several options:

  • make a wide-open collar with mounting fixed against the LED
    • in throw mode, lots of light escapes the collar reducing throw
  • make a collar with large diameter, moving together with the lens
    • the light loss is at high angles, this is less of a problem than with fixed collar
    • collar comes out of focus quickly, making partially focused beam much less useful
    • complicates head construction
  • combine fixed and moving partial-collars
    • complex, costly
    • restricts zooming movement - they’ll hit each other quite early
    • less of a problem with stronger lenses
  • do Fresnel style collar
    • complex
    • is it reasonably manufacturable?
    • light collection will be good, but precision will suffer, reducing efficiency too

If we want to reduce collar-induced losses, white PCB comes immediately to mind.
But that produces ringy throw.

Kiriba_ru had the same thought as me. Osram offer an automotive LED which basically has “pixels”. If you use an aspheric lens you basically have a projector (finally making the bat logo possible whithout a lot of effort on the optics front).