HELP/w high lumens LEDs

240z

Newly Enlightened
Joined
Jan 16, 2010
Messages
5
I know nothing about leds, can anyone help?!! I started my research today and I am still lost on most of the terms. I believe i am a fast learner, so any direction all be most helpful. Here are my frst set of questions.

1. Is there an LED that can produce 120lm/w or brighter (i am assuming LM = Lumens and /w = Per Watt)

2. How can i make a panel of LEDs lights that produce about 17,500 lumens?

3. Can i use 60 8mm leds at produce 0.5watts, 70k mcd, at a 140 degree viewing angle?

Lets start with these questions....

THANK YO VERY MUCH ANYONE WHO HELPS!!
 
A string of very efficient LEDs which produce 17,500 lumens would not make much sense since they are only very efficient when driven at 350mA. To get this kind of output with a good efficiency and a reasonable cost you should look into high wattage HIDs. They have the added benefit of becoming more efficient at higher wattages. In addition to this the logistics needed to operate one or two HIDs are much less complicated then trying to power hundreds of power LEDs.

But I would still find it fascinating if you tried to achieve this using LEDs. :twothumbs
 
In addition to this the logistics needed to operate one or two HIDs are much less complicated then trying to power hundreds of power LEDs.

But I would still find it fascinating if you tried to achieve this using LEDs. :twothumbs

Not so much really, around 110-150 XP-G leds. As they are very very small, if run at 350mA they could be soldered nearly side by side. So, 17 500lm light wont be that huge.
 
8 mm LEDs give you 3-15 lumens each. 60 of them will give you 180-900 lumens, a lot less than 17,500 lumens. You will probably need over a thousand LEDs.

A problem with 3 mm/5 mm/8 mm/10 mm LEDs especially when clustered together is that there is no way to heat sink them to remove heat. So instead of 100,000 hours they often only last a couple of weeks.

To get 17,500 lumens you need 12+ 100 watt light bulbs. Allowing for a 4X improvement in efficiency for LEDs you still have to get rid of 300 watts of heat.
Incandescent bulbs run at 2700-3500°K. ~2400-3300°C. Can withstand high temperature.
LEDs die at 120°C.
Capacitors in your power supply die at 70°C.
The Evolux LED bulb at 1000 lumens already has a built in fan to cool it. To get 17,500 lumens you need proper engineering - optical, electrical, thermodynamics etc. It is no longer a simple matter of wiring a few LEDs together.
 
Couldnt i put 2 computer fans in the panel to cool the leds?

Also i though mcd and viewing angle produced lumens. Was i mislead?
how did u get 3-15 lumens from a 8mm led with 70k mcd and 140 degree viewing angel?

Sorry for my ignorance i just still do not understand leds yet. plz explain.
 
Custom high power LED panel

I want to make a custom led panel that will produce 17,500 lumens, wthout being too expensive.

1. Is this possible?

2. Where do i start?
 
Looks like stadium or huge car park. Its not THAT much actually :)

Pumping out 17,500 lumens from LEDs is quite a bit. Keep in mind that the footprint for the XP-G is tiny, 3.45 x 3.45mm. Thats a lot of light to come out of those little guys.

Not so much really, around 110-150 XP-G leds. As they are very very small, if run at 350mA they could be soldered nearly side by side. So, 17 500lm light wont be that huge.

About 51 XP-Gs @ 345 lumens (1 amp each) per LED will produce around that much (calculates to 17,595). 130-150 driven @ 350 mA would also do the trick. That cranks up the price though. About 20 SSC P7's would do it too, but they wouldn't be as energy efficient.

Also, In my experience, the closer together, the worse the heat management. I don't know how much of a concern this would be on the smaller XP-G/XP-E size emitters, but I know that the P7's have problems being too close to each other when cranked high.

Removing the heat on that amount of light output could prove to be a challenge though.

:thumbsup:
 
Using either of the light suggested like the x-g or p7. How would i power them?

Couldn't i use a computer fan to cool them if they are placed in a panel?
 
Using either of the light suggested like the x-g or p7. How would i power them?

Couldn't i use a computer fan to cool them if they are placed in a panel?

You can power them using constant current power supply, wich can be bought in many online shops. If you need, let us know, we will give you some names.
Yes, you can, but you still need to solder them onto PCB, then use some thermal compound to attach PCB to heatsink(can be bought online or on ebay).
 
I just want to chime in about the themal solution:
Cooling leds is just difficult if you want to do it in a flashlight envelope.

As soon as you are stationary, and able to use active cooling, 300W isnt that challanging at all. One cheap way would be putting all those led on a thich enough aluminium plate (lets say 5-10mm), which provides mechanical stability and heat spreading.
And then just put a few cheap CPU coolers you can get for $10 an the backside.

High performance alternative:
The baseplate of a _good_ heatpipe cpu cooler is usually 40*40mm, and very very nicely polished copper.
Those are able to draw away >150W while keeping a CPU below 60C or something, so 100C+ juntion temperature are not that challanging.
You could put A LOT of XP-Gs directly on such a cooler if you just fill the baseplate...

(just if you dont know about them, take a look here : http://www.guru3d.com/article/heatpipe-cpu-cooler-roundup-review/). Because of high volume production, they are rather cheap compared to other, more specialized solutions.
 
Now this is what i am looking for 2 great ideas. I can actually get my hand on a large piece of diamond plate aluminum found in cars.
I can also pull a couple cpu heat sinks from these old computers i have.

how many leds do u think i can fit on one cpu heat sink and fan? 4?
 
Re: Custom high power LED panel

You don't get power like that from LED without it being costly. I'd look at HID or traditional Incandescents for that sort of light personally.

In terms of possible - yes, it's possible.

Where to start - probably by looking at LEDs like the SST-90, Cree MC-E or SSC P7. You've got a cost per light / area to be illuminated / colour preference budgeting job to do. You'll also need some serious heat sinking for that lot.

You'll then want drivers for them. Just search for drivers on here - there are lots of threads and many custom made that you wont find elsewhere.
 
Re: Custom high power LED panel

To do that, you'll need 50 XP-G LED's, each running at 1 Amp. Expensive, bulky (massive thermal mass, and tons of fins, bordering on a need for active cooling), and you're going to need multiple high output drivers, which will further increase energy and cooling requirements.

In short, I would take a look at HID or incandescent.

Also, what's your intended usage?
 
Cree XP-G, then I would say 5 x 5 = 25

You underestimate how small xp-gs are. I got some 3 XP-G stars, and thexy consists of mostly empty space.

I just did check at my scythe fan, the base-plate is 38*38mm (the lapped area).

A cree has is about 3.5mm*3.5mm big. So would be able to fit 11*11 on the backside.
If driven at 350mA, it would be able to deal with that load. 1A though, not so much...

But I just realized that there is a tiny problem with XP-Gs for mounting directly on a metal surface: You would need to disconnect that contact pannels on the buttom of the LEDs, or use a custom MC-PCB.
 
how did u get 3-15 lumens from a 8mm ledl?
The early LED dies from the '90s from Japan's Nichia are 1/4 mm*1/4 mm and generate about 3 lumens. The Chinese copied them but without their own R&D was not able to improve on them much.
When people need red/green or red/yellow/green indicator lights they put 2 or 3 different colour dies into 1 LED. The Chinese copied that idea except using all white LEDs giving more lumens. Small improvements in LED and overdriving to produce more output makes 15 lumens available. To get rid of the extra heat and make assembly easier the makers increased the outside case to 8 or 10 mm. So a 5 mm or 8 mm or 10 mm LED may have 1 to 3 dies. Depending on the number of dies output varies from 3 to 15 lumens.
Also the early Arc AAA with Nichia BS has been measured at 3 lumens and multi-die SMJLED has been measured at 15 lumens.

In the early '00s Lumileds made a 1 mm*1 mm die LED called the Luxeon. This has 16X the area and can produce 30 lumens at 1 watt, up to 42 lumens at 350 mA. Further improvements brought the Luxeon III with 75 lumens at 700 mA 3 watts.

Cree came along and improved the LED design in the XR-E so much they got 80 lumens later 113 lumens at 350mA and over 200 lumens at 1A. 4 watts.

Cree increased the size of the die to 1.4 mm*1.4 mm or 32X the die size of the original 5 mm LED in the XP-G. With all their other improvements this new LED provides 139 lumens at 350 mA and 345 lumens at 1A. 4 watts

Before the XP-G came out Korea's SSC put 4 XR-E dies into one case and called it the P7. This produces 700-800 lumens at 2.8A. 10 watts
Cree also put 4 XR-E dies into one case and called it the MC-E.

Luminus Devices decided to push the big die concept.
Their SST-50 is 2 mm*2 mm similar to the SSC-P7 and MC-E but without the gaps between the dies.
Their SST-90 is 3 mm*3 mm or 144X the area of the early LEDs. It can produce 2250 lumens but you have to figure out how to to feed 9A to it safely. And get rid of 30 watts of heat.

The XP-G die can fit into an 8 mm LED no problem. So why does someone not do that and get 345 lumens out of an 8 mm LED instead of 15 lumens?
The problem is heat removal. The die is in the middle of the LED with an insulating air bubble on top and an insulating air bubble underneath. Only a small amount of heat can get out. Drive it harder and the LED will overheat and go :poof:.
With the Luxeon and later designs, the die is attached to the back of the case which can be attached to a larger heatsink. Now the heat can get out.

-----

70k mcd and 140 degree viewing angel is not enough info to calculate lumens. The problem is 70K is only true for 0°. We do not know how it drops off to the sides. Even if we do the integrating calculation is huge. Usually easier to send it to a lab with an integrating sphere to measure it.
 
Top