Help! (scientific experiment) Cheapest/easiest way of 150 watts per square foot

Help! Find a way to cut the cost of this as much as possible.
Cheapest way to do a 1 to 2 square feet array of LED's.

I am doing some scientific experiments that require me build a very bright, thin, even backlight strobe illumination device.
I will be needing the following:

- Even lighting. (Diffuser cover)
- Spread over two square foot
- Very shallow depth (2" or less, thinner the better).
- Can't use few light bulbs, need even lighting over many LED's
- As much wattage as possible
- At least 150 watts minimum per square feet
- Strobe speed capability (1/10,000second illumination)
- Other forms of lights (CFL) is too slow. LED is fast enough.
- I need row illumination capability, light up one row at a time. Need at least 16 independent rows
- approx 6500K (+/- 500K)
- approx 80 CRI
- I will build the circuit myself
- I will build the enclosure myself
- I don't need column illumination capability (don't need expensive dot matrix)

How to do this as cheaply as possible, without too much excess soldering?

The cheapest "easy" method I've discovered so far:


Search keyword "LED ribbon". These are found cheaply off places like Alibaba, eBay, Google that goes as little as 25 cents a watt (On eBay, they are available for as little as $11 for a 5 meter roll of 600 LED 50 watts). I've since found that these are available in 6mm widths, and I've found out I can squeeze 240 watts in a rectangle that is 288 millimeters by 500 millimeters. The 6mm width of ribbon, 120 LED per meter, 10 watts per meter, means I can squeeze 48 rows of 5 watt segments (0.5 meter) for a total of 240 watts. That requires 24 meters of ribbon, or about 5 reels of 5 meters, and each reel is $11 each off eBay, for a total price of $55. But CRI of these are only 65 to 70 for many of these.
However, my main catch is that I need neutral lighting, which is 6500K and 80 CRI. For some sources, that dramatically raises the price by 4x or more by some places.

Know the cheapest way to solve this scientific problem, while still getting 6500K (+/- 500K) and 80+ CRI?

For your application you could use far less high power leds like XM-L for example and overdrive them way beyond their max current as long as you need them to be turned on 1/10000s. So about 10 XM-Ls at 5 A would suck approx 150+ watts and would cost you about 80$. But you will have troubles finding 80+ CRI at 6000K and cheap.

I'm not sure if you can even get 6500K and 80+ CRI with normal LEDs. Maybe if you mix and match really cool white LEDs with warm white ones and some red LEDs. The highest I've seen it Osram's Golden Dragon + with 80 CRI at 5000K temp. Are you looking for 150w light output or to drive the LEDs at 150w?

I did a quick calculation and it takes ~38 top bin cool white XM-L @ 3A to hit 150w light output, or ~51 XM-Ls for their 80 CRI 4700K LEDs.

Not sure why we're talking in terms of wattage per foot when XM-Ls are likely throwing double if not more lumens per watt than chinese strip LEDs. What is lumen requirement here?

Rebels are easily available in 5000k 80-85 CRI bins. Cree's should be, but the bins seem trickier to find in higher color temps.

Unless you drive XM-Ls over 2 amps they are an expensive lumen source, albeit an efficient one.

You may want to do some research into the time constant of phosphors on white LEDs. If you need 1/10000 on/off, white LEDs are not going to cut it. You will need color LEDs.

Semiman

I need to edit what I said about the cost of the XM-Ls. I'm seeing the prices drop a lot in the past couple of months, and with T5 neutrals going for $6.50 retail they are no longer expensive (relative to other premium LEDs).

Just remembered that Luminus has 90 CRI LEDs at 5700K color temp. They are $40 each before shipping though (on mouser, avnet has them for cheaper it seems).

Thanks for the useful answers here. You've given me some scientific variables I need to be concerned about. Some extra comments:

- Good point about time-constant of white phosphor of LED's. I had covered that already. On that suspicion, I tested a photodiode with an oscilloscope on single white LED's, and found that small SMT 3528 / 5050 LED chips use very short persistence phosphors -- fast enough for 1/10,000sec strobes. There is still some glow, but not worrisome. I do need the phosphor afterglow to be less than 1% after approximately 1/2000sec (not set in stone). An array of unfiltered color LED's is harder to diffuse into an even white light. I do observe that raising CRI also often raises phosphor afterglow.

- If necessary, some CRI can be given up a little (e.g. CRI ~75), if it means I save hundreds of dollars. Most computer monitor LED backlights are worse than that, anyway. I don't have the equipment to measure exact CRI's, but I don't need to be ultra-precise about it. Some specifications I have seen, report CRI of 80 or better, but those could be inflated numbers. The higher the better, but uniformity *and* maximum wattage is more important.

- If I use fewer, but larger, LED's, I need to diffuse them extremely well for near-perfect uniformity over 1-to-2 square foot. Non-uniformity not perceptible to human eye, when covered with a good diffuser. This is difficult to do in a shallow depth, but I'm considering my options.