Modern LED scrubber light

Thanks for the reply

"More sane would probably be a handful of 1W LEDs spread around heatsink."

More sane maybe but also a lot more work and to some extent cost per lumen.

I have yet to install but the 10 watt warm led I just got is blinding. (same construction as the 100watt) I am curious to see what a light meter would read.

Could possibly go with something not so strong (50 watt) ???

Could anyone give an opinion about how far a 100 watt or 50 watt should be place away from a screen?

Would not some distance solve some problems with the light being to intense in any one area?

Move it back until it covers the whole screen and try it.

Quote:

---

Originally Posted by inkidu

I got 4 of these heat sinks for $64 shipped off of ebay. Aluminum Heat Sink 10"x7"x3"
Do you think a heat sink that size could handle a 100 watt led?

---

Probably.
A quick check on a standard 9x6x2 heat sink = 0.5 degC/Watt without fan.
Unfortunately, I don't know the exact efficiency of Light/Heat for those big LEDs. Lets guess at a fairly standard
70% heat, 30% light.
I also don't know the package thermal characteristics. Lets guess at 0.75 degC/Watt.
So you need to dissipate 70W. 70*1.25 = 87 degrees C.
Assuming you have 80 degF ambient around fishtank, that is 26 deg C.
So you are at 26 + 87 = 113
You absolutely need to keep them below 135, or they start to fail, so some margin there.
But you will lose a lot of efficiency and longevity as things heat up.
I strongly suggest a fan.

Also note: In researching this, I found two really interesting things of note.
1) Due to the way they are built and thermal issues, expect a 50% reduction in efficiency from a single LED.
Remember, they are specified cold, but running very hot.
2) The output light is not as dispersed, and very intense. It can cause cloth to burn and plastic to melt!

Consider a water-cooled version

Thanks for the reply. Well I just hooked up the 10 watt red and 10 watt warm to one of the heat sinks that had 6 rebels on them and after a few hours touching

the integrated heatsink of the led its seems to be pretty cool although my house is fairly cool. Kind of subjective but I am guessing something like body

temperature. Thats making me happy as well I am now starting to see a fair amount of algae growing after 12 days. I wish my camera was not broken because

there seems to be quite assortment of different algae ( some green and brown hair) Mainly because starting from my inlet the water passes under a 10

watt warm 10 watt red, then a cluster of 5 rebel cool white and one red, and then some common phillips 2700 k cf which gradually transition to a mixed line of

3 rebel red, 2 rebel blues and a rebel cool white. Brighter green under the cool whites transitioning to darker brown towards the end. Some of the results so

far might also be effected by the flow i.e. approx. 400 gph over 6" wide and 36" length horizontal screen. The flow seems more turbulent in the beginning. I am

also missing the baffles that came with the scrubber that increased the flow through the screen. I am working on that. Still a way to go to determine what will

work with my setup but I feeling better that I am starting to see results. Eventually I will add a fan I agree with rygh the heat sink will work many times better

with one. I am running out of places to plug things in. As far as efficiency I am trying to do some math too. I am far from believing all the limited specs from

ebay but this is what is I get from the 100 watt warm white led. (The way I am trying to compare efficiency is not conventional or maybe

not even right because I use forward voltage multiplied by current.)



forward voltage of 36 * 3.5 amps = 126 watts which supposedly put out 6000 lumen

6000 lumen / 126 watts = 47.6 lumen per watt

As a comparison a rebel warm typical forward voltage 3.4 * .7 amps = 2.38 watts which puts

out 110 lumen at .7 amps

110 / 2.38 = 46.2 lumen per watt

another comparison a cree warm forward voltage 3.4 * .7 amps = 2.38 watts which puts out 130 lumen ( my approx. its 93.9 at 350 mA)

130/2.38 = 54.6 lumen per watt

There must be a different way of calculating these #'s because I do see things like 160 lumen per watt efficiency for a cree ( the cool white are brighter 122

lumen compared to 93.9 at .35 amps but nothing that would make that much difference.)

I must be comparing something different. Like I said this is just how I am comparing them. After a little reading, wilkpedia, they do seem to use a much different equation.

http://upload.wikimedia.org/math/8/d/d/ ... ebf141.png

Not that its terrible important seems that this could be a fair way to compare different leds. Not sure????

power in light out

Anyone weigh in?

Yes, lumens/watt is a VERY good way of getting efficiency. As is lumens/dollar.
And you seem to be calculating it right.
I think you used the 350mA luminance values for 700mA current in one spot though. Not sure.

Cree Q5, XR-E Star, Cool white:
107 lm@350 mA
3.3Vf @ 350 mA
So power = Current * Voltage = 350mA * 3.3V = 1.155W
Thus power efficiency of Lum/W = 107 lm / 1.155W = 92.64 lm/W

Now to save cost, you will probably end up running at 700mA, which does drops it down a bit.
And they are a bit sneaky in not reporting all 700mA numbers in their data sheet.
But it is pretty close to double, and Vf only goes from 3.3 to 3.5.

Thanks for the reply I agree rygh Cree for some reason does do some odd things with there specs.

This is some more info I found. Its validity like any thing on the internet should be ? here is some more math

(not warm white they produce less lumen)

- True Cree Q5 bin emitter
- Cree Flux/Color BIN: Q5 WG
- 3.7V typical driving voltage
- Manufacturer Rated 228 Lumens at 1000mA input current:
350mA: 107~114lm
700mA: 171.2~182.4lm
1000mA: 214 ~ 228lm

3.5 * .7 = 2.45 watt

176 / 2.45 = 71.8 lumen per watt
avg.

3.7 * 1000 mA = 3.7 watt

221 / 3.7 = 59.7 lumen per watt
avg.

As you increase your current(amps) through the led the forward voltage increases ( thats why I used 3.5 for .7 and 3.7 for 1 amp

although the specs don't clearly state what current the 3.7 forward voltage pertains too. (must likely 1000 mA 3.7 is pretty high)

There is some variability to begin with. Usually given as a min, typ, amd max #'s. I personally find rebels specs a little better.


Something I want to point out, as you try to run leds like a rebel or cree xr-e at higher currents they start to be less efficient.

There is also a dramatic loss of over all life with the higher currents and the corresponding temps.

So instead of essential overvolting these would not undervolting a 10, 20, 50, or even a 100watt led make more sense?

I am far from an expert and I would appreciate anyones opinions.

Cree Vf:
350mA=3.3V
700mA=3.5V
1000mA=3.7V

Yes, efficiency drops significantly the higher the drive.
BTW: There is an extra factor that is not in the datasheet.
- The higher you drive it, the hotter it gets. The hotter it gets, the lower the output. But lumens are all specd at the same temp.

So driving at 350mA versus 700mA is far better. --- except for that pesky issue of ending up at double the
cost due to purchasing twice as many LEDs.

Datasheets I usually use: ledsupply.com/docs/cree-xre.pdf

On the 100W LEDs: These are not really a single 100W diode. They are an array of diodes. Maybe 7x7 each at 2W. Not sure on detail.
There is an interesting failure mode that they don't mention much.
In an array, in row dimension, they depend on the LEDs being all about the same, and sharing current evenly.
As they deteriorate over time, the LEDs start changing Vf a bit, and current is less shared.
That can end up over driving other rows.
Also, if any cell completely dies, that row stops drawing current, and that current goes into the other rows.
You can get a cascade effect.

At any rate, it can be analyzed endlessly. But then, analyzing it is the fun part.
- Light efficiency = Lumens/Watt. Highest Q, lower current to maximize.
- Cost efficiency = Lumens/Dollars. Upper Q, higher current to maximize.
- Overall build hassle: Generally the fewer the LEDs the better, until fancy heat sinks required.
- Longevity: Top brand name, keep temperature way down, keep current well below max.
- Spectrum control: More LEDs mean you can pick and choose colors. (Affects "useful" lumens)

Thanks for the reply

I count 10 x 10

I am thinking there must be some cost savings because the manufacture doesn't need to cut them all up

and can keep them all in one package. I assume the downside would be you get some diodes that are better than others.

Such is life. The good with the bad.

I might buy a higher watt one once I can see how the 10 watt led I have does. I figure its worth a try.

Thanks for the help.

Inkidu

Make sure you look up the chromacity group (e.g. WG, WC) on the cree datasheets. The binning on the leds can be a very big range - 5000K to 10000K from memory. One of the graphs in the datasheet shows where these sit if you know the grouping. I'm not qualified to comment on how this may affect performance in a scrubber or main tank but it should certainly affect aesthetics just as a 5000k tube would look substantially different to a 10K one.

edit: If you are looking for efficiency and are prepared to do the work to connect up multiple leds you may find some deals on the cree xp-e, xp-c or xp-g chips. Get them pre-mounted as they are tiny (about 3mm square). Some hit more than 130 lumens per watt and I have seen pre-mounted verrsions giving abouto 70 lumens per euro in terms of cost efficiency.