The Planted Tank Forum banner

1 - 20 of 23 Posts

·
Registered
Joined
·
1,121 Posts
Discussion Starter · #1 · (Edited)
Here is some data I got on two somewhat similar output LED combos that I wanted to test.

Here is the data, the story follows (I blew a driver in the process).

Data (NO optics used on any of these).


Single Cree XML T5 Cool White, 3,000mA -- roughly 10 watts

24"---8 micromols (sensor offset by 10" to the side, 7 micromols)
14"---25 micromols (offset by 10" to the side, 13 micromols)
7"----47 micromols (blew the driver before I could take offset numbers)


4 Satistronics "pure white" 3watt, 650mA-- roughly 9 watts

24"---3 micromols (10" offset, 1 micromol)
14"---9 micromols (10" offset, 5 micromol)
7"--- 20 micromols (no 10" offset taken, my pride too hurt from blowing the XML driver).



Anyhow, here are more detailed descriptions of the the two lights. Both were mounted to beefy aluminum heatsinks.

A single Cree XML T5, cool white, at approximately 2800-3000mA (I cannot find my multimeter so I can't say for sure). Driver rated for putting out 3,000mA and it's been tested by other folks to put out between 2800 and 3000. Total wattage about 10 watts.


4 Satistronics 3 watt "pure white" (looks identical to the eye to the cool white XML). Spaced with about 3/4" in between them in a square pattern. These again are on a driver I can't test without my multimeter, but it's rated to put out 650mA. Total wattage about 9w.

I blew the driver on the XML. It was a flashlight driver (3,000mA) that wasn't covered in any way--- lots of exposed conductive parts. I let it touch the heatsink, something went "zzzt" and it smelled burnt. No magic smoke though. Power for this driver was a 14.4v Makita drill battery, and it handled a 1 hour burn-in test just fine. Takes up to 18v input. Fortunately the XML driver was only about $5, which was worth it to get this data. But sucks for me because I planned on using that XML as a lantern on my next night fishing trip. :whaaa!:


Now this is what I want to demonstrate. A single Cree XML at a high current rate can give roughly double the PAR of some of these off-brand LEDs at a similar energy consumption.


Here is the take-away.

One Cree XML T5, cool white. $8
Four of these common import type, 3-watt LEDs--- $8. (at about half the PAR of the Cree XML!)

One Cree XML at 3,000mA means 1/4 of the LEDs to solder, 1/4 of the drilling/tapping, 1/4 of the thermal compound, and potentially less heatsink material.


Now. The hard part. There aren't many drivers commonly available for us that will run 3-10 Cree XML's at 2 or 3 amps. But they are out there, if they aren't out of stock.

There are a few Meanwells that do this, and Dealextreme.com has a PILE of drivers that can crank out that much current, but the problem is, they have HUGE voltage minimums, meaning you'd have to wire up a BUNCH of XML's.



.
 

·
Registered
Joined
·
1,985 Posts
I used O2Surplus' Driver design. It's nice because you can gang the drivers together to get whatever current you want and it puts out 24v so you can run 6-7xml. I decided to use four boards and ganged together two of the four drivers on each board so I can drive the xml at 2000mA.
 

·
Registered
Joined
·
1,985 Posts
As far as I can tell, you should be able to do it with any identical driver. You are simply paralleling them, so the voltage stays the same and the current stacks.

From what I can gather, most people think the heat produced with over 1000mA will be too much to handle. I don't really expect it to be a problem myself, given that nobody using a proper heatsink is even getting them more than warm to the touch, but I figure they should stay warm at best. Even if they are hot to the touch, nobody should worry. As long as you can hold the headsink while the XML are running full blast everything should be fine. Sure the expected life of the LEDs may drop from 12 years to 7, but even then I wouldn't be concerned. The longest I've used a fixture so far is 5 years with a Nova power compact fixture. The technology, even moving slowly, advances fast enough these LEDs will be cheap at the least and completely outdated at the worst.

What did you notice for the heat the XMLs were given off? I'm really wondering if I will have an issue.
 

·
Registered
Joined
·
8 Posts
redfishsc, VERY cool data. Thanks for sharing.

I'm running the same 3W LEDs and drivers you've tested and with 10 square inches of hardware store heatsink (aluminium angle) per W they don't get much above ambient. (they're underdriven at 650, so this in no surprise)
"beefy aluminum heatsinks" raise the price considerably, so it would be nice to know if the same W/surface area principle applies when trying to move more heat from each LED.

Sadly 20 no-brand LEDs (using 12) and two drivers still cost less than a single Meanwell driver in my part of the world.
Still, I think you've shown that as far as LEDs go you really can get less than you pay for, and I'll be keeping an eye out for inexpensive XML drivers.
Thanks :proud:

ps. what are the $5 drivers? if they'll run off 12V...
 

·
Registered
Joined
·
1,121 Posts
Discussion Starter · #7 ·

·
Registered
Joined
·
8 Posts
in case anyone wondered.
I did, repeatedly :tongue:

If my current setup works I think it may be worth moving to the XML, not only because of efficiency, but because I think it may be easier to make some nice pendants out of them (current setup needs some serious hiding)

Thanks again for the info.
 

·
Registered
Joined
·
1,121 Posts
Discussion Starter · #9 ·
I wouldn't say that the two heatsinks on the two different arrays I just compared are "equal", but "similar".



I don't really think the XML is a total heat demon like some LEDs can be. That's part of the benefit of efficiency:---- less energy is being converted to wasted useless heat.

They definitely need beefy heatsinking, and I personally would NOT consider running them at 3000mA on a heatsink that wasn't a minimum 1/4" thick with some 1" fins. Better safe than sorry. 3000mA is an incredible amount of current to run through such a small emitter. They're quite the wonder.



I also think that small clusters of XML's over larger tanks will do a LOT to producing a more pleasant light shimmer instead of they psycho-disco multi-source shimmer that arrays using lots of smaller LEDs can do.
 

·
Registered
Joined
·
8 Posts
I've always thought the problem with a relatively small area light source would be hotspots in the tank.

It turns out it's not as bad as I'd thought. Provided most of the light falls within the top of the tank there should be considerable brightening around the edges due to reflection (not shown).

edit: data in %, not PAR.
shows bottom of tank.
circles are 1/2 angle of optic, which is about 50% intensity for optic modeled.
 

Attachments

·
Registered
Joined
·
148 Posts
I'm running the same 3W LEDs and drivers you've tested and with 10 square inches of hardware store heatsink (aluminium angle) per W they don't get much above ambient. (they're underdriven at 650, so this in no surprise)
"beefy aluminum heatsinks" raise the price considerably, so it would be nice to know if the same W/surface area principle applies when trying to move more heat from each LED.
My feeling here is that people tend to overestimate the amount of heatsinking required for a given setup. I'm using 28" of 1/8" aluminum angle (probably same as you) for 6 CREE 3W LEDs run at 1A. No fans and very little air movement, and they merely get very warm.

The ability to heat transfer from your die/star to the heatsink makes a difference too. Scaling this contact area up with increasing heat sink area (and making sure to get excellent thermal contact) probably matters a lot.
 

·
Registered
Joined
·
1,121 Posts
Discussion Starter · #13 ·
My feeling here is that people tend to overestimate the amount of heatsinking required for a given setup. I'm using 28" of 1/8" aluminum angle (probably same as you) for 6 CREE 3W LEDs run at 1A. No fans and very little air movement, and they merely get very warm.
And your feeling would be right. Overkilling the heatsink and using good thermal bonding gives us complete confidence that heat isn't going to be an issue.

I'm glad you are getting such good results with so little heatsinking but I must emphasize that not everyone has this same result.


I was running 6 XRE and 6 XPG at 1A on a heatsink from heatsinkusa.com that is 2" wide, 12" long, 3/8" thick at the base, with some 1" tall fins.

The entire heatsink, with no fans, got quite uncomfortably hot, even though it had considerably more mass and passive cooling capabilities that 1/8" angle can.

So folks who replicate this need to monitor the LEDs, particularly in the spots each LED is located.
 

·
Registered
Joined
·
148 Posts
And your feeling would be right. Overkilling the heatsink and using good thermal bonding gives us complete confidence that heat isn't going to be an issue.

I'm glad you are getting such good results with so little heatsinking but I must emphasize that not everyone has this same result.
Totally agreed! I just meant to convey that people who are making fairly "standard" setups in terms of LED density and power shouldn't overly fret about having to source an expensive heatsink. With some care in design and construction it is entirely possible to get by with aluminum you can get at the hardware store, which is cheaper and easier to use. This is especially true if you add fans.

To get a little bit more on topic, my concern with the XM-Ls wasn't the power output--I'm sure they can replace a lot of smaller emitters just fine in terms of both raw output and value. My issues were:

(1) Hard time finding optics

(2) Placement to avoid dark spots in the aquarium obviously becomes trickier with fewer emitters and fewer optics choices.
 

·
Registered
Joined
·
1,121 Posts
Discussion Starter · #15 ·
my concern with the XM-Ls wasn't the power output--I'm sure they can replace a lot of smaller emitters just fine in terms of both raw output and value. My issues were:

(1) Hard time finding optics

(2) Placement to avoid dark spots in the aquarium obviously becomes trickier with fewer emitters and fewer optics choices.

Yeah it can get tricky. LEDgroupbuy.com carries optics for the XML's. Fortunately for tanks under 20", optics really aren't needed as long as you stick to 1 or 2 rows of LEDs centered in the tank, closer to the surface.


Over a deeper tank, you'd have to fiddle with the sweet spot of the right optic angle and right suspension height.


Fortunately you can figure out the light footprint of an LED using basic geometry, or perhaps using an online triangle calculator.

For a 60 degree optic, every inch away from the optic makes roughly a 1-inch diameter footprint. 20 inches away makes a 20 inch footprint, etc.

As long as we remember that the inner half of this footprint will be more intense, regarding PAR.
 

·
Registered
Joined
·
87 Posts
@Redfish, is the XML data you published µmol/m² or µm/ft² (i'm guessing ft²)

Thanks for your time in doing this :)
 

·
Registered
Joined
·
21,012 Posts
@Redfish, is the XML data you published µmol/m² or µm/ft² (i'm guessing ft²)

Thanks for your time in doing this :)
The standard measurement of PAR is in micromols per square meter per second. That is what PAR meters are calibrated to read.
 
1 - 20 of 23 Posts
Top