Heatsinks for DIY LED lamps

You can use PWM dimmer to control output of a driver only if the driver supports PWM input (most of cheap drivers don't). However, a lot of drivers do have an internal pot to trim its output. Often this fact isn't even mentioned and the driver is being sold as non-dimmable.
Disclaimer: If your driver indeed does have such a pot, I don't encourage you to use it. It might not been engineered to adjust the output and overloading might cause damage to your driver and/or LEDs.
 
Greengenes707 said:
@SupraSPL...just to clarify those effciency numbers are still taking the fan power into account right?
Click to expand...

I'll keep stressing it until it happens; we need people to replicate Supra's tests for comparison and accuracy. Everyone makes mistakes, so we need to eventually create a general consensus of the various CPU HS/Fan combos in order to validate our assumptions of these modes of cooling.

(Just think, a list comprising of all the practical HS/Fan combos for COB purposes, listing cost vs. efficiency, uses in differing environments, etc., with generally agreed data to support future choices)

I'll volunteer, I just need pointed in the right direction in regards to lab equipment. Photos help tremendously :-D
 
Greengenes707 said:
@SupraSPL...just to clarify those effciency numbers are still taking the fan power into account right?
Click to expand...
Yes, the numbers at the end of each post represent the efficiency with both temp droop and fan power consumption factored in (fan dissipation * 1.2 for PSU losses) Those efficiency numbers are for comparison purposes only, not an exact quantitative measurement of efficiency. The actual efficiency numbers would be higher, because the 42% baseline is based on Tj 50C but we are dealing with Tj 23C as a baseline for the pulsed measurements.. Also, these efficiency numbers are based on the minimum figures from Cree so probably a bit higher in practice.

Here is how I figure it:

If dissipation is 53W and baseline efficiency is 42%, we get 22.26 PAR W. So if we measured a light loss of say 2%, we have to penalize the PAR W by 2%. So multiply the PAR W by .98 = 21.81W. And if we used 1.87W of total fan power to get it done, I add that directly to the dissipation power. 53+1.87=54.87.

So to factor in temp droop and fan losses (for heatsink comparison purposes only), the efficiency in this example changed from 42% to:

21.81 PAR W / 54.87 Dissipation W = 39.75%
 
Last edited:
It is true, some of these tests will have a certain margin of error in them. This test is particularly difficult when using a pair of drivers because I cannot monitor the current of each driver as it varies with driver temperature ( I am working to remedy this). In order to get a really accurate test, the driver has to be warmed up before hand and the current should be noted as you take the pulsed light measurment. Once the COB/heatsink is thermally stable, I cool the driver back down by hand until it matches the current at the beginning of the test, and then take the light reading. Matching the driver current is critical to get accurate repeatable results. Another option is to use a dimmable driver, a big time save but the biggest one I have on hand is only 1050mA and I needed a pair of 1400mA.

Another issue with my testing is the lack of airflow during the passive cooling tests. It really is the type of test that should be done in the grow space but the other lights would interfere with the results one way or another. So I believe the passive cooling numbers I posted would be better in practice. When I check the temps of my 10.08 heatsinks in use, they are 27C. In the test, they were 33C.

So to perform the test, you would need a luxmeter, an accurate current measuring multimeter or dimmable driver, and an accurate digital thermometer (helpful but not necessary). It is critical that neither the COB or luxmeter has a chance to move between the pulsed measurement and the warmed up measurement. You can attach the meter to a tripod and hang the LED above it. Or you can perform the test on a flat surface, a silicon mat or something non skid to keep the components still. You can get a heavy silicon mat from Walmart for $6, just got one for my soldering bench.
 
Last edited:
When it comes to analyzing the fan power. You need accurate voltage and current measurement. The current measurement will change quite a bit when you change the voltage, so you need a new current measurement for each voltage. This also varies from fan to fan. For example, some fans are high RPM and will draw more current at a given voltage. They will dissipate more power at any given voltage than a lower rpm fan of the same size.

So once you have the voltage and current, multiply that to get dissipation power. Then multiply * 1.2. That represents the best case scenario for small fan power supplies, a high efficiency adapter running above 60% of its max current rating should be about 80% efficient. They are relatively cheap and a single adapter can power quite a few fans, so in most cases we should be able to achieve 80% efficiency.
 
Skaumannen said:
I found this heatsink on ebay
http://www.ebay.com/itm/Large-Alumi...341?pt=LH_DefaultDomain_3&hash=item2a17f8fa0d
And i am wondering if it is large enough to use with 2x vero 29 or 2x cxa3070 @1400mA?
I have some 120 or 140mm fans that i can use to cool them.
Click to expand...

That is a nice heatsink but about twice the price that to what we normally pay. It is about 2800cm² so for active cooling that is good for about 93W. I think it would be OK with a pair 105W of Vero or CXA as long as you have a fan blowing into all the fins. On the other hand you could probably score a pair of heavy duty CPU coolers for $20 or less at a local CPU shop or eBay and it would give you much more spread.
 
SupraSPL said:
That is a nice heatsink but about twice the price that to what we normally pay. It is about 2800cm² so for active cooling that is good for about 93W. I think it would be OK with a pair 105W of Vero or CXA as long as you have a fan blowing into all the fins. On the other hand you could probably score a pair of heavy duty CPU coolers for $20 or less at a local CPU shop or eBay and it would give you much more spread.
Click to expand...
The shipping is the reason that one might be worthwile for me. I alreadu have some artic alpine 11's running 4x cxa3070 z2, but i somehow managed to break two of them, so i will be replacing them with vero's, as they are cheaper. But with the tools i have available using two large heatsinks means i can build a better looking and easier to use light. I have not been abled to find anything as large as this for that little, so i actually thought this one was really cheap :oops:
 
...i miss a test with an blower fan and tubular heatshink ... ...steveleds style... dreaming again...:joint:

12954691999471426255104.jpeg
1295714245745-2121258012.jpeg

1343837238616-1766345699.jpeg
1302661661382-1086306243.jpeg
1338177445383-1024788029.jpeg
13327916321801597953476.jpeg


fans_and_cable.jpg


...and with cobs too...;)

Pd ...and maybe with others blowers fans and heatsinks with cover adapted piece...:bigjoint:

saludos
 
Last edited:
Experimenting with the Rosewill dimmer. It works with 5V or 12V adapter. Using the 5V adapter I was able to dim a high rpm 140mm fan to 2.8V before it stopped, but it could not start below 3.5V. I have not tested the dimmer efficiency yet but from what I can tell so far, it does not seem to be very bad.
 
SupraSPL said:
There are similar heatsinks available in the US from Nuventix, selling on Digikey. I think the vertical orientation of the fins would be awesome for passive cooling, as another poster here mentioned a few pages back. The downside though, they are just about the same price as the Alpine 11 and they may not be up to the task of a 53W COB?
http://www.digikey.com/product-detail/en/NX301101/1061-1092-ND/3505755
http://www.digikey.com/product-detail/en/NX301107/1061-1135-ND/3878753
Click to expand...

Up to 82 watts for the Vero 29.
http://www.nuventix.com/led-cooling/led-heatsinks/
 
You must log in or register to reply here.
Back
Top