Heatsink design?

Twitch

Newly Enlightened
Joined
Sep 3, 2007
Messages
9
I'm really new to LED, especially the high powered ones. I was rcently testing a generic LED I bough on ebay

Here were the spec:

Product Number
LE-LL001
Product Name
High Brightness 3W 80 lm White Lumen LED, 1 pcs
Emitted Color
White
Intensity Typical
[FONT=Arial, Helvetica, sans-serif]70 ~ 80 Lm [/FONT]
Viewing Angle
160°
Forward Voltage
[FONT=Arial, Helvetica, sans-serif]3.6V ~ 4.0V [/FONT]
Forward Current
650mA
Item Net Weight
10g / 0.4 oz​


Now I was running it on a bench at 3.7V with a regulated power supply. Being new to this, I just screw the MCPCB onto a a 2 inch^2 0.1 inch thick piece of aluminium. It was running fine and all for 10 minute, temperature reach around 60C, then poof gone. It must have overheated:laughing:

Now I no longer have any LED to play with :p, I'm waiting for my order of Cree XRE P4 to come in.

So I'm wondering anyone have and idea how big of a heat sink I need?(flat piece of aluminium) It doesn't need to be super precise, but just enough and not too large. I won't be running it in a flashlight.
 
Last edited:

znomit

Enlightened
Joined
Aug 1, 2007
Messages
979
Location
New Zealand
At 60 degrees it should be fine.
Did you have thermal paste between the led and the aluminium? This lets the heat pass from the led to the Aluminium. From there it gets dissapated to the air...so having it sit above the bench so air can flow around it helps. The more surface area the better.
The term "heat sink" is not really good...you want a "heat pipe" to transfer the heat to the air. If you cant direct air flow over it then surface area is the only practical way to improve things.

If you can comfortably hold it its probably not too warm...but the cooler the better.
 

James Jackson

Newly Enlightened
Joined
Aug 6, 2007
Messages
59
ummm... there is more to these LEDs than just the voltage spec.

Did you also limit the current to it? It is spec'ed at 650mA - so you need to have a constant current driver that limits the current to a 650mA (or less).

If you don't... *poof*

I am currently working on a constant current driver pcb for 4 Luxeon rebels... this outta be fun.

Regards,

James Jackson
Oztronics
 

evan9162

Flashlight Enthusiast
Joined
Apr 18, 2002
Messages
2,639
Location
Boise, ID
This is what happens when you drive an LED with a voltage source (without any kind of current limiting) and not a current source.

The critical factor for driving LEDs properly is current, not voltage. With a voltage source, you have no idea how much current is flowing through the LED. Since you have no idea how much current is flowing, you have no idea if you're overdriving it to a dangerous level.

If you are going to use a voltage source, you need to put the appropriate current limiting resistor in series with the LED.

Otherwise, you need a proper current regulator to supply the right amount of current to the LED.

Understanding the relationship between voltage and current in LEDs is vital to properly using them.

http://candlepowerforums.com/vb/showthread.php?t=77221
http://candlepowerforums.com/vb/showthread.php?t=72528
 

Mash

Enlightened
Joined
Dec 18, 2006
Messages
378
Learnt this the hard way recently.
Was playing with crees and a PC power supply. Wanted to see if I can power two LEDs in series from the 5V rail. Was sure nothing will happen since, two LEDs in series should handle up to 7.8- 8Vs. Next thing I know, bright light and POOOF, two Crees busted. The PSU was supplying 4.5 AMPS at 5 Vs. Is this even possible? To overdrive the current at a very low voltage?
Oh well! Funny thing is they both blew instantly, wheras I would expect one to fail and cutoff the circuit therefore saving the second one, but not this time!
 

evan9162

Flashlight Enthusiast
Joined
Apr 18, 2002
Messages
2,639
Location
Boise, ID
What you're describing isn't possible.

What is possible is one or more of these:

A) You had them wired in parallel and not in series, so each one was connected to 5V. Each LED was probably subjected to 3-4A of current.

B) You had them wired to the 12V rail and not the 5V rail.

C) Your 5V rail was somehow supplying 8-10V. The vast majority of the time this is not possble, since most 5V rails in PC PSUs have an overvoltage protection circuit that limits the rail voltage to 6V or so.
 

Mash

Enlightened
Joined
Dec 18, 2006
Messages
378
Definately Series wiring,
Mistake on connecting to 12V rail possible,
Need to check with multimeter the exact voltage output of the 5v rail.
Also found out something interesting with an oldish PSU which has "automatic noise control" or some such. Unless there is a porper load on the PSU, it doesnt supply full power; I had to connect an old hard drive, to make sure it was working at full spec!
 

Melchior

Newly Enlightened
Joined
Feb 4, 2006
Messages
192
Back on track

So to get this thread back on track;

What kind of 'rule-of-thumb' are there for selecting a proper heat-sink with power leds?

For instance:

What size of heat-sink would be needed with any old
1W Power Led;

an array of Power Leds? (4-5W Total thermal dissipation)

20 Watts, 50?

There should be some kind of formula that specifies the dimensions of a 'regular' finned aluminum heat-sink.

As these will probably be natural convection heat-sinks reusing forced convection heat-sinks is a bad idea?
 

James Jackson

Newly Enlightened
Joined
Aug 6, 2007
Messages
59
I believe this gets into thermal analysis. Not too many folks have the proper software to do this - and even if they do - it's kinduva Black Art. Not everything is as simple as "use this size heat sink for a 1-watt LED".

I have designed several high-power LED printed circuit boards in the past, and have used a variety of heatsinking methods.

There is even an aluminum (or copper) type of board material that is bonded with prepreg and a top copper trace layer. This is effective in dissipating the heat - but it too has to have thermal analysis done - or else one can run the LEDs hotter than they should be.

Start with getting the application notes and datasheets for the LEDs that you are going to use. Many times they will include thermal data for the LED and tips on what sort of heatsinking can be used.

One final thought on this. If you use too small of a heatsink - you will run the LEDs hotter than they should be run, and eventually they will fail - short of the 50,000 hours that many will claim for lifetime of the LED.

I hope this helps.

Regards,

James Jackson
Oztronics
 

laserblue

Newly Enlightened
Joined
Oct 27, 2007
Messages
37
Well, for LED heatsinks the larger the better but of course the problem with large heatsinks is that you have a limited space inside a flashlight.
 

LEDite

Enlightened
Joined
Dec 6, 2002
Messages
324
Location
Dallas, TX
Twitch;

The problem is that:

1. as the LED heats up, the voltage across it drops (up to 1/2 volt).

2. That leads to increasing current thru the LED.

Use a resistor (1-2 ohms) to set the current,
let it operate to reach steady-state heat levels,
check the final current level and adjust the resistor accordingly.

On my fixed lighting designs, I use aluminum heat sink weight as a 1st approximation.
For one 3-watt LED, I would use about 2oz alum. heat sink with fins.

I do not use a MCPCB. I use a filled-epoxy for direct mounting on anodized surface.

Larry
 
Last edited:

hank

Flashlight Enthusiast
Joined
Apr 12, 2001
Messages
1,561
Location
Berkeley CA
I found the heatsink I want for my next really really bright light (like one of the 3x Rebel stars).

Unfortunately right now it's for sale on a 110v light
http://store.lsgc.com/Assets/ProductImages/r20_300.png
http://store.lsgc.com/R20-AMBER-FLOODLIGHT-MEDIUM-EDISON-BASE-P54.aspx

(I own half a dozen of these amber lights for evening/night home use, they're really bright for the money; the heatsink gets comfortably warm after they're on for a while.)

But - anyone recognize these heatsinks as something available as a component?


They'd look just fine attached to something like 3xC or 4xAA battery tube.
 
Last edited:
Top