Super bright LED heat

Discussion in 'The Projects Forum' started by BobW55, Sep 14, 2013.

  1. BobW55

    BobW55 Thread Starter Member

    Joined:
    Jul 22, 2008
    Messages:
    26
    Just how much heat do these new super bright LEDs produce?

    For example:
    If I took a 1 watt white one, drove it at max output, just how hot would it get?
    Could I hold it in my fingers, or would I still need a hefty heat sink for it.

    Bob
  2. THE_RB

    THE_RB AAC Fanatic!

    Joined:
    Feb 11, 2008
    Messages:
    5,435
    If you drive the LED with one watt, the heat dissipated is determined by it's efficiency.

    If it is 50% efficient then 50% of a watt is output as light, and 50% of a watt is output as heat.

    Generally higher power LEDs like 1W versions require a heatsink, they cannot be operated for any length of time at full power without a heatsink.

    The LED manufacturer should offer info on recommended heatsink etc. :)
  3. Metalmann

    Metalmann Active Member

    Joined:
    Dec 8, 2012
    Messages:
    684
    Location:
    IN
    "If I took a 1 watt white one, drove it at max output, just how hot would it get?"



    I did a few rough tests for heat, and it varied around 145*-185* F, before burnout.

    Like I said, rough testing. Did not document......;)
    BobW55 likes this.
  4. BobW55

    BobW55 Thread Starter Member

    Joined:
    Jul 22, 2008
    Messages:
    26
    Metalmann,

    Thank you, that was the real life info I needed. Want to in-bed these in some epoxy resin to make a big lens, so heat is an issue. Will have to track down some 10mm standard ones and just use a few extras.
  5. John P

    John P Senior Member

    Joined:
    Oct 14, 2008
    Messages:
    1,321
    Location:
    Massachusetts
    I've been experimenting with 1W LEDs in the important field of model railroading. Knowing that they'd need heat sinks, I planned to mount them 4 at a time on 8in long aluminum bars, with a U-shaped cross section 3/4in square. Here's a picture--note the constant-current drivers, one a switching supply ordered off eBay (as were the LEDs) and the other hand-wired using an LM317:
    http://tmrc.mit.edu/progress/reports/2012/05/IMG_2059.JPG

    This picture was entirely lighted with a bank of 4 LEDs:
    http://tmrc.mit.edu/progress/reports/2012/04/IMG_1811.JPG

    As far as heat output is concerned, the 4 LEDs get the metal bar warm, but not painfully hot. I'm confident that it's enough heat sinking, but I'd be very doubtful about any setup that embedded that kind of unit in plastic. I think you either need a thermal path made of metal all the way to free air, or you have to use a heat spreader inside the plastic to make sure that the heat has an easy path to the surface.

    This is a floodlight I made for illumination of the worksite during track construction. Note the large area of metal:
    http://tmrc.mit.edu/progress/reports/2013/06/IMG_3486.JPG
  6. Metalmann

    Metalmann Active Member

    Joined:
    Dec 8, 2012
    Messages:
    684
    Location:
    IN
    Looks good, John.

    How thick is your aluminum heatsink in the last pic?
  7. trader007

    trader007 Member

    Joined:
    Feb 27, 2010
    Messages:
    217
    whats nice about led's is their thermal tolerance is right about where you cant touch it without burning your skin. Makes it easy to tell if your heatsink is enough. Things feel too hot to touch around 130F, and led's burn out around 160F... so if it feels too hot to touch, its getting too hot for the LED too.

    1w leds build up heat fairly slow though, so a good general rule is to let it run for at least an hour before you decide if the heatsink is working well enough.
  8. John P

    John P Senior Member

    Joined:
    Oct 14, 2008
    Messages:
    1,321
    Location:
    Massachusetts
    Can't remember. But judging by the thickness of the perf board, which is 1/16, I'd guess 3/32.
    Metalmann likes this.
  9. wayneh

    wayneh AAC Fanatic!

    Joined:
    Sep 9, 2010
    Messages:
    8,102
    Location:
    Roscoe, IL
    You may want to also consider allowing for replacement. LEDs fade and burn out eventually, especially if driven hard.
  10. THE_RB

    THE_RB AAC Fanatic!

    Joined:
    Feb 11, 2008
    Messages:
    5,435
    Even more so if someone has encapsulated a lot of heat generating devices inside an epoxy block WITHOUT a lot of de-rating for heat buildup. :)
  11. #12

    #12 AAC Fanatic!

    Joined:
    Nov 30, 2010
    Messages:
    9,586
    Location:
    15 miles west of Tampa, Florida
    Let's get the basics down. Any part which is heated by a current will get hotter until the radiated heat comes into equilibrium with the generated heat. If it was perfectly insulated, the temperature would theoretically rise toward infinity. In practical terms, if you make a nice wad of fiberglass insulation for the part to be contained, you can get most anything to smoke or burst into flames.

    So, how much heat does a 1 watt part make? One watt second per second. How hot will it get? It depends entirely on what it is surrounded with.

    Metalmann gave us a clue about how hot an LED would get when surrounded by air. That is a good starting point. You just have to decide what to surround the LED with to get the temperature where you want it.
  12. trader007

    trader007 Member

    Joined:
    Feb 27, 2010
    Messages:
    217
    #12 is correct. Take my custom troublelight as an example. The LED's are mounted to a steel pipe, which conducts heat fairly well. But with LED's, the heat sinks out the back, so the inside of the pipe warms up. That air in there just gets hotter and hotter. If you dont drill a couple holes in the bottom and top (convection cooling) the whole light will get too hot to touch after an hour or so, and the LEDs burn out eventually.

    If you drill just two 1/4" holes in the bottom, and two in the top, its more then enough to keep the whole light cool. I mean it still gets warm, but you can put your cheek next to the top holes and feel the heat pouring out. It works incredibly well.
  13. snav

    snav Member

    Joined:
    Aug 1, 2011
    Messages:
    73
    I used resistors to simulate the 8w of heat wasted by the 10w lamp in pic and found that a 25mm x 25mm x 25mm ASIC black anodized five fin heatsink rose to just under 200deg and 265deg at 11.9w, so for your application I would think any TO220 heatsink would work. The trick is to get good contact without shorting the leads.

    Attached Files:

  14. GopherT

    GopherT Well-Known Member

    Joined:
    Nov 23, 2012
    Messages:
    1,942
    Location:
    Western PA (USA) (GMT -5)
    There are three types of high power (>1W) /high brightness (gallium nitride) LEDs. The original type are. Made on sapphire substrate and are medium temperature. The newer, low cost LEDs made by Samsung and other companies associated with silicon semiconductors use, naturally, silicon as the substrate. These are generally more thermally sensitive. The newest type and used only by Soraa, use gallium nitride as the substrate. This so called GaN-on-GaN technology has no thermo expansion xoefficient mismatch and allows very high temp operation. Now, the funny part is, there is no general definition of low, medium and high temp because it really depends on the lifetime you define for the product. The new GaN on GaN operate near 200C at the die.

    Also, ther is no such thing as a white LED, they are Violet to UV and phosphors are used to generate a range of colors to make white. This is the Soraa PR. They are really impressive bulbs. About $80 each last year - target market was retail displays (everything from jewelry and clothing shops to grocery store freezer displays.



    [​IMG]
  15. Metalmann

    Metalmann Active Member

    Joined:
    Dec 8, 2012
    Messages:
    684
    Location:
    IN
  16. GopherT

    GopherT Well-Known Member

    Joined:
    Nov 23, 2012
    Messages:
    1,942
    Location:
    Western PA (USA) (GMT -5)
    Prices appear to have fallen to $35 range. Includes entire bulb (Power supply, heat sink, diffuser with high-efficiency organic phosphors on the diffuser (not on the LED), ...).

    [​IMG]
    THE_RB likes this.
Similar Threads
Forum Title Date
The Projects Forum Super Bright led flashlight Jul 17, 2010
The Projects Forum SUPPLY VOLTAGE SUPERVISORS and BUCK/BOOST and PUSH/PULL Sunday at 5:01 PM
The Projects Forum Limiting Super High Power (Amp) Draw (3KW+) Oct 5, 2014
The Projects Forum Supercapacitor, SLA Battery and Solar Cells in parallel Jul 20, 2014
The Projects Forum Changing battery to super caps Jun 27, 2014

Related Site Pages

Share This Page