Forum for heatsinks

Hello there,

I would like to know the criteria/procedure to figure out which heatsink I should use for my IC?

I have an IC and it gets very hot. It is compulsory to use a heatsink.

To find a heatsink, what parameters should I look into my IC's datasheet and once I get that, how do I calculate what parameter I should look for while researching for heatsink

8 answers Last reply
More about forum heatsinks
  1. consumer heatsinks dont offer much information, unless there is a thermal watt rating.

    cant decipher the thermal watt rating when it doesnt say if the heat is dissapated per hour or per minute (or the wild per second)

    size doesnt always matter.. it is highly dependant on the material used for the metal, and also important for the thermal material used to connect the two pieces together.

    if you want ANYTHING to look for.. use the thermal watt number.
    how hot does the IC chip get?
    a degree in celcius would be easy to compare it to PC coolers.
  2. Well, It gets pretty hot. About 55-65 Degree Celcius. How do i start my search.
    Lets assume a decent Thermal Watt number to start with.

    It will take me a while to get this number as it is not on the datasheet but I want to get started.
    Please advise.
  3. 55-65 degrees celcius is about the same as graphics cards for a computer.

    as you can see for yourself.. not many of them have a square and small heatsink.
    and, if you have ever seen a photo of a video card with the heatsink removed.. the processor isnt all that big.
    it could be about the same size of an IC.. but IC sizes vary.

    to say that you don't have much room sideways to use a video card cooler.. pc processors also get about the same degree celcius.
    not all PC coolers are the same.
    bigger with more fins will always help.. unless there is metal somewhere that isnt good at conducting heat.
    the point is.. if there isnt any room to the sides of the IC .. you could look to use a PC processor cooler to do the job.
    the taller you go, the better your chances of cooling the IC chip.
    i have a tuniq tower that used to get me idle temps somewhat close to water cooling.
    at maximum load was a different number.
    i was within 5 degrees of water cooling at idle.. but water cooling was usually 10 degrees less at maximum load.

    have a look at the tuniq tower, or find something similar.
    my temps idled at about 30 degrees celcius when i first installed it.
    now it is a bit dirty and the thermal paste is old (dont know which one is worse) and i am getting about 37 degrees celcius at idle.
    compare that to the 50 and 60 degrees that the stock cooler provides at idle.. the temperature drop is about half.
  4. Hi there....

    Thanks for the reply.
    However I was looking at something like a formula to calculate the parameters.
    There should be some formulas and calculations to figure out what kind of heatsink I would need.

    For example if I go to digikey website and search for "HeatSinks", I get a filter screen which asks me to enter few parameters to narrow down my search. One of the parameter here is "Thermal Resistance at forced air flow"..then there is "thermal resistance at natural" and also "power dissipation at temp rise".

    I would like to know how to get these values from the datasheet. I think there is a calculation to figure out all these parameters.

    If you can provide me some insights in this direction, that would be great.
  5. hard to say..
    thermal resistance at forced air flow could be temperature change with high barometer pressure, confused with thermal resistance at ____ cfm of air flow.
    it makes more clear sense to note the temperature changes with the barometer pressure, because it is a key indication of whether or not the thermal dissipation per hour needs to be high or low.

    to say you have something 100 degrees F that is as thin as a piece of paper, higher barometer pressure will cool that 100 degrees F

    if the thing was thick like a flashlight, the higher barometer pressure would cool it more.. but the final temperature would be significantly different than the hot thing that is as thin as a piece of paper.
    boils down to the desire of being reduced to ambient temperature.. and how stubborn the heated piece is.

    cant make any translations without the equipment to conduct thermal readings, and to change the forecast in the room the IC chip is in to again record those thermal readings.
    sounds like a plea to get you to buy some thermal imaging software, as well as obtain some ways to change the forcast in the room.
    just measure the temperature and match the number with thermal dissapation per hour.
    you should be able to find a conversion calculator to transform degrees into thermal watts.

    heat can be the same tricky situation as voltage and current.
    the paper thin IC chip would cool quicker than the 2 inch IC chip. (depending on materials used, this can be totally backwards)
    but you get the idea.. if it cools easy, you dont need a lot.
    if it is stubborn to reduce temperature, you need a lot more.

    quick wise words would say transform the degree to thermal watt and match the number with a specification listed for a heatsink (computer heatsinks say thermal watt dissipation all the time).
    if the heatsink isnt good enough.. you know you need to increase the thermal watt dissipation.
    when you take a note of how well the first heatsink worked (as long as the thermal watt wasnt a lie) .. then you should get a general idea of how much more you need.

    you will probably have to destroy the IC chip by letting it get as hot as it does normally without a heatsink.
    if it fails.. you know you need a heatsink.
    and if it fails before you get it to normal operating temperature.. try putting it in a freezer and letting it get up to normal operating temperature, then grab the temperature and add the temp of the freezer BEFORE you put the IC chip inside.
    that should get you a ballpark figure to get a heatsink.
    it might not be the perfect heatsink.. but it should be enough to allow the IC chip to get up to normal operating temperature long enough to reach a degree and stay.
    then check that temperature with the safety limit of the IC chip and get a better heatsink if you need it.

    what are we supposed to tell you when we dont know if the IC chip can get up to normal operating temperature before failing without a heatsink?

    the form from digikey isnt detailed and informative enough to describe exactly what number goes into those spaces.
    those forms need definitions before anybody can go about tackling some college degree of mathematics.
  6. i mean.. power dissapation at temperature rise is pretty simple.
    if you feed the IC 5 volts.. and there is a 5 volt output.. how much does that voltage drop with the increase of temperature?


    it could also mean.. how much does the heatsink kick into gear and begin cooling as the temperature rises.
    to say,
    the heatsink gets colder and colder as the temperature rises.

    maybe you need to convert those voltages into watts.. and then tell the chart how many watts are missing from the voltage output when the temperature reaches ___ degrees celsius.

    cant find a 25 watts @ 60 degrees C
    then you need 50 watts @ 30 degrees C
    that simply means the 50 watt version is going to perform the same.. except that the IC thickness would be more or less (or the material would be different), depending on the watts lost.

    power dissapation at temperature rise should be the most easy to use, but if the temperature rise means 'after reaching normal operating temperature'
    then you would need a heat gun to make the temps rise.
    the value should be either at natural operating temp, or value AFTER natural temp.

    what is the point?
    to keep your voltages the same, to keep the circuit working.
    heat will adjust the voltages, and it makes the circuit pointless.
    maybe you lower the temps to keep the voltages the same.. but the temps are still too high for the longevity of the IC chip.

    thermal resistance @ forced air flow.. perhaps you need to convert barometer pressure to ____ LFM
    and then apply 1 watt of power.. and take note of the temperature rise.
    this seems to be more accurate to keep the IC chip at ambient temperature.

    the other chart is.. give the IC chip 1 watt and notice the temperature.
    doesnt matter if the 1 watt is less than needed for the IC to work.. as long as the voltage is going through the entire IC chip, then you have enough flow to heatup the entire chip.
    but beware, the ambient temp of the room for the chart could be totally different than the ambient temperature of your room.
    so the temperature rise might be more for the chart, and less in your room.
  7. and when you are all done..
    you dont know if the IC chip wants to 'let it all out' .. meaning the heat.
    if it is better to keep some of the heat.
    if it is better to let out the heat faster than it can be generated.
  8. aluminum seems to be what i see used the most for heat sinks , you see fins allot because they increase surface area . my opinion is that provided this part doesn't require a minimum operating temperature , the cooler you can keep this the better .
Ask a new question

Read More