Calculating Thermal resistance of heat sink

Thread Starter

mayuresh arankalle

Joined Dec 16, 2017
2
Hello,

I have heat-sink sample with me, but I do not have documentation on what is thermal resistance of heat-sink.
Here is details on Heatsink I was able to find -
Material - Aluminium 6063-T5
No of fins - 4 on side for T0220 package TRAIC and 6 on opposite side
I also have dimension of heat-sink with me.
Can anyone help understand how I can calculate thermal resistance of heatsink ??
 

DickCappels

Joined Aug 21, 2008
6,386
Put another way, if you can find a similar looking heatsink with similar dimensions and color then you can start out with the assumption that the thermal resistances are similar. As already pointed out, you would want to confirm the temperature rise in your own system before ordering a boatload of them.
 

schmitt trigger

Joined Jul 12, 2010
241
I have done it, measuring the temperature rise difference of a TO220 resistor under constant power conditions, and measure temperature rise above ambient both unmounted and mounted on the heatsink in question.

The measurement accuracy depends on closely attaching a low-thermal-mass precision thermistor to the resistor. Close thermal coupling is necessary, and for that I recommend this product:
http://www.arcticsilver.com/arctic_silver_thermal_adhesive.htm

Of course, it goes without saying that thermistor's transfer curve must be known (or easy to calculate) and repeatable. A thermistor like this family would be recommended.
https://www.omega.com/en-us/sensors-and-sensing-equipment/temperature/sensors/thermistors/p/44000-THERMIS-ELEMENTS

Like all measurements, the accuracy depends on your instrument's quality AND MOST IMPORTANTLY, how well you prepare the experiment.
For instance, I found that even the slightest air current, from an overhead A/C unit, would significantly change the measurement. You have to shield the heatsink from air currents and allow pure convection and radiation to provide the heat dissipation.

Of course, one can add later forced air and determine the improvement over pure convection cooling.
 

Thread Starter

mayuresh arankalle

Joined Dec 16, 2017
2
Thank you Everyone for your help. I will order thermal paste and NTC thermistor. I will post my findings here for anyone else looking for similar thread in future.
 

Papabravo

Joined Feb 24, 2006
13,726
A possible attack on the analytical problem would involve a recognition of the modes of heat transfer which are:
  1. Radiation
  2. conduction
  3. convection
You could start with a block of aluminum, determine the mass and the surface area and come up with a model of how a hot semiconductor would heat up the block until equilibrium was established.
Then you could do the same for a block with fins and some airflow. You get the idea. Maybe the experiment is the way to go.
 

andrewmm

Joined Feb 25, 2011
323
I would suggest that you look in the catalogues for a similar unit, as that will give you a good ballpark start point,
What sort of accuracy are you after ?
remember that air moving / still , or humid or hot itself are all variables you will have to account for in any measurements.
 

DickCappels

Joined Aug 21, 2008
6,386
A further suggestion is that rather than buy a thermistor, you invest in a thermometer or digital multimeter (DMM) that suports a thermocouple.

I have a few UNI-T meters and a Fluke that came with K type thermocouples and are said to be accurate to better than 1 degree C. About $US $15 to $20 and a very useful tool. Much more accurate than a thermistor and much easier to use -just read the temperature shown on the display.

One example of many possibilities: https://www.uni-t.cz/en/p/multimeter-uni-t-ut33c-tipa-edition
 
Top