In industry we used to do it by weight, given
some allowance for your "goodness factor"...
In many cases heatsinks have a similar enough
efficiency factor that you can get a reasonable
estimate from the weight. It certainly is an
easy measurement to start from. :o)
-Roman



Russell McMahon wrote:
>
> Here's a potentially useful expression for the "rule of thumb" energy
> dissipation capacity of a free air cooled heatsink.
>
> Merciless (and possibly informed) criticism expected :-).
>
> This expression is an entirely empirical one based on inspection of heatsink
> profiles and degrees C / watt ratings in a catalogue. It is solely an
> attempt to fit the observed figures to reality and makes no claim to actual
> correctness in a given situation. That said, this is the sort of thing that
> helps make an engineer an engineer in the long term With suitable experience
> and noting where this falls down you should be able to pick up a heatsink
> and guesstimate its capacity within a factor of two or so. For those who can
> do this already by inspection (as can many who have much experience in the
> field), please move along, these are not the ones you want .... :-)
>
> The formula works ROUGHLY for heatsinks in the 0.2 to 40 C/W range with
> increasing inaccuracy at either end.
>
>         Degrees Celsius rise per watt = 1/(Volume in litres)^ 2/3 x G
>
> G is a "goodness factor" of your estimate based on the experience you are
> about to get :-). A heatsink with good webs for heat flow and ample finning
> within the volume profile will probably work better than a very thing unit
> with minimal fins. For typical commercial heatsinks G ~~ 1.
>
> ie Measure outer dimensions of heatsink profile in decimetres (0.1m = 10cm =
> 100mm = 1 decimetre)
> Multiply these together to get litres volume.
> Take 0.666 power (or square then cube root).
> Take inverse.
>
> eg 100 x 150 x 300 mm
> Typical design so estimate G = 1
>
> 1/(1 x 1.5 x 3.0)^0.666   x  1 = 0.22 C/W
>
> A Largish flag plus fins unit say 30mm x 30mm x 10mm =
>
> 1/(0.3 x 0.3 x 0.1)^0.666 = 23 C/W
> If this is not especially well endowed with fins or rather thin a value of
> 50% to 100% higher may be in order.

--
http://www.piclist.com hint: The PICList is archived three different
ways.  See http://www.piclist.com/#archives for details.