[time-nuts] Regulator choices

[Hal Murray]

bob at evoria.net said:
> The one thing that is missing is how to quantify the heat sink needs for a
> linear regulator.  Any thoughts?  IOW, is there some way to project how many
> square inches of heat sink needed for X watts to dissipate? 

In the simple minded case, you can add up the (series) resistance.  The units 
are degrees/watt.  Chips and heat sinks will have a parameter in the data 
sheet: often Theta something-or-other.

Mumble, round up for the connection between the chip and heat sink.

With fins-to-air, it depends on the velocity of air.  With a fan, that might 
be a constant.  If you are depending upon convection, it gets messy, aka 
nonlinear.  (I think, but don't have a good URL.)

If you are using the PC board rather than a heat sink, the key factor is the 
size of the copper plane that the chip is soldered to.  National had several 
app-notes with numbers for various chunks of copper.  With typical (digital) 
multilayer boards, you can also help a lot by using many vias to connect the 
small surface copper planelet to the ground plane(s) internal to the PCB.


-- 
These are my opinions.  I hate spam.