OK, I will be building three regulated power supplies... like so:
+/-25VDC power supply, feeding a +6V regulator and a +/-18V regulator. The +/-18V reg. is split to power one thing, while also being used to power the +/-15V regulator. All regs. are LM317/337. So, I ask... what type of heatsinking is best here?

For those of you with a Mouser catalog, on page 792 , they are listed. I was thinking of using N for all regulators. The part number is: 532-504222B00. Its thermal resistance is rated at 6.4. Is this good? Is the lower thermal resistance better? Or do I want higher?

I also will be using MOSFETS, and chose the shorter X heatsink. PN: 532-529802B25, thermal res. of 3.7. WOuld this work for my MOSFETs?

Finally, I will be using two more LM317s as current sources for the MOSFETs. I have been told they will generate a lot of heat. Which heatsink would be best for them? The X? A different one?

Thanks

This looks usefull:

http://www.coolingzone.com/Guest/New...esistance.html

So, according to this, a higher thermal resistance is better? Strange... all the heatsinks with lower thermal res. cost more... and the ones with high tr cost less. Ok...

Actually, the lower the better.


Temp res = T/W,

where T= difference between sink temperature and ambient temperature, and W is the heat energy given up by the sink.


So if the difference between the
heat sink temperature and the ambient temperature is x degrees Celsius and if the thermal resistance of the sink is R,
then the heat sink is giving up x/R watts
of heat. So you want R as low as possible for a given temperature difference, if you want to maximize heat output

In general, you want a sink that has the greatest surface area possible (the "figure x" sinks
in the Mouser page you posted are great).



[EDIT] Here's a much better article:

http://www.roassoc.com/aps/ap10.pdf

They are talking about heatsinking an "RO" module here; but the principles
would apply to any device.

How high is the input voltage?

How large are the currents (the loads)?

From this you can calculate how much power you want to cool away. For example, if you have 3 Watts power loss and a heatsink with 10 deg/W = 30 deg C temperature rise without any complicated calculations.

If you can sketch a scheamtic over the circuit, it much easier to give advises.

Thanks for the explanation- I totally misinterpreted that first article. (school's out... no need to thinkfor a while )

Basically, I am making a Szekeres amp with the LM317 current sources and an OPA627 voltage gain stage. The MOSFETs are biased to 250ma idle, as stated in the project addendum. This will be on a +/-15VDC power supply. So, I need sinking for both of these- the FETs and LMs- more sinking than what would probably be required in the PS. If I am doing this right, I take he 250mA and multiply that by the voltage to get the heat at idle. (I use the total voltage, right?-- 30V?) If I use 30V, I get 7.5W idle for the FET. Well... I guess I have a new heater. The "X" heatsinks are probably what I will use for the FETs and current source LMs.

In the power supply, I am just regulating, though quite a bit of current is passing through. I first have the +/-18VDC regulators, which is pushing 18mA at idle, and is also feeding the +/-15VDC regulators, which push about 700mA at idle. (about 700... I really don't know). Another regulation stage, not related to the first two, on a totally different tranny puts out +6V at about 370mA. I can figure those out... but now I must eat.

I do wonder about the current source regulators- how do I judge their current?

Thanks