[sdiy] VCS3/resistor heat

[Roy J. Tellason]

On Saturday 02 July 2005 06:21 am, Ingo Debus wrote:
> Am Donnerstag, 30.06.05 um 09:33 Uhr schrieb Magnus Danielson:
> > The power-supply of Mk I VCS3 and Synthi-A runs hot, since it uses
> > shunting regulation. In that powersupply the resistors run hot. My best
> > recommendation is to mount either larger resistors (more mass and
> > heat-radiating surface) and make sure it has alot air-contact for
> > convection.
>
> While we're on the subject, does anyone know why the power derating/heat
> sink requirements for power resistors are spec'd in such a funny way?
>
> I mean, with semiconductors usually the maximum die temperature is
> given, and the thermal conductivity from die to case.

Actually that's "thermal resistance"...

> If the power dissipation and the ambient temperature is known, it's easy to
> calculate what kind of heatsink is required.

Or how much of one,  anyhow.

> But recently I saw some datasheets of wirewound power resistors, and
> the specs were quite different: there's a "standard heat sink" assumed
> (with given size and material)

Do they specify either some mfr's part number or some specific thermal 
resistance from the heat sink to the surrounding air?

> and the thermal conductivity ("temperature rise" or so called in one data
> sheet) for the resistor/heatsink combo. Also the power derating is given for
> this special heatsink. This is convenient when this very heatsink is used,
> but what to do if I want to know which (different) heatsink I would need in
> a different situation?? Am I missing something obvious or do I interpret the
> data sheets uncorrectly? I'd think the mechanism in a power resistor is the
> same as in a power semiconductor: the die (or wire for the resistor) can
> stand only so much temperature, there's a thermal conductivity between die
>/wire and case and therefore for a given power dissipation of the die/wire
> there's a maximum allowed case temperature.

There's a maximum temperature whether it's in a resistor or a junction in a 
semiconductor,  before things don't work the way they should any more.  
There's a specific thermal resistance given,  in that as you start to 
dissipate power the temperature of that wire or junction will rise above 
ambient,  based on both the resistance to getting that heat out of there and 
the amount of power that you're dissipating.  For a semiconductor device in a 
package that's not meant to dissipate much in the way of heat,  that 
resistance is pretty high.  For medium-power and high-power packages it's 
significantly lower.  I have some parts salvaged here that are obviously 
meant to dissipate a little power,  they sort of resemble a TO220 package in 
that there's a metal tab in there, but it doesn't stick out very far and 
doesn't have a hole in it for mounting to a heatsink.  I guess this is one of 
those in-between packages.

Anyhow,  to figure out how hot the semiconductor or power resistor will get in 
a given configuration,  look at how much power you expect it to dissipate,  
and then add the thermal resistances of "junction-to-package",  
"package-to-heatsink" (this is the one we try and minimize with thermal 
grease),  and "heatsink-to-ambient",  that last one being given by the 
heatsink manufacturer on their data sheets.  The resulting number will be 
"junction-to-ambient" thermal resistance,  and will tell you how much of a 
temperature rise you will get above ambient for any given power dissipation.