Heatsinkometer?

Hey everyone!

So I was thinking, how do we go about telling how good a heatsink is? We don't, really, besides just guesstimating.

So what if someone designed a "heatsinkometer" (hahaha)? I figure that it would have two probes - one probe gives off a certain amount of wattage of heat (one side is insulated, the other side is the heated side so you can press it against the first surface of the heatsink) and the second probe would be a temperature measuring device, which would also be pressed against the first surface, so that one could effectively measure how well the whole system is heatsinked, right?

Or, maybe make that whole thing one probe (temperature measurement and heating element), and then have a temperature probe on the final surface of the heatsink?

I was thinking about this because it would be a good way to see how well the whole system interacts - it could essentially be used to figure out what kind of power dissipation could be handled by a whole system, instead of doing the math in order to figure out the junctions between components.

Any thoughts?

The performance of heatsinks can be measured in various ways. One obvious way is to heat the heatsink to a certain temperature and look at how rapidly it cools down towards ambient. But a more common method is to bolt a known-dissipation device to it (such as a power resistor) and measure the temperature rise over ambient for a given power level.

Obviously you can repeat such tests using your own equipment if you want to, and that may give you some insight into how to mount a heatsink for optimal performace under real word conditions.

Well, I mean, sure, that's easy, if you are just using a single surface for a heatsink. But what if you have three different junctions for the heat source to ambient?

For example:
diode case to Aixiz module
Aixiz module to heatsink
heatsink to ambient

How would you go about measuring the performance of the whole system? Measuring a single device is easy enough, but with multiple junctions... how do you know how high above ambient the heating element is getting? Obviously, we want to maintain the diode/IC/whatever temperature to be as low as possible, so....

This should be very easy to do without any devices whatsoever. Here's why:

We know the thermal conductivity of aluminum for 6061, it's 104 BTU/hr-ft-F or 180 W/m-°C.

And the shape we're testing is pretty easy to work with. It's basically a cylinder with a hole cut out of it. It's easy math. It's math I don't actually know how to do - but I know that it's easy

It should be relatively straightforward to come up with a formula that would take the height of your heatsink, the outer diameter of your heatsink, presume an inner cutout of 12mm, and then return some quantification of your heatsink's ability to pass a certain wattage of heat to it's host in a given period of time. There would be a different formula for Copper, but it would be the same general form.

EDIT: In fact, I would be willing to devote some time to coding a web-tool for this, if others can contribute the underlying math. I might not have time for a few weeks, but I think this would be a worth project.

Hmm... you know what? I will totally work on that when my brain starts working again. I'm good at math for a college freshman

Run the laser where it dissipates a known amount of heat and measure the diode/aixiz/heatsink with an IR thermometer?

The biggest variable would be the quality of the contact between components.

@Cyparagon: That's the issue - finding the quality of the contacts.

Pardon the double post, but it's not related at all to the previous one and it's been a few days. Anyway....

Just
@rhd: What would we do when it comes to non-cylindrical shapes, like heatsinks with lips, or with fins, etc.?

Wolfman29 said:

Pardon the double post, but it's not related at all to the previous one and it's been a few days. Anyway....

Just
@rhd: What would we do when it comes to non-cylindrical shapes, like heatsinks with lips, or with fins, etc.?

Click to expand...

Partial differential equations.

Aww Those aren't fun.

Wolfman29 said:

Aww Those aren't fun.

Click to expand...

Not if they need to be solved numerically, nor when you don't know how to do that, nor how to parameterize it all properly either, nor if its possible to solve or approximate it analytically and you don't know the more advanced PDE solving techniques required to do it
;_;

True dat. Not sure exactly what PDEs you were talking about, but maybe these can easily be solved analytically?

lol I don't know, actually. I just remember learning that the PDEs and ODEs we learn to solve, fall under a specific class that are analytically solvable.

But I guess if we figured out how to write an equation that describes the shape of a heatsink, then yeah maybe.

You know the heat equation right? Just use the usual heat equation but use cylindrical coordinates I guess. When I took PDEs, they only did cartesian x y, and spherical polar coordinates. What I mean is I've never solved any PDE in cylindrical coords so I am really unfamiliar with it lol

Hmm. I haven't done PDEs in months, and only briefly... so.... Maybe after this year's math class.

Yeah.. http://mathworld.wolfram.com/HeatConductionEquationDisk.html
That's at least a course or two beyond my knowledge :|

Not sure how that happened