Artificial heatsink tester

[krazyups]

So looking around the net at various sites' heat sink reviews got me thinking (which is sometimes a bad thing!); since most sites use a computer to test the performance of heat sinks, including different cases with various airflows, components, cable management, etc., would it be beneficial to create an artificial load that would be used to test the sinks instead?

My idea: basically it stems from getting my hands on an i7 IHS, which shouldn't be too difficult I would imagine, with all the people on XS removing them for direct die cooling. After that, I would get a case that would be sufficient to handle all heatsinks currently on the market (I was thinking of the CM Storm Scout), and something like the NZXT Sentry LX, which would give me control of all of the case fans, along with two additional channels for controlling the heatsink fan (or fans, in the case of the Nocuta coolers and the other coolers where push/pull is an option). The Sentry would also allow me to monitor the temps at 5 positions (intake, exhaust, IHS, heat sink itself and exhaust of the heat sink itself perhaps). For the load, I would grab a small cartridge heater (3/8" round by 1 1/4" long) embedded into a 2" x 2" x 1" block of copper capable of outputting 400W of power. This would be positioned in the area of the case where the CPU would normally sit, so as to try to create semi-real world conditions. Control of the heater would be relatively simple; a 555 timer circuit to send a PWM signal with variable duty cycle to a solid state relay would be sufficient, and inexpensive. A generic mounting mechanism would ensure that any heatsink for any platform could be tested (I haven't given this part much thought).

I'm only thinking about this because when I look around at all of the possible heat sinks to use, the results between any number of them are within a couple of degrees ... easily within the margin of error for the test, so it's difficult to declare one heat sink the clear winner when it only performs 1 - 2°C better than another. Also, I work in a lab so I'm used to seeing this type of thing and coming up with solutions to increase the precision of the testing.

Oh, I might as well mention before I get bashed that I've chosen the components that I did based on me forking over the ~$400 it would currently take to build it.

After all of this, I guess my main question is; would it be beneficial to HWC to have this type of tester available? I have no problem putting it together, but I as with a lot of people, money is quite tight right now and I can't justify the expense if it's not going to serve a purpose.

Thanks for reading all of this, and let the bashing begin!

-Steve

[JD]

I'm guessing it would be similar to FrostyTech's setup?

FrostyTech - Best Heat Sinks & PC Cooling Reviews

[MpG]

The problem with that sort of artificial tester is that one of the most important factors in a heatsinks performance is how well it mates up to the processor. Removing the processor from the equation, even if you're replacing it with a makeshift IHS, really makes it difficult to equate the results with real-world performance. Additionally, the IHS of the a CPU doesn't heat up evenly, nor the is heat source centered on the IHS, and a heatsink can easily rise or fall based on whether or not it takes advantage of those items. An artificial cartridge or tec heater doesn't mimic those kinds of conditions, and the usefulness of the testing suffers as a result.

Also, in regards to your statement that a few degrees difference is within margins of error - yes and no. Good reviews will incorporate multiple good mounts, and account for how consistent the achieved temperatures were. While this certainly doesn't eliminate the margin of error, it actually shrinks it quite a bit, to the point where the results take on some meaning.

[gingerbee]

Originally Posted by jdrom17

I'm guessing it would be similar to FrostyTech's setup?

FrostyTech - Best Heat Sinks & PC Cooling Reviews

beat me too it i really like they way they do it if your cooler sucks then it sucks period

[krazyups]

Originally Posted by MpG

The problem with that sort of artificial tester is that one of the most important factors in a heatsinks performance is how well it mates up to the processor. Removing the processor from the equation, even if you're replacing it with a makeshift IHS, really makes it difficult to equate the results with real-world performance. Additionally, the IHS of the a CPU doesn't heat up evenly, nor the is heat source centered on the IHS, and a heatsink can easily rise or fall based on whether or not it takes advantage of those items. An artificial cartridge or tec heater doesn't mimic those kinds of conditions, and the usefulness of the testing suffers as a result.

This is why I proposed that the actual interface to the heat sink would be an actual i7 IHS, salvaged from a damaged chip or from someone who removed it successfully to incorporate direct die cooling. Also, the cartridge heater, being only 3/8" in diameter, could be offset from the centre to mimic the loading of one or two cores, which is what happens in reality when only partially loaded. Really though, the bottom line is how effective the heat sink is at removing the heat and transferring it to the passing air, and I would think that an artificial, precisely controlled load, with no other influences and fully controlled fans would make for a more "apples to apples" comparison.

Don't get me wrong, I'm not advocating the complete replacement of the existing setup, because the mounting of the heatsink does play a major role, as well as other things like obstruction of the ram slots, rubbing capacitors, etc, I am just likening this to the same thing that we do for processor and GPU reviews, you do some synthetic benchmarks first, then follow it up with more real world conditions.

It's an idea that's been stuck in my head for a while, and I figured that I'd figure out one way or another if it's a good, feasible idea before I become obsessed with it.

Good points by the way.

-Steve

[krazyups]

Originally Posted by jdrom17

I'm guessing it would be similar to FrostyTech's setup?

FrostyTech - Best Heat Sinks & PC Cooling Reviews

Yes, I've looked at their setup, and it would be similar, I can't say that I didn't draw some inspiration from them, but but it would be leaning a little more towards real world. Plus, it would have much more control, being able to load the heatsink with any power ranging from near 0 to 400W.

-Steve

[MpG]

Originally Posted by krazyups

So looking around the net at various sites' heat sink reviews got me thinking (which is sometimes a bad thing!); since most sites use a computer to test the performance of heat sinks, including different cases with various airflows, components, cable management, etc., would it be beneficial to create an artificial load that would be used to test the sinks instead?

This didn't occur to me earlier, but now that I think about it, your proposed cartridge heater system doesn't actually address any of those variables you listed.

Another point to consider: mounting mechanisms are as much a part of the battle as other aspects of heatsink design. A generic mounting device would fail to account for that.

[krazyups]

Of course I was hoping that the tester would become sort of the "standard" as far as synthetic benchmarks go for heatsinks, I still look at it as sort of a business, being able to supply review sites with exactly the same setups to eliminate all of these potential issues. Also, the tester wouldn't have any additional heat generating components, and the airflow could be accurately controlled.

As I mentioned above, I'm not for removing the real world testing, I'm just seeing this as an addition to the testing arsenal.

-Steve