Science Fair Project

[skybolt_1]

Hey everyone. I am trying to think of a good, budget concious way to make the following science fair project...

I want to show the effects of heat on processor performance. I want to have 3 different models: Hot, Room temp and Cold. I am probably going to use two Peltier coolers for the hot and cold, but I only have one motherboard. What I want to do is have three 486 DX processors, and plug each one in turn. Does anyone have any better ideas? will it work? I plan to run benchmarking software to test the processors.

[Roy]

Sounds like a good demo!

You can just reverse the polarity on a TEC to make it change the direction it pumps heat. But you might not need to pump heat into the CPU, just turn off the TEC and quit removing it's own heat.

You could get it down to one processor by using a low current mode on the TEC to simulate a passive heat sink's cooling.

An independent temperature display using a probe touching the CPU would be nice, a Radio Shack indoor-outdoor thermometer might work.

[skybolt_1]

Sounds good, Roy, but I have just one question. How do you reverse the polarity of the TEC? By connecting the positive wire to the negative wire, and vice versa? Or is there something else im missing. Thanks.

[Mungla]

You got it. To reverse the polarity on a peltier, or anything for that matter, just hook the positive to the negative, and the negative to the positive.

[voogru]

and what ever you do dont let ANY one touch the CPU (you may get sued for burning someones finger off http://sysopt.earthweb.com/forum/smile.gif )

[Mungla]

Defently. Using a peltier, which only a few people know what one is, always attracts a crowd. Don't let someone steal one! lol. ;(

[Donkey]

Do the HOT demonstration last incase you go slightly too hot and ruin your CPU

[Roy]

I've been thinking about your project.

I'm not sure that benchmarks will change significantly ~ or at all ~ with the varying degrees of heat extraction. In my experience with overclocking, higher temperatures simply cause the system to shut down or fail to start in the first place.

If you can overclock the 486 setup, maybe timing the interval from boot to lockup at various temps would be useful. It also could be a faster way to show the importance of proper cooling.

Mungla's comment about the crowd's fascination with TECs is worth supporting. A TEC mounted so it can be grabbed between finger and thumb, and a momentary switch to connect a 9V battery will give people a good demo. A two-way switch that would demonstrate reversing polarity also might be interesting.

I have some miscellaneous parts and a few TECs you could use if it would help.

[skybolt_1]

Hey Roy, that may be usefull. How much do you want for a TEC? I've already got a complete Dell 486 lined up for my base system (plus i have a spare proc "just in case")

By "timing the inteval between boot and lockup" do you mean set a stopwatch the moment i turn it on till the moment it ceases to respond? That might work. Also, you say you dont think that the benchmarks would change signifigantly to the rise or fall of the temperature. Is this because the temperature differences I could achieve are not drastic enough? I was thinking of getting hold of some liquid nitrogen, my biology teacher could probably get me some. Would such a drastic drop create a noticable benchmark change? I should think so!

Anyway thanks for your help i appreciate it.

Joe
skybolt_1

[Roy]

Benchmark tests measure performance of a working system, not how well the system is working. A faster processor improves the results, but the same speed at different temperatures probably does not. Remember that the Scientific Method dictates that only one variable should change with each run.

Another demo might be to find a very demanding process and, again, clock the interval to lockup at two elevated temperatures. This assumes it works OK at some "normal" temp. I use the resizing of a gigantic graphic file to find out if a system is stable. If the resize completes, it's good.

The parts? They would be yours to use for the duration of the project. Let's continue this subject via email.