HOME
What's New

CPU Cooler
Programs
Introduction
AmnHLT
CpuIdle
Rain 
VCool 2.0 
Waterfall Pro 
Current Tests 
Test Archives

UDMA Driver 
Tests 
Windows 95
Windows 98
Windows NT

Way 2 Cool 
Projects 
Compaq 1800T
1800T Temp Monitor
Celeron-A
Going for 504
More on 504
Cheap Sandwich
BX6r2 Diode Fix
Condensation
Lap that Slug
Celeron PPGA
Journey to 1080
Building an NT Box
Heat Pipe 1
Heat Pipe 2
Heat Pipe 2.2
gp_Temp
gp_Temp4
gp_Temp5

Heat Sinks 
Aavid
A&C Prototype Sink
Alpha P125
Alpha P521
CPU Cooler
FullWay
Tom's Sink
Water Cooler 1
Water Cooler 2
Water Cooler 3
Water Cooler 4
*
Water Cooler 6
Water Cooler 7

Way 2 Cool
Thermometer
Thermometer 2
K-6 heat sink
Case Cooling
Heat Sink Tester
More Case Cooling
Klamath heat sink
Q500A Cooling
My Server Temps
Temp Calculators
CKS400 Rack Case
Fan Controller
Cooling Links
Downloads Page

Tips and Trix
Dual Boot Win95 / 98
Dual Boot Linux/NT4
D-B Linux/Win2k
Hot Flash your BIOS

Family Stuff
Prom / St. John
Providenciales

Email Jim


After my last few projects, I had come to understand a bit more on basic heat sink design and decided that it was time to move to what I considered the next step.  Up to this point, all of my heat sinks had been made of aluminum.  I had done some playing with copper for hot and cold plates and found how much better copper was for conducting heat. I decided to move to copper for the heat sink base for its better conduction.  While copper is a good conductor of heat, it is inferior to aluminum for getting rid of the heat it has absorbed.  The proposed solution was to use each of them for the properties they did best.  Copper for the base to collect the heat quickly and aluminum for the fins to get rid of it.
 
Thermal Properties of Aluminum and Copper
Property Aluminum Copper
Thermal Conductivity 
@ 0 - 100°C
237 W m-1 K-1 401 W m-1 K-1
Specific Heat @ 25°C 900 J K-1 kg-1 385 J K-1 kg-1
Linear Expansion Coefficient @0-100°C 23.5 x10-6 K-1 17.0 x10-6 K-1
Weight 2"x1"x5" block 0.9818 pounds 3.2443 pounds

Thermal Conduction
At  the molecular level, thermal conduction can be viewed as a transfer of energy (in this example, call it vibrations) from one particle of a substance to another particle of that substance due to collisions between the particles. As temperature increases, the vibration of the particle (molecule) increases and the collisions between molecules also increases. These molecules collide with each other and transfer energy from the more energetic to the less energetic molecules. The transfer of energy is always from active to less active - from hot to cold.  This net transfer of energy is known as a diffusion of energy.

Elements and compounds with closely spaced molecules will generally allow stronger and more frequent interactions between molecules.  Good electrical conductors like aluminum, copper, and silver generally have a higher thermal conductivity than nonconductors like wood and glass. But there are some exceptions to the rule, nonmetallic solids such as diamonds and beryllium oxide can transfer heat energy more efficiently than aluminum. 

Specific Heat
When we heat an object, we increase its internal energy, and hence its temperature.  How much the temperature increases depends on the amount of heating (Q, measured in joules), the amount of material (m, measures in kilograms) and the type of material.  The type of material is described by its specific heat capacity (c, usually measured in joules per gram per degrees Celsius).  The specific heat of a material is the number of joules required to raise the temperature of one gram of the material one degree Celsius. 

Because I wanted to build this in my workshop, I had to figure a way to bond the aluminum fins to the copper base.  This proved to be the first in a number of stumbling blocks I would encounter.  Between the library and the internet, I researched the possible ways to accomplish this.  There are a number of high-tech ways to do this - including using high explosives to literally "blow the two metals together."  While this did sound like something I would enjoy trying, I had to rule out the high explosives or move to a much less densely populated area.

I settled on a solution that, while not sounding like something very difficult to accomplish, took days of trial and error to come up with the proper measurements to produce the desired results.  The solution was to use pins instead of fins and to attach the pins by using the thermal expansion of metal as an means to get a good fit.

I started with a 5" by 2" by 1/4" piece of copper C110 bar and bored about 240 holes to a depth of 3/16".  The diameter of the holes was a couple thousandths of an inch less than the diameter of the aluminum pins that I would be installing into the holes. Here I spent many hours boring different diameter holes to check the fit of the pins.  I then heated the copper bar and froze the aluminum pins.  This was to expand the holes and shrink the pins.  One by one the pins were forced into the holes.  When the temperature came back down to room temp, the objective was as tight of a fit as I could get.  After getting the sizes right, I found that I could get about 5 or 6 pins in before the base would cool off too much and would have to be re-heated.  Talk about tedium....
 

Two rows completed - just a couple more to go.

Installing the pins went on for what seemed like weeks, but I finally got them all in.  I knew that I wanted to cut the heads off the pins to make the pin's profile a little smaller, but I should have done the cutting before I installed the pins.  I ended up cutting the pins assembled into the sink on the band saw.  Not the easiest task, as the pins tended to vibrate and I had to insert wooden wedges between the pins to keep them still.  I wouldn't make this mistake the next time.
 
 


The pins are in and cut to length.

To get some air moving through the pins, I needed to add some fans.  Fortunately, I HAD thought about that before starting to build.  The fan shroud that comes with the Alpha P125 was the route I decided to take.  I added a couple of 60mm, 12 volt, 3800 RPM, 25 CFM, Orion fans to the top and the usual array of temperature sensors - 1 on the edge of the Celeron slug and 1 on the sink itself.  My trusty Aavid heat sink took its place as the backing plate to anchor the Celeron in place without bending the circuit card.
 
 

Fans, shroud, Aavid rear plate, Celeron, and sensors
installed make for a busy looking package.

Performance for the sink was good.  It edged out the Alpha P125 is all categories, but not by any drastic margin.  It did respond more quickly to the changes in temperature the Celeron goes through in a normal session though.

Even with this sink, my stubborn Celeron still will not reliably run at 504 without additional cooling.  Though I will say that the longer I run the Cele at 504 with a peltier attached, the easier it is to keep it at 504 without a TEC.  Where I once couldn't even boot into Windows at 504 without the extra cooling of the TEC, I can now boot on most occasions and when I get lucky, even run a few tests.  But as usual, as soon as I stress it to any degree, it's lock up time.
 

The heat sink hardly notices that the processor is active
in the 4 minute boot up test.  That's cool!

Because I can't measure the internal diode temperature during boot up, the above CPU temperature is the reading taken from the side of the Celeron's slug.  From months of watching the correlation of the slug temperature to the temperature of the thermal diode, I have seen that the diode temperature can run as much as 25° F higher than the slug for short periods of time.  The slug temperature also lags behind the diode temp by 10 seconds or so.
 

In the 7 minute Quake2 loop, the sink easily handles all 
of the heat that the processor throws its way. 

I found, as I expected, that for non peltier applications, this thing is overkill.  But that was not the reason I built it.  So, it's time to throw a thermoelectric cooler on it and see what happens.

Add a TEC into the mix ==>

 
Page 1
Page 2
Page 3