Cohesive Air and an Obstinate Heatsink? 9/2/01

page 1 of 3

This cooling project does better with aluminum heat sink than with all copper!


You say you don’t want some strange looking blower hanging on the side of your case? OK, this kind of setup isn’t for everyone, especially if you are apt to tote your PC to LAN parties or wherever. For me, I needed to get better air cooling without the high pitched screech of high output fans.

So, the cohesive air cooling article by John ‘Hoot’ Hill looked like the way to proceed. See Hoot’s article, titled “Good Air/Bad Air, Good Fans/Bad Fans” and his follow up notes. I must note here that I do not own a digital camera so drawings and scanned images are the best I can do.

The overall system settings are as follows:

  • Motherboard: A7V133
  • CPU: Athlon 1000MHz AVIA stepping, running at 9.5 x 140 FSB or 1330 MHz
  • Mid-tower case
  • 10 gig Western Digital HDD – ATA100
  • Enermax 431 Watt PSU
  • BIOS is original Ver. 1004
  • Windows 98 STD

The temps were taken from less than scientific instruments – a good room thermometer for ambient temps, the average for the CPU temps from Motherboard Monitor 5 and an Enermax LCD probe positioned next to the CPU die.

Also, I do not favor the popular burn-in and benchmark programs as a means of determining stability in my case. Instead, I go about my regular use such as word processing, spreadsheets, photo editing etc. Actually, at one point I ran Prime95 for 45 minutes with the SK6 heat sink and Delta fan and saw the temps go DOWN by about 2 degrees! Go figure!

This cooling setup is the result of many ‘trial & error’ attempts to get the best cooling with an acceptable noise level. But then, isn’t that every overclocker’s goal? It is also an inexpensive setup in my case since I was able to borrow the blower used and spent only about $10 on the rest, not counting the heat sinks. Maybe I have been inspired by “Junkyard Wars”!


I use a 60 CFM, AC blower and a fan speed control and it has performed better than expected. It has an outlet about 2.5″ in diameter with 3 mounting holes.


Since I already cut a hole about this size in the side of the case, it would mount easily with the addition of 2 small holes for the bolts.
I then used 2 #8 stove bolts and nuts and mounted them as studs protruding out from the case side. This would allow me to place the blower onto the side as the final step and secure it with 2 wing nuts.


As a side note, I had cut this hole in the side directly opposite the HSF in an attempt to ‘port’ fresh air directly onto the heat sink fan. I used a stiff piece of paper rolled up to form the duct and positioned it just above the fan, getting too close at one point – but that’s another story. Anyway, it did help by about 2 degrees and I continued to use it until I installed the blower.

Now I needed to make a different duct so, being one to use anything that will work, I found some plastic water bottles that were about the correct diameter. Cutting off the ends and taping two of them together was long enough, so I slid one inside the other so that I could adjust the length to fit. Also, flaring one end, I glued it to a flange made from – what else – an old, unused mouse pad. Some ½” foam packing formed a nice airtight gasket and I was nearly there.


I needed to slow the blower, down so a trip to the store for a fan control did the trick. Mounting a workbox onto the blower and the fan control wired in – I was ready for a test. The blower worked just fine and I was able to regulate the speed easily.

Also, after realizing the vibration from the blower was felt throughout the desk, I got some rubber grommets and put them inside the blower mounting holes. This insulated the blower from the mounting bolts and quieted things down a bit. Since I am using an old blower, it is vibrating more than it should and I am sure I will be getting a new one later.

I have purchased several heat sink fan combinations over the past year and each has done a better job than the previous. That includes the Thermalright SK6, but the noise was intolerable for this overclocker! When I recently got my 1GHz Athlon and A7V133 mainboard, the temperature got too high at the 33% overclock.

That’s when I set up “Hoot’s cohesive air cooling”, which I call HCAC (with John’s permission). Before the SK6 heat sink, I was using a Taisol model CGK742092, which is an all aluminum heat sink. After mounting a 26 CFM fan from an FOP32-1, it performed OK but could not keep the temps below 50C, so I had high hopes for the SK6.

Continued on Page 2…

Dale DeHoff

page 2 of 3

Before I go any further, let’s have some discussion on the SK6 heat sink:

Upon looking at this new heat sink, there were very obvious machining marks on the bottom that could easily be detected by running a fingernail across them. Obviously these had to be removed. So “a lapping we will go”. At this point, I want to share my process of lapping the heat sink; skip it if you know all about it, but it may be of interest to some who may not fully understand it.

There is an image of the heat sink bottom (#1) as it came from the factory. The next seven images are the lapping progress – this is simply sanding the bottom of the heat sink to remove as much of the coarse machining marks while keeping the surface as flat as possible. The progress is as follows:

Image #2 is after 5 minutes of grit #320 using a figure-8 pattern.
Image #3 is after 15 minutes of the same paper and pattern.
Image #4 shows the progress after 30 minutes of the same paper.

I then change to a pattern of sanding in one direction from all 4 sides of the heat sink.

Picture #5 and #6 are after 10 minutes, each using this pattern and #400 grit sandpaper.

At this point, the machining marks are all but gone and about all you can see are the scratch marks from the sandpaper. This was all done with a 6″ square piece of tempered glass on a large piece of particleboard. The sandpaper is wet-or-dry automotive sandpaper and I use it dry, shaking off the dust periodically.

This is when I changed to the final #600 grit paper and another 15 minutes of sanding. Images #7 and #8 look the same, but there were the slightest traces of the machining marks in #7 to warrant the last bit of effort.

I agree with many that to go further than #600 produces little result for the effort. But, for those who prefer a mirror image, sand away!


The next image shows that the heat sink, when properly placed, leaves an imprint from the thermal grease that is the exact shape of the CPU die. I recommend doing such a trial placement to be sure the heat sink was properly placed and that the imprint shows the surface to be flat against the die.


There has been a great deal of discussion in the Forums about the way this heat sink mounts to the Socket A, so I won’t rehash it here. However, I do want to emphasize that since the clip is not ‘indexed’ into the heat sink, you must be sure it is all the way against the clip at the end opposite the screwdriver slot. I also note that this is not mentioned in the “Installation Guide”.

This “Installation Guide” accompanying the unit seems to be an attempt to oversimplify the installation but fails due to confusing information. Please note that I installed this thing twice without more than a glance at these instructions and only looked at them closely after deciding to do this write-up.

I read once that Thermalright Inc. had entered into the arena of heat sinks for overclockers with the introduction of the SK6. Well, here are a couple of quotes from the instructions that are, “in my humble opinion”, quite confusing.:

Step #4 says

“Insert the heat sink clip in the direction as shown in the below figure. First attach lever arm side, and then only use a flat screwdriver to put the cam box side.”

Now, aside from poor English, it would indicate the ‘lever arm’ and the ‘cam box’ were opposite each other, but they are not.

For attaching the fan with the included wires, step 5 says

“First insert the top fan clip into one side of the holes of the fan. Then clip to the hooks on to the bottom of the heat sink.”

The ‘hooks’ are on the wires, not on the heat sink, and I tried it this way; the only way you could do it is to use a pair of pliers to extend the wire far enough down to meet the holes in the heat sink. And pushing on any such tools near the motherboard is NOT recommended in my book.

It is much easier to engage the wire hooks into the heat sink first and then snap the clip over the edge of the fan. However, I found that there is not enough clearance on the “cam box” side to hook the wires in directly, so you must bend them at an angle first. What I don’t like about this entire process is the fact that unnecessary pressure is being put on the heat sink after it is mounted.

One last note about the instructions, and I promise this is the last: As confusing as they are, it also seems to me to be a bit insulting. A note near the bottom says, “As shown in the figure, fan at top, heat sink at bottom, fan clips on two sides.” I think the proper reply would be “DUH!” This doesn’t seem to be a sincere attempt to win the hearts of overclockers!

Continued on Page 3…

Dale DeHoff

page 3 of 3

Anyway, back to the project:

After taking the HCAC blower assembly off of the case and covering the hole, I mounted the SK6 with the Delta fan and reconnected the front intake 80mm fan and the top exhaust 80mm fan and fired it up. Well yea, it cooled like it was supposed to, but the noise really is unbearable.

While the SK6 with the Delta fan performed better than the Taisol with the 26 CFM fan, it could not win out with the HCAC setup. Going back to the HCAC system, I continued to chart temperature readings with the SK6 heat sink and it proved to be far less than useful for this setup.

The temps continued to be about 1 to 3 degrees higher than the Taisol heat sink, depending on what programs were being run. I have since abandoned the SK6 and have ordered the Taisol model CGK760092 with a copper insert. Perhaps the use of the copper and its capabilities over aluminum for the base will produce a more favorable result.

This chart shows some tracking of the ambient temperatures compared to the CPU temperatures.


Please note that because I could not stand the screech of the Delta 38 fan, there are only three plots of the SK6 with that fan, but the pattern is still apparent – aluminum outperformed copper. When the ambient temp is around 21 or 22 Celsius, the temps for the Taisol and HCAC stay at about 39c on the CPU.

As the ambient gets close to 28C, the CPU temp goes to nearly 48C which is a sign the cooling is effective but does not stay constant as the ambient rises. So, the change in room temp is just something I have to deal with. If it is summer and the weather is hot, I lower the speed to keep the temps from getting out of hand. Winter time – cooler – faster.

Perhaps an argument could be made here for the reason the fans force the air into the heat sink. It may be because there is as much of a need to push air from the center of the base plate toward the outside of the plates along the fins as there is to force air from the base plate to the top of the fins.

At one point I made an adjustment to the length of the chute so that it was within ¼” from the heat sink fins. My reasoning was that by concentrating more air to the center of the heat sink, it would move more heat away and more closely resemble the mounted fan. Unfortunately, this didn’t happen.

I also noticed that there was not as much air coming out of the hole in the top of the case. This told me that by concentrating the air into the center of the CPU heat sink, I was reducing the amount of air moving out of the case, thus not allowing the hot air from the heat sink to escape as fast as before. It was also decreasing the positive airflow within the case. I actually saw nearly a one-degree increase in temps.

The system is now running the aluminum heat sink and the HCAC setup. The transfer tube is about 1.25″ away from the heat sink and the to 80mm fan was left connected. It is working well for the warm weather here in the northeast.

My conclusion, and again this is an unqualified opinion, is that since the copper plates on the SK6 are very close together, the force of air coming from the Delta fan and its close proximity to the plates are the correct combination to get the needed performance. But, the tradeoff is excessive noise.

Unfortunately, the smaller spacing of the fins also inhibits the cohesive air from moving through the heat sink as well as it does from the high output fan.

When I get the copper based Taisol heat sink installed, I will post in the Cooling section of the forums with some results.

Thanks for your attention and happy computing!

Dale DeHoff

Be the first to comment

Leave a Reply