Five Myths about Cooling that You've Probably Read Online

By: pHaestus 8/22/02

Introduction-

I read a lot of different online forums and PC enthusiast sites daily. I typically hang out in the cooling boards because that is where my interest lies.  I have learned a lot over the years from overclocking and cooling nuts in many different online haunts, but I have also come across a fair amount of misinformation.  Since I have gotten (deservedly?) the reputation for being a hardass on the Pro Forums, I figure that I might as well earn the title by formally presenting and debunking some of the more egregious errors that I have come across online.

Myth #1. Aluminum cases act as giant heatsinks.

"These are my personal preference for a number of reasons, including light weight, durability, and, most important, the fact that an aluminum case acts as a huge heatsink drawing off the heat built up within the case. The difference in internal case temperature between aluminum and comparable steel cases is as much as 4 to 8° Fahrenheit or 2.2 to 4.5° Celsius."

 -SystemCooling.com
http://www.systemcooling.com/modules.php?name=News&file=article&sid=66

The entire concept as an aluminum case as a significant source of system cooling originated not in a laboratory or an engineer's plans, but with marketers. Any superiority of an aluminum case in cooling lies with improvements in airflow and increases in the number of fans; not with the function of aluminum cases as some giant heatsink. If you think about it logically, then this is clear. The case is not in physical contact with the hot parts of your system, so for the case to serve as a heatsink then heat would need to be transferred from warm air to the aluminum case and then dissipated.  This is NOT efficient, and for it to happen to any appreciable extent then there would need to be very little airflow in the case.  In reality, virtually all of the cooling of PC cases occurs due to air movement. Coolermasters and Lian Lis ship with a better stock fan configuration than most other mid towers, so they appear to cool better.  The fact that they are aluminum cases is incidental.

Myth #2. Copper absorbs heat better than aluminum, but aluminum "gives up" heat better than copper.

"The Z4 Aqua is the direct result of 3 yrs worth of ongoing R&D by CPU-FX .The Z4 Aqua has had several different maze designs prior to their new Gear Star maze design and while I really cant comment on how the older designs performed ,the new Gear Star maze design performs admirably. The newly designed Z4 Aqua combines a gear-star mazed silver plated copper alloy core for maximum heat absorption with an aluminium top for maximum heat dissipation"

 -AMDMB
http://www.amdmb.com/article-display.php?ArticleID=189

I am not sure where exactly that thermal conductivity was first split into two aspects, "taking in" and "giving up", but AMDMB are by no means alone in making this type of statement. Interestingly, Myv65 of AMDMB addresses this myth directly in their "General Heat Transfer Guide". To briefly summarize Dave's thoughts, the only real advantage that aluminum has over copper is its density.  If one is trying to remain within AMD or Intel's weight specifications then it is possible to produce an aluminum heatsink that is larger and with greater surface area.  However, if two heatsinks are of identical size and construction then the copper unit will perform better. The real reason aluminum heatsinks are common is the low cost and ease of producing them.  Copper isn't the ideal material to work with, as any machinist can tell you.

Myth #3. Northbridges need water cooling.

I can't trace this fallacy to any specific source, but you can find many makers and users of chipset waterblocks that are out there.

Chipsets don't generate very much heat, so there is no compelling reason to water cool it. On the other hand, the addition of a northbridge waterblock will cause a substantial loss of flow rate in the cooling loop. Flow rates can be considered to be the master variable in waterblock performance:

The above data was obtained with a 1600+ and an Eheim 1250/Maze2/Caprice heatercore/1/2" setup some time ago.  One can clearly see that lowering flow rates from the max obtainable (1.4 GPM) is detrimental to cooling performance. This is not a good trade.  Instead of plumbing in a northbridge block, just pick up a big heatsink, cut it to fit on the motherboard, and drill holes in it for mounting with good thermal paste.  If the heatsink alone isn't enough then by all means add a fan. Even an undervolted 50mm fan should provide sufficient cooling without having to resort to disrupting the CPU's cooling loop.

Myth #4. Radiate is an accurate program for estimating the heat produced by AMD CPUs.

Radiate is a program written by an avid overclocker to estimate the heat produced by many different processors when overvolted and overclocked.  It debuted on overclockers.com and has gained widespread acceptance in the community.

Virtually every hardware site that publishes C/W values for heatsinks uses radiate to determine the heat produced by their CPUs.  Unfortunately, the numbers that come out of it have little grounding in reality.  A bold statement?  Perhaps, but let's look at the supporting evidence. First of all, AMD's own technical documents for the "Morgan" Duron and Athlon XP processors report much different maximum power compared to Radiate. Furthermore, AMD implies that true maximum thermal load is not attainable with third party software and that "typical max" loads are around 88% of the theoretical maximum. If one simply compares the numbers presented by AMD to those obtained from Radiate then you get something like this:

Quite a big discrepancy. Some strong experimental evidence that XPs produce substantially less heat under load than Radiate predicts was provided by Cathar at O/Cers Au forums:

http://forums.overclockers.com.au/showthread.php?s=&threadid=60614

In short, Cathar estimated the total heat capacity of his cooling system and then used the initial rate of water temperature rise to calculate CPU heat production. He observed a heat input of ~ 66W for an XP that was rated by Radiate at 100W.

C/W values calculated by sites based upon these inflated watts are highly suspect.

Myth #5. Positive case pressure (more intake than exhaust) is good for PC cooling.

This will be the most unpopular statement that I make: most of the noise and air produced by highly modified fan-heavy systems is purely for show.

Many serious server cases operate with negative pressure, and most of the major OEMs have sold cases for years (some for decades) with no intake fans. AMD's design specs call for ONLY an exhaust fan in the rear of the case and an ATX 2.03 spec PSU (fan pulling air from case on bottom and through the power supply). Sure, the intake fans on Antecs and Lian Lis will keep hard drives cooler than a case with no airflow over the drives.  Are they actually improving the circulation of air through the case as a whole though?  Or are they disrupting the exhaust fan's ability to pull air through the system? Without a smoke test it is hard to say for sure, but I tend to trust Liteon, Chenbro, Palo Alto, and many other seasoned manufacturers of high end workstation and server cases more than I trust someone with a holesaw, a box of fans, and a garage.  There are other methods to directing airflow where needed besides massive CFM and noise. Many cases have louvres in the side panels that are designed to allow cool air to enter the case and be pulled across the hot CPU and then out the exhaust.  Intake fans can disrupt this design and may actually hurt CPU temps.  Dell servers and workstations often have elaborate ducting to pull heat from the CPU out the exhaust, but still operate sans intake fans.

The typical rationale for positive pressure in cases is that it protects cd-roms from dust build-up. This seems to be"a good thing", but can be pretty easily picked apart.  Plextor cd burners have shipped with exhaust fans for years.  These fans keep the inside of the drives cool by pulling outside air in and through the cd-rw.  This is the worst case scenario for positive pressure, but Plextors are among the most reliable of optical drives.

Conclusion

The internet is an amazing invention, with tons of information available on every conceivable topic. With this wonderful information superhighway, everyone can be a writer and share their knowledge, observations, and theories.  Unfortunately, there are very few editors.  A good rule (as always) is caveat emptor.

-Derek

If you have any comments or Questions please email me at pHaestus@ProCooling.com