Case Cooling Considerations
This content was originally featured on Amdmb.com and has been converted to PC Perspective’s website. Some color changes and flaws may appear.
Case Cooling ConsiderationsWell, we’ve beaten the concept of chip cooling to death. Along the way, we noted that the temperature of the air surrounding the chip cooler is of great importance. Here is where the discussion of case cooling begins. First of all, the obvious points must be mentioned. If using air or water cooling, the components inside your case can be no cooler than the air or water used to cool them. The air or water inside your case can be no cooler than the means used to cool that air or water. The vast majority of users today circulate room air through their case. The air in a given room is generally cooler near the floor than near the ceiling because air density decreases as temperature increases. The air in a given case is generally cooler near its bottom than near its top for the same reasons. In my particular situation, the air at my keyboard tends to be about 1.5°C warmer than the air at my PC’s intake fans.
As air circulates through a case, it picks up heat from the components in the case. This decreases the density of the air and helps it rise to the top of the case. When it exits the case, it will be warmer (T2 > T1) than when it entered. How much warmer the air gets depends entirely on the quantity of air circulated through the case and the amount of heat generated within it.
Here’s the relatively simple part. The density of air and its ability to hold heat vary with temperature, but this variation is pretty small over the typical range that applies to PCs. In the range of 300 to 350 Kelvin (roughly 80°F to 170°F or 27°C to 77°C), the density of air drops from 1.1614 kg/m^3 to 0.9950 kg/m^3 and the specific heat goes from 1.007 kJ/kg-°C to 1.009 kJ/kg-°C. What this means is that the mass of air circulating through the case and its ability to carry away heat are not particularly variable over this temperature range.
Knowing this, we can estimate how much airflow is needed to carry away all the heat while maintaining a decent temperature differential in the case. As a ballpark figure, let’s say you’ve got (2) 42CFM inlet fans and (2) 42CFM exhaust fans along with a power supply exhaust fan. The total airflow through the case should be somewhat higher than 84CFM, say 90CFM as an approximation. 90CFM equals 2.549 m^3/minute. Given an air density of about 1.16 kg/m^3, this means you are circulating about 2.96 kg of air through the case each minute. Now let’s assume you’re using 220 watts of power inside your computer at steady-state conditions. Air holds about 1.007 kJ/kg-°C and you’re pumping 2.96 kg/minute. Multiply these two and you get 3.163 kJ/°C-minute. Well, one kJ/minute equals 0.016667 kJ/second. Since one joule per second equals one watt, this last number equals 16.667 watts. So 3.163 kJ/°C-minute equals 3.163 * 16.667 = 52.717 watts/°C. Since we need to get rid of 220 watts, our air going through the case must heat up by 220 divided by 52.717 equals 4.17°C
Of the 220 watts in our example, say 70 watts comes from the CPU. This is about 1/3 of our total heat, so the air should be about 2.84°C warmer before any heat from the CPU is added. The constant value 1.71 in the equation above handles the air properties plus unit conversions. It is the correct constant only at the conditions stated. As air temperature increases, this constant will decrease slightly resulting in a higher temperature change for a given power and airflow.
Important Note: All of the above calculations are extremely simplified. The biggest assumption is that the air moves cleanly through the case and that all air maintains the same temperature at any given location. The fact is that airflow inside a typical case is anything but uniform. While we can be confident in the average exit temperature based on the total airflow and power, we can’t be certain that there aren’t hot spots within the case. So what happens if we decide to double our case airflow? The total differential will be cut in half from 4.17°C to 2.08°C and the air around the CPU will drop by about 1.42°C. Probably not as large a drop as one would expect from doubling the airflow. If this is true, what’s the deal with people that install large quantities of case fans and extol the virtues of high airflow? Or how about those people that claim that it’s better (cooler) to run without a case cover?
Well, there are a few things to consider. First is the total system power. If you’ve got multiple hard drives and are forever burning CDs while your five PCI cards and Geforce3 suck down power, you may be using over 400 watts inside the case. The same 90CFM would mean a temperature rise of nearly 8°C in this instance. More than 90CFM would definitely be warranted. More often than not, however, the problem is lack of proper circulation within the case. In an ideal world, cool air would enter at the bottom of the case, flow uniformly over all the motherboard’s hot spots including the HSF, and exit through the exhaust fans at the top of the case. More commonly, flow through the case is sporadic with large fluctuations in flow velocity that result in dead spots and hot spots.
For those without good circulation or enough volume for their power use, it’s often true that removing the case cover will result in lower motherboard and CPU temperatures. This is far from the preferred solution, though, due to rapid dust accumulation, the risk of static discharge, and the unsightly look of a “naked” computer.
One also needs to be aware that there is such a thing as too many fans. Each fan draws power to create airflow. The fan converts this electrical power to heat through motor inefficiency and direct compression heat in the air. If a fan draws 5 watts to generate 42 CFM, it will raise the temperature of that air by about 0.22°C. This is a small number to be sure, but certainly not zero. So how many CFM are enough?
A respectable target is to maintain the motherboard temperature within 1°C of the ambient room temperature. Since this temperature sensor is near the south bridge on most motherboards, it is measuring the temperature fairly near the air inlet before much of the heat gets added. This means the exhaust temperature will be more like 3-4°C higher than the inlet. For a typical system consuming around 220 watts, about 100 CFM should be sufficient. As a rule-of-thumb, airflow equal to about 1/3 to 1/2 the power should be sufficient. Note that for this rule, power is in watts and airflow is in CFM. As airflow exceeds 1/2 of the power, expect rapidly diminishing benefits from the added airflow.
The reason more airflow yields diminishing returns lies is the energy storage capacity of air. 100 CFM will carry 220 watts of power in a temperature rise of about 4°C. 200 CFM will carry this same energy in a temperature rise of 2°C. Even though airflow doubles and temperature differential is cut in half, the gain in system temperature is no more than 2°C.
If you see high system temperatures despite having sufficient airflow, investigate flow restrictions like fan grills/covers and the internal arrangement of your case. Often times simply opening up fan covers and/or rearranging power and drive cables results in dramatic temperature improvement. Finally, some people have noted that an increase in case temperature of 5°C results in a CPU increase of greater than 5°C. There are a number of factors involved that cause this. Among the most influential is air density. As air warms, its density decreases. This decreased density translates to lower mass airflow. This lower mass airflow means a higher differential is needed to carry the same amount of heat away from the computer. Aside from this, the conductivity of materials tends to decrease slightly with increasing temperature. This adversely affects the CPU temperature, too.
