I have seen such a horrid trend in commercial and diy cases in the last few years. Airflow in computer cases has become one of the most misunderstood and horribly confusing elements in computing.
The “more is better” idea seems like a good starting point, and to some extent more airflow will reduce temperatures. But when I was doing research for a new case design of my own a year or so ago, I found out that with the exception of a very few case manufacturers and designs, most cases are not optimized for proper airflow or cooling characteristics. What I am presenting here is just the results of my personal research and I welcome anyone else’s experience that may contradict or be an extension to my own.
The initial research
I figured there had to be research done on cooling constant enclosed air volumes, but it took me quite a while to find it. The research I used as the basis for my experiments and ultimately my case design was based on research in the commercial HVAC industry. For those unfamiliar with it, HVAC is Heating, Ventilation and Air Conditioning. It is a huge industry and has some amazingly well done research on closed volume heating and cooling.
What HVAC people have known for years
There are four main factors that contribute to the cooling effectiveness of any closed air volume.
- Pressure of the air within the space,
- Leakage of pressurized air from the space,
- Incoming air temperature and
- Flow over time (the cfm we all know and love).
The flow of air is broken down into three sub categories, total flow, flow per volume (how much of the air in the enclosed space is actually being moved), and heat source location(s).
There is a fifth factor used in commercial HVAC, system impedance, but this is nearly impossible to measure theoretically. Every case and setup will have different impedance characteristics, this is where experimentation is necessary to obtain optimal cooling.
The case should be as well sealed as possible. Many small form factor cases come with passive side vents and various holed and cutouts on the rear panels. Passive heat escapes are only effective in extremely low airflow situations though. Since most of us are concerned with performance, holes are the enemy. The more tightly enclosed your case is, the better it will hold pressure, which is key for effective cooling.
The case should be at as low a pressure as possible. This means you want more exhaust than input. This seems totally contradictory to almost every computer case on the market, and it is. Heat is dissipated much more quickly at low pressures. This is extremely true with air. Overall you want to achieve about a 2:1 differential of incoming to exhaust air. The math behind this simple ratio will be explained further down in the article.
Incoming air temperature is something we have very little control over. Unfortunately it’s also one of the major factors of cooling a space with air. One interesting, but pretty untested idea is to route hot exhaust air to a heat exchange, converting it into electricity and using that electricity to power a condenser to cool incoming air. For the rest of us, just keep your air intakes in a well ventilated room and keep the temperature in your house at a moderate level.
More airflow is not necessarily better. The idea is to keep all of the air within your case moving and exhausting. The reality is that in most stock cases you will only be moving 30-40% of the full air volume. Top blowholes and side blowholes will usually worsen overall cooling, and should be avoided. The idea is to have a single path of airflow. The problem as we all know is that it pretty darn hard to do in a computer case. The drive bays cause a tremendous amount of air blockage, the peripheral cards obstruct airflow to exhaust fans, the power supply often generates more heat than it exhausts and the CPU cooler exhausts its air perpendicular to the desired flow pattern.
How to make air flow properly.
Since most of us don’t have the luxury of designing a case for proper airflow characteristics, the next best thing is to improve what we have. In a traditional ATX case we will have to use a modified 2 path approach in order to get truly optimum airflow. In order to achieve this we will have to use a side mounted exhaust fan. Also since the front intakes are often blocked by one or more hard drives, we will have to beef up the intake cfm in order to get closer to an actual 2:1 airflow ratio.
The reason for the 2:1 intake to outflow ratio is VERY generalized. This rule of thumb can only be applied to cases with fairly unobstructed intake fans, with a reasonably clear airflow path to the outflow fans.
So where do I get 2:1? In general, the static pressure of a system increases in a parabolic curve as airflow increases. An intake fan’s airflow decreases fairly linearly as pressure increases.
So in order for a system to cool effectively, we must keep the pressure impedance on intake fans as low as possible. The lower the impedance, the more actual airflow inside the case.
The opposite effect happens at the outflow fans. Since we want the inside space to be at lower pressure than outside to aid in air intake, the exhaust fans will have very a very high impedance. In order to assure that they can exhaust air as fast as it is being drawn in, we must be able to overcome the impedance. This means more airflow at the rear in order to maintain internal low pressure. To actually get the case to a lower pressure to begin with, the exhaust fans must exhaust more total volume than the intakes bring in.
Luckily for us, under general conditions, at low altitudes, the difference is only about 1.3-1.5:1. This data was obtained from my empirical testing. If you live at high altitude, or do not have a fairly standard ATX mid tower to work with, this will be different.
So a 2:1 ratio provides enough overhead to produce low pressure and maintain full exhaust.
I am going to assume the case in question has room for two 80mm intake fans in front of the case, one side blowhole mounted directly over the peripheral cards and a single read 80mm exhaust fan in addition to the power supply fan.
The typical 80mm case fan puts out 20-25cfm of air. If at all possible to reduce noise levels, I would suggest modding your case to use 90mm or 120mm fans instead. But we will go with 80’s for this guide.
If you are only using a single hard drive (meaning only one drive blocking two intake fans) then you can skip the beefing up the intakes.
I am also going to assume this case is using two drives running in a nice raid0 array (nice and HOT that is). The first step is to make sure the intake fans have a nice open space to draw air into the case. Most cases cover 80% or more of the fan surface with grating or grills in front. Get out your dremel and get rid of it. If you want to hide the fans themselves, get some nice high flow air filters (you should really use filters anyway to prevent dust buildup in the case). On the other side of the fan we have hard drives. Each drive positioned directly behind an 80mm fan cause the fan to lose ~25-30% of its effective airflow. We are going to say we want 50cfm of inflow, so we will need two fans with a rated airflow of at least 34cfm each. To be on the safe side, we will go with two ~40cfm intake fans because a lot of this flow will be lost from interference.
Next there needs to be a divider put into the case. It should direct one intake fans flow to the upper component area (cpu, memory) and the other fan to the peripheral area. This can be as simple as a piece of foamboard, cardboard, plexiglass etc. A good principle is for it to line up with the rear of the AGP card on one end and the end of the drive bays at the split from the intake fans on the other end. Fastening this divider to the side panel is a very good way to manage it.
In the CPU area we need to look at how the CPU fan works. Most fans are configured to blow air down across the heatsink fins and out all sides. In order to make this process more efficient we need to put a shroud over the cpu fan. It should face toward the FRONT of the case, to ensure the air being drawn in is from the intake air and not from recirculated hot air that needs to be exhausted. This can be done very very easily, just build a 3 sided cage blocking the rear and sides from the CPU fan, this too can be secured to the case side panel for ease of use (can you tell Im not a big fan of side panel windows?) However to prevent turbulence, the shroud should be highly rounded and directed.
Here is a very cheap and quick example of a well constructed fan shroud for the CPU:
The exhaust fan for the side blowhole should be either a 90mm or 120mm fan rated at at LEAST 80cfm, 100cfm would be ideal. A 100cfm 120mm fan is very quiet in operation and so would be my recommendation.
The exhaust fan for the rear of the case should ideally match the side exhaust, however since most cases only have room for a single 80mm fan we will have to improvise. The power supply generally exhaust about the same as a typical case fan, 20-25cfm. So if we can get a 40-50cfm case fan, it should be adequate since we have a large enough side exhaust to promote low pressure within the case. If you have a full tower case and two rear exhausts, get two 40-50cfm exhaust fans, or cut them into a single 90 or 120mm exhaust.
Get a smoke gun. You can find them at most hobby shops or you might be able to use anything similar. If you smoke you can also use cigar smoke, cigarette smoke or ahem marijuana smoke if so inclined. Start up your system (filters removed) and send some smoke into the intakes. Let the system run for 3-4 seconds and watch the smoke come out of the rear of the machine. If you do have a window, look inside after a few seconds and look for pockets of stale (smokey) air. There should only be one, that is up in the top area of the case near the optical drives. This is okay as there is very little heat put off by these devices. If you see other “dead” spots, then you need to adjust the locations of your dividers accordingly.
The other test is for pressurization. Without specialized equipment it’s pretty hard to get an accurate reading of your case pressure, and it’s not going to be very high or low no matter what you attempt to do. For computer cooling a pressure of .8 to .5 ATM is about the best you can hope for and most people will see very good results with even .9-.8 (1ATM is the pressure at sea level for reference). Those of you at high elevations have a natural advantage. But to perform a simple test to see whether you have overall positive or negative pressure, go buy a 99 cent piece of foamboard. Cut the board to slightly larger than the size of your side panel and rig the exhaust fan onto the foam board so we dont lose it’s flow.
Power on the system and lightly hold the board to the case. if you notice there tends to be a pull on the board trying to suck it into the case, you have succeeded! If it’s being pushed out away from the case you need to rework your intakes to draw less air in, and beef up your exhaust fans (may be as simple as just reversing them!).
So how much does this really matter? In testing my case using a passively cooled PIII450 processor I found that moving from a standard ATX configuration to one with a 2:1 flow ratio and properly aligned flow patterns resulted in an average 13-20 degree F reduction in case temperatures. That is a HUGE reduction, especially for a stock, non overclocked setup. The more heat your system puts out, the bigger the delta will be.
Also do not forget the importance of proper cabling. All the planning in the world will go to hell if you have cables blocking your air routes. Use rounded cables, and whenever possible push the cables behind the motherboard and behind the drive bays to completely remove them from the airflow.
See a Proof of Concept Case using the theory in this guide here: WindTunnel PC1