Eric Brasseur Home    |    Links    |    Contact    |   
   



Dealing with computer dust





One problem with computer fans is they get stuck in dust. Whether it be a little fan on a graphic card or the computer's huge power supply block fan, after a year or so they are smeared in fine dust filaments. If a heatsink is placed beneath a fan it can accumulate quite a bunch of dust. This lowers the cooling performance of the fan and even can make it become noisy. Some people do open their computer once a year and vacuum clean all that dust inside and around the fans. Other ways round can be these:


Self-cleaning fans

The obvious way to make a self-cleaning fan is to make it able to turn in both directions. By blowing the opposite way the fan will blow away at least most of the dust that accumulated. There are variants to this basic principle:
Blowing the other direction requires the fan blades to be flat and not curved. In order to be able to blow efficiently in both directions. This in turn requires the fan blades to be not too much inclined. Which requires the fan to run faster to get a same air flow. Or to have two propeller stages and maybe two static stages like a compressor. The fact the fan has to be able to blow a least a short time at high speed requests for a stronger motor and a little more sophisticated electronics. This ensures for a more reliable functioning at standard speed which is a good thing.

Next puzzling is what should trigger the fan to blow the opposite way. Many criteria can be proposed:
What will happen with the dust blown away? Part of it will fall to the bottom of the computer which is not a problem. Another part will be propagated to the room the computer is situated which is more problematic. It may also trigger health accidents like asthma. Opening the windows can help. A partial solution could be to synchronize the fan cleanings. First the computer's main fan situated in the power supply block should do a short and powerful cleaning. This will generate a dust cloud inside the computer case. The fan should immediately stop. Then the other fans, situated on the motherboard and cards should clean, adding their dust to the the dust cloud inside the case. Then the computer should go into low power consumption or shut down, till most of the dust did deposit on the bottom of the computer case.


Aspirate the air from the heatsink instead of blowing towards the heatsink

If the ventilator blows towards the heatsink the dust will accumulate between the ventilator and the heatsink. This makes the dust quite hard to clean. A solution is to turn the ventilator 180° and make it aspirate the air from the heatsink. That way the dust will accumulate on the open sides of the heatsink and it will be much easier to clean. I wrote a Web page about this.


Filtering the air

The air that goes into the computer can be filtered. This can be done by making wide openings in the computer case (covered with conductive grids to disallow RF noise). Then tape pieces of tissue on all openings. Probably they should be replaced or washed once in a while. The more you're using your computer the cleaner the air you breath.

Another way round is to turn the computer mains fan so it blows the air towards the inside of the computer. Then place sort of a car air filter at the entrance of that fan. This can be done a neat way. The advantage is the rest of the computer case can remain the way it is.

A friend used to let an electrostatic air purifier on in his dark chamber. This allowed him to get a clean room and produce neat photographs. All devices inside the dark chamber remained dustless. Such a purifier is not very expensive. Yet be careful not to take risks for your health. Bad purifiers can produce dangerous ozone gas.

Another friend is going to tape tissue filters around the air inputs of his processor heatsink. These filters need to have a lot more surface than the surface of the heatsink inputs.


Liquid cooling

A high-tech way to cool a computer's hot parts is liquid cooling. The advantage is you don't get the parts dust in by an air flow. Even if they should get dust in this won't hamper. The liquid circuit's heatsink itself can get dust of course. But it can be placed a way that makes it easy to clean it.


Static cooling

A huge heatsink does not need a fan. They imply slow air speeds which cause few dust deposit. A compromise is to use such a huge piece to thermally link the warm part with the case. The problem with this solution is it implies huge, expensive and heavy heatsink parts.

Another compromise is to use a regular heatsink, just a little bigger, and a slow air flow by a big ventilator turning at low speed.


Mechanically cleaning heatsinks

Insects clean themselves continuously using their legs. Maybe a fan and heatsink system can do the same. Sort of a comb system interlaced with the heatsink can be kept slowely moving and scrape most of the dust away.


Closing the computer case

The computer case can be completely closed. That way no or few dust can penetrate and spoil the fan and heatsink blades. This can be done by shifting the power supply block a few centimeters to the inside of the computer and tape close most openings and grids. The point is a reasonably strong fan should be present inside the case, to make the air move and mix fast inside the computer. The outer surface of a computer case is quite huge. To me this is enough to evacuate the heath provided the inside fan ensures the hot air is continuously smeared all over the inner surface. Maybe the power supply block fan is enough to achieve this, perhaps helped by cardboard shapes. 'Think I'll try that on my computer.

Should the case surface not be sufficient then that surface can be rippled like heatsinks are. An outside huge and very slow turning fan can be also be directed towards a part of the surface.

This approach has also to advantage to lower the computer noise. But this means the inner sides of the computer case should not be too much covered with thermally insulating sound-deafening materials.

I didn't try out with my own computer because it's not in a case. Its components are spread all over my table. I used an old Pentium II computer a friend gave me. It has a horizontal case which makes the experiment a little easier too. I unscrewed and pushed the power supply block about 5 centimers towards the inside of the case (see the drawing below). I put a cardboard wall (red) above it in order to force the air to circulate inside the case and not be swallowed immediately back again by the power supply block's fan. I closed all holes of the case with removable tape (especially the 9 centimeters diameter wide hole left by the power supply block fan) (orange). There just remain a few half a millimeter wide slits around the case's surface (actually there is no need to make the case hermetically closed and this could even be dangerous because of water condensation). The result is perfect. I launched a very long and power-consuming calculation. After half an hour time most of the case surface simply is cold. The power consumption of that old PC is about half that of today PC's. Anyway a little less cold than perfectly cold would still be very cold. The only tepid part I could feel is a little surface part at the rear of the case. That's because the mains' fan, which is now enclosed inside the case, is blowing towards towards that part of the case. The temperature is that of the hot air that fan always has blown. Two decimeters to the left of this tepid part, above the motherboard, the case temperature is cold. It really is a success. With no design effort. The computer also makes less noise. No opening or expansion possibility has been compromised. I don't understand why this design is not widely used. Since a human generation time personal computers have gone dirty in dust and go malfunctioning. Just shifting the main power block 5 centimeters aside solves the problem. This also generates a little stronger air circulation inside the computer, which makes the components a little cooler. There is less fire hazard since a fire cannot start inside the computer, be fed by air and spread to the outside. There is also less danger children put an electric wire inside the power supply block.

Mains displacement drawing

A friend told me I should put a temperature probe inside the case. I did so, just in the middle of the air inlets of the power supply block. After a long period of heavy calculations the temperature had risen to 39 °C. This is more than I tought yet still acceptable. A dedicated air circulation design would be a good thing.

I removed the cardboard wall above the powersupply block and put a fan to force the air to circulate inside the case. The result was obviously better. The temperature at the power supply block inlets rose only to 33 °C. There was no tepid zone feelable on the computer case. Simply the whole case is no more "ice cold" like metal normally is when you put your hand on it. Below are a top view and side view drawings. I let the computer on for two hours without calculations, the temperature didn't exceed 10 °C above room temperature. Options to lower the case temperature down to near room temperature are to paint the case black, put it vertically or direct external silent slow turning fans towards the case. This bears a market for design.

top and sideways view of power supply shifted, holes closed and internal ventilator added

I thought to build a cardboard labyrinth to force the air coming out of the power supply to first pass along the case's right side surface. That way it would loose most of its heath before going to the rest of the inside of the case. But I allready have good performances so I probably won't try it out.

Placing a fan loosily above the computer surface didn't help. The case surface feels colder yet the inside temperature probe shows no significant temperature decrease. This is because the air inside the case does not move fast on the inner surface of the case. The global air motion inside the case may be fast but a millimeter above the inner case surface it stands still. I solved that by placing two fans on each side of the case, forcing the air to move fast on the case surface at that given place (picture below). This allowed for a much better heat transfer and temperature decrease. The inside temperature is now only 6°C above room temperature (no calculations).



This design became somehow ridiculous. It uses four fans and one of these fans is on the outside so it accumulates dust. Obviously there should be only one fan inside the case, that is the power supply fan. It should be bigger and placed a way it makes the air circulate fast inside the computer case. The heat transfer towards the outside should be achieved by using a heatsink shape on both the inside and outside of the case. For example aluminum cylinders or blades placed wherever there is empty space or there where the fan air flow is strong. There should be no fan on the outside, just aluminum cylinders or blades pointing out. I'll try to get hold of such pieces of heat conducting metal; aluminum or copper.

The problem with a big aluminum heatsink structure is it will be heavy, cumbersome  and expensive. Since we asume an internal forced air flow, a more lightweight and thiner structure can be used: just a rippled surface. My best bet would be a vertical case with a rippled structure on the right side (drawing below). There would be a thermal insulation on the inside (orange). The inside side of the rippled plate forms air ducts, leading the air towards the bottom. The air is forced through these ducts by the fan (black) of the power supply block (blue) at the top of the case. Maybe a flat plate can be place on the whole to form a chimney structure. Provided that plate can be removed easily to wipe away the dust.




I whish to thank Dimitry Gathy, Pierre Backers, Didier Bizzarri, Frédéric Cloth, Philippe Thoma and the designers of early Russian space probes for their ideas, additions to this text, technical remarks and gifts of hardware.


Links:

http://www.xlx2.ca
http://www.computerairfilter.com/
http://www.purapc.com/
http://www.computerdust.com/


Lend your processor power for valuable scientific research: http://boinc.berkeley.edu



Eric Brasseur  -  January 21 2004  till  April 16 2006
www.000webhost.com