Air cooling of semiconductor assemblies is often aided by the use of a fan. A typical fan might look like the following.


Figure 1
A typical fan

A fan like this could be placed in a tunnel-shaped heat sink which might carry some number of power semiconductors whose heat has to be carried away to somewhere else. A single fan of this kind will be capable of moving such and so many cubic feet per minute (CFM) of air, taking air in from one end of the tunnel and expelling it from the other end.

You might think that if one fan were capable of moving say 50 CFM, that you could get greater air flow by using two fans in cascade as in the following sketch, but sometimes, that may not be the case.


Figure 2
Cascaded fans

Although tunnel structures with cascaded fans such as seen in Figure 2, operating under a condition of high back pressure, might benefit from the actions of two fans placed in cascade within a tunnel, under the pretty common condition of low back pressure, I have been told that using two fans in cascade will not confer any air moving advantage.

The product for which this information was given to me was a heat sink tunnel which supported eight type 2N6578 power transistors. Each 2N6578 comes in a TO-3 package.

The thesis that was presented to me by the manufacturer of the heat sink tunnel assembly was that even if there were two fans are of the same model so arranged, one of them would be slightly less CFM-capable than the other. Whichever of the two was less CFM-capable would dominate the overall CFM air flow to its own limit and actually obstruct the CFM delivery of the more capable fan.

I have been told that if the first fan on the left had the higher CFM capability, the second fan on the right would dominate the first with slightly higher air pressure in the zone between the two fans. Conversely, I was told that if the second fan on the right had the higher CFM capability, the first fan on the left would dominate the second with slightly lower air pressure in the zone between the two fans.

This issue is quite complex when looked into in depth, but this was the application advice that I followed. Any comments?

John Dunn is an electronics consultant, and a graduate of The Polytechnic Institute of Brooklyn (BSEE) and of New York University (MSEE).

Related articles: