Cooling fan direction

Question

I am building a temperature controller. The controller is built in a metal rack mount chassis. The components inside the controller generally consist of ACDC power supplies, an arduino and small signal handling PCBs, fuses, and a bunch of wiring. We have designed a cooling fan on the rear panel of the chassis. We also put venting cutouts on a separate region of the rear panel. Should the cooling fan blow air out of the chassis or suck air in? Why and what is more common?

There is no particularly hot item in the controller chassis. I am generally more concerned about the overall temperature in there.

Answer 1

A full-on 'professional' design would model airflow and thermal dissipation using a tool like Flotherm. It seems like this isn't being done for your design. That’s understandable since the software tools are kind of expensive, and require a lot of effort to get good modeling.

Regardless, the ‘one weird trick’ to achieving a good air-cooled design is to ensure enough velocity past your hottest components to achieve your target junction temperature.

As far as pull air in vs. push air out, both approaches get used. A small system with just one fan and modest power favors the 'inhale' approach, focusing the highest flow velocity on the power supply.

Example: a typical desktop PC will pull air in through the power supply and blow it directly onto the switching regulator's power transistor heatsink. This ensures the highest local velocity as well as the coolest possible air for that area. The CPU (and GPU) will also have heatsinks and local fans, but otherwise rely on the gross air movement through the box coming through the PSU to keep in-case ambient temps down.

Gamer PCs will often add more case fans to improve air movement to further reduce in-case ambient air temp.

Another example: DVR set-top boxes (something I have personal experience with). The ones I worked on 'inhaled', and like a PC, blew high-velocity air across the power supply. As part of our product design we did airflow and thermal modeling, then followed up with thermal testing to ensure the critical hot spots (disk drive and SoC) were in spec.

Based on your system description, seems like your power needs are modest and you won't have significant hot spots. The DVR-like 'inhale' approach should be fine, so long as your air movement through the rest of the chassis is adequate.

If your testing does reveal hot spots you'll need to revise your mechanical design, possibly adding local heatsinks and fans. In fact, it's a good idea to plan for this, as you could delete them later if you don't need them.

Other things to consider:

• ducting/channeling to direct airflow to internal hot spots
• thermal contact with the case for passive cooling

Using the case as a radiating element is a popular approach with consumer electronics, especially for 'silent' devices.

Answer 2

Air direction depends on the design of the device you wish to cool. Typically you want to maximize the airflow over surfaces like heatsinks to carry heat away. This can be accomplished with fans in either direction, but which is more optimal depends on the design of the chassis.

Some things to consider in no particular order:

• If there's a heatsink or component that is near an intake or exhaust fan, it will have higher airflow than if it is further away. However, having it near an exhaust fan may not be ideal as air cooling the component may already have been warmed by the rest of the device.
• Is air expected to be cooler in front or behind the device? This plays a major role, for example, in rackmount equipment where the air in front of the device (where the humans are) will be cooler than the rear of the device (where the exhaust air goes).
• Is dust a concern? You may want to design the chassis so that air is blown into it through a filter, which ensures that the pressure inside the case is greater than ambient, which helps prevent dust from entering every opening.
• In devices where heat can build up quickly, blowing air in one side and exhausting it out the other may be necessary. You can still prevent dust entry be ensuring the intake airflow exceeds that of the output airflow somewhat.

Answer 3

No formal justifications here.
These are just what would influence my decision.

• Fast air will cool better than slow air.

• Filtering energy concentrated air at a fan will be easier than filtering low energy per area air at inlet slots.

• Directing high velocity incoming air to key areas is easier than directing low speed air.

All these combine to make a fan blowing in the best choice in most, but maybe not all situations.

eg If you have multiple fans in an enclosure and the air has been blown inwards through a capable filter then adding a fan for a selected sub-enclosure would allow you to blow pre-cleaned air outwards. This would remove the cleaning aspect from consideration.