[hpsdr] To suck or to blow

[L. Van Warren]

This is rocket science, and it turns out I do that.

Fan mounting is all about dynamic pressure. There is an instantaneous
pressure spike across the face of the fan, and this causes the blowing side
of the fan to emit a narrower and more focused blast of air to the board,
assuming the ambient temperature is lower than the internal temperature of
the chassis.

If hot components are in direct line with this cold flow, they will
experience a higher rate of cooling, compared to the low velocity they will
experience if the fan is sucking air out of the chassis, from vents that may
or may not be in line with the flow.

That said, internal flow in a mounting cabinet is extremely complex, and
insufficient venting can cause the pressure to back up and reduce the
effectiveness of the fan. Your fan will only be effective if the cross
sectional area of peripheral vents are greater than or equal to the cross
sectional area of the fan itself. You want at least one vent in addition to
the fan mounting hole, or the fan is just blowing hot air around and
contributing to the problem. More vents slow the flow of a sucking fan down,
in the limit, reducing its cooling effectiveness to that approximating an
open board. The situation is reversed in the blowing fan case, increasing
cooling effectiveness, because the back pressure is reduced. The limiting
case of that is the open board where components in the vicinity of the fan
are being cooled in proportion to their distance from the fan.

That said, there is no substitute for measurement. Use a logging pyrometer
for hot components, especially expensive DAC's and ADC's that are difficult
to replace. Use the fan configuration that keeps critical components the
coolest, but my first choice is blowing air into the chassis, rather than
sucking it out, although the latter has one important advantage:

Second order issues include dust build up. Dust filters greatly reduce the
effectiveness of the fan, by lowering the velocity of the flow, increasing
drag, and reducing the cross sectional area. It is preferable to put the
filter on the upstream side of a blowing fan, and ideally the filter would
be much larger than the fan to mitigate its negative effects.

There is the open question of whether dust buildup is worse in the sucking
or blown situation, in which case we are not building electronic equipment
but rather microparticle sorters which collect certain sizes, compositions
and stickiness of particles!

Liquid cooling is an alternative used by some high performance CPU's, but
this simply moves the problem from the board to the radiator where these
problems repeat themselves. Immersive cooling has also been done, but you
still must have sufficient surface area to dissipate the heat. Immersion in
a solid then suggests deletion of the fan altogether.

High surface area fanless designs where the heat is dissipated over a large
area and moved around with thermally conductive strips. Thermal conductivity
travels with electrical conductivity to first order, a pro and a con, watch
your grounds, etc. But this can also look like shielding which can be
advantageous.

One could consider non-planar mounting topologies where critical components
are mounted on ducts that physically channel the flow to the component in
question. Special purpose ducting can also be designed.

There is also fan noise, both acoustic and electromagnetic and I despise
both. I have power supplies where the principal noise source is the PM motor
powering the fan. Fan acoustic noise is broadband in nature and thus
difficult to filter.

When in doubt, measure, measure, and measure.

Van / AE5CC / wdv.com