Archived from groups: alt.comp.hardware.homebuilt (
More info?)
Best is a compromise - filtration, size, airflow etc...
o Airflow resistance rises with filtration level
---- HEPA = sub-micron filtration and thus very high resistance
---- HEPA = high pressure radial (blowers) + 48-55dB(A) + few 10s cfm
o Airflow resistance rises with airflow
---- filter presents 15Pa resistance at low cfm, 45Pa at double that cfm
---- compensating for resistance by a merely faster fan is undermined
Coloco Centres filter via the HVAC plant - so racks don't have to...
o HVAC plant uses huge multi-kW high pressure radial (blower) fans
o Racks then try to keep the top (hot) part of the rack within spec
PCs impose several restrictions on cooling system...
o Fans are Axial -- PCs present low airflow resistance & demand high cfm
---- Axial fans -- small size, shallow depth, low noise, high cfm, low pressure
---- Radial fans -- large size, high depth, high noise, low cfm, high pressure
o Filters are omitted -- filter size limited by fan/case size & use of axial fans
---- Axial fans produce high airflow, but little pressure
---- Filter resistance exceeding an axial fans stall pressure results in zero airflow
PCs impose implementation restrictions on a filtered solution...
o All intake air must come through the filter
---- yet PC storage bays reduce the frontal area for filters
o Intake airflow must exceed exhaust fan airflow
---- positive case pressure prevents air intake thro non-filtered holes
---- again PC storage bays reduce the front area for fans
So there has to be a compromise...
o Dissipation of N watts limited to X temp rise requires a mass of air A
o Filters add resistance, reducing airflow, so increasing temp rise
o Balance via adjustments to filter system
---- increase filter size = less airflow velocity = less resistance
---- crease filter level = less resistance
o Balance via adjustments to cooling system
---- increase fan size, fan depth, fan speed
---- increase fan number re serial operation with exhaust fans
In practice?
o Most PC filters simply stop the fan blades dusting
o They will not prevent skived copper heatsinks from clogging
o They will increase the time between clogging - perhaps usefully
It is simpler to clean the CPU/GPU heatsink once in a while.
This is more of an issue with laptops - which use high density skived
copper heatsinks of low height, low airflow, non self-cleaning designs.
Heatsinks are frankly the sole device affected by "dust bunnies"
o Blow through, large fin, high velocity heatsinks are preferable
o Multi-fin (copper sheet flower) or asterisk (alloy extrusion) work well
Comes down to how much thermal dissipation you have.
o Easy to filter (& cool) a VIA C3 PC quietly
o Harder to filter (& cool) a Dual-Opteron 5-SCSI-RAID Twin-GPU Server
Most filters do merely stop fan blade dusting.
o Yes filters will stop the large particulates
o However higher heatsink fin density is a structure for dust matting
o In laptops fans can be roaring in 6 months from clogged heatsinks
In summary...
o Axial PC fans will lose ~40-60% of their airflow with a clean filter
---- the figure gets worse as the filter clogs, and HDs get hotter
o Fitting 38mm depth fans or larger fans can cause "fan count runaway"
---- velocity does not help -- resistance rises with higher airflow fans
---- area can assist -- it reduces airflow velocity, reduces resistance
---- fan count assists -- it increases fan area, reduces velocity & resistance
If you suffer hayfever, go buy a Honewell HEPA filter unit.
They use near operating theatre quality HEPA filter drums inside, they
are relatively quiet and are a better ROI than trying to filter a PC
--
Dorothy Bradbury