When I see a large difference between the CPU and the MB temps, it's
because of dirt in the heat sink. After I clean the heatsink, the temp
difference is a couple degrees at most.
1) It depends on what is being measured as the
"motherboard" or "system" temp. Some boards have discrete
sensors that are simply mounted to the board, others are
integral to a larger chip which may produce more or less
heat (density, and better or worse cooling itself), and
finally, some boards have more than one secondary thermal
feedback sensors, the one being reported is only a function
of how the reading (software) was set up).
2) At idle the two temps may be closer, even as close as
you saw. At full load, the CPU has a much higher change in
temp, most systems will not have the CPU temp within a few
degrees of so-called system or MB temp near a true full load
condition.
I am tempted to get a piece of non-conducting foam that is maybe an
inch larger than the heat fan assy and hollow it out like a mitten.
Then I could slide it over the fan assy so that it would filter the
air flowing over the heat sink. As long as I monitor the CPU temp, I
would not be concerned with this homemade filter clogging.
Non-conductive foam is any foam actually, the typical
varieties found almost everywhere don't conduct. If you
meant ESD rated foam, don't use it, because it is specially
treated so it _does_ conduct (only slightly, but any degree
of conduction is an unnecessary and possibly negative
quality of it).
There are filter panels that screw onto standard fan frames
though, if you had a standard fan instead of one of those
proprietary ones. It may be unnecessary though, ideally the
entire case would be filtered and that is enough, usually,
but with the entire case filtered it can mean larger filter
panels are used so they take longer to clog up and the
service interval is extended.
