I have assembled a PC with a Pentium IV, 3 GHz, HT processor. Its
performance is good; but since the system does not have the rear 92mm
case fan yet, the processor becomes overheated when it is used at 100
percent capacity.
It would be wise to not run a system before you have it
finished. Cooling is not an optional part, odds are your
system lifespan is already degraded some by running hot.
Probably not enough to matter within it's usable lifespan,
but possibly... we can't see the rest of the case or temps
of other parts that don't have a thermal reporting function.
I wish to buy an efficient, low-noise,
temperature-controlled case fan.
Don't bother with a temp controlled fan.
Even when specific parts change a lot in their temp due to
load, the air temp in the case will only vary a dozen
degrees or so (unless the case airflow was horrendous, in
which case the solution is not thermally controlled fan but
to increase the case air intake passage area).
With only about a dozen degrees or less change in chassis
ambient air, there is not enough of a difference in air temp
to get an appropriate (useful) level of control from a
thermally controlled fan. If the fan had a control line
plugged into the motherboard's CPU (heatsink fan) designated
pin header, it would then gain more resolution per CPU temp,
but it should not, the necessary fan speed to change with
CPU temp is only that on the CPU.
The room temperature here ranges from
15 degrees centigrade in winter to 32 degrees centigrade in summer.
You need some insulation, a heater, and air conditioning.
Except if you're in a log cabin in the woods, there is no
need to have such inhospitable temp changes in this day and
age. Further, a 3Ghz P4 is not a very good CPU to use in 32
degree ambient, unless you replace the heatsink with a
quality aftermarket model. Again this will result in not
much change occuring in chassis fan RPM.
Essentially what getting the thermally controlled chassis
fan will do is make it harder to target the RPM you want,
not easier. It won't adjust enough from ambient temp
changes but the fact that it is adjustable at all means
you'll have to consult spec sheets to determine what it's
nominal RPM will be in the actual use, so it is much harder
to buy a low-noise model with any assurance it will provide
enough airflow, unless you over-spec the RPM and then use a
further method of speed reduction to tweak it once installed
and while temp measurements are taken during the 32C summer
ambient temp scenario.
After a long research over the Web, I found these two fans close to my
requirements:
1. Antec 92mm Smartcool Case Fan
temperature control range: 20 to 50 degrees centigrade
RPM: 1450 to 2070
noise: 23.49 to 31.23 dBA
air flow: 28 to 40 CFM
Not fast enough, with a 92mm fan you will need more than
2070 RPM to keep that CPU cool at full load in 32C room
ambient temp.
2. Vantec 92mm Thermoflow Case Fan
temperature control range: 25 to 50 degrees centigrade
RPM: 1860 to 3100
noise: 29 to 40 dBA
air flow: 34.5 to 58.5 CFM
This is closer to the right max RPM, and possibly about as
good as it gets for thermal control based only on case
exhaust air temp. However, it would ideally be at 3100 RPM
if the air temp had risen to 42C, 10C rise over ambient.
Keep in mind that this is all relative- the more passive
intake the case has, the more air will be exhausted with "X"
RPM of the exhaust fan. That is the key to maximizing
cooling effectiveness, seeing the fans as only 1/2 (or less)
of the case airflow design.
This fan suffers the same problems as most though, that it's
response to temp isn't great. 25C change in temp only
results in 1240 change in RPM. The ideal fan would run at
slightly lower than 1000 RPM at 25C, and be ramping up to at
least 4000 RPM at 60C if not 50C. Blame it partly on
cheap fan design- if it had 2 or more thermal sensors in
parallel, there would be a much better response to temp
changes. You could try to modify a fan to do this, choosing
sensors at roughly 2 x the resistance per C of the
originally spec'd part but that is beyond the scope of what
most would consider reasonable.
On one hand, I want my system to be quiet; on the other hand, I want it
to be properly cooled in every season.
Everyone has their own idea of how quiet, "quiet" is. Even
so, in a 32C room the rear fan will have to spin fast enough
most people would not call it quiet. Ideally you would
lower the room temp or use a 120 x 38 mm rear exhaust fan,
even changing the case (different case or modify that one to
accomodate the larger sized fan) towards that end.
In 15C wintertime use, it will be very easy to keep the
system cool enough, even the 1860 RPM of the Vantec is more
than needed, it could be ran lower than 1000 RPM and be
sufficient (again, providing the case intake area is good).
However, it is expected that below the 25C spec'd for the
Vantec, it would run slower than 1860 RPM... how much slower
I cannot say, it depends on the topology and component
values used in it's thermal feedback control which would
have to be dissected, reverse engineered OR actually tested
for reliable figures.
I am confused which of the above
two fans I should order because the first one seems to be quieter,
whereas the second one appears to be more effective. I shall highly
appreciate any suggestions to help me out of this seemingly trivial but
reasonable confusion.
The first one is not fast enough at max speed for the 32C
environment... not a good choice at all. The second one is
closer, but even it may be too slow. To put it another way,
who doesn't want a quiet system? Everyone does, but at 32C,
concessions have to be made if you dont' want parts to have
a decreased lifespan from running hot.
How about two fans? If you could fit two of the Antec fans
on the rear of the case, that might work well. It might be
louder than necessary in the off-season, non-summer use and
have a little extra margin in summer so a 22-47 ohm 2W
resistor could be placed in series on each fan's power lead.
