Re: Have you changed a a PSU fan?

The message <5137b64f-3af1-40f7-a115-34fd4e31a3bd@xxxxxxxxxxxxxxxxxxxxxxxxxxx>
from w_tom <w_tom1@xxxxxxx> contains these words:

On Jan 5, 12:22 pm, "Michael Chare" <MunderscoreN...@xxxxxxxxxxxxxx>
wrote:
One issue I am trying to decide is what is a sensible temperature to
run the
E6600 CPU at.  The Intel web site says that the upper limit for the
temperature is 60C but is less clear as to what would be  a sensible
working
temperature, and comments on this would be appreciated.

Add 20 degree C for when the computer works in a 40 degree C room.

That's a rather extreme example of "Heatwave" room temperature
conditions (but it's what my own system is specced for :-). More likely
(if improbable :-) summer heatwave room temperature in the UK will be in
the 30 to 35 deg range unless the room is a 'sun trap', in which case, a
bad choice in which to set a PC system up.

That is a perfectly normal room temperature for a PC.

No, it isn't! A more normal room temperature would be around the 20 to
23 deg mark, with 23 to 26 deg being more probable in the summer months,
only rising to the 30 deg mark during those rare 1 day to a fortnight's
worth of 'heatwave' conditions so typical of a UK summer.

Will your CPU
still remain below 60 degree C when its temperature increases 20
degree C? If yes, then your CPU is at perfect temperatures. Why do
we worry about high temperature? Because a CPU too hot crashes - will
not execute software accurately - can be crashed by reduced signal to
noise ratios inside the chip.

60 deg is stupidly low for a cpu. A more realistic Tcase rating for a
cpu would be 70 to 80 deg (Tsubstrate rating of 90 to 100 deg C). Modern
hard disk drives are now following Maxtor's lead in upping the 55 deg
rating to 60 deg. Are you sure this 60 deg rating isn't for the hard
drive rather than the CPU?

Anyhow, this is starting to look like a cut'n'paste from a website article.

If CPU is too hot, the problem typically has no relation to chassis
air flow. The problem is in thermal resistance between CPU and
internal chassis air. The analysis starts with a numeric spec called
degree C per watt as any minimally sufficient CPU heatsink will

This 'numeric spec' is simply known as 'Thermal Resistance', the lower
the better.

provide. Confirm this spec number is being met by your system. If
not, well, why does the heatsink fan assembly not do what the
manufacturer stated? Did he lie? Or is your heatsink misaligned on
CPU?

The 'space heater/hot air recirculation' effect is the usual culprit,
coupled with insufficient changes of air per unit of time. The main
culprit for a less than sufficient flow of air is excessive restriction
on both the incoming and the outgoing ventillation interfaces (intake
and exhaust ventilation grillework).

As has already been stated by Dorothy Bradbury, the type of fan
commonly employed in PC cooling, whilst excellent at shifting
impressively large amounts of air when there is no resistance to
overcome, provide very little pressure difference to squeeze air past
any sort of restriction.

The problem is often confounded by the fact that the restriction is
quite often a close coupled obstruction to the fan itself (ventillation
grillework), creating turbulence at the the fan blades, further reducing
their efficiency. A commonly quoted rule of thumb for these fans is that
a minimum of one diameter (of the fan) be employed to seperate it from
any such obstruction.

Mounting extra rear panel extractor fans will quite often make no
improvement whatsoever if the intake ventillation restriction isn't
dealt with (usually a tinsnips on the front panel fan location
'grillework' and 10mm dia drill job on the plastic facia immediately in
front of said fan location - some 42 such holes in my own case (plus the
existing, and already opened up vent slots on the underside of the
plastic fascia) :-)

Others will hype thermal paste as if some major difference
exists. The only major difference is price and hype. Thermal past
should make a minor improvement in the thermal conductivity (reduce
the degree C per watt measured number). Too much thermal paste (any
paste squeezed out to outside edges) may even increase the degree C
per watt performance.

Agreed! (where that leaves the 'toughened bluetack' type of thermal pad
that was so often employed more for conveneience than heat transfer
performance, God only knows - unless someone has got some test result
figures hidden away somehwere :-).

One chassis fan (with sufficient CFMs) should be more than
sufficient for any typical computer. Don't get caught up with the
hype of "More Fans". Discussed are how and where to discover heat
problems - or if it really does exist.

Yep! Definitely a cut'n'paste job! However, I have to agree that a
properly designed (or modified) case should allow the fan cooling to be
held to a minimum of just the PSU fan, _assisted_ by the CPU cooling fan
(with perhaps a small 7 volted 60mm slimline free standing fan within
the case to quietly divert a cooling flow of air around the hard drives,
which should each have a good half inch clearance top and bottom for
just such airflow).

The real upper temperature environmental limitation is set by the hard
disks rather than the cpu. In my own setup, the disk temperatures track
the room temperature by running some 9 to 11 degrees hotter. An upper
limit of 40 deg room temperature leaves me with a 4 deg margin on the 55
deg upper limit rating of the hard drives (and a worst case upper
temperature peak of 65 deg on the CPU - tested to 67 deg for stability
under prime95's stress test).

A point worth keeping in mind when fitting additional extractor fans to
the case rear panel in a futile attempt to reduce system temperature, is
that you could end up with so much 'suction' as to stall the airflow
through the PSU and have it 'go out with a bang' due to overheating.

Whilst there is no disputing that mounting the second extra case fan as
a front panel intake rather than as an additional rear panel exhaust one
is a better 'balanced' solution, an even better solution is to simply
remove the front panel fan 'grillework' altogether and open up the
fascia vent slots to reduce the intake airflow restriction. Not only
does this save on the cost of an extra fan, it also saves the noise
penalty that results from the fitting of such a fan (even when the 12v
fan is '7 volted').

You can go further by removing such 'grillework' on the exhaust fan
locations and even on the PSU fan if it is of the classic rear panel
80mm type - often a more effective cooling solution than the 120mm fans
fitted on the interior panel of the PSU (nice idea except spoilt on two
accounts, 1 being turbulence due to close proximity of the PSU
components and 2 the 'Exit Stage Left' airflow path in this layout).

There's no great secret to keeping the innards of a PC cool (and
quiet). It essentially boils down to this: Eliminate intake and exhaust
airflow resistance. This means, for quietness, no intake dust filtering
unless you're quite happy to use your PC as a room dust collecting
vacuum cleaner (and we all know how noisy a vacuum cleaner is :-)

--
Regards, John.

Please remove the "ohggcyht" before replying.
The address has been munged to reject Spam-bots.

.