Re: Hot PC

The message <46399121$0$21837$db0fefd9@xxxxxxxxxxxxxx>
from Trevor Best <googlegroups@xxxxxxxxxxxx> contains these words:

Rob wrote:
I agree. Over the years, I've found that adding a front case fan
to blow cool air into the case is very effective. Most cases
already have a mounting point for one, but only the more expensive
case manufacturers seem to actually bother fitting a fan in place.

Or a grill, I had space for 2 80mm fans at front of my case, they suck
against the front wall of the case, great.

That's a good point. My current build (3 years old now) was an upgrade
from a BX440 MoBo/P2/350 (oc'd to 466MHz and blowing P3/1GHz boxes away,
performancewise) to a KT600 MoBo with XP2500+ Barton CPU retained the
then 6 year old squat midi tower case[1].

IOW, the case is now some 9 years old (on it's 3rd PSU - 1st change was
an upgrade to a noisekiller type PSU which had never had any cause to
speed the fan up _until_after_ I did the MoBo/CPU upgrade - a spooky
experience when I heard this happen for the first time late in the dead
of night after 4 years of almost silent service :-).

That P2 build, in spite of using a pair of 12 volted 60mm slimline cpu
heatsink fans, was remarkably quiet in operation, so much so that the
noisier fan cooling in the new build really bugged me into considering
how to improve airflow to eliminate the need to use fans at noisy
speeds. IOW, I didn't want my upgrade to carry a noise price tag and
spoil the quietness of my 'office'.

Two things helped me to achieve may aim of a performance boost without
the penalty of objectionable fan noise. The first was the MoBo's
reluctance to play nicely at the 200MHz FSB oc speed required to realise
the full oc potential of the CPU.

This might seem a strange statement, but it was whilst trying the
overvolting of the CPU to eliminate the instability that I came to fully
appreciate how dramatic an effect this has on cpu temperature and hence
cooling requirements.

Once I realised I was onto a loser with the overclocking exercise (no
fault of the CPU), I realised I could turn the oc/ov concept on its head
and translate the oc capability of the cpu into less heat at stock speed
by undervolting (luckily, the MoBo, a Jetway model, _fully_ supported V
core adjustment and allowed me to drop the stock 1.65v setting to 1.35v
which dramtically reduced cooling requirements). Unfortunately, this
still left me with another serious problem which was my second in the
list of things that helped achieve my ambition of performance without
fan noise penalty.

That second thing being the layout of the PSU within the case. The
squat design of the tower was the result of its extra inch of width over
the 'standard' ATX designs permitting the PSU to be located alongside
the MoBo rather than 'above'. Whilst this hadn't significantly impacted
on the CPU fan cooling arrangement of the slot 1 P2, it now represent a
blank wall of metal within 6mm of the fan on the heatsink required to
effectively cool the socket A cpu I was now using.

At first sight it looked like the case had finally 'had its day', but a
little bit of lateral thinking proved that the case had in fact reached
its full potential once fitted with an appropriately vented PSU (i.e.
vent slots in the _large_ panel of the PSU rather than the smaller
'opposite the back panel' panel.

My inspiration came from noticing such vented PSU designs in my PSU
'scrapheap'. All I had to do was transplant the PSU innards to a 'proper
PSU case' (modifying the vent slots by twisting the metal into
alignement with the airflow, thus making two improvements - increased
CSA _and_ massive reduction of airflow turbulence :-).

The close proximity of the CPU cooling fan to the PSU went from being a
liability to being an asset by the simple expedient of reverse mounting
the fan. At a stroke I was able to almost totally eliminate the 'Space
Heater' effect typical of most normal ATX boxes[2] and provide a boost
to the , now, half volted PSU fan[3].

I now revisited the front panel fan experiment I'd tried 5 years
earlier which had revealed the need to replace the slim rubber feet with
standard sized ones to increase the gap between the bottom the case
front panel and desk to reduce the restriction on intake airflow via the
underside vent slots I'd previously enlarged to a less restrictive size.

Even at 7 volts, such a front panel fan was still adding noticably to
the noise level and only reduced the 'system' sensor reading by 4 deg C
courtesy of the incoming air being wafted towards it before being sucked
out the case via the cpu fan. The hoped for reduction in cpu temperature
never materialised indicating (since a good 80 to 90 percent of the
total case airflow was now via the cpu heatsink) that it was making a
negligable improvement for its noise penalty.

Taking Dorothy Bradbury's comments to heart with regard to the
extremely limited 'suction power' of the fans used in PC cases (i.e.
extremely low pressure airflow), I not only removed that front panel
fan, but also the 'collander style' ventillation holes, replacing them
all with a single 70 odd mm diameter hole which had a more beneficial
effect (especially after opening up the 3mm diameter holes in the
plastic facia to 10mm diameters with a drill :-).

The main defect with most computer cases, even today, is woefully
inadequate intake ventillation slots or grilles. Most such cases can be
dramatically improved by judicious use of a pair of tin snips and a 10
to 12mm drill bit (for those parts of the job where a 70 odd mm dia hole
is just going to look out of place :-).

The remaining weak point is the lack of attention to hard disk cooling
requirements. NEVER stack hard disks one on top of the other. You need
at least a half inch seperation (assuming a reasonable flow of air
through this gap - use a strategically mounted 7 volted fan, if the case
doesn't induce such a flow naturally).

Your main concern should be the hard disk temperatures. Exceeding the
maximum rating (55 deg C for all except Maxtors -60 deg limit in this
case) can be a matter of irretrievable data loss. Exceeding the much
higher limit of a CPU is more a case of irritation than the tragedy of
data loss.

My own system, as described above, (with a pair of 320GB SATA drives)
is now rated for full spec operation at a max environmental temperature
limit of 40 deg C (the hard disk delta Ta of 9 to 11 deg C being the
limiting factor in this instance).

Provided the case has ample intake ventillation (by design or courtesy
of a modding exercise), the PSU fan alone will, quite often, suffice for
the job for a typically equipped desktop PC, rarely requiring more than
the addition of a 7 volted 80 or 120mm rear panel fan (avoid front and
side panel mounted fans if at all possible since these are 'noisy'
locations).

If you need a quiet cooling solution for a high power box, you're left
with the expense of liquid cooling or, for more extreme systems,
cryogenics (the liquid nitrogen option). When building or specifying a
system, it's best to avoid these latter two cooling options if possible.


NOTES:

[1] This case might even be as old as 10 years. I'm pretty certain I was
using it for the SKT7 P166 mmx (oc'd to 292MHz) predecessor to the P2
upgrade. The intake vent slots (ludicrously small) lived on the
underside of the front panel.

Even though I opened up these out of sight vent slots, the very low
clearance with the desk due to the very thin feet origionally fitted to
the underside of the case was severely restricting the airflow as I
later discovered when I tried a front panel fan and noticed the
increased flow when the case was tilted whilst trying to take a closer
look at the innards. Needless to say, I replaced the feet with standard
height ones to reduce the restriction (and removed the fan :-).


[2] Space heater effect is a very apt description. If you've ever
wondered at just how detrimental high temperature operation can be to
the service life of a PC, I can tell you that an identical motherboard
to the one in my 3 year old PC which had seen a similar service life,
was showing blown caps due to the much higher temperatures they'd been
subjected to in the traditional spaceheater configuration.

I was so impressed by the sight of the 'blown' caps on this ( a
customer's) motherboard, that I just had to take a peek at my own, just
to confirm that it was still in the pristine state I'd recalled seeing
it in just less than a month before.


[3] About a month or two after transferring the innards of that 300Watt
'Noisekiller'(tm) PSU into the reclaimed side vented PSU case, it very
quietly died (actually, overnight whilst the mains power was turned
off). Most annoyingly, it had gone from 100% functionality to emulating
a PSU attached to a short circuit overload. So much so, in fact, that a
replacement PSU was the very last option I tried.

I landed up canabalising a brand new 350watt unit to refurbish the PSU
(it _was_ most important to stick with a side vented PSU).
Unfortunately, this was no 'Noisekiller' thermostatic controlled fan
design. The 80mm rear panel fan ran at noisy speed so I had to design
and fit a thermostatic controller to this unit, setting the fan low
voltage point at the 6v mark. The controller is designed to increase the
fan voltage in proportion to temperatures exceeding a set point rather
than alternate between half and full speed settings.

I'm not sure how effective this thermostatic control is. If there is
any speed increase over the ambient temperature range 20 to 30 deg C,
it's a very subtle one. However, the 40 deg max environemental rating
for the system is very conservately based on there _not_ being an
increased airflow rate due to my homebrewed fan speed controller.

--
Regards, John.

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

.