OT: The fan club (was: Re: [Going further OT] Re: Leaving a server on all day)

• Previous message: Louis LeBlanc: "Re: Printing to a network photosmart printer"
• In reply to: Jason Taylor: "Re: [Going further OT] Re: Leaving a server on all day"
• Next in thread: Robert Storey: "Re: Leaving a server on all day"
• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

To: Jason Taylor <jason@infinitebubble.com>
Date: Tue, 08 Jun 2004 18:07:24 -0500

On 8 Jun 2004 at 13:45, Jason Taylor wrote:

> Bill Moran wrote:
>
> > Nico Meijer <nico.meijer@zonnet.nl> wrote:
> >
> >>BTW - I'd make sure I'd get/have a decent computer case with a decent
> >>PSU with enough room for some decent 80mm or larger low noise fans
> >>rather than opening up the side panel. Perhaps an aluminum (Chieftec
> >>Dragon, anyone?) case with some Enermax and Zalman coolers and PSU might
> >>do the trick. I've {b,s}een told a good airflow (front to back) is king.
> >
> > I saw an article recently by a guy who had a degree in thermal dynamics or
> > something that was dispelling the common myths about PC cooling.
> >
> > His conclusion was basically that airflow is king. You need to move air across
> > the heat sinks that is cooler than the heat sinks are. Sounds simple, but the
> > overall conclusion was that you could improve cooling without increasing noise
> > by ensuring that air from _outside_ the case was flowing directly over the
> > processor heatsink. Reason this works well is becuase the air inside the case
> > is usually considerably warmer than the air outside the case, and moving warm
> > air across the heat sink doesn't accomplish much. By drawing cool air in from
> > outside the case, things stay cooler.
> >
> > Anyway, his suggestion was that the best thing you could do for your cooling
> > rig was to purchase/fab one of those little duct kits that allows the cpu fan
> > to pull air from outside the case. Some cases even have the duct built in (my
> > brother's computer does).
> >
> Ok, I'll chime in here. Here's what everything I ever learned about
> heat transfer and fluid flow tells me:
>
> Everything Bill is saying is correct. The best way to cool is to move
> as much fluid (air is a fluid for the purpose of this discussion) as
> fast as possible across whatever is hot. Of course, the fluid has to be
> cooler than whatever is being cooled. A fan rotating at certain speed
> is going to push a given volume of air in a given amount of time. By
> leaving the case covers on and providing only a few small "holes" for
> the air to travel through, you're going to force the air coming through
> those holes to travel through the case faster.
>
> That being said, if the case design, component placement, etc. is such
> that leaving the the cover off actually allows a significantly greater
> volume of air to get to the heatsink(s) in a given amount of time, then
> leaving the cover off is a good thing.

Okay, no degree in thermal here, but I used to design these things for
a living (Dell, Tandem, Datapoint). Sorry I missed the start of this
thread, but I'll jump in here and see how much confusion I can
generate.

The generalities above are generally true, generally. :-)

Leaving a cover off may help or it may hurt, depending upon what's hot
in the case and how leaving the cover off affects airflow over those
items. What you're interested in is a maximum of airflow (volume more
than velocity) and a maximum of temperature delta specifically at the
hot components. (This assumes the temp of the air is lower than the
hot component. If it's warmer than the hot component your house is
probably on fire and you've got bigger problems.)

You're also interested in things like maximum surface area at the
heatsink/fluid interface, but that's a function of heatsink design, not
fan design or placement, and there are other factors influencing the
design of that interface. Obviously, if heatsink blades are crosswise
to the airflow the heatsink will be much less efficient.

If the case is really well designed, the incoming air is directed at
the hot components. Since cases are generally generic and motherboards
don't always put things in the same place, this may or may not be
achieved. This matchup issue is one of the reasons generic cases
usually don't have ducts, since a misdirected duct is worse than no
duct. If you're Dell or HP and control both the MB and the case, you
can use good, cheap ducts to allow the use of cheaper heatsinks because
you know where everything is. If the incoming cool air is not directed
at the hot components, leaving off the cover may actually help, but if
the case and motherboard are a good match leaving off the cover can
disrupt the planned flow.

For moving a lot of air with low noise, go for the largest fan you can
and run it slow. The cases I'm using these days to build workstations
are Antec Sonatas, and I mount two 120 mm fans, one in front and one in
the rear, one exhausting and one intaking (therefore in series). I
wish they had proper ducting like the Fong Kai 603 I used to use, but
our components are staying cool enough and the noise level is low. If
you prefer aluminum, the Antec Super LANboy is very similar to the
Sonata, and we have one of these for a machine we carry around quite a
bit. Aluminum is a great help for weight, but I doubt it adds much to
cooling unless you're mounting heatsinks directly to your case. Most
of the heat will leave with the air.

Other fan generalities: ball-bearing fans normally last longer than
sleeve-bearing fans, but are usually noisier. Blade design is quite
critical to noise level, but it's difficult to tell by looking at a
blade whether it will be noisy.

This is a complex issue, so exceptions abound. YMMV.

--
Jerry Dunham
M3 Design, Inc.
Round Rock, Texas
jdunham@m3designinc.com
(512) 218-8858
_______________________________________________
freebsd-questions@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-questions
To unsubscribe, send any mail to "freebsd-questions-unsubscribe@freebsd.org"

• Previous message: Louis LeBlanc: "Re: Printing to a network photosmart printer"
• In reply to: Jason Taylor: "Re: [Going further OT] Re: Leaving a server on all day"
• Next in thread: Robert Storey: "Re: Leaving a server on all day"
• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]