Re: Going to one HD
mechdan_at_yahoo.com
Date: 08/25/05
- Next message: John Hasler: "Re: Going to one HD"
- Previous message: Martin Gysel: "Re: Lying For LieNix? It Seems To Be Very Common."
- In reply to: Peter T. Breuer: "Re: Going to one HD"
- Next in thread: John Hasler: "Re: Going to one HD"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Date: 25 Aug 2005 10:05:53 -0700
Peter T. Breuer wrote:
>mechdan@yahoo.com wrote:
>>Jean-David Beyer wrote:
>>>Not at all. If you blow down instead of parallel to
>>>the heat sink fins, you need a larger heat sink, and
>>>more air flow from the fan, to get the same amount
>>>of cooling.
>>Put bluntly, you don't know what you're talking about.
>He knows as much physics as you or me, I daresay.
Bah, physics. If you REALLY want this discussion, then
I can explain in detail the physics of why the traditional
HSF layout generally provides better cooling performance
than a sideways fan (for a non-heatpipe cooler).
>If you blow DOWN then each (half) volume of air passes
>over at max HALF the available surface as it spreads
>out both left and right on hitting the center of the sink.
The airflow reaches most of the heatsink, except for
a relatively small dead spot in the center.
And it's actually good for the airflow to only pass
through "half" of the available surface. Air which
gets heated up becomes less and less effective at
removing heat. Ideally, you'd want fresh air to hit
all of the available surface. That's not physically
possible, but a traditional fan layout gets closer
than a sideways fan layout.
>In actuality, some will be bounced off the air
>already rising and bouncing back up, so it won't
>contact the surface at all.
That bears no resemblance whatsoever to the way fluid
flow actually works.
>If you blow SIDEWAYS, however (or even better,
>from below - but that's impossible ;-), then each
>volume of air travels across the whole plate,
>achieving maximum heat transfer through maximum contact .
If your goal were to heat the air, then this would
be good. However, your goal is to cool the heatsink.
For heating the air, you want low airflow and to have
the air stay in contact with the heating surfaces as
long as possible. For cooling the heatsink, you want
high airflow and to have the air stay in contact with
the heating surfaces as short as possible. For
various reasons, traditional heat sinks tend to be
shorter than they are wide, so a traditional fan
position allows for greater fan area and greater
airflow. But let's assume a "cubical" heatsink...
With this cubical heat sink, let's assume you use the
same fan and the same level of airflow. The traditional
fan position still works better, because the way heat
conduction within the heat sink works.
The source of heat is the CPU, and this makes the
heatsink fins hottest near the bottom and increasingly
cooler toward the top. The rate at which this heat
gets transfered to the air is directly proportional
to the temperature difference between the air and the
metal.
The situation with a sideways fan is not very good.
The air traveling near the top of the fins has
limited cooling potential because the fins are too
cool. The air traveling near the bottom of the
fins has limited cooling potential because the air
gets heated up quickly and provides less and less
cooling the further it goes.
The situation with a traditional downward pointing
fan far better. The airflow looks a bit like this:
[===fan===]
<-'||| |||`->
<--'|| ||`-->
<---'| |`--->
<----' `---->
The part where the air goes downward is very nearly
ideal, based on the principle of "counterflow".
The heat within the fins flows upward, such that
the fins are coolest at the top and hottest at the
bottom. The air is flowing in the opposite direction,
downward. This means that as the air gets hotter
and hotter, the metal its in contact with also gets
hotter and hotter. This means useful heat transfer
is occuring over the entire distance of travel,
as long as the air is travelling downward.
Then the airflow bends sideways and exits. Here, the
counterflow principle no longer applies; the airflow
is still useful but is no longer optimal.
In addition to this, there's also the matter of
the fact that the CPU core is actually at the center
of the base of the heatsink, instead of spread
across the entire side. With a sideways fan, the
best cooling occurs near the edge of the heatsink,
far from the CPU core. With a traditional fan,
the best cooling occurs in a ring around the core.
>>There are reasons why the overwhelming majority of
>>heatsink fans are designed with fans blowing downward
>>onto the heatsink fins. For one thing, they actually
>>do indeed work, amazingly enough.
>But they don't - not very well, anyway.
Yes, they do. I'm not talking about physics theories,
but actual real world tests. For example, the stock
Intel P4 coolers are actually very effective, in
actual tests.
>It would indeed be much better
>if there were a tall finned cooling tower and the air were blown
>sideways through it.
With plain metal heatsinks, no.
Tower cooling does look really good with at heatpipe
coolers, though. Heat pipes are amazingly efficient
at moving heat from the CPU base to a large radiator.
But we haven't reached a point where heatpipe coolers
are absolutely dictated.
Isaac Kuo
- Next message: John Hasler: "Re: Going to one HD"
- Previous message: Martin Gysel: "Re: Lying For LieNix? It Seems To Be Very Common."
- In reply to: Peter T. Breuer: "Re: Going to one HD"
- Next in thread: John Hasler: "Re: Going to one HD"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Relevant Pages
|
