On Sun, 31 Aug 2003 22:14:41 -0500, David Maynard
wrote:


The issue is not one of achieving "the coolest we can get it", but
rather noise/heat ratio, that a duct impedes exhaust, making the rear
fan noiser at the same exhaust flow rate. To keep the CPU at the same
(high) operating temp as in the Dell ducted system, a good active
cooler's fan can run at low enough RPM, low enough turbulence, that
what litte noise it creates, being buffered by the enclosure, is less
audible than the increase in ducted exhaust fan noise. Running the
rear fan at higher RPM is necessitated by the duct, else there is more
heat buildup in the system.

Nice theory but not true. Case temp is cooler because the heat is directly
exhausted rather than circulated. You also fail to take into account that the
Dell motherboard is designed with the CPU socket located to minimize duct air
impedance.

Anyone with a Dell system can open their system, feel the heatsink,
and know that it is true. The CPU socket location does minimize the
impact, but that impact is still a great reduction in airflow.
Reduced airflow results in higher heat retention, there's no fancy
combination of ducting theories that changes that. I'm not
suggesting that Dell systems usually run hot enough to be instable,
but neither do other equally thought out, quiet systems.


The total system heat generation can be considered constant so the
total airflow through the system must also remain the same else the
Dell system runs hotter in more areas than just the CPU.

Directly exhausting the heat requires less airflow than first mixing it then
exhausting it.

Yet there is less airflow, significantly less going though the
heatsink, enough less that even a percentage of recirculated air on an
active cooler keeps the CPU at lower temp. The goal is not to reduce
airflow, to get the maximual cooling benefit from each CFM of airflow,
or rather it shouldn't be. The goal should be maximal cooling benefit
from each db of noise, and give the system tolerance to user
alteration, component additions, etc.


Plus it results in cooler air into the heatsink because the case
air is cooler: not being first preheated with the heatsink exhaust.

Reduction in intake airflow (if the ducted system is operating with
rear fan at same noise level) causes the air to become more heated on
it's way to the duct. Heat can't magically escape... given a
radiation rate the airflow rate dictates total heat retention over
ambient external room temp. No matter how great the duct is, it
reduces airflow, which increases heat retention.


Lower air
intake into the system will then result in hotter air entering the
heatsink.

There isn't necessarily lower air intake into the system but it could be as it's
easier to cool the remaining components when you don't have the CPU heat being
vented into the case.

There IS necessarily lower air intake. It is impossible to add a duct
to the rear fan and have the same exhaust rate, same intake rate as
before it's addition, unless the fan is operated at higher RPM, which
is louder, and of course increases airflow on the ductless system
further.


Proper operation within the temperature spec (instead of lowest
possible CPU temp) is the notion that allows using a low-RPM fan on
the heatsink instead of the noisey fans most people compare to the
Dell ducted system.

True, but no fan is still less noise than a 'low noise' fan.

Not true from a user's perspective. Fans mounted on the chassis allow
a far greater percentage of sound to escape the chassis. The increase
in rear fan noise to attain same airflow as a ductless cooling system,
is audibly lower to the user than the heatsink fan running at low RPM.



If we are considering a custom-designed OEM
cooling system it has to be compared to an active cooling stategy with
similar forethought, not just a cheap/junk $5 'sink with a
tornado-speed fan on top.

No fan is still less noise than a 'low noise' fan and, with your active HSF, you
STILL need the case fan; with more airflow because it has more volume to get rid of.

Noise is not additive, two lower RPM fans are usually quieter than one
higher RPM fan, particularly when the higher RPM fan is mounted on the
wall of the chassis, and even further the higher-pitch of the
higher-RPM rear fan is more noticeable to human ears. I will mention
again that proper fan selection is crucial.


There are a lot of variables involved with design, production, parts,
but in general a high-volume production piece of plastic should be
less costly than a second fan and more elaborate heatsink.

It's your assumption than an active heatsink is necessarily "more elaborate" for
the same cooling that I dispute.

I could likewise dispute that a passive cooler is more elaborate than
an equally well-designed active cooler. Actually I feel the opposite,
that it takes a much more elaborate and expensive passive cooler to
come anywhere near the cooling efficiency of a cheap, low-end active
cooler. The problem is, they generally aren't more elaborate when
comparing to a good active cooler, not mid-to-low end.

Which should cost more to produce in volume? Granted there are some
really junky fans that may hardly be worth the plastic they're made
out of from a reliability or noise standpoint, but instead considering
a fan of similar quality to the rear fan, the main issue would be
volume production. Since fans are already produced in volume,
production lines are in place to produce them regardless of what Dell
is doing. If Dell buys enough ducts they're going to be even more
inexpensive, and since these are likley supplied by the case
manufacturer it may be a "free" duct. I don't recall any case
manufacturers providing free heatsink fans ecxept in rare cases like
thoose miniature Shuttle systems or similar where space constraints
make a specific heatsink size necessary.

There can
be exceptions but practically we can only consider the ducted systems
being used by OEMs, not theorectical, nonexistant systems. A ducted
cooling system does tend to be the best noise/heat ratio if there are
budget constraints, at least for an OEM who can buy in bulk.

It's also the best technical noise/heat solution between the two regardless of
cost. It's simply a more efficient use of the airflow.

A ducted passive cooling system is a more efficient use of a given
CFM, but it's not like there's only "X" amount of airflow allowed,
rather the primary issue is usually noise. Introducing the duct
causes a significantly lower CFM to noise ratio. It's unavoidable
with an axial fan.

In other words, it can be the best budget-optimized solution for an
OEM, but optimizing as much as possible for cooling or noise, a ducted
passive 'sink cooling system cannot attain as low a noise/heat ratio.

I have no idea what makes you think that simply recirculating hot CPU air inside
the case, rather than expelling it, is 'better'.

Recirculating airflow isn't better than not recirculating, but since
that's not the only issue it has to be weighed against others, like
total airflow, and a greater airflow though the CPU 'sink. The
greater the airflow though the CPU 'sink on a passive cooler (given an
optimal duct for that purpose) the greater the reduction in exhaust,
and intake airflow. Moving the opposite direction, reducing
backpressue by diverting less air through the CPU 'sink will somewhat
increase chassis airflow, but the CPU runs hotter.

Ducting CPU heat out is inherently a better cooling/noise solution.

Combining the two strategies, having a duct AND an active cooler,
would be the choice for lowest temps but again the duct necessitates
an increase in noise else *something* will run hotter, and of course
it will tend to be the most expensive alternative.

You simply refuse to recognize the benefit of expelling the heat directly rather
than first mixing it with case air and then having to expel the entire case
volume to remove it. You don't NEED the same airflow to remove the heat if it
goes directly out.

At the same airflow rate it is of benefit to remove the heated air as
directly as possible. At a greater airflow rate the benefit of it is
quickly overcome. Higher exhaust rate also reduces mixing of heated
air, reduces the pre-heating of air from other components prior to
reaching the CPU heatsink.


Not to mention one can use larger, quieter, fans on the rear mount than can be
shoved into the socket footprint.

The size of rear fan is a constant, since the same chassis can be used
for either cooling configuration. Given this same sized fan, it will
produce more noise at same flow rate OR same noise at lower flow rate
with a duct on it.

A fan mounted on the CPU heatsink need not be as large as the case
fan, will move a lot more air through the heatsink due to proximity,
because axial fans cannot create significant pressure.


It is possible with a high-quality active cooler.

I don't care HOW 'quality' your active cooler is;

Perhaps this is why you haven't seen the same results. Passive
coolers don't even come close, not with a duct, not with a 120mm fan.
They usually provide just enough margin to operate stabily, little
more than that.

it's still dissipating the CPU
heat INTO the case and not OUT of the case.

A great portion of the CPU exhaust from an active cooler is directly
removed, especially with a 120mm fan.

And after you've dumped the CPU heat
into the case with an active HSF you STILL have to have a case fan to expel it.

So you don't really care about performance, you're just opposed to
using two fans?


Even though some
heated air is recirculated the primary source of audible noise in an
optimized configuration is the rear fan, which is less efficient with
a duct on it.

You have to expel the heat sooner or later and it takes less airflow if the CPU
heat is vented directly instead of having to expel the entire case volume to get
rid of it.

The system 'creates' a given amount of heat. There is no duct that
changes that. The lower the exhaust rate, the hotter the air is...
nothing changes that either. The ducted system has lower exhaust
rate at same noise level, because of the duct... exactly what this
duct does, concentrate and redirect airflow, is exactly why it reduces
airflow. It can't be argued that all this air was heated passing
through the heatsink on it's way out because with less cool intake air
the rest of the components are hotter, heat the intake air more. Many
such ducted systems even feel noticably warm on the outside, a clear
indication that ambient temps are quite high.

By optimizing both methods there might be similar
noise/heat ratios, but then the ducted system is more dependent on
preservation of the chassis airflow model.

'Preserving' the 'chassis airflow model'? What then heck does that mean? The
chassis has to be ventilated regardless. MORE so if you dump the CPU heat into it.

Dell (and others) have to move a cetain amount of airflow through that
CPU heatsink, and likewise in other areas of the chassis. At a bare
minimum I'll bet they want it to last until the warranty is expired.
This is an engineered cooling system to achieve the desired result,
not just the duct but the whole chassis... The proposed solution is
an environmental model of an actual system, before it ever goes into
production. If the intake, exhaust, or flow pattern is altered, the
target temps deviate from the model. Our OP, Peter, may be
experiencing exactly that when he added the exhaust fan for his video
card, which reduces exhaust though the duct even further. He might
see improvement by using an intake fan, except that I dont' recall
seeing any slot-mounted intake fans, but if there are any...



Passive 'sinks are usually less densely populated with fins and have a
greater percentage of weight at their base, which is less leveraged
force against the socket or retention mechanism.

Whether true or not it's not a passive heatsink. The fan is just located in a
different spot.

Yes, but it goes to the point of stress against the socket, retention
mechanism, or CPU during shipping... just another motivation for Dell
to use this passive 'sink design.



Doesn't need the same airflow over the CPU heatsink as the heatsink is operating
with cooler case air, since it's not preheated with the exhaust from the CPU.
Plus it's venting the case in addition to whatever case cooling you'd require
when dumping CPU heat into the case.

It doesn't need as much airflow given a 'sink of same efficiency,
which is another issue that could expand this discussion into
something more fit for a book rather than a newsgroup thread.
Historically and today, passively cooled CPUs run hotter, even though
they (all, AFAIK) use a duct. To many people this is considered
common knowledge. Dumping CPU heat into the case with an active
cooler isn't an issue of the significance that you suggest, since the
airflow is moving in one direction, up and out of the case. Some
heated air is mixed and recirculated though the heatsink but the
airflow rate is SO much greater than even slightly warmer air results
in a lower CPU temp.


"Optimized" is precisely what an active HSF is not.

On the contrary, it is optimized to cool the CPU, the exact goal.
When a high quality sink is paired with a low-RPM fan, it is then
futher optimized to have maximum cooling with minimum noise.

It's the 'norm' for generic
systems because one cannot make any assumptions about the system design: I.E.
what else is in it and where anything is located.

It has widespread adoption because it works well in comparision to the
alternatives of similar cost.

You can't count on there even
being a rear exhaust port or, if it exists, that it's located where a 'standard'
duct (that doesn't exist) would lead to, not to mention you don't know the exact
location of the socket as that is left open, within reason, to the motherboard
designer.

Yes, the active 'sink works SO well that it can overcome all these
obstacles most of the time. Not because it's "generic" but becuase it
has such a huge margin over minimal efficiency required, that when a
decent chassis is used the active sink works even better, even with a
silent fan on it.

But you CAN just slap an active HSF on it and then leave case cooling
to someone else; with the typical solution being to slap gaggles of fans
everywhere to 'cool' it.

It's true enough that some people purchase, build, redesign, and
outfit their systems in ignorance, but it has little to do with this
topic since these same folks could either improve or degrade a ducted
passive CPU cooling system using the same methods as on one actively
cooled. We both know there are plenty of crap cases out there too,
but that's not an argument towards a ducted cooling system, but rather
to choose the case carefully regardless of which cooling system will
be used.

It's usually the situation that user mods to a passively cooled system
have a greater impact, positive or negative, depending on where the
fan is located and whether it's intake or exhaust.

I suppose we just won't come to an agreement... it happens.
I do despise long threads so if you want a thread where we can all
argue about passive/ducted vs active cooling instead of hijacking
Peter's thread, that's fine, but I'm done with this thread.


Dave