CPU Cooling Fan Question

[Bruce Murray]

As the warmer weather approaches New England, the CPU fan has to work
harder.
I have a P5GDC De luxe running with a P4 3.2GHz 800 MHz FSB
I am using the Intel fan suppled with the CPU.
Although it does keep the CPU at reasonable temperatures, say 100 deg F,
the fan is really quite noisy and it is running at 4500 RPM at the moment.
The side of the PC case is removed for maximum ventilation.
Since I am a long way from dangerous CPU temperatures I would have
thought the fan would be running more slowly. I don't see a way to
schedule the system gain to trade CPU temperature for lower fan speed.

Also I note that the fan is running to push air into the heat sink,
rather than draw the air out. Is this correct? I thought most fans
acted to draw air out of the heatsink area and not blow it in. The way
it runs now ,clockwise rotation, is very noisy since the fan blades are
chopping the wakes from the support struts, thus making a siren noise.

So what is the received wisdom for the quietest and most effective
cooling system for the P4?

Thanks

Bruce

 

[Paul]

As the warmer weather approaches New England, the CPU fan has to work
harder.
I have a P5GDC De luxe running with a P4 3.2GHz 800 MHz FSB
I am using the Intel fan suppled with the CPU.
Although it does keep the CPU at reasonable temperatures, say 100 deg F,
the fan is really quite noisy and it is running at 4500 RPM at the moment.
The side of the PC case is removed for maximum ventilation.
Since I am a long way from dangerous CPU temperatures I would have
thought the fan would be running more slowly. I don't see a way to
schedule the system gain to trade CPU temperature for lower fan speed.

Also I note that the fan is running to push air into the heat sink,
rather than draw the air out. Is this correct? I thought most fans
acted to draw air out of the heatsink area and not blow it in. The way
it runs now ,clockwise rotation, is very noisy since the fan blades are
chopping the wakes from the support struts, thus making a siren noise.

So what is the received wisdom for the quietest and most effective
cooling system for the P4?

Thanks

Bruce

XP-120 0.18C/W with inaudible fan
XP-90 0.18C/W with inaudible fan
Zalman 7000B 0.22C/W with fan at full speed (reasonably quiet)
Zalman 7700 0.21C/W with fan at full speed (reasonably quiet)

This article says the Intel S478 retail HSF is 0.33C/W
http://www.silentpcreview.com/article138-page1.html

http://www.overclockers.com/articles1094/ (Review of the XP-90)
http://www.overclockers.com/articles1043/ (Review of the XP-120)
http://www.silentpcreview.com/article194-page1.html (XP-90 review)
http://www.thermalright.com/a_page/main_product_accessories.htm#acc_775
http://www.thermalright.com/a_page/main_support_installation_lga775.htm

http://www.zalman.co.kr/eng/product/view.asp?idx=142&code=005009
ZM-CS1 Adapter for Zalman 7000x, for use with LGA775:
http://www2.newegg.com/Product/Product.asp?Item=N82E16835118219
http://www.zalman.co.kr/eng/product/view.asp?idx=146&code=005009
http://www.zalman.co.kr/product/cooler/7700-775MBlist_eng.htm

These are large heatsinks, and some of them overhang the edge of
the motherboard. As well as having to worry about clearance to
DIMMs, Northbridge heatsink, and capacitors, some space is
typically needed between the top edge of the motherboard and
the PSU.

On the Thermalright, you will need to purchase a separate fan.
(Which is a nuisance, finding a fan with a tachometer output
signal, for monitoring by the CPU_FAN header.)

The above products are not too tall, so there is not a lot of
torque on the socket when the heatsink is in place. You can avoid
excess weight on the Zalman heatsinks, by using the AlCu version.

I don't recomment tower coolers, due to the tendency for them to
pull on the motherboard, when the computer is upright. It is not
just the weight which is an issue, it is what happens if the computer
case receives a shock, and how the tower will behave then.
Generally, the heavier the assembly, or the more susceptable
to shock, then the higher the clamping pressure needed to keep
the heatsink on the socket.

There are a number of other heatsinks out there, some of which
will not have an overhang issue, but the noise and performance
may not be much better than the Intel retail product.

(review list)
http://www.overclockers.com/topiclist/index15.asp
(recommended product list)
http://www.silentpcreview.com/article30-page1.html
(review list)
http://www.silentpcreview.com/article_index.php#cooling

HTH,
Paul

 

[Bruce Murray]

XP-120 0.18C/W with inaudible fan
XP-90 0.18C/W with inaudible fan
Zalman 7000B 0.22C/W with fan at full speed (reasonably quiet)
Zalman 7700 0.21C/W with fan at full speed (reasonably quiet)

This article says the Intel S478 retail HSF is 0.33C/W
http://www.silentpcreview.com/article138-page1.html

http://www.overclockers.com/articles1094/ (Review of the XP-90)
http://www.overclockers.com/articles1043/ (Review of the XP-120)
http://www.silentpcreview.com/article194-page1.html (XP-90 review)
http://www.thermalright.com/a_page/main_product_accessories.htm#acc_775
http://www.thermalright.com/a_page/main_support_installation_lga775.htm

http://www.zalman.co.kr/eng/product/view.asp?idx=142&code=005009
ZM-CS1 Adapter for Zalman 7000x, for use with LGA775:
http://www2.newegg.com/Product/Product.asp?Item=N82E16835118219
http://www.zalman.co.kr/eng/product/view.asp?idx=146&code=005009
http://www.zalman.co.kr/product/cooler/7700-775MBlist_eng.htm

These are large heatsinks, and some of them overhang the edge of
the motherboard. As well as having to worry about clearance to
DIMMs, Northbridge heatsink, and capacitors, some space is
typically needed between the top edge of the motherboard and
the PSU.

On the Thermalright, you will need to purchase a separate fan.
(Which is a nuisance, finding a fan with a tachometer output
signal, for monitoring by the CPU_FAN header.)

The above products are not too tall, so there is not a lot of
torque on the socket when the heatsink is in place. You can avoid
excess weight on the Zalman heatsinks, by using the AlCu version.

I don't recomment tower coolers, due to the tendency for them to
pull on the motherboard, when the computer is upright. It is not
just the weight which is an issue, it is what happens if the computer
case receives a shock, and how the tower will behave then.
Generally, the heavier the assembly, or the more susceptable
to shock, then the higher the clamping pressure needed to keep
the heatsink on the socket.

There are a number of other heatsinks out there, some of which
will not have an overhang issue, but the noise and performance
may not be much better than the Intel retail product.

(review list)
http://www.overclockers.com/topiclist/index15.asp
(recommended product list)
http://www.silentpcreview.com/article30-page1.html
(review list)
http://www.silentpcreview.com/article_index.php#cooling

HTH,
Paul

Paul

Thanks once again for your deep insights into these systems. I am sure
other list members will benefit from your summary.
As is so often the case I was able to improve my situation by reading
the manual!
I did not have the BIOS set up the best way. I had disabled the Q Fan
Control. Now I have it enabled and also have the Fan Mode in PWM and
the fan ratio set to Auto.
It's pretty quiet now and the CPU temperature is reasonable.
I had no idea about the PWM fan control system by Intel and even after
spending a little time on the website, I am still not completely clear
on the logic being used.
Is there a "plain man's guide to the Intel Fan PWM control" that you
know of?


Thanks

Bruce

 

[milleron]

XP-120 0.18C/W with inaudible fan
XP-90 0.18C/W with inaudible fan
Zalman 7000B 0.22C/W with fan at full speed (reasonably quiet)
Zalman 7700 0.21C/W with fan at full speed (reasonably quiet)

This article says the Intel S478 retail HSF is 0.33C/W
http://www.silentpcreview.com/article138-page1.html

http://www.overclockers.com/articles1094/ (Review of the XP-90)
http://www.overclockers.com/articles1043/ (Review of the XP-120)
http://www.silentpcreview.com/article194-page1.html (XP-90 review)
http://www.thermalright.com/a_page/main_product_accessories.htm#acc_775
http://www.thermalright.com/a_page/main_support_installation_lga775.htm

http://www.zalman.co.kr/eng/product/view.asp?idx=142&code=005009
ZM-CS1 Adapter for Zalman 7000x, for use with LGA775:
http://www2.newegg.com/Product/Product.asp?Item=N82E16835118219
http://www.zalman.co.kr/eng/product/view.asp?idx=146&code=005009
http://www.zalman.co.kr/product/cooler/7700-775MBlist_eng.htm

These are large heatsinks, and some of them overhang the edge of
the motherboard. As well as having to worry about clearance to
DIMMs, Northbridge heatsink, and capacitors, some space is
typically needed between the top edge of the motherboard and
the PSU.

On the Thermalright, you will need to purchase a separate fan.
(Which is a nuisance, finding a fan with a tachometer output
signal, for monitoring by the CPU_FAN header.)

Quite true. I started a thread on this a few weeks ago, but I didn't
get many suggestions. I'd like to use a Panaflo fan, but the 120mm
models have no tach output. I'm still looking for suggestions about
the best and most silent 120mm fans for the XP120 that DO have the
tach output. Some suggested that I'd not need the RPM-monitoring
function on the CPU fan, and I'm sure that's true, but I'd just like
to be able to do it. Any suggestions from users??

The above products are not too tall, so there is not a lot of
torque on the socket when the heatsink is in place. You can avoid
excess weight on the Zalman heatsinks, by using the AlCu version.

I don't recomment tower coolers

I'm not sure what you mean by "tower coolers." I guess you're saying
that the XP120 is not a "tower cooler" when you state that the "above
products are not too tall." Am I inferring correctly? If so, what's
an example of a tower cooler?

due to the tendency for them to
pull on the motherboard, when the computer is upright. It is not
just the weight which is an issue, it is what happens if the computer
case receives a shock, and how the tower will behave then.
Generally, the heavier the assembly, or the more susceptable
to shock, then the higher the clamping pressure needed to keep
the heatsink on the socket.

There are a number of other heatsinks out there, some of which
will not have an overhang issue, but the noise and performance
may not be much better than the Intel retail product.

(review list)
http://www.overclockers.com/topiclist/index15.asp
(recommended product list)
http://www.silentpcreview.com/article30-page1.html
(review list)
http://www.silentpcreview.com/article_index.php#cooling

HTH,
Paul

Ron

 

[Paul]

Paul

Thanks once again for your deep insights into these systems. I am sure
other list members will benefit from your summary.
As is so often the case I was able to improve my situation by reading
the manual!
I did not have the BIOS set up the best way. I had disabled the Q Fan
Control. Now I have it enabled and also have the Fan Mode in PWM and
the fan ratio set to Auto.
It's pretty quiet now and the CPU temperature is reasonable.
I had no idea about the PWM fan control system by Intel and even after
spending a little time on the website, I am still not completely clear
on the logic being used.
Is there a "plain man's guide to the Intel Fan PWM control" that you
know of?


Thanks

Bruce

The Intel fan has four pins. +12V and GND are power for the
fan. The Tachometer signal is used by the motherboard, to
monitor whether the fan is spinning or not. Those are
standard on both the old three pin and new four pin fan
headers.

First, a little background. You've likely noticed a whole
raft of products that adjust the fan speed. Basically they
use two methods for adjusting the fan speed. Both methods
reduce the voltage, but the implementation affects how
efficient the schemes are.

One kind of scheme, puts a resistance in the +12V line.
This drops some voltage, so that less voltage is made
available at the fan terminals. The only problem with a
scheme like this, is the device doing the "resisting"
gets hot. It dissipates power equivalent to the dropped
voltage times the current flow. For example, if the
resisting device drops the voltage from 12V to 7V, and
the fan draws 0.1 amp, the power wasted in the resisting
device is 0.5 watts. This is generally an issue if the
resisting device doesn't have a nice flow of cooling air
over it or wasn't given a heatsink of some sort.

The second scheme is called Pulse Width Modulation. The
control signal in this case, is at a constant frequency,
and the frequency is likely a lot higher than the fan
speed. The pulse is repetitive, and only the width of
the pulse changes. It is obvious, if the pulse is 12V
high, and has 100% duty cycle, that the resulting voltage
can only be exactly 12V. If the duty cycle is less than
100%, and the pulse is filtered with a capacitor, then
the time averaged voltage developed across the capacitor
is proportional to the pulse width. The voltage on the
capacitor will have some ripple on it, and sometimes the
result is a little extra noise coming from the fan hub.

The beauty of PWM, is that the power switching device
making the +12V square wave, is either completely
turned on (saturated) or completely turned off (open).
In either state, it dissipates virtually zero power
(since in one case, the voltage drop is zero, and in the
other case, the current flow is zero.) That means, the
PWM transistor can be soldered to the motherboard, without
worrying about cooling.

OK. Now we return to the Intel fan header. The +12V line
feeding the fan, can be fed by either of the solutions
mentioned above. If a motherboard chooses to only touch
the standard three pins, then either kind of solution
can be added to the motherboard (linear resisting device
or PWM switching, feeding the +12V line to the fan).

A motherboard can also offer to use the fourth pin on
the Intel fan. This pin also uses the PWM method, only
some of the electrical components are inside the fan
hub. Why Intel thinks this is the "bees knees" is a
good question. The fan will be a few more cents
expensive than before, due to the PWM transistor and
filter capacitor being inside the fan hub. The motherboard
still has to generate a square wave, with a variable pulse
width, as determined by the fan speed that the BIOS or some
Windows driver/program decide. Effectively, the Intel
solution is the same as the motherboard version of PWM,
only the solution is distributed between the fan
and the motherboard.

So, what is the difference between using a motherboard-only
PWM scheme, and the Intel in-the-fan-hub PWM scheme ?
Should be no difference. If both the motherboard and
the fan implement PWM for themselves, then there is
unnecessary duplication of functionality. Maybe someone
can explain why this is a good thing, 'cause I don't
get it. The old way was getting the job done...

I suppose if the motherboard designer sticks with the Intel
provided solution, they don't need a PWM transistor and
a capacitor, but if the user plugs in an after market
three pin fan, then there would be no fan speed control
available.

As for where the PWM signal comes from, the hardware monitor
circuitry usually has the fan control PWM signal generator
inside it. So, if the motherboard is going to have a hardware
monitor anyway, the ability to do PWM can come for "free".

Paul