power supply overheating?

[Dan Brill]

Hi,

This question began with my attempts to find out why the CPU fan in my new
homebuilt PC was rotating so fast (5000+ RPM) and, as a result, causing so
much noise. I have one of the newish Intel P4 3E (Prescott/90nm)
processors - the core temperature of which seems to vary between 60 and 65
Celsius (with the Intel supplied heatsink/fan). Initially assuming that the
reason for the high fan speed was the high CPU temperature, I emailed Intel
product support who claimed that the cause was probably a failure to
maintain the internal chassis temperature at or below 38 Celcius (and that
this *chassis temperature* was what the CPU fan was responding to).

As part of this ongoing investigation, I've noticed that the case of my
power supply (an Antec TruePower 430) gets very warm - between 43 and 50
Celsius (depending on load). I can imagine that a large block of metal
emanating heat at 50 degress might make it difficult to maintain an internal
temperature of 38 Celsius! But... does this indicate a fault with the PSU?
What should the approxiate temperature be? (For instance, this review
http://www.anandtech.com/showdoc.html?i=1774&p=8 at AnandTech measures the
temperature *inside* the unit after 30 minutes at 33.7C)

As a comparison, I have another computer with a lower spec Antec TruePower
380 PSU. The temperature of its case holds rock steady at about 40C
irrespective of usage or apparent load - this is in contrast to mine which
fluctuates without really stabilising (at about 10 degrees higher).

It seems to me that I am rather stuck in a cycle between various
manufacturers/vendors claiming that there equipment is not to blame and I
definitely need some independent advice. The actual cooling system is two
120mm Antec case fans (one intake, one exhaust) in an Antec Performance One
P160 aluminium chassis. There are also the usual number of other fans in the
system - on the graphics card (ATI Radeon 9800 Pro) and North Bridge
(Gigabyte GA-8KNXP Rev 2.0 motherboard) etc.

-dan

 

[w_tom]

Ignore the power supply chassis temperature. The Intel
response was correct. With two 120 mm fans, you are moving so
much air that the internal chassis *AIR* temperature should be
under 38 degrees. Is that same air also moving in the
vicinity of CPU? No ribbon cables obstructing airflow? Air
temperature - irregardless of PSU chassis temperature - is
only important.

Measure air temperature - not power supply temperature - as
the Intel rep suggested. Don't speculate about something
irrelevant as power supply chassis.

Furthermore, Prescott runs so hot that newer Intel designs
were scrapped - due to heat dissipation problems. The new
Intel CPUs will be multiple chips to radiate more heat. At 60
some degrees, you are still more than 20 degrees below max
temperature. That is significant so that your system works
just fine in a 100+ degree F room.

 

[kony]

Hi,

This question began with my attempts to find out why the CPU fan in my new
homebuilt PC was rotating so fast (5000+ RPM) and, as a result, causing so
much noise.

Max RPM of a fan is dictated by the design of the fan, or rather, the
resistor value used in the fan circuit, generally same circuit is used but
different value for one or two resistos is the determination of fan speed.
This is not a user-changeabe parameter, BUT the user can change the fan
(and heatsink if necessary) to one with more efficient thermal dissipation
and lower noise. In other words, a high-end aftermarket heatsink with a
800-92 x 25 mm fan on top, running at sub-3000 RPM.

Likely your motherboard also fan RPM control, but that fansink which it
controls, is the ultimate determination of resulting noise level per CPU
temp.

I have one of the newish Intel P4 3E (Prescott/90nm)
processors - the core temperature of which seems to vary between 60 and 65
Celsius (with the Intel supplied heatsink/fan). Initially assuming that the
reason for the high fan speed was the high CPU temperature, I emailed Intel
product support who claimed that the cause was probably a failure to
maintain the internal chassis temperature at or below 38 Celcius (and that
this *chassis temperature* was what the CPU fan was responding to).

It's difficult to build a system that is so poorly ventilated that the
chassis air temp is 38C. I mean, VERY difficult, you might have to take
extra measures to block all fan exhaust to achieve that temp. Even the
temp monitoring chip itself, which is it's own source of heat and always
reads higher than ambient temp, seldom gets much higher than 38C.

As part of this ongoing investigation, I've noticed that the case of my
power supply (an Antec TruePower 430) gets very warm - between 43 and 50
Celsius (depending on load). I can imagine that a large block of metal
emanating heat at 50 degress might make it difficult to maintain an internal
temperature of 38 Celsius! But... does this indicate a fault with the PSU?

The PSU is hot because it is creating (converting) so much power, and
isn't 100% efficient so the "waste" heat is being produced. That is not a
sign of a power supply fault but rather a sign of a (relatively) highly
loaded power supply, that the system it's powering is using a lot of
current. Any modern power supply has exhaust fan, the heat created is
being removed from the system by this path rather than heating up the
chassis except in rare cases where generic manufacturers use a bottom
intake fan on PSU that has substantially higher airflow than the
exit/exhaust fan on PSU... it is possible to "cause" this but it's not a
situation seem in any decent name-brand like Antec unless your case had an
overly restrictive stamped-in-metal grill guard behind the power supply
fan...more often seen with OEMs like Compaq in the past years but it's
easy enough to see for yourself by airflow rate out of power supply.

What should the approxiate temperature be? (For instance, this review
http://www.anandtech.com/showdoc.html?i=1774&p=8 at AnandTech measures the
temperature *inside* the unit after 30 minutes at 33.7C)

It is a meaningless test, since it varies based on degree of load as
mentioned previously, and actually the higher the temp from same load, the
better... that means the regulator to 'sink junction is more efficiently
transferring heat, and to a lesser extent, that the heatsink is smaller
and heat more concentrated, inherant in any PSU using a 2nd bottom-mounted
fan due to heatsink clearance issues. In other words, the temp may be
more of an indication that your system uses a lot of power more than
anything else, and that you need good quality heatsinks and as much
chassis cooling as reasonably possible if you'd like to minimize temps,
which isn't so necessary as keeping on eye on temps of sensitive parts of
the system, for example the CPU stability, HDD temp, and capacitor, fan
temp as they affect lifespan of those components as well.

If you want PSU to run cooler then you might focus on improving chassis
air intake, either increasing air passage size or installing fan in
addition to increasing air passage, but this may not be enough to achieve
significantly lower CPU temp without a heatsink swap. By significant, I
mean a dozen degrees or so.

As a comparison, I have another computer with a lower spec Antec TruePower
380 PSU. The temperature of its case holds rock steady at about 40C
irrespective of usage or apparent load - this is in contrast to mine which
fluctuates without really stabilising (at about 10 degrees higher).

How did you measure case temp? Keep in mind what I wrote previously about
the temp shown by motherboard bios or software usually being reported from
a thermal sensor in a chip, typically eing higher than actual chassis air
temp or having an artificial numerical offset that is a guesstimation, not
accurate across different systems that that used to determine the offset.

It seems to me that I am rather stuck in a cycle between various
manufacturers/vendors claiming that there equipment is not to blame and I
definitely need some independent advice. The actual cooling system is two
120mm Antec case fans (one intake, one exhaust) in an Antec Performance One
P160 aluminium chassis. There are also the usual number of other fans in the
system - on the graphics card (ATI Radeon 9800 Pro) and North Bridge
(Gigabyte GA-8KNXP Rev 2.0 motherboard) etc.

Your 120mm fans certainly have the potential to move sufficient air, but
for two things...

1) Make sure case bezel is unobstructed enough... any case bezel WILL
decrease airflow, but it's the case designer (or user's will to modify
that bezel) that determines the tradeoff between a sleek front without
large hole, and how much noise escapes from the front.

2) Stamped-in-metal fan grills on case walls... Every system I own has
these cut out. Typically there will be at least a 50% improvement in
airflow by just getting rid of (cutting them out with metalwork tools)
these stamped guards, give or take a little when adding a chrome grill
guard(s) in their place.... which I would do for a 120mm fan because the
larger area increases chance of wires, fingers, pets, etc, getting
damaged.

One thing not mentioned often enough, is that on smaller or newer cases
(often with thinner metal) a large percentage of chassis structural
integrity may be lost by cutting out a 120mm hole! This can be quite
significant with a rear fan because the rest of the width of the rear may
be a snap-in I/O shield, removable motherboard tray, etc, such that
there's little uncut metal left. In such cases you must assess the
resulting integrity and plan accordingly... sometimes it's beneficial to
only cut out part of that stamped-in grill, or cut slits and fold the
metal backwards, so it's not blocking the fan anymore but still providing
rigidity.

Folding can be less asthetically pleasing though, it is difficult to get
professional looking results compared to using a fairly rigid fan and a
couple of beefy chrome grills with screws mounted though the metal instead
of a plastic slip-in fan mount.... each case must be assessed
individually, there is not one solution that applies to all of them plus
your willingness and ability to work with metal also makes a difference.
In some situations the easiest fix for an already-built system is to add a
(left) side-panel fan, since the side panel can be removed and modified
without disassembly of rest of system. Perhaps your case has such a fan,
but again the air passage to that fan is mostly obstructed. Cutting out
that area or a larger area and larger fan will help lower chassis temp,
but be mindful of potential changes in HDD bay airflow.

Bottom line is as w_tom suggested, Intel's Prescott runs quite hot, for
optimal temps you have to take extra measures compared to most users (of
different CPU).

 

[DaveW]

The problem is that the Prescott CPU's put out over 100 WATTS of heat! And
Intel has not wanted to admit to the world that their latest products are
the cause of the overheating problems. Short of replacing the Prescott with
a much cooler and just as fast Northwood., you're looking at putting in a
high capacity cooling unit on the CPU, and installing the maximum number of
case fans that your case allows.

 

[Dan Brill]

Hi, thanks for the responses (to all three of you).

Just to be precise - the thermal guideline for this processor is 89.0 watts
and the thermal spec is 69.1 Celsius. The Intel tech support guy wrote:

"This processor's maximum temperature is 69.1 degrees Celsius"

From which, it seems like the CPU is running only a few degrees below
maximum (not much!).

One point I forgot to mention is that if I open the side of the case, the
air temperature *immediately* drops by ten degrees (say from 45C to 35C)
which had led me to believe that it might be a heat dissipation issue. There
are no ribbon cables obstructing airflow and the 120mm exhaust fan is only 2
or 3 inches away from the processor. As to measuring points, I have one
probe mounted on the CPU fan hub and (prior to moving it to check the PSU
temperature) another in the air about an inch vertically from the first. The
temperature of the second probe was generally a degree or so cooler than
that of this first (but still in the mid-40s). I'm pretty new to this so if
someone could let me know where the appropriate points would be, I'd be
grateful.

The max RPM of this fan is 6000 (again according to Intel) but it is not the
speed which worries me (or even the surrounding temperatures) - it was that
the fan makes a hell of a lot of noise whenever it exceeds 4000RPM (which is
almost all of the time, at 5000RPM it sounds rather like a jet aircraft
taking-off). I have be looking at all this from the perspective of
preventing the fan from reaching those speeds but maybe I am looking in the
wrong direction!

I can't easily install more fans (there are only two 120mm mounts) but I
could buy an aftermarket heatsink and fan. I believe, however, that this
will void my warranty - shouldn't the Intel supplied one at least be up to
the job? If I was overclocking it would be a different matter but...

Also, with this motherboard (Gigabyte GA-8KNXP) there is a riser card
(DSP2), with its own fan, very close to the CPU and apparently this prevents
the fitting of many third-party cooling solutions. Since the card is
specifically intended to regulate the power provided to Prescott CPUs I
would be loathe to remove it (and have been specifically recommended
elsewhere not to do so).

All three of you seem to think I should lay off the power supply angle which
is fair enough (though its behaviour still, to me, seems odd compared to
the, if anything, more heavily loaded Antec TruePower 380 I have in another
computer).

Finally, it has been brought to my attention that this case (Antec
Performance One P160) may not be 'thermally advantaged' - unfortunately it
certainly isn't advertised as such. It is fairly newly released, aluminium,
and expensive - so I'm potentially quite unhappy about this. However, before
I shell out another $150, I want to explore all other avenues.
Unfortunately, if there is nothing 'wrong' with any of the components (and
therefore nothing to return/replace), then this might be my only real
option. That and I'm really begining to wish I'd bought a Northwood for
around the same price!

Cheers,

-dan

 

[kony]

One point I forgot to mention is that if I open the side of the case, the
air temperature *immediately* drops by ten degrees (say from 45C to 35C)
which had led me to believe that it might be a heat dissipation issue. There
are no ribbon cables obstructing airflow and the 120mm exhaust fan is only 2
or 3 inches away from the processor. As to measuring points, I have one
probe mounted on the CPU fan hub and (prior to moving it to check the PSU
temperature) another in the air about an inch vertically from the first. The
temperature of the second probe was generally a degree or so cooler than
that of this first (but still in the mid-40s). I'm pretty new to this so if
someone could let me know where the appropriate points would be, I'd be
grateful.

It seems pretty clear from this that your fan grills are a large part of
the problem. Two 120mm fans "can" result in a LOWER temp with the cover
on than with it off, due to reducing amount of air recycled though
heatsink by moving air in different path.

The max RPM of this fan is 6000 (again according to Intel) but it is not the
speed which worries me (or even the surrounding temperatures) - it was that
the fan makes a hell of a lot of noise whenever it exceeds 4000RPM (which is
almost all of the time, at 5000RPM it sounds rather like a jet aircraft
taking-off). I have be looking at all this from the perspective of
preventing the fan from reaching those speeds but maybe I am looking in the
wrong direction!

I can't easily install more fans (there are only two 120mm mounts) but I
could buy an aftermarket heatsink and fan. I believe, however, that this
will void my warranty - shouldn't the Intel supplied one at least be up to
the job? If I was overclocking it would be a different matter but...

Well, Intel didn't make any claim that it would run at low noise, right?
So, it's running cool enough to be stable, but their (as always) marginal
fan design is the problem. Frankly it has amazed me for years that they
can make great CPUs (not to discount AMD's products, but in the larger
picture...) yet have failed to understand the simple fan... it's not even
an issue of cost really, since a standard larger fan could only be a few
cents more expensive if not cheaper than their proprietary design.

Perhaps you will have to void your warranty... I still have a few brand
new P3 1GHz heatsinks lying around because I wouldn't use them, even a $6
volcano 5 heatsink with a voltage-reduced fan was a better alternative,
but that was then and this is now... your CPU needs a higher-end cooler
and/or chassis changes. It would likely help to remove heatsink, scrape
off TIM with a credit card and lap it, use heatsink compound instead, but
again the warranty is voided.

Also, with this motherboard (Gigabyte GA-8KNXP) there is a riser card
(DSP2), with its own fan, very close to the CPU and apparently this prevents
the fitting of many third-party cooling solutions. Since the card is
specifically intended to regulate the power provided to Prescott CPUs I
would be loathe to remove it (and have been specifically recommended
elsewhere not to do so).

WITH that 2nd VRM, it's a great motherboard for supporting a Prescott, but
without it merely average. There is still a bit of clearance, you can
find a good 'sink that fits if you pay attention to measurements, but it's
really not THAT close unless I'm misjudging how tall (thick) the VRM card
is with the fan on it.

All three of you seem to think I should lay off the power supply angle which
is fair enough (though its behaviour still, to me, seems odd compared to
the, if anything, more heavily loaded Antec TruePower 380 I have in another
computer).

The power supply's heat is directly removed, and you have two of the
largest fans any "PC" case can support without extreme (excessive)
measures. If anything the primary problem with the power supply would
seem to be that your case has inadequate airflow because of the fan
grills. That is, when a lot of the heat off the CPU is expelled though
the power supply, it'll certainly run hotter. It simply means you need
the case fans to be more effective.

Finally, it has been brought to my attention that this case (Antec
Performance One P160) may not be 'thermally advantaged' - unfortunately it
certainly isn't advertised as such. It is fairly newly released, aluminium,
and expensive - so I'm potentially quite unhappy about this. However, before
I shell out another $150, I want to explore all other avenues.

Frankly, after cutting out the fan grills it'll have superior cooling to
most cases at any price. If the two 120mm fans are being throttled back
too much then you might also use a fan controller to choose a higher RPM,
or if they're thermally controlled fans then either replace the fans or
short their thermal sensor which will result in full-speed operation,
which is also noisey so it'd make a fan speed controller even more
important. The HDD rack in your case is convenient, but also reduces
airflow more than the traditional longitudinal arrangement. That cannot
be easily overcome but it can be the area of greatest restriction instead
of only one of many, so overall airflow rate can be be higher with that
bay than it is at present.

Unfortunately, if there is nothing 'wrong' with any of the components (and
therefore nothing to return/replace), then this might be my only real
option. That and I'm really begining to wish I'd bought a Northwood for
around the same price!

Replacing heatsink is the easiest solution.

 

[Guest]

why the CPU fan in my new homebuilt PC was rotating so fast
(5000+ RPM) and, as a result,
I have one of the newish Intel P4 3E (Prescott/90nm)
processors - the core temperature of which seems to vary
between 60 and 65 Celsius (with the Intel supplied heatsink/fan).
I emailed Intel product support who claimed that the cause was
probably a failure to maintain the internal chassis temperature
at or below 38 Celcius (and that this *chassis temperature*
was what the CPU fan was responding to).

I've noticed that the case of my power supply (an Antec
TruePower 430) gets very warm - between 43 and 50 Celsius

does this indicate a fault with the PSU? What should the
approxiate temperature be? (For instance, this review
www.anandtech.com/showdoc.html?i=1774&p=8 at AnandTech
measures the temperature *inside* the unit after 30 minutes
at 33.7C)

I have another computer with a lower spec Antec TruePower
380 PSU. The temperature of its case holds rock steady at
about 40C irrespective of usage or apparent load - this
is in contrast to mine which fluctuates without really
stabilising (at about 10 degrees higher).

50C is unusually hot for the case temperature, and I've never seen
that except in hot rooms or when the air flow was bad because of
blockage, a bad fan, or fans not being set up properly, or hot air
blew directly over the temperature sensing chip. If removing a side
panel makes the case temperature drop more than about 5C, then you
need to change something. It's possible that a card or cable is
blowing hot air over the temperature sensor or making hot air get
sucked right back into the fan, but sometimes the case fan blows in
the wrong direction for best cooling, especially in setups with more
than 2 case fans.

Antec power supplies tend to spin their fans slowly, causing higher
temperatures, but, contrary to claims made by all manufacturers, ATX
power supplies aren't actually highly efficient but only about 70%
efficient. So a computer that draws 150W will generate almost 50W of
heat, which will make its case fairly warm even with a fan running.

 

[Matt]

One point I forgot to mention is that if I open the side of the case, the
air temperature *immediately* drops by ten degrees (say from 45C to 35C)
which had led me to believe that it might be a heat dissipation issue.

Presumably you have one 120mm blowing in and the other blowing out.
Please verify.

 

[Lou]

Hi, thanks for the responses (to all three of you).

Just to be precise - the thermal guideline for this processor is 89.0 watts
and the thermal spec is 69.1 Celsius. The Intel tech support guy wrote:

"This processor's maximum temperature is 69.1 degrees Celsius"

From which, it seems like the CPU is running only a few degrees below
maximum (not much!).

One point I forgot to mention is that if I open the side of the case, the
air temperature *immediately* drops by ten degrees (say from 45C to 35C)
which had led me to believe that it might be a heat dissipation issue. There
are no ribbon cables obstructing airflow and the 120mm exhaust fan is only 2
or 3 inches away from the processor. As to measuring points, I have one
probe mounted on the CPU fan hub and (prior to moving it to check the PSU
temperature) another in the air about an inch vertically from the first. The
temperature of the second probe was generally a degree or so cooler than
that of this first (but still in the mid-40s). I'm pretty new to this so if
someone could let me know where the appropriate points would be, I'd be
grateful.

The max RPM of this fan is 6000 (again according to Intel) but it is not the
speed which worries me (or even the surrounding temperatures) - it was that
the fan makes a hell of a lot of noise whenever it exceeds 4000RPM (which is
almost all of the time, at 5000RPM it sounds rather like a jet aircraft
taking-off). I have be looking at all this from the perspective of
preventing the fan from reaching those speeds but maybe I am looking in the
wrong direction!

I can't easily install more fans (there are only two 120mm mounts) but I
could buy an aftermarket heatsink and fan. I believe, however, that this
will void my warranty - shouldn't the Intel supplied one at least be up to
the job? If I was overclocking it would be a different matter but...

Also, with this motherboard (Gigabyte GA-8KNXP) there is a riser card
(DSP2), with its own fan, very close to the CPU and apparently this prevents
the fitting of many third-party cooling solutions. Since the card is
specifically intended to regulate the power provided to Prescott CPUs I
would be loathe to remove it (and have been specifically recommended
elsewhere not to do so).

All three of you seem to think I should lay off the power supply angle which
is fair enough (though its behaviour still, to me, seems odd compared to
the, if anything, more heavily loaded Antec TruePower 380 I have in another
computer).

Finally, it has been brought to my attention that this case (Antec
Performance One P160) may not be 'thermally advantaged' - unfortunately it
certainly isn't advertised as such. It is fairly newly released, aluminium,
and expensive - so I'm potentially quite unhappy about this. However, before
I shell out another $150, I want to explore all other avenues.
Unfortunately, if there is nothing 'wrong' with any of the components (and
therefore nothing to return/replace), then this might be my only real
option. That and I'm really begining to wish I'd bought a Northwood for
around the same price!

Cheers,

-dan

I have the Antec P-160 also, but with an Athlon 64 3400 and the Zalman
70000A-cu hsf ,Coolmax 400w PS with 120mm fan (on low speed), only 1 case
fan placed in the back for exhaust and running on 5 volts for low speed, ATI
9800 pro with Zalman HSF at 7 volts, no other fans in the PC ,DVD and DVD R
and two HDD's. My case temp under load measured behind the top DVD drive is
about 37c max. My CPU under load is about 50c and 40c at idle. I could run
my fans faster for more cooling, but I wanted to build a quiet PC and
compromise temp for low noise (although I think its still cool enough).

My PC is very quiet with this setup and because I removed the front filter
and I cut out the grill the air flow is very good and that allows me to run
the fans slow and quiet. Also my HDD's are very quiet Samsung SpinPoint
mounted on Sorbathane rubber so they are nearly silent and even though I can
just about see them through the front case slots, I don't hear them.
I have read that the Intel P4 3E can run hot, but I think you also have an
airflow problem.
I would try a better CPU hsf that you can run quiet or even try a larger fan
using a fan adaptor on your stock CPU hs.
http://www.svc.com/fanad80to60.html .
Also try taking the front filter out for more air flow.
More info on making PC's quiet and air flow here
http://www.silentpcreview.com/

Lou

 

[kony]

My PC is very quiet with this setup and because I removed the front filter
and I cut out the grill the air flow is very good and that allows me to run
the fans slow and quiet.

NOW it's starting to make sense why a pair of 120mm fans can't keep his
rig cool enough... the filter

I have a shedload of fans and every combination of fan and filter that
I've tried (of any density even slightly high enough to stop dust, be more
than a decoration) cuts airflow by well over 50%.

Right now I'm building a system that has filter area over 300% greater
than the intake area of the fans, and still I expect significant airflow
reduction from their use. Unfortunately with the OP's system, having the
two exhaust fans and the intake filtered, the filter isn't even going to
be optimal, with a greater exhaust than intake potential all the cracks
(and drives) will have negative pressure and air will flow in though them,
somewhat defeating the purpose of a filter since it's quite easy to blow
out many areas of a pc that isn't filtered, but the small cracks and
drives are the harder parts to clean.

 

[Matt]

Presumably you have one 120mm blowing in and the other blowing out.
Please verify.

I mean to say that you _should_ have the lower front fan blowing in and
the upper rear fan blowing out. If they both blow in or both blow out,
they fight each other and you get poor air flow and poor heat removal.

 

[Timothy Daniels]

NOW it's starting to make sense why a pair of 120mm fans
can't keep his rig cool enough... the filter

I have a shedload of fans and every combination of fan and
filter that I've tried (of any density even slightly high enough
to stop dust, be more than a decoration) cuts airflow by well
over 50%.

Right now I'm building a system that has filter area over 300%
greater than the intake area of the fans, and still I expect
significant airflow reduction from their use. Unfortunately with
the OP's system, having the two exhaust fans and the intake
filtered, the filter isn't even going to be optimal, with a greater
exhaust than intake potential all the cracks (and drives) will have
negative pressure and air will flow in though them, somewhat
defeating the purpose of a filter since it's quite easy to blow
out many areas of a pc that isn't filtered, but the small cracks
and drives are the harder parts to clean.

Yeah, but....
You better blow out the system frequently without having a
filter and with all that air flowing through the case. The reason
I distrust those humongous heatsinks with all the narrowly
spaced fins is that dust gets in between the fins and blankets
their surface, reducing their capacity to conduct heat to the
passing air.

*TimDaniels*

 

[w_tom]

To put numbers to what Matt has posted. The three
parameters that define case internal air temperature are watts
consumed (which is less than what power supply is rated at),
incoming air temperature, and CFM. For example, if you have a
122 mm fan blowing in and another blowing out, then your CFM
is the CFM rating for one fan. That CFM number is provide by
the fan's manufacturer. CFM is the volume of air moved
through chassis. However if both 122 mm fans blow in, then
fans cannot move rated volume of air. Resulting insufficient
CFM means higher internal temperature.

Even worse, too much air means major dust problems. No
computer in a standard room environment should have dust
problems. Excessive dust buildup occurs when chassis moves
too much air - has too many high CFM fans. Most cases move
more than enough air with one 80 mm fan.

Hang and read a thermometer inside the case. That is the
'less than 38 degree C' that Intel cited. That temperature
would only be excessive if insufficient CFM are pushed through
case or if room temperature is too high.

Further note. That entire system must work just fine even
when room temperature is over 100 degree F. IOW measure CPU
temperature when room is at 70 degree F. That CPU temperature
must be 20 or more degrees C less than the maximum rated
temperature for the CPU. If not, then system will not work in
a 100 degree F room; therefore system is defective.

Therein defines how hot a CPU can run. It must remain below
manufacturer's maximum rated temperature even when room is 100
degree F. That means a CPU needs only run at 20 or more
degrees below maximum operating temperature (from Intel
datasheet) when room temperature is 70 degree F.

 

[Matt]

To put numbers to what Matt has posted. The three
parameters that define case internal air temperature are watts
consumed (which is less than what power supply is rated at),
incoming air temperature, and CFM. For example, if you have a
122 mm fan blowing in and another blowing out, then your CFM
is the CFM rating for one fan. That CFM number is provide by
the fan's manufacturer. CFM is the volume of air moved
through chassis. However if both 122 mm fans blow in, then
fans cannot move rated volume of air. Resulting insufficient
CFM means higher internal temperature.

I expect he has both case fans blowing out, so that they are fighting
the PSU fans, which are also blowing out, so that there is reduced or
even reversed flow through the PSU.

Dan, I believe your problem will be solved by making the lower-front fan
blow in instead of out. Maybe you can even just removed it to reduce
noise and still have enough air flow.

 

[Timothy Daniels]

...For example, if you have a
122 mm fan blowing in and another blowing out, then your CFM
is the CFM rating for one fan. That CFM number is provide by
the fan's manufacturer. CFM is the volume of air moved
through chassis. However if both 122 mm fans blow in, then
fans cannot move rated volume of air.

You assume that the volume of air per minute (CFM) is
independent of the resistance and internal pressure of the
case. If it were, two fans in series (i.e. in-line) would
provide the same CFM as a single fan, as you state.
But the one blowing in raises the pressure inside the case,
easing the load for the exhaust fan, and the exhaust fan
lowers the pressure, easing the load on the intake fan.
IOW, they help each other. And if there were no internal
resistance to the air flow, it wouldn't matter *how* much
voltage one supplied to a fan - it would always produce
the same air flow - which is not the case. The conclusion,
therefore, is that air flow is a function of the speed and the
size of the fan, and it varies with the amount of electrical
power supplied to the fan, and fans can help each other
increase air flow even when arranged in series.

*TimDaniels*

 

[kony]

I expect he has both case fans blowing out, so that they are fighting
the PSU fans, which are also blowing out, so that there is reduced or
even reversed flow through the PSU.

Dan, I believe your problem will be solved by making the lower-front fan
blow in instead of out. Maybe you can even just removed it to reduce
noise and still have enough air flow.

If the fans were pre-installed it's doubtful that the front fan is blowing
out. I suppose the factory might've accidentally installed one backwards
but odds are certainly against it.

I can simply look at the front of that case and see where the problem
lies, with the relatively small intake area, having a filter behind it,
and futher the more obstructive drive bay arrangement, the case is likely
restructing intake below what is seen with an unfiltered, unobstructed
80mm fan... and if the case is sealed enough, few if any aux air ports or
gaps, etc, then the rear fan is pretty much irrelevant, constrained by the
intake flow rate.

The quick and easy way to see if this is true would be to remove the front
bezel and filter, leaving side panels on the case, checking temp
difference running in that state. Also while the front bezel is off it
might be a good opportunity to take a nibbler and cut out the front fan
grill, if it's overly constrictive.

 

[Dan Brill]

Hi,

Both fans are blowing/sucking in the correct direction - inwards at the
bottom-front of the chassis, outwards at top-rear.

Today, feeling the spirit of adventure, I cut a rectangle of cardboard the
size of the side panel, then cut a 2-inch-square hole roughly over the
processor, and taped it onto the chassis in place of the side panel. Presto!
The case temperature never exceeds 36C.

Now I just need a proper side panel with this 'feature' built in and maybe
everything will be sorted. I've also been looking at Intel-designated
'thermally advantaged' cases (with a 'Chassis Air Guide'). There is a list
at
http://www.intel.com/cd/channel/res...t_inst_info/dsk_tested_source_lists/53211.htm
but most of them look so bloody ugly. Has anyone any experience with these?

Thanks for all your responses - most instructive.

-dan

 

[Matt]

If the fans were pre-installed it's doubtful that the front fan is blowing
out. I suppose the factory might've accidentally installed one backwards
but odds are certainly against it.

Yes. I thought the case came with only one case fan and guessed that
the OP added a second one in the front.

 

[Matt]

Hi,

Both fans are blowing/sucking in the correct direction - inwards at the
bottom-front of the chassis, outwards at top-rear.

Dang it, I woulda bet ten bucks.

Today, feeling the spirit of adventure, I cut a rectangle of cardboard the
size of the side panel, then cut a 2-inch-square hole roughly over the
processor, and taped it onto the chassis in place of the side panel. Presto!
The case temperature never exceeds 36C.

Now I just need a proper side panel with this 'feature' built in and maybe
everything will be sorted. I've also been looking at Intel-designated
'thermally advantaged' cases (with a 'Chassis Air Guide'). There is a list
at
http://www.intel.com/cd/channel/res...t_inst_info/dsk_tested_source_lists/53211.htm
but most of them look so bloody ugly. Has anyone any experience with these?

Functionality shouldn't be ugly.

Looks like you could just take your side panel to a glass shop to have
them cut a hole in the plexiglass. And buy a grill and/or an 80mm fan
to put over the hole. Or take it to somebody who does case
modifications. Or, if you are so inclined, practice a little with a
plexiglass cutter (cheap) and some scrap plexiglass and do it yourself.

Then you could keep the oh-so-attractive P160.

 

[~misfit~]

It's difficult to build a system that is so poorly ventilated that the
chassis air temp is 38C. I mean, VERY difficult, you might have to
take extra measures to block all fan exhaust to achieve that temp.
Even the temp monitoring chip itself, which is it's own source of
heat and always reads higher than ambient temp, seldom gets much
higher than 38C.

Depends on the positioning of the thermistor that is monitoring the case
temp. My case temp regularly gets over 38°C (On a hot day, 25°C ambient or
hotter) but then again the thermistor is positioned right in the exhaust
air-flow from my graphics card. My case has plenty of ventilation.