Re: Power supply fan airflow direction

kony wrote:

> A passively cooled system needs air vents, but a actively (fan-cooled)
> system shouldn't have any except in very strategically located spots,
> and then only if the case is designed so that internal pressurization
> forces air out those vents. Many systems/cases are designed just the
> opposite, with a slightl depressurization inside the case, reducing
> effective cooling of the fans, and pulling in dirt though every tiny
> crack, crevace, and optical, floppy drive.

That's something I've noticed with my other computers, that air is sucked in
through the floppy drive and the like. I think one of the reasons why this
is not happening with this case is that is is packed with cables. It didn't
even do that with the original fan in the PSU.

> Having air vents in top would result in the power supply pulling in
> outside air, which again, would be good for the power supply, but
> reduce system cooling. This is another area in which many poorly
> designed cases are causing problems, drawing air in through vents
> directly above and below the PSU, instead of the front face of the
> case.
>
> If I had your system I'd put a second fan on the power supply, blowing
> in, perhaps slightly higher RPM than that Pabst you're already
> using... the internal PSU fan's noise should be less noticeable than
> an exhaust fan.

I'll experiment with that. BTW, do you mean by a fan with a slightly higher
RPM than the Papst that the intake fan should blow harder than the
"outtake" fan, or just that the Papst doesn't blow hard enough?

I'm testing an extra exhaust fan now (see my reply to Wheat Muncher's
message), inspite of what you said, the PSU seems cooler now. The case
definately is, but not enough (4-5 degrees C).

Halfgaar
--
To send email, change nospam.com into yahoo.com.

On Tue, 24 Jun 2003 00:35:53 +0200, Halfgaar <voor_achter@nospam.com>
wrote:

<snip>

>
>I'll experiment with that. BTW, do you mean by a fan with a slightly higher
>RPM than the Papst that the intake fan should blow harder than the
>"outtake" fan, or just that the Papst doesn't blow hard enough?

I mean both, that a Papst fan with that low of a noise rating can
safely be assumed to produce too little airflow for heat removal in a
PSU producing significant heat as it will in a modern
(higher-performance) system.

Perhaps only changing the Pabst fan to a higher CFM fan would work,
but the rear fan in a PSU, being at the wall of the casing blowing
out, will cause more noise to escape the case. A lower-noise solution
would be to keep that fan but add a second fan, blowing into the PSU.
It's possible that another similar (low-airflow) Pabst fan would be
sufficient addt'l airflow, but only you can make that determination
and decide how hot you want to allow the PSU to be.

>
>I'm testing an extra exhaust fan now (see my reply to Wheat Muncher's
>message), inspite of what you said, the PSU seems cooler now. The case
>definately is, but not enough (4-5 degrees C).
>
>Halfgaar

Where are you taking your case reading? If it's the motherboard's
integrated sensor, the reading you get is from a chip, often the
southbridge, and is not directly related to actual ambient case temps,
can largely be ignored. The Southbridge will be fine running at ~45C,
but consider the other components like the HDDs, effecitveness of the
CPU heatsink, motherboard power supply area (capacitors temp).


Dave

kony wrote:

> Where are you taking your case reading? If it's the motherboard's
> integrated sensor, the reading you get is from a chip, often the
> southbridge, and is not directly related to actual ambient case temps,
> can largely be ignored. The Southbridge will be fine running at ~45C,
> but consider the other components like the HDDs, effecitveness of the
> CPU heatsink, motherboard power supply area (capacitors temp).

The motherboards sensor, that right. I'm not sure where it is located, I'm
running a topic on it on a MSI group . I don't think it's the southbridge
temp, because my linux sensor tools can read a third sensor which is called
the southbridge, but it is normally disabled because it mostly gives bogus
readings. Someone on the MSI group said that on the K7T-Turbo2, the sensor
is below the AGP. I have a K7T-Turbo.

BTW:
HDD Temperature_Celsius 39
CPU Temp: +41.7°C
Sys Temp: +32.2°C

As I said to my reply to your other message, I made some changes. This is
the exact result. It' the result from hours of mostly being idle.
Unfortunatly, I don't know the temp at the top of the case.

Halfgaar
--
To send email, change nospam.com into yahoo.com.

kony wrote:

> Now we're getting somewhere... I had an MSI
> K7T Turbo and still have a K7T Turbo2. There's no quesiton about it,
> the sensor is in the southbridge. No sensor below the AGP or anywhere
> else, just the Southbridge. As for why your Linux program mislabeled
> them, well that's common, there's just a few temp monitor "circuits"
> and they are often in reversed labeled "positions" in misc apps...
> same goes for the fan RPM sensors. The Southbridge reading is not an
> accurate representation of the system temp, is more related to how
> close the voltage levels are to spec, and how much air is flowing
> directly over the southbridge. The southbridge needs no heatsink, but
> if you were to attach a fanned heatsink similar to on a northbridge,
> you could watch the temp drop as a result, it would more closely
> represent the ambient case temp but still be a bit higher.

Funny you should mention "a fanned sink like on the northbridge". My
northbridge only has a sink and it gets finger-burning hot. It doesn't seem
OK to me, but it's what MSI decided to do.
And you're right about the mislabling, the fan labels were also swapped. So
I guess your right about my systemtemp sensor too.

>
> I wouldn't worry about those temps, they seem fine, but you might take
> note of the temps on the capacitors next to the row of voltage
> regulators onboard (touch-test them). If they're more than slightly
> warm expect them to have reduced lifespan. When I got my Turbo2 I put
> heatsinks on the regulators, a socket 7 'sink on the northbridge, then
> sort of "draped" the fan half on the northbridge 'sink and half
> hanging off to the left of it, to move air through the 'sinks for the
> regulators. You may not need to do this, I was taking precautionary
> measures prior to highly overclocking a Palomino Athlon in a small
> case.

At least two caps on my board get very warm, not finger-burning, but still
quite warmer than the rest.

> Putting a second fan on the power supply should be sufficient to keep
> the top of the case cool, providing there's an adequate air inlet.

Intake air is supplied by a 12 cm fan at 1300 RPM, not very very much, but I
don't think that's going to be a problem.

If you don't mind (and if you do ), I'll answer some questions of your
other reply here as well:

My 300W powerman PSU has a few small inlet slots at the back and the side
and many of them are covered by components inside. I think what I must do
is get myself a powerfull PSU really built for low noise but still produce
a good airflow, like the one in my other computer I talked about. This PSU
is also labelled as "low noise" but the only reason for that seems that it
has a fanspeed controller which lets the fan spin faster if it's hotter.
Not that it helps, because it only slows down the fan when the PSU is
practicly freezing.

About the livespan of my hardware, I care about long livespans. I also make
sure I don't turn my computers or other electronic devices on and off many
times, to prevent power surge damage.
But I also like low-noise. I think I have to mix the two in this case.

And about the papst, I must say that I don't know what bearing it has. I had
thought a ball bearing, because in my experience only very cheap fans use
sleeves, but you're making me doubt.

Halfgaar
--
To send email, change nospam.com into yahoo.com.

On Thu, 26 Jun 2003 00:42:36 +0200, Halfgaar <voor_achter@nospam.com>
wrote:


>Funny you should mention "a fanned sink like on the northbridge". My
>northbridge only has a sink and it gets finger-burning hot. It doesn't seem
>OK to me, but it's what MSI decided to do.
>And you're right about the mislabling, the fan labels were also swapped. So
>I guess your right about my systemtemp sensor too.

My Turbo2 also had only the tiny 'sink, but since it sits on the board
at a somewhat 45 degree angle, it allowed a few slow 60mm fan to be
mounted diagonally across half of a traditional 40-50mm 'sink... half
the reason I changed the 'sink was so I had a nice location to mount
the fan for cooling the voltage regulator/capacitor area to the left.
I was using a relatively low-CFM CPU heatsink fan so a bit of extra,
directed airflow did seem to help.

Hmmm, I just realized something, I've been getting my MSI motherboad's
names mixed-up... I was thinking of a K7T266 Pro2 half the time, and
the other half I was thinking about a K7T Turbo2 (I've had both). The
Turbo2 already had heatsinks on the voltage regulators from the
factory, and was unusual in that it used a 12V P4 connector for CPU
power. What I previously described about "draping a fan" to cool the
reguloators/caps, was on the K7T266 board. On the K7TTurbo2 I cooled
the northbridge passively, used 1/3 of an old P2 Katmai heatsink, cut
off.. was one of those OEM-passive-type with really tall tines. It
must've done well enough, I recally that the board was stable over
150MHz FSB, around which point the USB ports crapped out.

>> I wouldn't worry about those temps, they seem fine, but you might take
>> note of the temps on the capacitors next to the row of voltage
>> regulators onboard (touch-test them). If they're more than slightly
>> warm expect them to have reduced lifespan. When I got my Turbo2 I put
>> heatsinks on the regulators, a socket 7 'sink on the northbridge, then
>> sort of "draped" the fan half on the northbridge 'sink and half
>> hanging off to the left of it, to move air through the 'sinks for the
>> regulators. You may not need to do this, I was taking precautionary
>> measures prior to highly overclocking a Palomino Athlon in a small
>> case.
>
>At least two caps on my board get very warm, not finger-burning, but still
>quite warmer than the rest.

I do recall some people reporting that a couple of caps right above
the AGP slot were prone to failure, the one in the middle and the
other next to the memory slots, right below the northbridge.


>
>> Putting a second fan on the power supply should be sufficient to keep
>> the top of the case cool, providing there's an adequate air inlet.
>
>Intake air is supplied by a 12 cm fan at 1300 RPM, not very very much, but I
>don't think that's going to be a problem.
>
>If you don't mind (and if you do ), I'll answer some questions of your
>other reply here as well:
>
>My 300W powerman PSU has a few small inlet slots at the back and the side
>and many of them are covered by components inside. I think what I must do
>is get myself a powerfull PSU really built for low noise but still produce
>a good airflow, like the one in my other computer I talked about. This PSU
>is also labelled as "low noise" but the only reason for that seems that it
>has a fanspeed controller which lets the fan spin faster if it's hotter.
>Not that it helps, because it only slows down the fan when the PSU is
>practicly freezing.
>
>About the livespan of my hardware, I care about long livespans. I also make
>sure I don't turn my computers or other electronic devices on and off many
>times, to prevent power surge damage.
>But I also like low-noise. I think I have to mix the two in this case.
>
>And about the papst, I must say that I don't know what bearing it has. I had
>thought a ball bearing, because in my experience only very cheap fans use
>sleeves, but you're making me doubt.
>
>Halfgaar

Here's a Pabst spec sheet that seems to indicate sleeve bearings:
http://www.papst.de/pdf_dat_e/S17gb.pdf

There are decent fans that use sleeve bearings like Pabst and Panaflo,
but it is true that often a sleeve bearing is a sign of a cheap/junk
fan. Pabst and Panaflo fans have pretty good bearings that'll last
near as long at room temps, in a vertical mounting position, but even
so an equivalent quality dual-ball-bearing fan will last a lot longer
in high-heat environments. The main thing is to lube the
sleeve-bearing before it dries out and starts grinding into dust,
getting noisey.

Dave

kony wrote:

> My Turbo2 also had only the tiny 'sink, but since it sits on the board
> at a somewhat 45 degree angle, it allowed a few slow 60mm fan to be
> mounted diagonally across half of a traditional 40-50mm 'sink... half
> the reason I changed the 'sink was so I had a nice location to mount
> the fan for cooling the voltage regulator/capacitor area to the left.
> I was using a relatively low-CFM CPU heatsink fan so a bit of extra,
> directed airflow did seem to help.
>
> Hmmm, I just realized something, I've been getting my MSI motherboad's
> names mixed-up... I was thinking of a K7T266 Pro2 half the time, and
> the other half I was thinking about a K7T Turbo2 (I've had both). The
> Turbo2 already had heatsinks on the voltage regulators from the
> factory, and was unusual in that it used a 12V P4 connector for CPU
> power. What I previously described about "draping a fan" to cool the
> reguloators/caps, was on the K7T266 board. On the K7TTurbo2 I cooled
> the northbridge passively, used 1/3 of an old P2 Katmai heatsink, cut
> off.. was one of those OEM-passive-type with really tall tines. It
> must've done well enough, I recally that the board was stable over
> 150MHz FSB, around which point the USB ports crapped out.

I remember Katmai as a P3 core, but never mind that...

You know, by looking at my board, it looks like it was designed to have have
the really hot parts be cooled by CPU-fan airflow. The northbridge, voltage
regulators and caps which get hot, are all located around the socket. As
you know, I don't have a CPU fan anymore, and some of the parts get very
hot. I told you yesterday that I have two caps which get a bit warmer than
the rest, but I tried again today and they actually get very very hot. I
burn my finger on them. My alcohol-thermometer says they're 52 degrees C. I
don't think a stronger airflow from a PSU fan is going to fix this. I think
I need a fan blowing over the parts. I tried this with my desk-fan and then
the caps are hardly hot at all.

Perhaps you could explain something to my BTW. On the caps is written "SM105
degrees C" (but with a degree sign). Does that mean that they're capable of
running stably at 105 degrees C?

> I do recall some people reporting that a couple of caps right above
> the AGP slot were prone to failure, the one in the middle and the
> other next to the memory slots, right below the northbridge.

The caps on my board which get so hot are at the top-edge of the board (in
tower mount), above the CPU.

> Here's a Pabst spec sheet that seems to indicate sleeve bearings:
> http://www.papst.de/pdf_dat_e/S17gb.pdf

There's only one 12 db fan, and it uses a sleeve, so I guess my fan uses a
sleeve

> There are decent fans that use sleeve bearings like Pabst and Panaflo,
> but it is true that often a sleeve bearing is a sign of a cheap/junk
> fan. Pabst and Panaflo fans have pretty good bearings that'll last
> near as long at room temps, in a vertical mounting position, but even
> so an equivalent quality dual-ball-bearing fan will last a lot longer
> in high-heat environments. The main thing is to lube the
> sleeve-bearing before it dries out and starts grinding into dust,
> getting noisey.

Are the fanblades removable? I mean, how else can I grease it up when the
fanblades can't be removed.

BTW. I noticed that when I open my cdrom drive, hot air is coming out of it.
I seems that the excess pressure inside the case can't flow out through the
PSU. This confirms a bit that my PSU is too full of components. But on the
other hand, the back of my computer has airholes in it, though which
pressure can escape. Any thoughts on this?

Halfgaar
--
To send email, change nospam.com into yahoo.com.

On Fri, 27 Jun 2003 00:20:29 +0200, Halfgaar <voor_achter@nospam.com>
wrote:

> Perhaps you could explain something to my BTW. On the caps is
> written "SM105 degrees C" (but with a degree sign). Does that
> mean that they're capable of running stably at 105 degrees C?

Yes, within 2000 hours :-(

On Fri, 27 Jun 2003 00:09:59 GMT, kony <spam@spam.com> wrote:

> It's an exhaust fan, right? The sticker you need to peel back should
> already be facing the outside of the system, so you don't even need to
> open the case (unless your PSU still has a fan grill on it... if it
> does, i'd remove/cut-out that grill. Just peel the sticker back, then
> there may be a rubber or plastic plug, which when removed exposes the
> fan shaft... just a drop or two of oil, the thicker the better (will
> also reduce it leaking back out if the plug doesn't make a perfect
> seal). Synthetic gear-oil is good, grease mixed with motor oil is ok
> too, but avoid light oils or non-lubricants like WD-40.

Images.

http://w1.857.telia.com/~u85710476/...ample/fan01.jpg

http://w1.857.telia.com/~u85710476/...ample/fan02.jpg

kony wrote:

> People who use cheap heatsinks which compensate for lack of metal with
> a few high-flow fan have it better in this regard, there's more,
> cooler air coming off the 'sink, while I always use highly efficent
> large 'sinks with low-flow fans, so there's less but hotter air. For
> me a 50-60mm fan running at 7V helps keep that area cooler than any
> other method, with negligable noise.
>
> There are other "hard-core" methods of keeping the caps cooler, but
> most aren't worth the effort... for example if the board has vacant
> spots for more caps in parallel, adding those caps would decrease the
> ripple current, and the ESR as a whole is lower, with the heat
> generated being spread among more caps. Another option is replacement
> with lower ESR caps, either larger in the similar low-ESR Aluminum
> variety, or something like an organic cap.
>
> Yes, but not really... They can function properly up to that temp,
> without being rapidly destoyed, but will have a decreasing, very short
> lifespan. Think of capactiors like incandescent light-bulbs, in that
> they wear out and need replaced on a regular interval... what that
> interval is in an otherwise-static environment, depends on how hot
> they run, and how much the circuit is over-engineered, so that even if
> the cap is below it's acceptable value after a period of time, it
> still functions enough for the circuit it's in.

Ken said about 2000 hours, which is not very much... My two year old HD has
already 7000 hours on it, according to SMART.


> Sounds like you need a fan ;-)

[...]

>
> It's an exhaust fan, right? The sticker you need to peel back should
> already be facing the outside of the system, so you don't even need to
> open the case (unless your PSU still has a fan grill on it... if it
> does, i'd remove/cut-out that grill. Just peel the sticker back, then
> there may be a rubber or plastic plug, which when removed exposes the
> fan shaft... just a drop or two of oil, the thicker the better (will
> also reduce it leaking back out if the plug doesn't make a perfect
> seal). Synthetic gear-oil is good, grease mixed with motor oil is ok
> too, but avoid light oils or non-lubricants like WD-40.
>
>
> When the interior is pressurized air will escape any and everywhere it
> can. That's actually a good thing, provided there's enough airflow.
> With the power supply's rear exhaust hole cut out and another hole in
> the "rear" of the PSU (towards the front of the case) or under the
> PSU, and a second fan added to this location, the density of the PSU
> interior shouldn't be so great as to impede proper airflow... after
> all, in theory they designed it to stay cool enough with proper
> airflow rate. I'd rather have a full power supply than one with a lot
> of extra space in it, a higher-wattage PSU with good electrical
> properties will indeed leave little extra space inside a standard PS2
> sized PSU casing. Considering your motherboard capacitor temps, I'd
> either replace the PSU with one having a fan underneith, or cut out a
> fan hole, stategically placing the fan (and additional ductwork if
> necessary) to move some air past that region of the motherboard).
> Adding a fan may only work if you have at least 11mm of clearance
> between the motherboard and PSU, for a 10mm thick fan, though
> certainly as thick a fan as possible would be preferrible, greater
> CFM/noise ratio.
>
>
> Dave

I don't even have the fangrill cut out, that's a good idea. It probably
decreases turbulence as well, won't it? When I'm at it, I'll also cut a
hole for an 80mm intake fan. I have enough space for a fan on the outside
of the PSU (inside of the computer), the PSU is mounted on it's side next
to the mainboard, not above it. Perhaps the hole itself and the removed
fangrill will help enough, but I don't think so. I just hope there is room
for the hole for the intake fan. There are parts mounted on every side of
the PSU on the inside.

BTW, are those 12 to 7 volt adapters widely available?

On Sat, 28 Jun 2003 00:58:09 +0200, Halfgaar <voor_achter@nospam.com>
wrote:


>> Yes, but not really... They can function properly up to that temp,
>> without being rapidly destoyed, but will have a decreasing, very short
>> lifespan. Think of capactiors like incandescent light-bulbs, in that
>> they wear out and need replaced on a regular interval... what that
>> interval is in an otherwise-static environment, depends on how hot
>> they run, and how much the circuit is over-engineered, so that even if
>> the cap is below it's acceptable value after a period of time, it
>> still functions enough for the circuit it's in.
>
>Ken said about 2000 hours, which is not very much... My two year old HD has
>already 7000 hours on it, according to SMART.

True, the lifespan is short, but hopefully the board is overengineered
enough that it's lifespan exceeds it's useful life. I always "try" to
keep temps low enough that I can leave my finger on the cap
indefinitely.


>I don't even have the fangrill cut out, that's a good idea. It probably
>decreases turbulence as well, won't it?

Yes, and increases airflow. If it's a stamped-in fan grill, removing
it alone may greatly increase the airflow.

>When I'm at it, I'll also cut a
>hole for an 80mm intake fan. I have enough space for a fan on the outside
>of the PSU (inside of the computer), the PSU is mounted on it's side next
>to the mainboard, not above it. Perhaps the hole itself and the removed
>fangrill will help enough, but I don't think so.

Don't cut a hole underneith unless you're committed to installing a
fan there... otherwise, the rear portion of the PSU will receive even
LESS airflow. COnsidering that you still had the fan grill on the
rear I'd expect that adding a second 12db Pabst fan will be enough,
but with the PSU in that location it does little to help cool the
motherboard.

>I just hope there is room
>for the hole for the intake fan. There are parts mounted on every side of
>the PSU on the inside.

What brand of PSU is it again?
Since your PSU is mouted in front of the mobo, you may or may not have
room for mounting a fan on the outside of the PSU casing... might
depend a lot on the configuration of the water-cooling block but I'd
expect it to be more accomodating than a large traditional heatsink.

>BTW, are those 12 to 7 volt adapters widely available?

No, there may be some places that sell them ready-made for a premium
price, but IMHO, the best thing to do is find the lowest price in your
region for a regular 4-pin-to-3-pin adapter, then just swap the wires
in the connectors, the ground for the fan with the 5V wire (just be
sure to swap both ends, to minimized the risk of frying something
later attached to the adapter, be sure you have the right voltages on
the right pins at each end-connector of the adapter. Here's an
example of the adapter I mean:

http://www.svcompucycle.com/3pinto4pinad.html


Dave