Posted by Bob on January 14, 2010, 11:39 pmPlease Register and login to reply and use other advanced options
I just installed a new ATX Power Supply, 275 watts, model Silencer by
PC Power and Cooling, Inc. It's several years old but never used.
The fan was blowing yesterday and sometime in the 8 hours the computer
has been running, it stopped.
(I turned it off and Im using another computer now. )
I'm running Asus Probe, which monitor fan speeds and temperatures. It
says the speed of the "power fan" is zero now. It's been below the
threshold speed since yesterday.
Is it possible something in or on the mobo could cause the fan not to
run? Asus Proble?
Did failure to connect it right stop the fan from working, even though
the computer is working?
Did it just fail, and if I replace it the next one won't?
Thanks for any help you can give
(The first fan failed because it was 8 years old and maybe because
it's a hostile enviorment wth a lot of dust.)
I'm running win98SE and XP on the same box and the symptoms are the
same in each.
Posted by Bob on January 14, 2010, 11:48 pm
BTW, could this have anything to do with the power supply not being
big enough for the load? The power supply which broke was rated 450
I have an Athlon 800 Mhz CPU and one gig, and a 64 meg video card,
ATI 7000 series, and not that much else. At leastr I think not.
Posted by Paul on January 15, 2010, 12:24 am
+5V @ 30A. +12V @ 10A, -5V @ 0.3A, -12V @ 1A, +3.3V @ 14A, +5VSB @ 2A
I can't find a picture of the supply. Some supplies have a "fan curve"
printed on the side, intended to indicate whether the fan shuts off
if the unit is cool enough. Some fan profiles are set up, such that
the fan spins at a low constant speed, when the inside of the PSU
is cool enough. Others are set up, such that the fan can stop, if the
power dissipation is not too great. On some dual fan units, the external
facing fan may continue to run, while the internal fan remains stopped
most of the time. So policies vary, as to how the fan can behave.
The above power supply is 70% efficient. If the power drawn was
150W (gaming mode), 150/0.70 = 214W total. Of that, 150W is
dissipated by the load, while 214-150 = 64 watts is waste heat
from the power supply. That should be enough heat to require
the fan to remain on. Convection wouldn't be sufficient to
handle that. If the computer case has a powerful exhaust fan, with the
PSU fan stopped, a sneak air path through the power supply, may cause
the power supply to be cooled by the main cooling fan on the computer
If it was an 80%+ efficient PSU, I might understand the fan going
off, but at 70% efficient, it seems a little less likely.
The other question is, what is the idling power of a computer with
an Athlon 800 in it. I don't know the answer to that. Athlon processors
don't have Cool N' Quiet, but they do have Halt and some kind of Bus
Disconnect ("Stop Grant"?), which are lower power states. So there is
some opportunity for power saving, but perhaps not as great as more
modern processors. My last two processors draw about 12W at idle. And
with that low a power draw for the processor, I suppose a fan could
switch off on the PSU.
The fan in the PSU, can only be monitored if the power supply has a
two wire cable with a three hole connector on the end. That allows
the fan inside the PSU to be monitored, when the cable is connected
to a monitored fan header on the motherboard.
The hardware monitor chip, measures the period of time between
two successive pulses from the fan RPM signal. Taking the inverse
of that time period, gives the RPM rate. The time is measured by a
register that counts clock ticks. If the register is "full", then
the monitoring circuit has come upon a fan rotation rate which is
too low to be measured. The monitor doesn't know if the fan is
300RPM or 0 RPM, since the register has overflowed, and it isn't
possible for the hardware to say what happened. As a result, the
BIOS (or Asus Probe), could sound an alarm or give an indication
that the fan has stopped.
Some Asus boards have a fairly high threshold. My old P4 board,
registers "0 RPM" when any fan falls below 1800RPM. If I use
Speedfan, Speedfan can reprogram the divider in the hardware monitor
chip, and monitor fan speeds as low as perhaps 700RPM or so.
In that case, it is the BIOS not using the proper divider setting,
that makes the hardware monitor a bit on the useless side.
Take the side off your PC, and place your hand on the side of the
power supply. If the power supply is cool to the touch and the
fan is stopped, then presumably all is still well. IF the power
supply is scalding hot, then you know the fan has actually
failed (or the cable has fallen off inside) and you should not
continue to use the supply until it is fixed. Most supplies will
have some kind of overtemperature circuit, so that the supply
will eventually switch off if it is too hot.
Some power supplies, use a connector for the internal cooling fan.
So you can actually change out the fan, without needing a soldering iron
or without touching the electronics inside. There are dangerous
voltages inside, so if you take it apart, try not to touch
anything except for the fan connector.
I've changed the fan in one power supply here (it was on a
Macintosh computer). It would have been hard to get an exact
replacement for that old Mac, so I decided to swap the fan out
and see if it all held together. I don't use the machine any more,
but the fan surgery appears to be OK. I changed all the fans
in that computer, because they were getting pretty noisy
If a fan power cable is soldered into the PCB of the power supply,
then replacing the fan (if that was what was needed), would be
a lot more dangerous. The main capacitor in the power supply has
a bleeder resistor, to remove the dangerous voltage on the cap,
but you never trust a bleeder to be working (if you expect
to remain safe). Some capacitors contain so much energy, that
you should not throw a dead short across them (the noise is
deafening). A high value resistor (like the bleeder) should be used,
to drain the cap over a 30 second period. Even when a cap is "bled",
if you come back in five minutes, you can find more voltage on it,
so you still have to be careful. I've only been thrown across
the room by high voltage once, and it is a great learning experience.
One minute you're here - the next you're there - and you can't
figure out how you got there.
Posted by Bob on January 15, 2010, 2:47 am
Thanks again for such a detailed reply.
There is a lot to absorb in your post, but let me make a 3-part reply
to these 3 parts.
What got me is that the breeze was noticeable the first time I turned
the computer on with the new power supply, but not at all the last
three times. I keep forgetting to check the breeze when I first start
it now, but I will.
It seems no matter how much I think I've told a complete story, I
never do in the first post: The machine wasn't idling but doing some
things when there was no detectable breeze. (I'll admit I couldn't
see the fan, but if it was running, it was very slow.)
The computer wasn't idling
Well I don't have a two-wire cable with a 3 hole connector. Nothing
like that. This probably accounts for why the "Power fan" was listed
as "Not monitored" in the Asus Probe screen. And at first I thought
that it wasn't monitored because of that. But when I found no breeze,
I guess I started to panic a bit.. I put a checkmark for the Power
fan, and immediately the warning beep started and the fan speed showed
0 rpm. So that really made me concerned, but now I realize I was
right in the first place, that the fan wasn't monitored because ther
was no way to monitor it, and checking the box for that fan could only
be misleading. It said 0 because it had no info.
But the other thing that worried me is that the CPU temperature was
rising, and when I went to rebooted to win98, it acted strangely, as
it sometimes does. The first time it just stopped in the 5 second
countdown, when I have a choice between safe start, regular start,
etc. It jjust stopped! Also I have a gig of RAM but every 6 or 9
months, the regular memory test only counts off 400 meg, or today 100
Meg. Then I do a hard reset and it counts off less, about 1/3 as much
each time as the previous time, once getting all the way down, in
about 7 steps, to 2000K instead of 1 million K. Oh, and when I
press Reset the tone that it makes early in startup, instead of one or
two seconds, lasts 5 seconds and is much much louder, and today it
lasted 10 seconds and wouldn't stop until I turned the computer off.
Then when I turn it on it goes to the CMOS screen, which I don't
change and I restart and it starts. Because letting it rest seems to
help, today I figured the computer was too hot, although it started at
130 or 140 andt was only up to 165 F, I think, at the CPU, well under
the value that Asus Probe calls dangerous.
I also imagined I smelled something, like two days ago when the PSU
had actually failed.
I realize now there is little if any relationship to the temp of the
power supply and the temp of the CPU.
But all these strange things together made me turn off the computer
and go to the win2000 one, which fwiw is still not fully set up.
Tomorrow it will be fully cold, and I'll check the breeze at the
start, and watch the temp more closely before concluding it's getting
Posted by Paul on January 15, 2010, 3:44 am
I would aim to keep the Athlon below 65C (149F), for stability. The
Athlon can run too hot because
1) Either the multiplier (on an unlocked processor) was set too high,
or the bus frequency was cranked up. For example, my AthlonXP
runs at 200x11, where 200MHz is the bus frequency and 11 is the
multiplier. In CPUZ, I'd see 2200MHz as my core frequency. I can set
the multiplier in the BIOS, to lower values, which makes the PC slower
and makes the processor run cooler. I also have the option to reduce
the input clock, but that slows down the front side bus (FSB). A 200MHz
input clock gives a FSB400 rate. 400 million times a second, the processor
can transfer 64 data bits, and that is 3200MB/sec of data (enough to keep
PC3200 RAM busy). That is the best that the S462 family can do. The slowest
Athlon, transfers at half that rate, or 1600MB/sec, and uses FSB200. You
should know the proper multiplier and input clock rate, so you can verify the
results later with CPUZ.
http://www.cpuid.com/cpuz.php (32 bit no install version)
(A screenshot, to show what info CPUZ can show you.)
2) A user can crank up Vcore in the BIOS, which makes the processor hot.
According to this, your Athlon 800 might normally use 1.75V for Vcore.
(The motherboard Vcore circuit is what creates the 1.75V, derived from
the 5V rail of your power supply.) Now, a welcome sign here, is the table
lists your processor as 40W max, so it isn't as hot as the processor I had.
That should make it easier on the power supply (less excuse for a burning
3) Improper cooling. If the CPU die doesn't have thermal paste on it,
to displace any air gap between heatsink and CPU, that can allow
the processor to get hot, without the heatsink dissipating the heat.
The CPU silicon die has to be in good contact with the heatsink, and
this is especially important with the Athlon. Also, you have to take
great care to not chip the edges of the bare die. Some processors
have rubber bumpers glued to them, to help protect the silicon die.
In other cases, a "shim" can be used for the same purpose. The idea is,
to prevent the heatsink from rocking from side to side, when you're
setting it in place. if it can't rock, it won't crack the edges of the
I'm beginning to wonder, if that burning smell isn't your new 275W power
supply running really hot, because the fan has failed ?
It sounds like the operating frequency of your hardware is wrong for some
reason. But proving something like that is happening, is pretty tough.
Years ago, you could get devices like this "RD2", and could measure
the PCI bus frequency with it. That might be a way to check one of
the fundamental frequencies on the PC (i.e. without buying a more
expensive frequency counter to do it). But I don't think they sell
these any more. They were discontinued.
If you can get the computer booted into Windows, you can try CPUZ, and
that is the quickest way to see if some frequency is not correct.
My guess is, your Athlon 800 should be 100 x 8, use FSB200 for
the front side bus, and run from a Vcore value of 1.75 volts.
The multiplier (8x)and input frequency (100) should also be visible
in one of the A7M266 BIOS screens. I can see such a screen in
the PDF manual.
I notice in a picture of the motherboard, there is a block (3x4 pins)
for VID in the lower left corner of the motherboard. That would be
a way of manually controlling Vcore. "Jumper Free Mode" on page 19 of
the PDF manual, shows how to use that. The processor picks the voltage,
if all the jumpers are in columns 3 and 4. If you choose to set
the voltage manually, the table on page 24 shows how to set the
voltage. You only use as much voltage as is needed to keep the
processor stable. The default for your processor is 1.75V.
If the previous owner of the board, set the jumper block manually,
it is up to you to verify how it has been set up. It will either
be in JumperFree mode (intended for BIOS control), or if in
Jumper mode, you can use the jumper block to set Vcore to whatever
value you want.
If the previous owner was overclocking, then the input clock might
have been set to 133MHz, instead of 100MHz.
And if you think this is complicated, then don't ever buy a modern
motherboard :-) Some of them are like flying the space shuttle.
Please Register and login to reply and use other advanced options
- Posted In
- Subscribe via RSS