Antares said:
I have an older desktop computer that I built a few years back. I'm
still running Windows XP Pro SP3 on this computer. It has a Gigabyte
GA-P35-DS3L motherboard with an Intel Core 2 Duo 2.66 GHz CPU setup.
This computer has two Western Digital WD5000 hard drives and both have
about half the drive space available.
This computer has worked very well over the years but recently, the
CPU fan will do a start - stop - start - stop over and over again when
I try to start this computer up or bring it out of hibernate. When
this occurs there is a loud beep that is synchronized with the CPU
fan's on again, off again pattern. I can remove the side cover of this
computer and use my finger to twirl the CPU fan when it stalls out.
This usually gets it going and all is well until the next time I start
this computer.
Is this a failing CPU fan or is it a motherboard fan control circuit
failure?
Any time I put my finger on the fan and stop it the same beep pattern
occurs. Then, if I twirl the fan and get it going the beeping stops.
Thanks for any suggestions. I can go buy a new fan if this seems the
most likely source of the problem, but I don't know about trying to
repair or replace the motherboard. Gordon
Based on your description, I'd say the BIOS is detecting the low
fan speed, and the "beep" is a cry for help.
Fans are controlled two ways. The old way, was a three wire fan, with
voltage control on the +12V lead to the fan. If the +12V copper track
gets burned, or if there is a transistor near the header that
is bad, that can cause the fan to run slow. Accumulated
dirt can also cause the fan to run slow.
The new control method, uses a four wire tiny connector, with the
supply voltage lead sent a fixed +12V signal, and the new fourth wire
carrying a PWM (pulse width modulation) signal. If the PWM signal is
disconnected, it floats to a logic 1 on the fan end, and that causes
the fan to run at 100% speed. A PWM controlled fan can run slow,
if the +12V copper track gets burned (as before), or if the PWM signal
is grounded, that would indicate a request for "low speed". And the low
speed, may be too slow for the BIOS speed check. The PWM signal would
have a 25KHz 50% duty cycle square wave on it, if the motherboard
wanted the fan to run at 50% speed (half way).
The best way to handle this, is to bypass the CPU fan header. But
there is a problem with this approach. You can get an adapter
cable which is included in the box some cooling fans come in.
The last "Vantec Stealth" fan I bought, has a cable with
a male Molex 4 pin, a female Molex 4 pin (for daisy chaining
connectors(, and an LP3 male to connect to a fan. The fan spins fine
using one of those (you can connect an LP3 to a four pin Intel
fan, as the fourth PWM pins floats and that signals a request
for 100% speed).
What's wrong when you do that, is while the CPU is now cooled
properly, the motherboard CPU fan header has nothing connected to
it. The BIOS expects to see thirty pulses per second
on the tachometer pin of the CPU fan header. You then need to
run a wire, from the adapter cable, to the CPU fan header.
I own precisely one of these adapter cables which was designed
to do exactly that, and you can't normally just buy them at
the store. If I needed a second one now, I'd have to custom
make it from parts. Luckily, my local electronics store
(active123.com) carries male and female LP3 connectors,
so I have all the parts I need.
LP3 male
Moles_male -----+--------------> +12V --- connect
+12,GND,GND,+5 | --------> GND --- to Intel CPU
| +-------- Tachometer --- four pin fan
Molex_female ---+ | ----/ (PWM floats)
+12,GND,GND,+5 |
+--------> LP3 female --- Plug onto
/--- (plug onto --- motherboard CPU
/--- --- fan header
----/ (PWM is no connect)
The adapter cable in this case, comes with four connectors. Two
Molex disk drive connectors are provided, so you can "daisy
chain" power and steal power from a hard drive connector, without
"using up" any ATX power supply connectors.
One LP3 connector goes to the Intel CPU fan. It is a three pin
male connector, with alignment tab to prevent wrong insertion.
The Intel four pin CPU fan connects to this, and the fourth (PWM)
pin on the Intel fan is not connected. This causes the Intel
fan to run at 100%, giving good cooling to the CPU.
The second LP3 connector is a female. It has only one wire on
it, and the other two holes are blank. You plug that three pin
female, into the motherboard four pin (LP4) CPU fan header. The PWM
pin doesn't need a connection (it's an output from the motherboard,
which in this case isn't being used). There is a grand total
of only one signal being connected via this LP3 female. And
that is the tachometer or RPM signal, flowing from the Intel fan
to the motherboard header. That RPM signal is needed to fool
the BIOS into thinking the CPU fan header is normally
connected.
My local electronics store, carries LP3 connectors and crimp
pins, so I can make an adapter of that type. I could, for example
buy a Vantec Stealth fan for $12.77 at a local computer store,
buy a single wire and add an LP3 female to the
cable assembly, to make a complete cable assembly
to bypass the power coming from the motherboard CPU
fan header. By connecting the LP3 female with the RPM
pulse signal on it, to the CPU fan header, that keeps
the BIOS convinced all is well with the fan.
By bypassing the CPU fan header that way, my hope would
be that the fan can be properly powered and run at
full speed. There would then be no need to turn it
with your finger to get it started.
Don't forget to check the BIOS fan settings, in case
you've selected some fancy automated control. Set it
to manual mode, in such a way as to request 100% CPU
fan speed at all times. That will give the fan the best
chance of working properly for you. I don't normally
use the built-in speed controls on the motherboards -
if I need to slow down a CPU fan, I use an external
adapter for that.
*******
There is one other explanation for this problem. It
has nothing to do with the fan, and has to do with
the motherboard design itself. You need to find some
reviews for the motherboard, to see if other users
have experienced this problem as a motherboard failure.
If the motherboard is bad, then bypassing the CPU fan header
won't be enough to fix it.
One of the reviews on the Newegg site, notes that
the CPU fan doesn't start to spin right away. That
is probably the PWM signal, not being set to a
high enough duty cycle by the BIOS. Apparently,
when the CPU gets warm enough, the fan then starts
to spin. But that's not the other kind of problem,
which appeared to be a design problem with the
motherboard. And in the Newegg reviews, I can't find
any examples of that kind of failure. So your
symptoms probably aren't a motherboard design issue,
as I don't see any examples on Newegg like that.
http://www.newegg.com/Product/Product.aspx?Item=13-128-059
HTH,
Paul