Do you have a question? Post it now! No Registration Necessary. Now with pictures!
- Posted on
- New CPU Fan Prevents PC From Running
- Gary Brown
May 10, 2008, 8:37 pm
rate this thread
A newly installed CPU heat sink and fan prevented my PC
from running. The PC started up then turned itself off after
a few seconds. The HSF appeared to be seated properly
with more than adequate heat sink compound. The heat
sink was considerably larger than its predecessor but no
shorts to adjacent components were apparent. I put the
old heat sink back on and the PC runs again. Any clues
why this could happen?
The new HSF is a MassCool 5F263B1M3 and replaced
the stock AMD unit. The CPU is an AMD Athlon 2800+.
Re: New CPU Fan Prevents PC From Running
How many wires does the new fan use ?
When a fan has two wires, they carry +12V and GND. That
basically powers the fan. That is the minimum a fan can have
and do something for you.
When a fan has three wires, the third wire carries the RPM
signal from the fan to the motherboard. Many motherboards
sense the RPM signal as delivered to the CPU fan header
plug. If the RPM signal is missing or the fan is at
zero RPMs (stalled), the motherboard may shut down. The
shut down would be based on the execution of BIOS code.
If, for any reason, the BIOS cannot run (execution error),
then the missing RPM signal will be ignored. The next
safety feature in that case, is THERMTRIP, an overheat
detection feature present in both Athlon64+ and Intel
processors. If the processor overheats because of fan
failure, in fact the processor can shut down the motherboard,
without any BIOS code needed. If a processor overheats
because you forgot the heatsink, then shutdown can be
in a couple seconds. With heatsink still in place, it
takes a lot longer to hit THERMTRIP.
When a fan has four wires, the fourth wire is called PWM
(pulse width modulation) and travels from the motherboard
to the fan. The motherboard uses that to control the fan
speed. Either the SuperIO hardware monitor, or some other
piece of hardware, sends a control signal. PWM is optional -
if it is cut, or not present on the interface for some reason,
the fan runs at 100%. If the PWM signal is grounded at the fan,
the fan should run at its minimum speed, but perhaps not
exactly zero RPM (the spec actually defines different fan
classifications, so the behavior can actually differ between
products). You can plug a four wire fan into a three pin fan
header, without a problem. And in that case, the RPM signal
is still present, so the BIOS is happy.
Based on your description, it sounds like there is no
RPM signal getting to the CPU fan header. Make sure the
CPU fan header, has some fan plugged into it, with an
RPM signal present. (If the CPU fan only had two wires,
there is no need to despair. You can connect some other
fan that does have an RPM signal, to the CPU fan header,
to keep the BIOS happy.)
Many case fans lack the RPM signal. The fan that comes
with a processor, generally always has the RPM signal.
Aftermarket coolers should have the RPM signal (if the
company has half a brain), but of course anything is
possible when you're saving a nickel here and there.
Case fans lack the RPM signal, because of the money
saved by the manufacturer. I think the extra component
is a tiny transistor inside the fan hub, plus that
third wire in the cable.
Re: New CPU Fan Prevents PC From Running
For the new fan and heatsink, is the new fan a 3-wire or 4-wire fan?
Are the 3 or 4 pins on the motherboard header marked "CPU" to which you
are connecting that fan?
Intel decided to come out with a newer scheme to control fan speed. The
old scheme was use change the duty cycle for the fan controller in the
super I/O chip. That's how, for example, Speedfan and Motherboard
Monitor worked to control speeds for 3-wire fans. The new scheme uses a
separate control line to monitor the temperature and decide the speed of
the fan rather than old duty cycle (pulse) scheme the power on and off
I believe a 3-pin fan will work with a 4-pin header but you need to be
sure to align the pins correctly. The result is that the sense wire
from the fan will be seen by the BIOS to know the fan is spinning but
you will not have any control over the fan's speed (i.e., it will always
spin a max RPM and always be noisy even when a slower RPM could be used
to sufficiently cool the CPU under less than max load).
http://tinyurl.com/6zznx7 (Google Groups copy of my prior post)
The problem with the old 3-wire, super I/O, power duty-cycling scheme is
that it could faster burn out the windings in a fan because of a higher
inrush current when the fan got restarted (when the power pulsed on).
Some fans aren't designed for that mode of operation. It was also not
accurate for controlling speed, especially when you got down to 5%, or
less, for duty cycle; for example, I can go down to 3% on my CPU's fan
so that it is sure to start spinning (has enough current) and to keep
spinning but below that and the fan might not start or it will stop. So
I run at 5% duty cycle to make sure it keeps running okay at a minimum
You might have a new 4-wire fan that won't work on a 3-pin header. You
might have a new 3-wire fan but mispositioned it on the 4-pin header.
- » Newest MemTest86, MemTest86+ incompatible with Sandy, Ivy chipsets?
- — Next thread in » Computer Hardware