Bob said:
Computer is hp pavillion 700, pentium 4, 2.53ghz. I have replaced the
heatsink retaining bracket and when booting get the message that cpu fan has
failed, press F2 to continue. Thing is, the fan is running, working fine. I
replaced it with fan directly connected to heatsink, still get same message
even though its working. If I use the F2 option, computer boots into XP and
works fine. Right now, using everest program, shows 50 degrees celtsius, 122
degrees fahrenheit. Processor is pentium
478. Question is, why am I getting the fan message, what can I do to fix.
Also, what temperature is to high to run this. Thanks for any help.
Fans come in two wire (case cooling fans) and three wire (typically CPU)
cooling versions. The third wire is an RPM signal. The RPM signal pulses
twice per revolution of the fan blade. The motherboard hardware monitor
interface, measures the time between pulses. Software takes the inverse
of that number, to come up with an RPM number. The hardware monitor has
a "lower limit", as to how slow a fan it can measure (since it measures
the time between pulses). If a fan runs slower than the lower limit,
the monitor reads that fan speed as zero. The monitor cannot tell the
difference between a fan running slightly below the lower limit, and a
fan running at zero RPMs (stalled).
So there are a few explanations. The third wire might not be connected.
The fan might be running slower than 1800 RPM (the threshold on my
machine - more modern machines are adjusted to read as low as 500
RPM or so). The RPM signal could have died (but that is not a common
failure mode). The RPM signal is driven by a simple open collector
transistor, which is very reliable. Shorting the fan signal to
ground or +12V would also affect the ability to get a pulse from
the fan.
50C is not too high. If the computer has been adjusting the fan speed
according to CPU temperature, it could be running at 50C on purpose.
(Qfan from Asus does that, and in order to keep the fan quiet,
attempts to maintain a 50C CPU temperature.) If the processor was getting
up to 65-70C, then I'd become more curious as to what was wrong with
the cooling. For processors of that era, THERMTRIP is probably around
135C, and the throttling point (CPU slows down) is a lot lower than
that. More modern processors have a THERMTRIP of maybe 90C or so
and a throttle temperature of 70C.
CPU reliability should also vary with temperature (Arrhenius reaction
rate equation applies to RAM-like devices), which means running
a CPU at a high temperature, affects its theoretical life. But there
is no empirical evidence that temperature is an issue for the average
user. The news groups are not rife with stories of dead processors,
so whatever the field reliability numbers are for modern processors,
they seem to be pretty good.
For the sake of completeness, there are cooling fans now, that ship
with processors, that have four wires on them. The fourth wire is a
PWM signal at 25KHz. It can be used to control the fan speed under
software control. The pulse width of the signal is varied between
0 percent and 100 percent, to create an adjustable voltage for the
cooling fan. Such a scheme moves the control transistor for the
fan, from the surface of the motherboard, to inside the fan hub.
But your older computer will not have one of those, and will have
the 3 wire variety with RPM instead (+12V, GND, RPM).
As for repairing it, you have to figure out what you've changed
during your repair. Is the fan blade moving freely ? (Check with
power off.) The connector must be connected properly, otherwise
it would not be spinning right now. So there is no reason to
suspect the RPM signal is disconnected. That leaves the possibility
of the fan running too slow, as the only other practical explanation.
Paul
