Faking a PWM fan

  • Thread starter Thread starter Keve Nagy
  • Start date Start date
K

Keve Nagy

Hello Everyone,

I need your help with a hardware electronics question.
There is a standard PWM 4-pin male connector socket on my motherboard
for case fan. I replaced the factory case fan and CPU fan with a far
more efficient and silent CPU cooler, so now there is nothing connected
to the case fan connector as the case fan is no longer needed.
Because of this, every time the computer boots the POST stops with a
warning "No case fan detected" and I need to press F1 to continue.
And this is bad if I need to reboot the machine remotely, or if it
restarts automatically after a power outage, etc.

There is no BIOS option to ignore this or any other warning.

I need to create an electronic dummy fan, a circuitry that the POST will
accept as a PWM case fan.

PWM pinout specs says:
1 - black wire - GND
2 - yellow wire - 12V
3 - green wire - sense
4 - blue wire - control

Question:
Can I just short blue and green to imitate the presence of a PWM fan?
Is that safe to do?
Or do I need to build a circuit that feeds back pulses of signals?

Regards,
 
Keve said:
I need to create an electronic dummy fan, a circuitry that the POST
will accept as a PWM case fan.

PWM pinout specs says:
1 - black wire - GND
2 - yellow wire - 12V
3 - green wire - sense
4 - blue wire - control

Question:
Can I just short blue and green to imitate the presence of a PWM fan?
Is that safe to do?
Click to expand...

Most likely not.
Or do I need to build a circuit that feeds back pulses of signals?
Click to expand...

That is what I would do. A 555 circuit is very simple to build, needing
only a 555 chip and a couple of other parts (capacitor and a couple of
resistors). Ideally you want a 5V output for the squarewave. Here is one
page with some information on the tachometer output:
http://petervis.co.cc/cpu fan tacho/cpu fan tacho.html

Alternately, you could just connect a small (AKA quiet) fan and hide it
somewhere in the case, or run a case fan at reduced voltage.

Jon
 
Keve said:
Hello Everyone,

I need your help with a hardware electronics question.
There is a standard PWM 4-pin male connector socket on my motherboard
for case fan. I replaced the factory case fan and CPU fan with a far
more efficient and silent CPU cooler, so now there is nothing connected
to the case fan connector as the case fan is no longer needed.
Because of this, every time the computer boots the POST stops with a
warning "No case fan detected" and I need to press F1 to continue.
And this is bad if I need to reboot the machine remotely, or if it
restarts automatically after a power outage, etc.

There is no BIOS option to ignore this or any other warning.

I need to create an electronic dummy fan, a circuitry that the POST will
accept as a PWM case fan.

PWM pinout specs says:
1 - black wire - GND
2 - yellow wire - 12V
3 - green wire - sense
4 - blue wire - control

Question:
Can I just short blue and green to imitate the presence of a PWM fan?
Is that safe to do?
Or do I need to build a circuit that feeds back pulses of signals?

Regards,
Click to expand...
Not clear exactly what you're doing. If I interpret it as, adding a fan
to the cpu, all that does is take the heat out of the CPU and put it into
the box.
You NEED a fan to exhaust the heat from the box.
A hard drive meltdown is not a cheap thing.
 
Keve said:
Hello Everyone,

I need your help with a hardware electronics question.
There is a standard PWM 4-pin male connector socket on my motherboard
for case fan. I replaced the factory case fan and CPU fan with a far
more efficient and silent CPU cooler, so now there is nothing connected
to the case fan connector as the case fan is no longer needed.
Because of this, every time the computer boots the POST stops with a
warning "No case fan detected" and I need to press F1 to continue.
And this is bad if I need to reboot the machine remotely, or if it
restarts automatically after a power outage, etc.

There is no BIOS option to ignore this or any other warning.

I need to create an electronic dummy fan, a circuitry that the POST will
accept as a PWM case fan.

PWM pinout specs says:
1 - black wire - GND
2 - yellow wire - 12V
3 - green wire - sense
4 - blue wire - control

Question:
Can I just short blue and green to imitate the presence of a PWM fan?
Is that safe to do?
Or do I need to build a circuit that feeds back pulses of signals?

Regards,
Click to expand...

So you are relying solely on the presence and working of the PSU fan?
Not smart, especially considering you are not always present on this
host as you mention accessing it remotely. The case fan should be
considered the backup to the PSU fan (if it isn't also considered needed
to expel the hot air load that the PSU fan cannot handle). Since it is
a PWM fan for the case fan, the case fan doesn't need to spin unless the
temperature goes up (based either on the CPU or case temperatures). So
the case fan won't be adding noise to the host while you are at the host
unless you make the host so busy that its heat exceeds a threshold not
manageable by the PSU fan alone.

Yeah, you could maybe figure out a PCB on which you add components to
emulate a spinning fan (assuming there is no BIOS option to disable that
motherboard fan connector) but then why would you go through all that
work and expense. It'll cost you more in time and money to build a
custom PCB controller than just to add a fan. Add a case fan is the
cheap solution and the better solution.
 
You NEED a fan to exhaust the heat from the box.

OK, I didn't make this clear enough. My mistake.
There IS a fan to exhaust the heat from the box. It is just the CPU fan
and the case fan are now the same.
The new huge heatsink I now use has a silent 120mm fan attached (CPU
fan). The other side of the fan housing is touching the back of the
case, so the fan is essentially between the case and the heatsink.
Hence no more need to a case fan.

Regards,
 
So you are relying solely on the presence and working of the PSU fan?

No. I didn't explain this well enough. My mistake.
There is a fan inside the PSU.
And there is another huge 120mm PWM quiet fan "jammed" between the new
large heatsink and the back of the case (where the exhaust holes are).
Yeah, you could maybe figure out a PCB on which you add components to
emulate a spinning fan (assuming there is no BIOS option to disable that
motherboard fan connector) but then why would you go through all that
work and expense. It'll cost you more in time and money to build a
custom PCB controller than just to add a fan. Add a case fan is the
cheap solution and the better solution.
Click to expand...

I agree with all your points.
However, due to the large size of the new heatsink and the 120mm PWM fan
that replaced the old CPU fan and case fan, there is no room left to
place another separate case fan there. I could add one and position it
somewhere else, but that ruins the point, it would only make additional
noise with no cooling purpose.

Besides, adding a cheap fan is always NOISY. A silent PWM fan is quite
expensive. Costs more than the components to build the fake circuitry.
Regarding the time and effort invested in the circuitry however, I
cannot and will not argue with you. This is a price I need to pay during
my battle for silence.

Regards,
 
Keve said:
Hello Everyone,

I need your help with a hardware electronics question.
There is a standard PWM 4-pin male connector socket on my motherboard
for case fan. I replaced the factory case fan and CPU fan with a far
more efficient and silent CPU cooler, so now there is nothing connected
to the case fan connector as the case fan is no longer needed.
Because of this, every time the computer boots the POST stops with a
warning "No case fan detected" and I need to press F1 to continue.
And this is bad if I need to reboot the machine remotely, or if it
restarts automatically after a power outage, etc.

There is no BIOS option to ignore this or any other warning.

I need to create an electronic dummy fan, a circuitry that the POST will
accept as a PWM case fan.

PWM pinout specs says:
1 - black wire - GND
2 - yellow wire - 12V
3 - green wire - sense
4 - blue wire - control

Question:
Can I just short blue and green to imitate the presence of a PWM fan?
Is that safe to do?
Or do I need to build a circuit that feeds back pulses of signals?

Regards,
Click to expand...

The control signal (blue wire) runs at 25KHz. It is variable pulse width, with
the pulse width in proportion to what speed the computer wants the fan to work at.
If you loop that back to the tacho input, that's a pretty short duration
of pulse. And the thing is, what happens if the PWM duty cycle goes to 100% and
the signal sits at 5V for 100% of the time ? Then, you get no usable pulses
for your RPM signal. (I don't know what all the legal states for that
PWM signal are.)

Intel fan spec for four wire fans (i.e. not a Molex 1x4, the smaller LP4 style fan connector)

http://www.formfactors.org/developer\specs\REV1_2_Public.pdf (Page 9)

Example of the fan connector from that spec, the small kind.

( http://ca.mouser.com/search/ProductDetail.aspx?qs=RQ9zk%2b50r5G5REkKmXuucQ== )

The tachometer wire, pulses twice per revolution. I'm not sure whether
there is an authoritative standard on what the interface would consist of.
It is likely open collector (as stated in the formfactor doc). The
pullup on the motherboard could be to 12V, to 5V, or some other circuit
along those lines. The 555 may be able to drive something like that,
as long as the output can stand a 12V level on it. The 555 wouldn't have to
sink too much current.

If I was going to connect the PWM output to the RPM input for a test,
I'd probably put a zener across it, along the lines in the following
circuit. I might select a 4.7V zener, to ensure the clipping is
pretty close to 5V, so if there is a 12V pullup resistor on RPM on
the motherboard side, the PWM output is protected from it.

(A loopback circuit, to be used for testing this theory...)

PWM --- 5V, 25KHz, variable width ------+-------- RPM input
|
===
/ \ 4.7V zener,
---
|
GND

( Zener, white band marking cathode, upwards in diagram )
http://ca.mouser.com/ProductDetail/...X4V7C133/?qs=sGAEpiMZZMuKA4j4J4t1THuJeeKQ0Jdc

*******

This web page, shows how to retrofit an RPM signal, to a two wire fan.
So this allows you to add a third wire to a two wire fan, using a
transistor to pick off a signal and convert it to open collector.

http://tipperlinne.com/fan-tach.htm

On the motherboard, they sometimes have a clipping circuit, to reduce the
potential 12V amplitude of the RPM pulses, to 5V or less for the logic IC
that checks them. If the 555 has an open collector output, that may be
sufficient to do the job. (Or, you can drive a 2N2222 from the 555,
and use the 2N2222 to make the open collector driver. The 2N2222 would
isolate the characteristics of the 555, whatever they are, from the
outside world.)

The motherboard circuit, is sometimes intended to allow the fan tacho
output to swing the full 12V, and then not interfere with it. Something
along these lines. The reason for doing it this way, is some fans may want
a full 12V across their RPM signal output stage, so the transistor
switches properly. Otherwise, I don't see a strong reason for doing it
this way. The pulsed waveform on the left of the series resistor would have
12V amplitude, while the signal on the zener diode side is clipped to 5V.
The series resistor, prevents the clipping of the zener, from pulling
excessively on the 12V pullup resistor. I've seen a circuit like this,
in some Super I/O datasheet (and this drawing is just from memory).
You'd want to find a SuperI/O datasheet, and double check this. So
this is one of the variations, on how the motherboard fan speed input
is designed. The only reason for showing this diagram, is to show
how a motherboard may present 12V to the outside world, but retain
5V compatibility so the SuperI/O isn't damaged.

+12V
|
resistor
|
RPM >-------+--------- resistor -----+------ SuperI/O fan speed input
Open | 5V max input
Collector ===
/ \ approx 5V
--- zener
|
GND

A motherboard may choose to do it this way, which would be cheaper, but
here, the fan open collector transistor only sees 5V applied to it. This
might not be the best way for the motherboard designer to do it, although
is keeps the SuperI/O chip safe. The pulse amplitude is 5V (at least,
as long as the RPM transistor in the fan can switch properly).

+5V
|
resistor
|
RPM >-------+------------------------------- SuperI/O fan speed input
Open 5V max input
Collector

Just make sure you don't apply 12V to the wrong thing :-)

HTH,
Paul
 
Keve said:
No. I didn't explain this well enough. My mistake.
There is a fan inside the PSU.
And there is another huge 120mm PWM quiet fan "jammed" between the new
large heatsink and the back of the case (where the exhaust holes are).


I agree with all your points.
However, due to the large size of the new heatsink and the 120mm PWM fan
that replaced the old CPU fan and case fan, there is no room left to
place another separate case fan there. I could add one and position it
somewhere else, but that ruins the point, it would only make additional
noise with no cooling purpose.

Besides, adding a cheap fan is always NOISY. A silent PWM fan is quite
expensive. Costs more than the components to build the fake circuitry.
Regarding the time and effort invested in the circuitry however, I
cannot and will not argue with you. This is a price I need to pay during
my battle for silence.

Regards,
Click to expand...
Measure some temperatures.
I have a Dell 4600.
Got a Dell 4700.
Looks identical.
What I didn't realize was that they reversed the flow
of the CPU fan.
The net result was NO air flow to the upper drive bays and a DEAD
hard drive.
There are exceptions, but in general, lower noise is accompanied
by lower air flow. If your CPU fan measures CPU temp to determine
the speed, you may be in for a surprise.
Don't expect a PSU fan to be able to exhaust the heat.
Measure the temps around the box.

If you're dead-set on doing this,
First check bios settings.
Then, I'd take a scope and look
at the signal on the sense terminal of the fan. It's likely
some repetitive waveform who's frequency is determined by the
fan speed. Emulate that with a 555 timer stuck on a plug.
Watch the signal levels so you don't smoke the sense chip.

It's possibly risky, cause you don't know how either circuit works,
but you may be able to jumper, with some resistance, an existing
fan sense to the one you removed. Check levels with a scope first.
Don't blame me if the motherboard smokes.


Are we having fun yet?
 
Keve said:
No. I didn't explain this well enough. My mistake.
There is a fan inside the PSU.
And there is another huge 120mm PWM quiet fan "jammed" between the new
large heatsink and the back of the case (where the exhaust holes are).


I agree with all your points.
However, due to the large size of the new heatsink and the 120mm PWM fan
that replaced the old CPU fan and case fan, there is no room left to
place another separate case fan there. I could add one and position it
somewhere else, but that ruins the point, it would only make additional
noise with no cooling purpose.
Click to expand...

So mount the case fan on the OUTSIDE of the case. The wiring probably
runs along one of the "legs" or supports which you may have to peel out
so you can keep the wires inside the case. Else, use a hacksaw to notch
out the fan so the wiring can go back into the case (and without the
wires getting pinched when you screw the fan down tightly to the case),
or use a punchout or remove a slot cover if available and you don't care
about sloppy wiring behind the case. Just be sure to orient the air
flow correctly for the exterior mounted case fan.
Besides, adding a cheap fan is always NOISY. A silent PWM fan is quite
expensive. Costs more than the components to build the fake circuitry.
Regarding the time and effort invested in the circuitry however, I
cannot and will not argue with you. This is a price I need to pay during
my battle for silence.
Click to expand...

Since it is a PWM case fan, it won't be making ANY noise when it isn't
needed for cooling and is NOT spinning. The point of having regulation
for the fan is that it does NOT need to be spining all the time. When
it does spin means more cooling is needed which would be absent if you
omitted the fan. As the when the case fan starts to spin to add the
cooling needed under a high load would depend on what threshold you
configured in whatever temperature monitoring software you installed.
 
So mount the case fan on the OUTSIDE of the case.

How often one hears the saying "Think outside the box"!
And when I literally work on a box, do not even consider the outside.

Great pointer, Mr. Vanguard.
Thank you!
 
HTH,

A wealth of useful information.
Thank you, Paul!

Ultimately, it seems I do need to borrow a 'scope and analyse those
signals coming back from a PWM fan. I was trying to avoid this, even
though it was the most logical step to do all along.

Regards,
 
Keve said:
How often one hears the saying "Think outside the box"!
And when I literally work on a box, do not even consider the outside.
Click to expand...

With the fan on the outside of the case, make sure you add a fan guard
on the now exposed fan blades.

To maximize air flow (or minimize restriction to air flow), you might
consider if the grill opening in the case for the fan should remain.
You might want to nibble it out so the fan can exhaust without an
immediate restriction at its intake.
 
Back
Top