resitor on cpu fan ?

[Skeleton Man]

Hi,

Just wondering is it appropriate to put a resistor on the +12v for the cpu fan
to slow it down (I'm thinking drop to 5v or less) ? Would this interefere with
the fan speed monitor on the board ? ( I would like to do it to a s370 P3 and a
s478 P4)

I don't want to go out and buy a fancy hs/f combo when all I need is a 20c
resistor..

Regards,
Chris

 

[Paul Murphy]

Hi,

Just wondering is it appropriate to put a resistor on the +12v for the cpu
fan
to slow it down (I'm thinking drop to 5v or less) ? Would this interefere
with
the fan speed monitor on the board ? ( I would like to do it to a s370 P3
and a
s478 P4)

I don't want to go out and buy a fancy hs/f combo when all I need is a 20c
resistor..

Regards,
Chris

It usually not a good idea because the heatsink will be designed to
dissipate a set amount of heat for a given airflow across the heatsink - if
you lower the flow rate, you lower the heat dissipated and raise the CPU
temperature and consequently reduce its life. The thing about the "fancy
hs/f combo" is that if its any good it'll have a very high surface area for
heat exchange to occur (lots of big fins) and they're often made out of more
thermally conductive metal than the cheap ones. Many 12 volt fans won't even
start at less than 5 volts and have a stall voltage of around 5 volts.

The only situation when I think using a resistor is safe is when you're able
to carefully monitor the effects on CPU temperature and a larger slowing
turning fan is available or fitted. Often the noisiest part of a PC is the
PSU fans(s) and doing a resistor mod there is something to be extremely
careful about for many reasons and is best avoided.

Paul

 

[kony]

Hi,

Just wondering is it appropriate to put a resistor on the +12v for the cpu fan
to slow it down

yes, so long as you're sure it'll remain fast enough for the
worse thermal condition possible.

(I'm thinking drop to 5v or less) ?

No, bad idea. Most fans will not run reliably below 5V,
many won't under 6V, and some even make more noise (a
pulsing sort of sound) at around 6.5V and under. The best
strategy is to start out with lowest RPM fan possible.
Another alternative is to use a series of diodes instead of
a resistor, but you might still have issues with new noise
or unreliable operation below the voltage threshold.

Would this interefere with
the fan speed monitor on the board ? ( I would like to do it to a s370 P3 and a
s478 P4)

IIRC, you should put the resistor on the power lead. Some
motherboards have trouble sensing fan RPM once it falls
below a certain level. For example a board might detect a
fan @ 3000 RPM ok, but not at 800RPM. YMMV, you'll have to
try it on your parts.

I don't want to go out and buy a fancy hs/f combo when all I need is a 20c
resistor..

I've done it several times, it is a fine solution providing
you don't go overboard. If adjusting fan speed to cool a
CPU based on it's idle temp, then that CPU gets stuck in a
busy loop somehow, there'll be a lot more heat to account
for.

Also consider that some motherboards have their own
fan-speed reduction feature., but that the feature doesn't
kick-in immediately upon power on. A later
motherboard-controlled reduction or a spin-up after a power
management state are two areas to keep an eye on to be sure
the fan RPM is at least high enough to allow reliable
operation. In general it's good for a cheap alternative,
the resistor values I've most often used are in the 22-110
Ohm range, though these were usually with fans having
moderately low RPM already.

 

[ElJerid]

Hi,

Just wondering is it appropriate to put a resistor on the +12v for the cpu fan
to slow it down (I'm thinking drop to 5v or less) ? Would this interefere with
the fan speed monitor on the board ? ( I would like to do it to a s370 P3 and a
s478 P4)

I don't want to go out and buy a fancy hs/f combo when all I need is a 20c
resistor..

Regards,
Chris

You'd better put a wired 10 K potentiometer on the power line of the fan.
Costs a little more but allows you to set the speed to your needs. And yes,
it will interfere with the on-board monitor. If this one asks for full speed
and your potmeter is at 50 %, you will never go above 50 %. So, watch your
temps.

 

[Skeleton Man]

You'd better put a wired 10 K potentiometer on the power line of the fan.

Costs a little more but allows you to set the speed to your needs. And yes,
it will interfere with the on-board monitor. If this one asks for full speed
and your potmeter is at 50 %, you will never go above 50 %. So, watch your
temps.

Might do that to the old P3 box.. I have it right next to my bed as a jukebox
(plugged into my stero).. drill a small hold in a 5.25" blanking plate to put
the pot.. with the fan disconnected it's 10x quiter.. but I don't wanna rely on
thermal throttling..

I didn't wanna have to drill holes and have something sticking out, hence a
fixed value.. I may yet invest in a new cooler for the P4.. that way fan
control is at the rear..

There's no thermal fan control on either board, and neither support sofware fan
control.. (hence a hardware hack to lower rpm)

Regards,
Chris

 

[kony]

You'd better put a wired 10 K potentiometer on the power line of the fan.
Costs a little more but allows you to set the speed to your needs. And yes,
it will interfere with the on-board monitor. If this one asks for full speed
and your potmeter is at 50 %, you will never go above 50 %. So, watch your
temps.

A "pot" of sufficient power capability to handle a fan would
typically be called a Rheostat.

10K is WAY off, not a value one could use reasonably.
Anything higher than 330 Ohm would have far too little
travel over the effective range. Most fans don't even need
110 Ohm. A ~ 100 Ohm Rheostat might be ideal then.

 

[Roby]

A "pot" of sufficient power capability to handle a fan would
typically be called a Rheostat.

10K is WAY off, not a value one could use reasonably.
Anything higher than 330 Ohm would have far too little
travel over the effective range. Most fans don't even need
110 Ohm. A ~ 100 Ohm Rheostat might be ideal then.

Here are the results of testing my Intel P4 heatsink fan:

V RPM ma ohms watts
12 2670 145 0 Loud
11 2420 125 8 0.13
10 2230 115 17 0.22
9 1985 105 29 0.32
8 1730 95 42 0.38 Noise greatly diminished
7 1460 80 63 0.40 Barely audible
6 1155 75 80 0.45 CPU sweats!
5 890 65 108 0.46
4 680 55 145 0.44
3 (fan stalls)

V = Set using external variable-voltage p/s
ma = Measured current drawn by fan
ohms = calculated R = ( 12 - V ) / ( ma / 1000 )
watts = power dissipated in that resistor

Roby

 

[Noozer]

Here are the results of testing my Intel P4 heatsink fan:

V RPM ma ohms watts
12 2670 145 0 Loud
11 2420 125 8 0.13
10 2230 115 17 0.22
9 1985 105 29 0.32
8 1730 95 42 0.38 Noise greatly diminished
7 1460 80 63 0.40 Barely audible
6 1155 75 80 0.45 CPU sweats!
5 890 65 108 0.46
4 680 55 145 0.44
3 (fan stalls)

Now shut down the PC and then restart with fan at 4 volts. Does it even spin
at all?

....and was this the stock Intel fan?

....and, finally, what were your Prime95 temps after leaving the fan at each
of these set voltages for 15 minutes?

 

[Roby]

Now shut down the PC and then restart with fan at 4 volts. Does it even
spin at all?

Unlikely: depends on static friction (stiction?) of the fan bearings.
Assuming it breaks loose, it ought to accelerate since a fan load torque is
proportional to RPM squared. I didn't test it: it's kinda academic, since
the silicon is already toasty at 6 volts.

...and was this the stock Intel fan?

Yup. Intel A80856-003 DC12V 0.28A sitting on a P4 at 2.8gHz

...and, finally, what were your Prime95 temps after leaving the fan at
each of these set voltages for 15 minutes?

I don't know. Results would depend on CPU load and ambient temperature,
case and case fans, etc.

I ended up running the fan at about 9 volts. For *my* hardware, that
produces sufficient cooling (no overtemp alarm, no thermal throttling) for
*my* worst-case load (compiling a new kernel). YMMV

Roby

 

[kony]

I'm a bit too lazy to do the math at the moment but don't
believe one can come to resistor values in this fashion.
Fans aren't constant-current devices. The resistor values
might be roughly corresponding to real-world results due to
the linearity of the RPM vs voltage response but to get a
better idea of the actual resistor value the resistance of
the fan itself must also be factored in.

Unlikely: depends on static friction (stiction?) of the fan bearings.
Assuming it breaks loose, it ought to accelerate since a fan load torque is
proportional to RPM squared. I didn't test it: it's kinda academic, since
the silicon is already toasty at 6 volts.

It's not usually a matter of bearing friction, rather the
controller circuit on the fan itself, and/or the magnetic
field needed to overcome the force of the fixed magnetic hub
(when fan moves but won't spin). Most decent/modern fans
will "attempt" to spin, moving a bit, then fail near the
threshold voltage.

 

[William W. Plummer]

I'm a bit too lazy to do the math at the moment but don't
believe one can come to resistor values in this fashion.
Fans aren't constant-current devices. The resistor values
might be roughly corresponding to real-world results due to
the linearity of the RPM vs voltage response but to get a
better idea of the actual resistor value the resistance of
the fan itself must also be factored in.

The poster's method is fine, assuming a steady-state DC circuit. You
are probably thinking of the "back EMF" generated by the motor when it
is spinning that has the effect of decreasing the current into the motor
and is a function of RPM. No matter. The poster _measured_ the current
with various actual resistors. There is no arguing with that.

 

[kony]

The poster's method is fine, assuming a steady-state DC circuit. You
are probably thinking of the "back EMF" generated by the motor when it
is spinning that has the effect of decreasing the current into the motor
and is a function of RPM. No matter. The poster _measured_ the current
with various actual resistors. There is no arguing with that.

I read it differently, that current was NOT measured from
use of resistors but rather the resistor value was
calculated based on a test with a variable voltage power
supply.

 

[Zotin Khuma]

I'm a bit too lazy to do the math at the moment but don't
believe one can come to resistor values in this fashion.
Fans aren't constant-current devices. The resistor values
might be roughly corresponding to real-world results due to
the linearity of the RPM vs voltage response but to get a
better idea of the actual resistor value the resistance of
the fan itself must also be factored in.

This would make a nice little project for an electronics
DIY-er. Instead of arriving at the resistor value by
guesswork or trial and error, one could dispense with
a series resistor and use a standard 3-terminal voltage
regulator IC. Say you want to run the fan at 9V, a $0.25
7809 IC plus a couple of $0.05 capacitors will be all
that's needed. The IC will supply a fixed 9V output
whether the fan is drawing 50mA or 200mA.

And an LM317 plus a few more parts would be ideal
for making a variable but accurately controlled output.

 

[Guest]

Just wondering is it appropriate to put a resistor on the
+12v for the cpu fan to slow it down (I'm thinking drop to
5v or less) ? Would this interefere with the fan speed
monitor on the board ?

I don't want to go out and buy a fancy hs/f combo when
all I need is a 20c resistor..

5V is too slow for many fans; they may run at that voltage but often
need at least 6V to start spinning. 10K ohms is also far too high a
resistance because its equivalent resistance is 12V/current rating,
typically in the range of 100 ohms, and fans typically draw only half
their rated current. The resistor must be rated for at least twice the
power flowing through it or it'll overheat (watts = current^2 x
resistance). A resistor wired in series with the fan's positive lead
won't affect any tachometer because 5-6V is enough voltage for the
circuitry, and this modification doesn't raise the ground as the
typical 7v modification does (positive to +12v, negative to +5V).

Instead of a fixed resistor, consider using a 50-100 ohm thermistor
(room temperature rating) and attaching it to the heatsink closest to
all the cable leads to automatically increase the fan speed with
temeprature. It's important to attach it to that heatsink, which
contains only low voltage components, and not the one next to the large
200-400V cylindrical capacitors, which not only has high voltage
transistors attached to it but is often itself connected to high
voltage. Glue the thermistor securely with RTV silicone rubber.

 

[kony]

This would make a nice little project for an electronics
DIY-er. Instead of arriving at the resistor value by
guesswork or trial and error, one could dispense with
a series resistor and use a standard 3-terminal voltage
regulator IC. Say you want to run the fan at 9V, a $0.25
7809 IC plus a couple of $0.05 capacitors will be all
that's needed. The IC will supply a fixed 9V output
whether the fan is drawing 50mA or 200mA.

And an LM317 plus a few more parts would be ideal
for making a variable but accurately controlled output.

Well it would be a good beginner's project and more in line
with what a control freak (or someone wanting
thermal-control) would want, but having that much tethered
bulk is a turn-off to some. If one wants to use resistors
instead of some other method of control, the easier way is
to buy a rheostat to dial in the speed desired then just
take a multimeter reading and you have the resistance value.
I've likewise done the same with a 12 position switch and
4001 diodes in series. One can build a single controller
with this purpose in mind and reuse to to size resistors or
diodes per each fan.

The diodes work better for lowest fan RPM (spinup) but
either, particularly the resistor, allows a streamline
unobtrusive package that can be heat-shrink tubed.

 

[ElJerid]

Here are the results of testing my Intel P4 heatsink fan:

V RPM ma ohms watts
12 2670 145 0 Loud
11 2420 125 8 0.13
10 2230 115 17 0.22
9 1985 105 29 0.32
8 1730 95 42 0.38 Noise greatly diminished
7 1460 80 63 0.40 Barely audible
6 1155 75 80 0.45 CPU sweats!
5 890 65 108 0.46
4 680 55 145 0.44
3 (fan stalls)

V = Set using external variable-voltage p/s
ma = Measured current drawn by fan
ohms = calculated R = ( 12 - V ) / ( ma / 1000 )
watts = power dissipated in that resistor

Roby

Just checked the value of the rheostat as delivered by ThermalTake with
their P4 CPU cooler. It' s 10 K and wired serial on the + lead !

 

[Spajky]

5V is too slow for many fans; they may run at that voltage but often
need at least 6V to start spinning. ...

A resistor wired in series with the fan's positive lead
won't affect any tachometer because 5-6V is enough voltage for the
circuitry, and this modification doesn't raise the ground as the
typical 7v modification does (positive to +12v, negative to +5V).

Resolutions for such problems can be found on my site under
electronics ...

 

[kony]

Just checked the value of the rheostat as delivered by ThermalTake with
their P4 CPU cooler. It' s 10 K and wired serial on the + lead !

Yes but that's not the same thing.
They are using a different control method then, a fan
typically has two internal resistors that determine it's
RPM... for example when a manufacturer offers a low, medium,
or high speed version of "same" fan, the difference is
typically the internal resistor values.

A variable resistance can be used in place of (or wired in
parallel or series with) these to change speed, but only
when wired into the circuit, not hooked only in series on a
power lead. If you take an old fan and tear it apart you'll
see what I mean, or many of the circuits on the following
page are examples:
http://www.pavouk.comp.cz/hw/fan/fan2.html

If the fan in question has no access to the internal circuit
via a control lead (or you take apart the fan and hack it
out yourself) then the control via series resistance must
use the lower ~ 68 Ohm +- 47 Ohm (roughly) method of current
reduction on the power lead, which of course means one can't
use a tiny POT due to heat dissipation issues.