LinuxQuestions.org

LinuxQuestions.org (/questions/)
-   Linux - Hardware (https://www.linuxquestions.org/questions/linux-hardware-18/)
-   -   chassis fan speed control (case fan) (https://www.linuxquestions.org/questions/linux-hardware-18/chassis-fan-speed-control-case-fan-602818/)

wastingtime 11-27-2007 03:53 PM
chassis fan speed control (case fan)
 
I have two desktop machines running pclos 2007 (2.6.18.8). Both have a chassis fan and it is running full speed making a lot of noise.
How can I configure the system to control the speed of the fan (i.e. slow it down so it's not noisy)

Machine 1: white box w/ MB Asus P5GC-MX, Celeron D, Intel 945 chipset
Machine 2: emachines T3302, AMD 3300+ , VIA chipset

Both machines are pretty basic, use on-board sound, video, 1GB ram, 1 HD, 1 DVD-CD/RW

Thanks

jiml8 11-27-2007 04:46 PM
easiest way is to replace the fans with fans that have built in controls. Antec makes a series of such fans.

Harder way is to obtain 3rd party fan speed controllers and wire them in.

Hardest way is to make your own fan speed controller. You need a square-wave oscillator with frequency of a couple hundred hertz (best if tunable; you should match it to the fan motor), a mosfet (or power darlington, but mosfet'd be better), and associated components.

Actually, as I looked into this a bit it would appear that most commercial speed controllers for computer fans are not using pulse width modulation controllers. That kind of surprises me because PWM gives full torque while modulating RPM. The commercial devices I see are varying voltage to the motor. The best choice I see is one that uses a variable voltage regulator to control the voltage to the fan. The simplest I see connects the fan between the power supply 12V line and the 5V line (rather than between 12V and ground), giving a net voltage of 7V to the fan. This would work just fine, but isn't adjustable.

Wrong way is to use a potentiometer. At least one controller I saw takes this route.

lazlow 11-27-2007 05:20 PM
A lot of motherboard have places on them to plug in fans. You usually can adjust the fan speed via the bios. A lot of times you can tie it to the temperature of something on the motherboard. Just make sure you are not drawing more power than the motherboard can provide.

Good Luck
Lazlow

onebuck 11-29-2007 07:28 AM
Hi,

Quote:
Originally Posted by jiml8 (Post 2972630)
easiest way is to replace the fans with fans that have built in controls. Antec makes a series of such fans.

Harder way is to obtain 3rd party fan speed controllers and wire them in.

Hardest way is to make your own fan speed controller. You need a square-wave oscillator with frequency of a couple hundred hertz (best if tunable; you should match it to the fan motor), a mosfet (or power darlington, but mosfet'd be better), and associated components.

Actually, as I looked into this a bit it would appear that most commercial speed controllers for computer fans are not using pulse width modulation controllers. That kind of surprises me because PWM gives full torque while modulating RPM. The commercial devices I see are varying voltage to the motor. The best choice I see is one that uses a variable voltage regulator to control the voltage to the fan. The simplest I see connects the fan between the power supply 12V line and the 5V line (rather than between 12V and ground), giving a net voltage of 7V to the fan. This would work just fine, but isn't adjustable.

Wrong way is to use a potentiometer. At least one controller I saw takes this route.
Man of my own heart!

Most of the devices that you will find are using the technique because of cost. And are not engineered properly. Sure PWM would be the way to go but not the simplest for the lay person.

You have to remember that most of these are just 'hacks'. I've gotten several deals from shops that put fans in their junkbox because of the 'pot' technique. Just redesign the interface and use it in one of my many personal systems.

Heck the fan is worth more that the buck/two that is charged. Sure my time is worth more but I don't have to pay the premium price for a well designed system. My charges are mine! Plus all the fun.

onebuck 11-29-2007 07:39 AM
Hi,

Quote:
Originally Posted by wastingtime (Post 2972581)
I have two desktop machines running pclos 2007 (2.6.18.8). Both have a chassis fan and it is running full speed making a lot of noise.
How can I configure the system to control the speed of the fan (i.e. slow it down so it's not noisy)

Machine 1: white box w/ MB Asus P5GC-MX, Celeron D, Intel 945 chipset
Machine 2: emachines T3302, AMD 3300+ , VIA chipset

Both machines are pretty basic, use on-board sound, video, 1GB ram, 1 HD, 1 DVD-CD/RW

Thanks
You will need to know the motherboards specifications. If the MB supports a chassis fan then that is one route but be sure the MB can support the fan specifications.

Google shows that the 'Asus P5GC-MX' supports a chassis fan. You can look further since I don't know the fan specs.

For the emachine, you need the motherboard manufacture not just the chipset.

I personally would prefer a separate fan controller for the chassis fans instead of the MB controller. Sure you won't know the fan speed via a system monitor but you won't load the MB controller with a potential error or damage if the match is not correct.

There are some good engineered fan speed controllers with active monitors but expect to pay some $$. The cheap 'hacks are not the way to go if you value your systems and time.

wastingtime 11-29-2007 09:34 AM
I found the spec for P5GC-MX here
http://dlsvr03.asus.com/pub/ASUS/mb/...mx_content.zip

on page 2-30 it says:

CPU Fan Speed [xxxxRPM] or [N/A] or [Ignored]
The onboard hardware monitor automatically detects and displays the CPU
fan speed in rotations per minute (RPM). If the fan is not connected to the
motherboard, the field shows N/A.
Configuration options: [Ignored] [xxxRPM]

The fan has 3-pin connector that matches the MB 3-pin connector and is connected to the MB
Yet, the bios shows [N/A] ?

onebuck 11-29-2007 08:51 PM
Hi,

Quote:
Originally Posted by wastingtime (Post 2974533)
I found the spec for P5GC-MX here
http://dlsvr03.asus.com/pub/ASUS/mb/...mx_content.zip

on page 2-30 it says:

CPU Fan Speed [xxxxRPM] or [N/A] or [Ignored]
The onboard hardware monitor automatically detects and displays the CPU
fan speed in rotations per minute (RPM). If the fan is not connected to the
motherboard, the field shows N/A.
Configuration options: [Ignored] [xxxRPM]

The fan has 3-pin connector that matches the MB 3-pin connector and is connected to the MB
Yet, the bios shows [N/A] ?
The 3-pin connector is keyed. Is connector inserted properly? Is that fan rotating? If the chassis fan has drawn to much current then the possibility of damage to the control circuit.

I prefer off board fan controllers for that very reason. I don't need feedback for a chassis fan. Just strobe blade and measure the current to set the speed.

lazlow 11-29-2007 09:38 PM
Some of the three pin connectors only have two wires going to the fan. If this is the case, the fan is not reporting back it's rpm.

Electro 11-30-2007 08:30 PM
Quote:
Originally Posted by jiml8 (Post 2972630)
easiest way is to replace the fans with fans that have built in controls. Antec makes a series of such fans.

Harder way is to obtain 3rd party fan speed controllers and wire them in.

Hardest way is to make your own fan speed controller. You need a square-wave oscillator with frequency of a couple hundred hertz (best if tunable; you should match it to the fan motor), a mosfet (or power darlington, but mosfet'd be better), and associated components.

Actually, as I looked into this a bit it would appear that most commercial speed controllers for computer fans are not using pulse width modulation controllers. That kind of surprises me because PWM gives full torque while modulating RPM. The commercial devices I see are varying voltage to the motor. The best choice I see is one that uses a variable voltage regulator to control the voltage to the fan. The simplest I see connects the fan between the power supply 12V line and the 5V line (rather than between 12V and ground), giving a net voltage of 7V to the fan. This would work just fine, but isn't adjustable.

Wrong way is to use a potentiometer. At least one controller I saw takes this route.
The problems using PWM is not the complexity. It is the problems that causes the fan to make more noise than it should. In order to create a PWM fan/motor controller, you need to know the inductance and the limits of the motor. PWM does include noise that is not mention in many books. The motor will produce an on and off noise that resembles the same when the user first flips the switch to on and motor jolts back. This noise will carry through out the case and chassis will amplify it. The best way is to use a voltage regulator and modify it to work well with motors or fans. Zalman did a good job with a 5 volt voltage regulator to control fan speed.

A potentiometer is not the wrong way. It is just another way. A a two or five watt potentiometer cost a lot and the customer have to read the manual to know the maximum limits.

The best way to create very minimal noise with a lot of efficiency it is to use stepper motors. Another way is to use a servo motor.

onebuck 12-01-2007 07:43 AM
Hi,

Quote:
Originally Posted by Electro (Post 2976200)
The problems using PWM is not the complexity. It is the problems that causes the fan to make more noise than it should. In order to create a PWM fan/motor controller, you need to know the inductance and the limits of the motor. PWM does include noise that is not mention in many books. The motor will produce an on and off noise that resembles the same when the user first flips the switch to on and motor jolts back. This noise will carry through out the case and chassis will amplify it. The best way is to use a voltage regulator and modify it to work well with motors or fans. Zalman did a good job with a 5 volt voltage regulator to control fan speed.

A potentiometer is not the wrong way. It is just another way. A a two or five watt potentiometer cost a lot and the customer have to read the manual to know the maximum limits.

The best way to create very minimal noise with a lot of efficiency it is to use stepper motors. Another way is to use a servo motor.
Not to get into a debate but to use a stepper to move air. An to state efficiency, less noise, is not correct. Talk about reverberation action. How do you mechanically insulate the motor from the chassis? Power for the stepper? What force of a stepper are you suggesting to move the air volume a small DC fan will move. The cost for the DC fan system vs the stepper? The stepper controller cost vs the DC fan controller cost? That one I would like to be presented!

As for the potentiometer use; power loss/waste which is not going to be the way I would suggest?

Quote:
Originally Posted by Electro (Post 2976200)
The problems using PWM is not the complexity. It is the problems that causes the fan to make more noise than it should. In order to create a PWM fan/motor controller, you need to know the inductance and the limits of the motor. PWM does include noise that is not mention in many books. The motor will produce an on and off noise that resembles the same when the user first flips the switch to on and motor jolts back. This noise will carry through out the case and chassis will amplify it.
The problem you state can be attributed to the design of the controller which can lead to complexity. If the timing is not proper then, yes the noise can develop. The precision of the control will depend on the timing of the PWM. If this is open loop then you must have some limits. As for your simple statement about on/off control. Then that type of design is simplistic end of the design spectrum for PWM. If the design is matched for the load, inductance and application then you should have something that would be feasible. Without a lot of cost in design and implementation of the system.

wastingtime 12-01-2007 09:02 AM
I checked the fan connector - now the bios shows the chassis fan rpm

I think that means the P5MX-GC has the hardware, bios code to monitor and control the fan's speed

It just does not do that. Perhaps I do need drivers to for the OS to do that.

jiml8 12-02-2007 09:54 AM
I don't want to get into a debate over controller technology here, but a PWM controller is quiet if it is properly matched to the load. Also helps to low pass filter it to round the leading edge of the pulse (gentler turn-on).

All other techniques are less efficient; the potentiometer technique is grossly inefficient.

Electro 12-02-2007 07:08 PM
Quote:
Originally Posted by onebuck (Post 2976568)
Hi,



Not to get into a debate but to use a stepper to move air. An to state efficiency, less noise, is not correct. Talk about reverberation action. How do you mechanically insulate the motor from the chassis? Power for the stepper? What force of a stepper are you suggesting to move the air volume a small DC fan will move. The cost for the DC fan system vs the stepper? The stepper controller cost vs the DC fan controller cost? That one I would like to be presented!

As for the potentiometer use; power loss/waste which is not going to be the way I would suggest?



The problem you state can be attributed to the design of the controller which can lead to complexity. If the timing is not proper then, yes the noise can develop. The precision of the control will depend on the timing of the PWM. If this is open loop then you must have some limits. As for your simple statement about on/off control. Then that type of design is simplistic end of the design spectrum for PWM. If the design is matched for the load, inductance and application then you should have something that would be feasible. Without a lot of cost in design and implementation of the system.
Just add a propeller to the stepper motor. The speed can be adjusted depending on the delays between steps. Do not need to go crazy with the force and voltage. Just a 12 volt and poor precision stepper motor is all is needed. Smaller the stepper motor the better because the spindle can be a smaller diameter. There are cheap stepper motor controller chips that can be used.


All times are GMT -5. The time now is 03:18 AM.