noisy fans + PWM alternative ?

[VJ]

Hello,

I'm considering the purchase of a MO display for both fan-control as displaying information. The main reason I'd like to control the fanspeed is noise (currently 4x80mm casefans, the system is a Supermicro X5DA8 in a LianLi PC70)... However, this thread got me worrying:
http://www.lcdforums.com/forums/viewtopic.php?t=1500

This is strange. I've tested out over twenty fans of many different manufacturers, sizes and speeds, and all of them were only slightly noiser with PWM mode on the fans. We'll have to pick up the fans and test them out for ourselves.

Does this mean that the PWM controlled fans don't run more silent in comparizon to when they run at full speed ? Also, does this mean that potentiometer controlled fans run more silent ?

Before the MX2, I had the idea of using the GPOs on a MO to address a digital potentiometer with up/down pins (I2C is too complex and expensive for home made stuff). Such a digital pot usually has a chip selection, a direction and a change pin. Basically, 5 GPOs could control 3 digital pots (3 chips select + direction + change). Has anyone played with this idea ?

Thanks !


J

[garysteel]

Hmmm.. I don't think you'll have any problem, my concern was probably me being overly pedantic with "Vatec Stealth fans" and expecting stealth!

I find the control the MX gives me is far better than what I had previously (Zalman Pots) and easier to control. I don't beleive there is an increase in Fan Noise due to the PWM. I just had higher expectations of these FANS.

I would be concerned if you were going the Vantec route but that doesn't appear to be the case...highly overated fans.

I've set my system up so that the MX controls the blow hole fan and two on the front, the M/B throttles the CPU Fan depending on temp and the rear fan is on a zalman pot (rheostat or whatever before I'm corrected).

If i'm not pushing the system, I can turn of the fans controlled by the MX and get a very quiet system. Silent PCs are an art. Best of Luck. Hope my comment was of use.


Wee G.

[VJ]

Hmmm.. I don't think you'll have any problem, my concern was probably me being overly pedantic with "Vatec Stealth fans" and expecting stealth!

Well, the quote was by Paradigm, and seemed more general than just the Vantec fans... I just want to be sure that PWM does not lead to more noise, some reviews mention this as a disadvantage of PWM (some fans can exhibit a "growling" effect).

My first idea to overcome this noise and still have digital control (before the MatrixOrbital MX-series) was a modification of this circuit:
http://www.maxim-ic.com/appnotes.cfm/appnote_number/843
The "pushbuttons" would be connected to the GPOs, but I'd still have a voltage issue (no digital pots go up to 12V). Over on 2CPU, it was suggested that I place three lower voltage pots in serial and address them together to spread current over them.
I would be interested in comments on doing this....

I've set my system up so that the MX controls the blow hole fan and two on the front, the M/B throttles the CPU Fan depending on temp and the rear fan is on a zalman pot (rheostat or whatever before I'm corrected).

My current setup is 2 80mm intake fans (front bottom), running at full speed (the LianLi case offers 3 speedsettings on these fans), but I have 3 10K SCSI harddisks behind them. At the rear of the case (between PSU and AGP slot) there are 2 80mm outtake fans, currently connected to the mainboard. The other fans present are the PSU fan (a PC Power Cooling 510XE, which I don't intend to change) and 2 Supermicro Xeon coolers. The case still allows for 2 more 80 mm fans (above PSU) and 4 40mm fans next to the PCI-slots.

I would like the system to be a bit more quiet (don't I'll go watercooled), so I'm thinking noiseblocker mats in combination with speedcontrolled fans and more quiet fans (Verax or Noiseblocker).



J

[garysteel]

Curious as to what you're CPU temp is with all those fans and you're CPU.

Mines is 3.06 O/c to 3.42 Ghz and idles at 45 and at load 57-60c Ambient temp 23. With SLK 900 Heatsink and a 92mm Zalman Fan.

Runing Stock Voltage, and in an CM ATSC 110B Case.

I've been trying to get the cooling sorted which is why I bought the 4 Vantec stealth at a rediculous price and was dissapointed with the noise. BUt now with the MX with the digital control, it is far easier to regulate and control.

Good Luck.

[VJ]

Curious as to what you're CPU temp is with all those fans and your CPU.

The system is a dual Xeon 2.4 GHz (533 FSB) on a Supermicro X5DA8 (passively cooled chipset), 3x 10K disks (IBM 36LZX, Cheetah 10K.6, Quantum Atlas - should return shortly from RMA)
Idle (room 23

[Paradigm]

Just to clarify that quote: I was comparing the noise of with PWM to the ideal circumstance with a pot. Because PWM is turning the fan on and off rapidly, you can get some mechanical vibration, which does lead to noise. Pot's (and other analog methods) don't have this problem since they deliver a consistant voltage to the fan.

Even so, this noise is only slight. In a quiet room with your ear close to the fan, you would only hear a faint extra noise. The vast majority of the noise from a fan is the air turbulance generated by the fan blades spinning. The increase in noise from the PWM is more than offset by the decrease in noise from the lower RPM.

Hope this clears that up.

[VJ]

Even so, this noise is only slight. In a quiet room with your ear close to the fan, you would only hear a faint extra noise. The vast majority of the noise from a fan is the air turbulance generated by the fan blades spinning. The increase in noise from the PWM is more than offset by the decrease in noise from the lower RPM.

Hope this clears that up.

Yes, that clears it up completely! Thanks!
I gather the faint sound added by doing PWM is in a different frequency range than the noise by the air turbulence. Is it lower or higher pitched ? Or does that depend on the RPM ? (just wanting to make sure the noise is not annoying than normal fan noise)

What influence does PWM have on the lifespan & reliability of a fan ?

Just curious: would the digital pot usage I described work, or could there be issues with it ?


J

[Paradigm]

The new sound is at the PWM frequency, about 37Hz. We chose this one because it is at the lower end of the human hearing spectrum (and we couldn't get it faster than 20KHz which is the upper end). This frequency is independant of the RPM since it is tied to the PWM frequency (which doesn't change).

On the occasional fan it sound almost like a very quiet growel or purr. But again it is still very quiet and you have to get quite close to it in order to hear it.

As for lifespan/reliability, I actually have no idea I've never really thought about it. I can't see why it would have any affect on that but I can't say for sure.

As for your idea on digital pots... I have a few ideas:

First, you won't be able to use our GPOs, they are output only, hence why they are not GPIOs. Why not use the 1-Wire bus. Go and have a look at:

http://www.maxim-ic.com/quick_view2.cfm?qv_pk=2934&ln=

It's a 1-wire digital pot . The next catch is that it only supports upto 6V. You would have to rig up an non-inverting op-amp circuit to amplify that fromn 6V to 12V (a very simple 2x amplifier). Just make sure that the op-amp can supply about an amp of current. It can be done, but it isn't straight forward, and unless you have a bit of experience in electronics, you might be in for a headache.

[VJ]

The new sound is at the PWM frequency, about 37Hz. We chose this one because it is at the lower end of the human hearing spectrum (and we couldn't get it faster than 20KHz which is the upper end). This frequency is independant of the RPM since it is tied to the PWM frequency (which doesn't change).

On the occasional fan it sound almost like a very quiet growel or purr. But again it is still very quiet and you have to get quite close to it in order to hear it.

Ok thanks (most likely, my harddisk will make more noise then )! Do some makes of fans perform better than others (e.g. the Vantec silent - accoring to the other thread don't seem to like PWM that much) ? And if so, is there a list of some kind ?

Does the PWM use equal on/off intervals (e.g. _=off, -= on)
--__--__--__--__
or does this depend on the speed set ?

As for lifespan/reliability, I actually have no idea I've never really thought about it. I can't see why it would have any affect on that but I can't say for sure.

Ok, no problem!

First, you won't be able to use our GPOs, they are output only, hence why they are not GPIOs.

Why isn't this sufficiant ? In the scheme I posted, the pot is connected to 2 pushbuttons to set its speed. I just thought the GPO could behave like a button...

It can be done, but it isn't straight forward, and unless you have a bit of experience in electronics, you might be in for a headache.

Hmm, I just know the basics, so guess this is too far fetched for me... On 2CPU, some guys suggested putting the pots in serial (and linking them speedwise to maintain equal resistance), as to split the current going through them and to avoid an op-amp...

Oh, are there ways of connecting more than 3 fans ? I found this thing:
http://www.directron.com/diverter.html
but haven't found any reviews on it. I think one would run into power issues when powering multiple fans from one header though...

I also noticed you guys have a 3-pin Y splitter. Does this still allow PWM ? And what happens with rpm monitoring ?


J

[Paradigm]

Ok lets start at the top:

Ok thanks (most likely, my harddisk will make more noise then )! Do some makes of fans perform better than others (e.g. the Vantec silent - accoring to the other thread don't seem to like PWM that much) ? And if so, is there a list of some kind ?

Does the PWM use equal on/off intervals (e.g. _=off, -= on)
--__--__--__--__
or does this depend on the speed set ?

PWM works by varying the on/off intervals. When you set the PWM to 128 (or 50%) the on/off intervals are equal. But when you set it to 192 (or 75%) the signal will be on 75% of the time and off 25% of the time.

Why isn't this sufficiant ? In the scheme I posted, the pot is connected to 2 pushbuttons to set its speed. I just thought the GPO could behave like a button...

Confession time, I didn't actually look at the link you sent Looking at it now I can see what you're getting at. I'd remove the MAX6817. All it's used for is debouncing the push buttons, but since our GPOs don't bounce, you don't need it. Instead connecting the GPOs directly to the INC and U/D signals would work. Sorry about that, next time I might actually have to follow the links

Hmm, I just know the basics, so guess this is too far fetched for me... On 2CPU, some guys suggested putting the pots in serial (and linking them speedwise to maintain equal resistance), as to split the current going through them and to avoid an op-amp...

First I made another small mistake, the digital pot's 1-wire operating voltage is upto 6V, but the pot side can work up to 11V. Still doesn't matter as fans operate at 12V. As for the series part, if you put them in series that could work, but you have to be absolutely sure that niether pot get's more than 11V across it. Working it out you have to make sure that:

R1 < (R1 + R2) * 11/12
and
R2 < (R1 + R2) * 11/12

Oh, are there ways of connecting more than 3 fans ? I found this thing:
http://www.directron.com/diverter.html
but haven't found any reviews on it. I think one would run into power issues when powering multiple fans from one header though...

I also noticed you guys have a 3-pin Y splitter. Does this still allow PWM ? And what happens with rpm monitoring ?

You won't run into power issues as long as you keep the total current to 1A or total power to 12W. That is the maximum specified current or power at 12V. All this device does is split the power and ground and only takes one RPM wire back. So you get to monitor the speed of one of the fans and the other RPM wires are ignored.

[VJ]

Sorry for the long delay...

Ok lets start at the top:
PWM works by varying the on/off intervals. When you set the PWM to 128 (or 50%) the on/off intervals are equal. But when you set it to 192 (or 75%) the signal will be on 75% of the time and off 25% of the time.

Hmyes, I don't know why I posted that question...

Confession time, I didn't actually look at the link you sent Looking at it now I can see what you're getting at. I'd remove the MAX6817. All it's used for is debouncing the push buttons, but since our GPOs don't bounce, you don't need it. Instead connecting the GPOs directly to the INC and U/D signals would work. Sorry about that, next time I might actually have to follow the links

Hehe... thanks!

First I made another small mistake, the digital pot's 1-wire operating voltage is upto 6V, but the pot side can work up to 11V. Still doesn't matter as fans operate at 12V. As for the series part, if you put them in series that could work, but you have to be absolutely sure that niether pot get's more than 11V across it. Working it out you have to make sure that:

R1 < (R1 + R2) * 11/12
and
R2 < (R1 + R2) * 11/12

Thanks again !

Oh, are there ways of connecting more than 3 fans ? I found this thing:
http://www.directron.com/diverter.html
but haven't found any reviews on it. I think one would run into power issues when powering multiple fans from one header though...

I also noticed you guys have a 3-pin Y splitter. Does this still allow PWM ? And what happens with rpm monitoring ?

You won't run into power issues as long as you keep the total current to 1A or total power to 12W. That is the maximum specified current or power at 12V. All this device does is split the power and ground and only takes one RPM wire back. So you get to monitor the speed of one of the fans and the other RPM wires are ignored.

Ok, thanks!
The reason I mentioned the fan diverter, is that it outps the RPM of the slowest fan. So if one uses this to check up on 2 identical fans, it is still possible to detect failure of one of the fans....


J

[Paradigm]

No problem Let me know if you have any more questions.