Hi I ran "sensors" and found the following

:~# sensors
dme1737-i2c-0-2e
Adapter: SMBus I801 adapter at 3400
5VSB: +0.00 V (min = +4.50 V, max = +5.50 V) ALARM
Vcore: +0.00 V (min = +0.00 V, max = +2.99 V) ALARM
+3.3V: +3.34 V (min = +2.97 V, max = +3.63 V)
+5V: +5.05 V (min = +4.50 V, max = +5.50 V)
+12V: +11.98 V (min = +10.81 V, max = +13.19 V)
3VSB: +3.28 V (min = +2.97 V, max = +3.63 V)
Vbat: +3.05 V (min = +2.70 V, max = +3.30 V)
fan1: 979 RPM (min = 0 RPM)
fan2: 0 RPM (min = 0 RPM)
fan3: 0 RPM (min = 0 RPM)
fan4: 0 RPM (min = 0 RPM)
temp1: +31.6°C (low = -127.0°C, high = -128.0°C) ALARM
SIO Temp: +31.1°C (low = -127.0°C, high = +61.0°C)
temp3: +45.9°C (low = -127.0°C, high = +76.0°C)
cpu0_vid: +1.088 V

My PC has 3 fans "Power supply" , "Case" and "CPU"

I'm guessing? fan 1 is my CPU..

Can anyone suggest a few good links for "modding" the other fans so I can pick up their rpms and possible a cheap way of picking up the power supply heatsink temperature.

I've built two physical variable controls for the "Power supply" and "case" fan speed regulation (to reduce noise) and it would be nice to see their rpms within terminal

If you dont have a reporting wire going back to the motherboard from the power supply fan or case, you wont get any reported speeds.

Even if you do have a reporting wire, if fan RPMs are low you may get no reported speed, or the wrong speeds.

It's possible to replace both fans with 3 wire fans. "Power supply" and "case"

The sensor wire is for fan speed.

So non linear or at least the curve will drop off at low rpm resulting in poor rpm feedback. (OK I got that)
("Afterthought....... is the sensing Hall effect?")

I will have to take a look later at the motherboard to see what fan connections exist.

Turning to temperature sensing, I'm thinking initially of popping in a couple of the £2 -£3 LCD thermometers. (the ones that look like the ones which come with the cheap radio weather stations with 30" wires"... putting the sensor on heatsink and display outside etc.

But I would really like to pickup a few cheap PWM digital thermostats with load switching to control the fans.

Anyone have recommendations?

I have one. Don't mess.
You're looking at banging in a few cheap scrappy(= less accurate) parts to oversee cooling on your pc, when you need a thoughtful design. If you were likely to succeed, you wouldn't have to ask us. You'd do the calculations, specify and tune or program.

Does anyone think this is going to be a good idea/learning exercise?

NO!

Can't agree with that...........

I'm an avid DIYer....... Radio Ham....... Homebrewer

I've been tweaking electronics since the 60's, and computers since the BBC B circa 1982.

In fact the PC I'm working on at the moment is a mongrel from several discards.

My workshop is where I get my kicks...........I always make before buy.

So to put this thread back on track does anyone have any positive ideas for this project.

I have to say that I somewhat agree with Business Kid, so you'll probably ignore all of this.

If by "Power Supply" fan, you mean a fan actually in the power supply (as opposed to a case fan that assists in power supply cooling) I'd solidly argue don't. Dangerous. These things are usually cr*p, on the edge of blowing up, and able to taken over that edge by the most minor of changes (or drawing the full specified load).

On the other hand, if you've got a good one, leave it alone.

You could blow more air in its general direction (if you've got room in the case to arrange that), but that doesn't imply opening the PSU.


Well, you won't be getting any feedback from your fan in your PSU, so it won't be that one. You could unplug the case fan briefly to check that it isn't that one, so that would only leave you with one option.

If you wanted to do something constructive, use your optical pyrometer to measure temperatures and use non-toxic smoke to have a look at the flow through the case. Use these results to fine tune the temperatures and eliminate hot spots and reduce temperatures generally. And then forget it, apart from checking back occasionally and it hot weather to check that you aren't running out of margin.

Thanks for the ideas.

The simplest of ideas are often missed,
I like the smoke idea for flow.
I have a suitable sheet of glass in the workshop and plenty of double sided tape to seal it over one side.

As I thought the "fan 1" shown via sensors is the CPU fan. (I stopped the case fan momentarily)

I had all the fans out last week, "power", "cpu" and "case" peeled back their labels removed their dust caps and put in two drops of very thin machine oil.
Remounted the "power" and "case" fans on some offcuts of a rubberish mat.
This has reduced the noise quite a bit but not enough.

I must admit I was very surprised at the basic switch mode power supply and agree with you on their lack of robustness.

If I get chance next week I'll pop the fan controllers in an throttle back the two fan's speeds.

Hi,
Most cheap add on fan speed controllers are using potentiometer(s). Not finite control. Look at PWM controllers for additional fans.

I dress my cables within the case to minimize breaks in air flow. Boundary flow is interrupted by the type of case and peripheral positioning within the case. You might consider the use of a top fan exhaust. When preparing for a flow study using smoke, be sure to seal the case parameter mount points. I have sealed several cases to control the air flow and it is doable. You want a positive flow so the inputs and outlets must provide this state. Choke the flow and you will have issues. Plus be sure to use caution with the type of smoke for testing.

Your rubber mat cuts for vibration limiting is just a minimal rejection. You should consider a active shock mount to prevent injection by the fan to the chassis or vice verse. Your mat type mount is transferring mechanical vibrations to the blades via reflection. Plus remember that there will be a resonant point for the speed change of the fan that will be dominant noise/vibration.

You can get PSU with sensor & control for the temp. Be sure the motherboard does support this option. Over heat the PSU and you will get more trouble than it is worth. As to other sensor monitor/control for the system, I would use a active monitor with ported information along with active display. Most of this type are used within racks to monitor/alarm rack interior flow and temperature. Not cheap but doable by a knowledgeable hardware hacker. arduino?? Linksys & arduino?? Two good hints.

The software angle: have you tried fancontrol?

I use this in Ubuntu.

I leave the CPU cooler to the BIOS. It is a big Noctua cooler with a couple of 140mm fans, and it runs slow and quiet anyway. I wouldn't take the risk of messing with this one; accidents would be too expensive. Likewise, the power supply is a hefty, quality one, and its built-in fan also runs cool and slow as it is.

Apart from that, only one of the other fan outlets on my MB can be controlled. I split that to two fans, and use fancontrol. The other two fans in the case are inblowing, and I just run them slowly using the supplied voltage selectors.

You can have hours of endless fun calibrating the temperature/speed of the controlled fans, and then just leave it to start up with the machine.

Having had those fun hours, in Ubuntu 11.04, my later 12.04 runs the fans slower without any fancontrol so I'll just leave it.

What's the point of all this anyway? Well, for me, it is to make the machine quieter. That is almost more important when I'm not listening to music than when I am.

Ultimately, that quest will go past tweaking and into buying better and quieter fans as and when. So far as an air-cooled system is concerned, I believe that I can't do better than the Noctuas I've ended up now, until someone re-invents fan technology. And that is more likely than you might think: how about a self-silencing fan that outputs the soundwave to cancel the soundwave it is creating? That is something that exists in labs: it will come.

I was looking at an Arduino project in one of the uk linux magazines last week.
Might be worth exploring further or perhaps a Raspberry? Thanks

Yes I agree the noise level is more important when your NOT listening to music.

Popping the case fan onto the top seems so obvious heat rises!
Likewise the power supply fan faces backwards....

The rubber mats were just an experiment .. I've seen some purpose made silicon standoffs for sale might invest in some.
(When I added a cooling fan to my dreambox about 3 or 4 years back I mounted that with some salvaged silicon grommets)

One thing I have read about but until NOW discounted is the dressing of the cables.

I have several (SILENT) high Ampege power supplies for my radio transmitters, its surprising that there are not "more" cheap fan-less PC ones.

Before I retired in our server rooms we kept noise at bay using bought in acoustic cabinets with baffles. (nice rooms to work in in the Winter!)

My Antec case has a top fan. It also has a bottom power supply in a separate compartment. As it runs at a small fraction of its rating, I also keep the two hard disks in the same compartment, with a gentle inlet fan blowing over them.

I believe the PSU has its own PWM control over its own fan, which is effective at keeping down speed/noise. Unfortunately there is no way of reporting temperature or fan speed from that unit.

The Gigabyte MB is good in providing fan headers, but dumb in that it only allows control of two of them.

However quite the fan[s] it seems impossible to move air completely silently --- although there are completely passive cooling designs, should one get fanatical enough

Considering that I live in the tropics, with ambient at not a lot below 30C for most of the year (nice for the winter, not that you'd notice it was winter!) my machine and I don't do too badly!

Solved problem, by 1. swapping the fan inside the power supply, 2. upgraded cpu cooler arctic freezer 13 and 3. some extra case padding. I will be remounting the hdds on either wood and/or rubber bands when I can get around to it.

Thanks everyone for your suggestions.

1 members found this post helpful.