Temperature Controlled PWM Fan Controller

[JDCentral]

Since I haven't been able to find any of these things, I've decided to design one...

I'll post schematics and such when I'm done... it looks to be pretty easy, thus far - following specs/suggestions for getting the die-temp from AMD's documentation.

Frequency will be regulated (per-fan) by a trimmer potentiomenter (variable resistor.. tuned by a screw) - this way, you can adjust the frequency of the pulse so the fan won't make any clicks or whining noises. Thermalcouple will read the core temp, and adjust the duty cycle of the pulses.

I was going to read the voltage directly off of the pins on the processor, but then each processor is different, and you run the gambit of pulling too much current and frying your chip/mobo. Thermalcouple is more accurate, anyway.

The 'hard' part will be in determining the target temps and so-on... but that'll come from some tweaking and asking around anandtech.

As of right now, I'm moving software platforms... PSpice is win32 only - trying to get gEDA built and installed on my machine.

Comments/Suggestions are welcome.

 

[BonzaiDuck]

JDCEntral, my hat is off to you.

You CAN find these items -- I hunted a few of the more notable entries down late last year. It's been so long I can't even remember the brand names. One is shaped like a little LED pyramid. They all interact with the system through USB or PCI. Earlier models had buggy software.

I decided to keep fan control simple and use the motherboard headers with Speedfan. The way my cooling system is set up, I have exhaust and intake fans set at a constant but "silent" speed so that the front intake fans feed the CPU fan, which blows through the CPU cooler and (into?) onto a ducted motherboard -- to be sucked out immediately by the exhaust. With the case fans at a constant speed, I need only to regulate the CPU fan.

I WILL be interested in seeing your schematics. I might like to regulate the other case fans to spin up and down with the CPU fan, if your design is simple enough. And I didn't like the prices and features of the three or so commercial offerings I found in my search.

 

[JDCentral]

Thanks for the support - I want a hardware only solution, so I don't have to reconfigure the OS every time I reformat .

Also... price is probably the most important factor!

Right now, the components come to just under $5 (w/o PCB material). But I only have about half of it designed... and the gEDA tools aren't working, correctly. PHFFT!

Right now I'm designing for support of four fans - as the ICs come in quad-packages.

 

[l Xes l]

a group buy? =P

 

[Elcs]

A hardware-based, temperature controlled fan controller would be great, although Id find it a bit difficult to sort out what temps to use as defining points for fan speed change.

At times I wish I had some electrical know-how to rig up my own circuitry, fan control would be nice.

 

[JDCentral]

Yeah... deciding on temps is the hardest part. I think I'm just going to draw up the circuit, and then set 'target' temps after I put the system together and OC it a bit.

Right now I'm targeting 50C... but that may change

 

[Elcs]

Originally posted by: JDCentral
Yeah... deciding on temps is the hardest part. I think I'm just going to draw up the circuit, and then set 'target' temps after I put the system together and OC it a bit.

Right now I'm targeting 50C... but that may change

An LCD output allowing user setting of temps and when the fans spin up/down would probably be the best way to do it if you plan for cross-pc compatibility.

I admire the DIY nature of some of the projects people undertake, especially when it is to solve a problem or its functional.

 

[JDCentral]

Originally posted by: Elcs
An LCD output allowing user setting of temps and when the fans spin up/down would probably be the best way to do it if you plan for cross-pc compatibility.

I admire the DIY nature of some of the projects people undertake, especially when it is to solve a problem or its functional.

Well.. the entire point of the project is for the fans to 'control themselves'. I don't want to have to worry that I may bump the knobs so my fans to shut off, causing the CPU temps to skyrocket and the mobo catches the system and shuts it off (hopefully).

It's supposed to be a 'worryless' solution. I plug it into the system, and the temps stay reasonable. And when I'm not using my computer, the fans turn down. Cross-pc compatability isn't really the focus - I was originally going to connect to the temp diode pins on the Socket A chips. I have since decided that's a bad idea .

I still plan on probably drilling a (very) small hole in the HS to thread the temperature probe down, as outlined in AMDs Socket A thermal documents... but I haven't gotten that far, yet.

 

[Elcs]

Originally posted by: JDCentral

Originally posted by: Elcs
An LCD output allowing user setting of temps and when the fans spin up/down would probably be the best way to do it if you plan for cross-pc compatibility.

I admire the DIY nature of some of the projects people undertake, especially when it is to solve a problem or its functional.

Well.. the entire point of the project is for the fans to 'control themselves'. I don't want to have to worry that I may bump the knobs so my fans to shut off, causing the CPU temps to skyrocket and the mobo catches the system and shuts it off (hopefully).

It's supposed to be a 'worryless' solution. I plug it into the system, and the temps stay reasonable. And when I'm not using my computer, the fans turn down.

The idea of mine was a 'worryless' solution after pre-setting the temperature thresholds and criteria but I was thinking more of cross-pc and cross-platform compatibility.

You wont have to worry about that if your designing it for the 1 pc which will remain constant in terms of environment and parts. I doubt a cpu would get significantly hotter under full load over its lifespan, dust-aside.

Go forth and create!

 

[Goi]

I actually made one of these myself for 1 of my microcontroller classes. Compared to what other people were doing, it was pretty trivial, but it was fun and useful if they actually mass-produce is on PCBs rather than breadboards that we were using.