Intelligent cooling system for overclocked PC

Hello all, Im new here to the forms and I have a task here ahead of me. I said what better than post on OCforms the best place for extreme modding

At the moment, Im in my year 5 of my Bachelor of Engineering degree. I must complete a project over the course of the year. The project I was allocated was….. Wait for it

Intelligent cooling system for overclocked PC (sounds cool)

Before I start explaining the project im not asking anyone to do the whole thing for me far from that but rather get ideas from you guys on how to interface it, maby embedding software in windows which I don’t know too much about and so on. So here goes here is the project overview
---------------------------------------------------------------------
---------------------------------------------------------------------
Topic:
The design and construction of an intelligent fan-based cooling system for personal computers. The system will automatically provide the correct level of cooling for the system depending on operating conditions. This will aid system stability and reliability especially in overclocked systems.


Rationale:
Overclocking is the commonly used term to describe the process of operation PC’s beyond their intended specifications. This can be applied to various components of a computer system and will usually result in increased performance. The drawback of overclocking is that system stability and reliability is usually compromised. The main cause of this is that components generate more heat than under normal operating conditions. For example an overclocked graphics card may run fine for 2D work, but generate excessive heat during intensive 3D operation. This extra heat needs to be removed to ensure stable operation, usually through the use of cooling fans. Traditionally, a PC would have a number of cooling fans sufficient for cooling the system. These are usually monitored by the mainboard and can be adjusted through software. An overclocked system requires additional cooling through the use of extra fans. The main problem with this is the increased noise level of the system, even when the extra cooling is not required. There are a number of commercial device that allow the manual adjustment of these fans, but this means one has to keep and eye on system temperature and then manually increase or decrease fan speed.

There is a need for a system that controls additional cooling fans intelligently by monitoring temperature of various components.


Project Objectives:
This project is primarily about design and construction of a system that can monitor temperature of various components (case, cpu, gpu, ram etc.) and adjust the rotational speed of the cooling fans to provide the correct level of cooling while minimizing noise.


Key objectives include:-
 In-depth literature review of digital electronic, motor control, device drivers and interfacing electronics with respect to the system to be developed
 Design and construction of a suitable interface that allows the control of four additional cooling fans with temperature feedback
 Implementation of circuit protection
 Programming of appropriate control software to achieve correct cooling
 Evaluation of the system performance through extensive real-world testing


Technical Description & Methodology:

Thesis: The design and construction of an intelligent fan-based cooling system for personal computers. The system will automatically provide the correct level of cooling for the system depending on operating conditions. This will aid system stability and reliability especially in overclocked systems

Research: The project will require the student to become intimately familiar with basic digital electronics interfacing and motor control. Basic research will include leading texts on electronic testing and contemporary papers in the same, the effects of cooling and overclocking.

Hardware Design: The hardware that needs to be constructed for this project can be broken down into the following parts: 1) multiple fan controller circuit, 2) temperature feedback circuits, and . The overall cost for this project is expected to e very low, with mostly standard components being required.

Software Design: The software requirements may be broken down into the following:
 Programming at low level for hardware interfacing and control
 Programming of decision making module
 Designing a suitable report and management user interface

Critical Success Factors The student will be required to achieve the project objectives in order to pass. Additional marks will be rewarded on the following criteria:-
 Innovation in balancing the need to minimize noise while guaranteeing system stability
 User-Friendliness and ease of installation
 Low cost and commerciality of finished product
 Quality of Research; reliability of system, fail safe methods
-------------------------------------------------------------------------
-------------------------------------------------------------------------

thats what i was given and so far i have a few options at hand
firstly how to interface it to the PC
at first i tought through the parallel port because of its ease of use but this would look crap and also limit the io pins needed so i would need to multiplex so im thinking of using the PCI BUS, you can get pci interface cards and im allowed use one of these

ill post more to this as i get along as im only been given the project but anyones ideas would be great as two heads are better than 1

when the project is finaly completed ill post all the specs, circuit design, data sheets, pics and test reviews here 4 u to play wit

Sounds great and I can't wait to see it, but don't simple temperature controlled fans do exactly what you're trying to do?

Yes m8 they do but I must design my own temp sensors and fan controller system in hardware (chips and stuff), what im trying to achieve is using 4 temp sensors giving feedback to the card and then the fans changing their speeds accordingly
Also im thinking of having the cpu fan speed controller not using temp but system usage like if the system performance is high the fan will increase in speed

I couldn’t believe that I got such a cool project. Its something I like doing for a change

the problem with controlling the cpu-fan by usage is that it would need to be calibrated, and if the system was (as you stated) overclocked. the callibration would need to be adjusted frequently based on change in cpu wattage. and the specific heatsink used, size of the fan, etc

thats very interesting m8 tanx for pointing it out. but by using task manager and interfacing to this with software could system usage not be monitored from there and possibly the fan could be easily controlled


oh right I understand, im overclocking so the temp will rise but not the system usage. lol, it sounded nice in theory but yes your perfectly right m8, as im overclocking the wattage and temps are more important for system stability than just concentrating on system usage to determine the correct speeds for the cpu fan

sounds like a dream project , shame in the UK u need maths to go into engineering, which i havnt got

i'm doing a similar degree, i might do something similar for my final year project

I've actually been toying with that concept for a while now myself, never got around to actually doing anything with the idea though. Good luck and keep us posted!

freekhead said:

Project Objectives:
This project is primarily about design and construction of a system that can monitor temperature of various components (case, cpu, gpu, ram etc.) and adjust the rotational speed of the cooling fans to provide the correct level of cooling while minimizing noise.

Click to expand...

Adjusting the fan speed - even downward - is an excellent way to draw attention to a noise people have grown oblivious to.

So address that, embarrass your instructor, and have found the solution of adjusting fan speed very, very gradually. That is unless you figure "commerciality of finished product" means (quite possibly I'm afraid) the idiotic bling bling of a revving PC.

Also beware of oscillating harmonics between fans running at different frequencies - this yields greater perceived noise than a simple decibel measurement would indicate.

Handbook for Acoustic Ecology

Start here: Perceived Noise Level

I think the relationship of voltage, heat, and clock speed is interesting.

Higher voltage = more stability (but more heat generated).
More heat = less stability.
More fan power = less heat (but more noise generated)
Faster clock = better (but more heat generated).

I think it would be interesting to have an intelligent device that searches for the "sweet spot" on this manifold, however "sweet spot" might be defined.

the wesson

I would put it in a 5 1/2" bay drive rather than build a PCI slot version.

Oh, and don't forget to overclock those fans, too

id like to do that 2 fuzzba11 but im going to have to interface this project with software maby c++, lab view, or VB so putting it in a drive bay would only mean lots of wires and maby USB interface so that would be untidy.... but ill look into it anywho... like your Radiohead sig m8 there the greatest British rock band around

TheWesson that’s a good point trying to find the optimal settings for the 2 most critical aspects of overclocking heat vs stability yes you can hav a fan going full rev with 120cfm but would u sleep beside it, no so im trying to achieve a fan speed heat ratio that will still give good stability without using manual fan controllers

Sean Lindstrom thanks for the links. there is a lot of technical info there for me to fully understand harmonics of different fans and also yes I have thought of addressing the issue of adjusting the fan speed gradually by using 16-bits for each fan using a mux as not to use to much i/o on the card this would give me 65336 different speeds per fan


thanks for all the help so far lads this project is still in the toughts stage but without your comments i would of never hit on some of the more important aspects of this project..

http://store.yahoo.com/svcompucycle/thermaltake-hardcano12.html

thermaltake beat you too it. Not software but does adjust speeds with temp

yeah it has bn done before but not in the way i will approach the problem anyway its a thermaltake product so its probably overpriced, tacky and inefficent lol....

does anyone know much about PCI prototyping

well there are programs like...CPUZ, which tells you your FSB speed, and everything.

now, what if a program was wrote, that if the Clock speed of the comptuer was raised above a certain point ( and you'd have to put in information for multiple processors, ram etc... ) that would cause them to speed up.

Also, this could get complicated.

But, why just stop at making the fan(s) speed up or slow down?

Have a program keep record of the Temps your processor ( or whatever component you want ) then, fan/s in the case, slowly move to certain directions, degree by degree, angeling up down, and even left and right.

It records all the temps for all those different angles. Then it picks the best angle for each fan, to get optimal Airflow.

Or even if you have to to move the fan/s yourself. have a program keep track of them all, record all the data. then use the data and tell you what exact direction to place the fans and all

would be kinda cool, but it would take quite a bit of work to do, mainly if you have it move the fans itself. But if can make or use devices such as a potentiometer, and use that as a way to keep exact movents of the fans, you can just write down the spots you had the fan angeled ( in Ohm's, but as long as the pot doesn't move, then 56 ohms will be in the same spot every time )

just an idea. good luck man!

the way this is going it's looking like more of a software project than hardware, only thing is with software monitoring you don't know how much the temps are out by, you'd need a combination of hardware and software. not just for accuracy by to distinguish it from programs like 'speedfan' that controls fans on MB headers

yes it looks like a lot of software for the project alright...what im going to try and do is maby model it using labview as its pritty easy to use then go from there

The design of this project involves fan controller circuitry for four fans, Temperature sensor controller circuitry for 4 sensors, operation and monitoring with Windows Software that runs on startup, connection via USB or PCI bus interface, Unit works independent from the motherboard: so as not to impede any performance loss, four outputs to the Motherboard for the tachosignal: Tachometer with delivered Motherboard monitoring tools such as ASUS Probe or PC doctor. Unit does not take up valuable space inside the PC, can be installed internally, also the CPU fan speed controller monitors the wattage input to the CPU This would give a indication that the processor has been overclocked or system performance usage has increased.

sounds good

i hate to be the poop head...

"Innovation in balancing the need to minimize noise while guaranteeing system stability"

system stability cannot be guarenteed, even with a non- overclocked system....no matter the heat.

yea its been done, (just think of a fan controller with temp sensors it can work with and no knobs), but not for cheap...

it sounds like whoever is guiding your project would have them pitch the project to the general computing world, rather than us OCers... consider a simplified explanation of the benefits (economical) of overclocking when/ifever you do pitch this.

regarding the angle idea - a product won't be easily marketable, even to OCers like ourselves, if it involves a lot of work like manually adjusting the fan angle a bunch of times... and even if you did make some mechanisim to mount the fan and change the fan's angle to eventually determine the ideal angle for a fan on a heatsink, i have a feeling that those results (if significant) would eventually get posted and shared via forums like this, and anyone who cares enough would probably just use that data rather than test it themselves by buying a product.

some more ideas to throw into the mix:

will you have an LCD readout (assuming you follow the 5.5" bay advice, which i consider a good idea) to show what temperatures are being read / RPMs of the fans under control?

what controls will the end user have over this device's functions? will you have a "noise knob" that will adjust fan speeds to use more cooling / make less noise overall? will there be an "off" switch, which will allow fans to return to their previous voltages?

should something fail (be it your device, or a fan breaks, or connections are bad...) and temps get to be ridiculously high, will there be some sort of backup (i.e. that "off" switch idea" / or some sort of alarm to alert the user so he/she may unplug the computer?

hope this brainstorm helps

to be sure, if you keep costs low enough to sell them for 30 USD or less (you may have to take a "no frills" approach and keep it internal, like in a HDD cage), and it gets the job done well, i'll bet your product could become a serious competitor in the market (assuming you wish to take your project into the market)... ^^