T.E.C. help?

[rusty1620]

Hey guys, I'm new to the site, got into the enthusiast side of computing about a year ago, have done a few custom builds with overclocking and a lot of playing around and researching various topics.
So I decided I want to try and make a stand alone T.E.C./liquid cooling unit. In this stand alone unit I would like to have 2 x 1.5L reservoirs with 4 outlets coming out of the unit. I have ordered 2 x 160W Tecs (1 for each tank) but have yet to decide on a feasible and effective way to cool the tecs.
Also I will be incorporating a PCB I would like to build with the following specs:
-5 temp sensors (1 unit air temp, 1 on each hot side Tec, and 1 in each reservoir)
-Power up to five 4-pin fans
-Touch screen fan speed controller
-power pump for liquid
-unit power button with l.e.d. halo
-heat alarm(buzzer+l.e.d.) x 2 (1 for each Tec)
-trigger for computer power at temp X (up to 4)
-power for the Thermoelectric Coolers (2)

My main questions are:
What size PSU will I need?
How do I know which PCB parts I need to buy to set this up?
But mainly how in the world do I go about Fabricating this circuit board?(never done this, attempting my first repair on an APC power supply board which has a broken trace)

Any help is greatly appreciated, feel free to ask me anything I left out.

Thanks

 

[m0r7if3r]

The custom PCB thing isn't something you're going to be able to do, I'll tell you that now...it's not as simple as just putting components in a board and it all working presto chango, you have to design the circuitry and what not...and if you've only ever done one repair then you're certainly not at the level where that is feasible for you.

Where do the reserviors go to? I don't quite understand that. It sounds like you're making a water chiller, not a tec cooler (your water chiller just happens to integrate tecs). I suggest you SLOW DOWN and figure out what you're doing before you kill any gear.

 

[rusty1620]

I do realize this is above my head, and I don't plan to build it overnight. Also I will not be hooking this up to any computer until I have thoroughly tested it first.
What about universal circuit boards?
Like this:
http://cgi.ebay.ca/1-X-PROTOTYPE-PC...396?pt=LH_DefaultDomain_0&hash=item335d5fdfcc
And the whole point of me doing this is so I can learn and figure it out as Im going.

No offense but I want to know how I go about doing it,not that I can't do it.
Thanks

 

[m0r7if3r]

How do you want this to interface? Because measuring temp and outputting it isn't exactly a simple task, neither is interfacing to a computer, especially if you don't have significant windows development experience. For a touch screen you need some sort of microprocessor to control it, so now you're into the realm of building a small computer (by soldering it yourself). Turning the TEC on separate from the rest of the computer is outright foolish...think about it, when do you want it on or off that it doesn't involve turning the computer on or off. All you're doing is inserting means for human error. Triggering power has to be done with a logic circuit via relay and it has to be SOLID...like, no way that thing could go wrong or your computer will just turn itself off. The power in for this board is going to be in excess of 300w for the tecs alone and will need to be split to multiple rails of a power supply. I'm an EE major, trust me, if you're just getting into this sorta thing you don't want it jeopardizing 100's in parts and $1000's in computer equipment.

This is all ignoring the fact that you seem to have no plan for cooling the hot side of the TECs, as most tec-chiller setups are actually dual water loops with the hot side of the TEC cooled by it's own, separate loop and the cold side in a loop with the components.

Oh, and that board you posted is EMPTY pcb, it's breadboard with solder pads. I'm not trying to tell you not to learn, I'm just trying to prevent you from killing a whole bunch of expensive gear in a single puff of blue smoke.

 

[rusty1620]

Okay so I understand a little more where you are coming from.
For the fan controller can I not just buy one already made for a computer and power that?
the reason for turning on the tec unit first would be to achieve below ambient temperatures before the comuter gets power,
For example:
My current processor is an AMD quad core 3.2 955. When it was Overclocked to 3.8 with stock cooling and running prime 95 for 1 hour the core sat at 60 Deg. C. (with a max temp of 62) lets say I overclock this system to 4.2 Ghz/core, Do I not have to chill the processor before I run power to it to avoid overheating?
That is where I am coming from for a trigger to turn on the computer because also the computer will never be turned on without the chiller.
I was thinking a 450W PSU will suffice? or should I go 500?
And before this gets attached to any of my main computers I have many older junk ones and also a heat pad to test the unit with.
As for cooling the Tecs to me the most efficient way seems to be a separate loop like you suggest, so then I need to add a second pump?
For the PCB in otherwords you are saying there are no pathways? how do people make prototype boards before they get them custom fabricated?
Now getting into microprocessors etc. I know that is too far above my head at this point so anything that involves things on that level I will have to leave out.
How do you suggest I go about making my own pcb for the first time to practice? without taking any classes etc. I want to teach myself.
Thanks

 

[m0r7if3r]

There are plenty of good fan controllers out there. You might want to look into the lamptron touch as it's a great unit for the price and is well built.

amd chips throttle at 60C, so you probably don't want to leave i there. What's your current cooling and why is just straight water cooling inadequate (and why do you need to run 4.2ghz 24/7?). You don't need to chill it beforehand, the loop will cool down very quickly.

Not sure on the PSU...450 should do it for JUST the chiller, not including the rest of the PC.

hot side of the tec gets a traditional water loop, strap a CPU block to the hot side and the cold side, one block goes into the cold loop which is just a pump, cpu block, tec block, and a res. The other side goes to the hot side which is the tec block, res, radiator, and pump. The hot side loop in particular needs to be rather robust as it will be dissipating the wattage of the TECs PLUS the wattage of the processor (well over 500W...which to put it in perspective is more than SLI'd gtx580's at load)

that PCB you post links to was just a bunch of copper pads...maybe traces. Look up breadboard, that's how most prototyping is done, usually goes straight to custom PCB from there. That pcb is probably laid out like breadboard or is a bunch of individual copper pads, I've seen both (the former being the more common)

Good place to start would be some 555 timer led flashers, if you can understand one of those you're doing pretty good...though for any temp monitoring you really need some sort of microprocessor unless you're doing some sort of dummy event (like turning something on when a thermistor hits a certain point or something of that nature)

 

[I.M.O.G.]

If you want to roll your own, its not that hard or impossible. It should be a challenge though.

Some of the guys in this thread may be able to help you with the circuitry:
http://www.overclockers.com/forums/showthread.php?t=641111

I would start a thread in the PSU/Electronic devices forum and request help. I would then send a PM to some of the guys in that thread to see if anyone may be able to offer a hand with your project.

I would also drop the touchscreen, unless you want to get into some basic programming also. You can do a basic controller and temp display without much trouble though.

 

[Mr.DLucey]

how do people make prototype boards before they get them custom fabricated?
How do you suggest I go about making my own pcb for the first time to practice? without taking any classes etc. I want to teach myself.
Thanks

You make them your self. I made a few many years ago. Use to be you could buy the stuff from radio shack. I see they still have the etchant available in store. Maybe the rest is still in the store. Used to make the path ways with a special pen and etchant off the rest.
http://www.radioshack.com/product/index.jsp?productId=2102868

Take a look here:
How To Make Your Own Homebrewed PCBs
http://aliben.wordpress.com/2006/12/02/41/

 

[I.M.O.G.]

You don't really need a PCB do you? Why not just do it with breadboard?

 

[rusty1620]

@m0r7if3r- My current cooling is a home made water exchange unit (no rad, or fans)and it keeps my system at 42C with no load. The overclocking was just to see if I could do it and backed it off shortly after, and I don't need to run 4.2Ghz all the time, I just want to see if I can build a cooling unit that can allow me to run at faster speeds, for fun really.
Breadboard was my first idea, but I thought something soldered would be better, more permanent. I don't mind spending a little extra time to do it properly either. Maybe I will start with a breadboard and then fabricate a pcb at home.
And a dummy event is exactly what I want to do, when any thermistor hits Temp X then the heat alarm will sound with an led indicating which thermistor was set off. but I also want to be able to view that temp. Really doesn't have to be touch, would just make it much sleeker.

@ Mr.DLucey- Thanks for the links, which method would you recommend, regarding lower cost/quality and time involved?

@I.M.O.G.- Also thank you for the links, like I said earlier in this post I was originally thinking breadboard then thought something more permanent would be better?
i.e. if I keep this cooling unit together for 2 years will the breadboard be good enough? or am I better off using it as a protoboard and fabricating a real one after?

Thanks again to all, this help is really appreciated.

 

[m0r7if3r]

@m0r7if3r- My current cooling is a home made water exchange unit (no rad, or fans)and it keeps my system at 42C with no load. The overclocking was just to see if I could do it and backed it off shortly after, and I don't need to run 4.2Ghz all the time, I just want to see if I can build a cooling unit that can allow me to run at faster speeds, for fun really.
Breadboard was my first idea, but I thought something soldered would be better, more permanent. I don't mind spending a little extra time to do it properly either. Maybe I will start with a breadboard and then fabricate a pcb at home.
And a dummy event is exactly what I want to do, when any thermistor hits Temp X then the heat alarm will sound with an led indicating which thermistor was set off. but I also want to be able to view that temp. Really doesn't have to be touch, would just make it much sleeker.

Viewing temp requires an interface to a chip that can interface to whatever you want it output to (be that software or hardware)...if you wanna avoid chips an output isn't doable...I'ma lurk this thread but I probably won't post much more

 

[I.M.O.G.]

That's probably a better characterization of the problem. I'd say thats a good delineator between casually interested project or really dedicated to doing something kinda of cool... if you want to get into interfacing with a basic processor.

 

[hokiealumnus]

Here's how I measure ambient temperatures for heatsink/water component testing:

For instance, I am measuring ambient using three thermistors, wired up to a Lab Jack U3-HV and polling every second right alongside Coretemp. So I have an ambient measurement for every core temperature measurement. (For the science-minded, the labjack measures voltage drop across a resistor, 10Kohm in this instance, then I use Excel to calculate the resistance from said voltage drop and use the Steinhart-Hart equation to calculate temperature from resistance.)

It requires a third-party part, but is more easily doable and you don't have to build one of those things by yourself, the possibility for which is doubtful without a solid background in somewhat sophisticated electronics.

The thermistors I use are actually from a Lamptron controller. you can of course get better than that if you want, and any two-wire thermistor should work, for instance I have a feeling one of these fittings would be good for measuring water temps in this manner.

The problem then becomes that you'll need some software expertise to enable you to use the labjack results realtime. It comes with software but I haven't delved into whether it can take action on a trigger. If it can, you could set a specific voltage measurement (which = non-extrapolated temperature) in the software to trigger an alarm through your PC.

Anyway...what I'm saying is your goals are very ambitious and from the sound of it, you'll need some help to get there. If you want to actually make something that can accomplish this, it's not going to be in the form of forum posts. Start reading electronics / circuit design / IC implementation & use books...lots of them.

I'd also stick with breadboard for now. As long as your solder joints aren't crap, it will last two years without issue.

 

[m0r7if3r]

I'd also stick with breadboard for now. As long as your solder joints aren't crap, it will last two years without issue.

You don't solder on breadboard, you're thinking of perfboard...or maybe printed circuit board. this is breadboard http://en.wikipedia.org/wiki/Breadboard

 

[hokiealumnus]

You don't solder on breadboard, you're thinking of perfboard...or maybe printed circuit board. this is breadboard http://en.wikipedia.org/wiki/Breadboard

Whoops...yep. Perfboard is what I was thinking. Thanks for the catch.

I would go with that over breadboard for the final design, even if you use breadboard for easy plug-and-play prototyping. It's not as complicated (or expensive) as a custom PCB and will last quite a while.

 

[Mr.DLucey]

@ Mr.DLucey- Thanks for the links, which method would you recommend, regarding lower cost/quality and time involved?

Time involved with out a drought "Modular IC Breadboard Socket", their $9.00 for a small one to $20.00 for a big one, easy fast make a mistake no big deal easy to fix.

Cheaper but takes a little more time to produce the part (my preferred method) "Pre-Punched IC-Spacing Perfboard" it's like $4.00. You solder everything up on the back

"Copper-Clad PC Board" is like $4.00 plus you got to get the special pen (Price?) plus PCB Etchant Solution $10.00, might as well go Modular IC Breadboard Socket when you add it all up. Most time consuming, make a mistake and you have to do a hole new board or start running jumper wires.


Oh, always use IC sockets and plug in the IC's after the socket is soldered up. (not required for the Modular IC Breadboard Socket)


All the above is available fron Radio Shack.
http://www.radioshack.com/category/index.jsp?categoryId=2032230

 

[Old H-Ware Tech]

These guys have pretty much got you covered, I just want to re-enforce 1 thing Mr D said. If you're new to soldering, and it sounds like you are, socket EVERYTHING! Avoid soldering directly on any component if you can and use a heatsink if you can't. You'll save yourself a lot of time trying to find problems that you caused yourself by overheating components. It's VERY easy to cook components without them showing any outward signs of overheating. Good Luck and have fun.

 

[canibalpenguin]

Sorry for digging up an old thread but I couldn't help it.

I'm surprised no one mentioned the arduino platform. As far as microcontrollers go I think it is by far the easiest to get going.
You could take care of most of what you want to do with a single at mega chip.
-Temp sensors, you can monitor these with the arduino, although I do not see a need for five as most tecs will only have a temp difference of around 30* any ways, your only real concern would be the cold side.
-the fans, easily done with a pot and an output to a transistor.
-screen and control, I as well would suggest not using a touch screen but you can use a serial lcd with the arduino output and just use knobs and what not for input.
-pump power, just a switch, you could use the same power button for the whole system but relays may be needed.
-temp alarm, programed into the chip and out put to a buzzer, led or whatever you want.

With 6 analog ins, 14 digital i/o, 6 of which are pwm out, I'm sure you can figure something out.

Here are some links for ideas.
http://fritzing.org/projects/arduino-temperature-based-servo-control/
http://www.element14.com/community/thread/7853
http://www.instructables.com/id/Control-your-arduino-from-your-PC-with-the-Qt-Gui/
http://forums.bit-tech.net/showthread.php?t=195753

You can use a breadboard to prototype and setup then make a pcb.

For the pcb you can draw the circuit with a sharpie or use the toner transfer method. For the etchant, well you can make that as well with hydrochloric acid and peroxide, do a google search. Instructables.com and the arduino.cc page are good places for info.

Enojy...

 

[ssjwizard]

arduino thats a name I haven't heard since the last time I was surfing hack-a-day. While I doubt this project ever got off the ground, thats a decent suggestion for anyone who might consider such a project.

 

[Bobnova]

It definitely could be done, the TEC itself can be switched by a couple (or few) mosfets via the Arduino, too.

I do love me some Atmel MCUs.