air cooled peltier setup

ok I've given up phase change for the time being I have a refrigeration book on order and I'll read that before I attempt any serious phase change projects But I just ordered a peltier and from what I hear you can get some very good temps... Now would sitting the peltier between my slk-800a and core work? its a very very small peltier 15mm by 30mm is there something I need to know? Any good sites you can link me to?

EDIT: Just rediscovered this... Reading this article, if you haven't already, is this most important and helpful thing you could do: http://www.overclockers.com/tips45/

What is the wattage of the TEC?

Your peltier needs to be able to transport enough wattage so that it acts as a thermal transporter and not a thermal insulator. If the TEC is not strong enough to do this, then the thermal energy will remain in the cpu die and your processor will quickly begin to overheat.

The common recommendation in the extreme cooling forum is usually that you need to get a peltier that is ATLEAST twice the wattage of your processor heat output - so if your processor outputs 60 watts of heat, you need atleast a 120 watt TEC. Remember that you will need to calculate the heat output of your chip if it is overclocked and that this method of estimating is considered a minimum requirement.

How are you powering the peltier?

If you need a dedicated power supply this article might be useful:
http://www.overclockers.com/tips260/

I have an air-cooled peltier. I dont have a thermal sensor on CPU so i dont know exact temperature but i will be getting one by the end of the week. I am using the swiftech MCX462+T which has a 226 watt pelt in it. The MCX462+T is dicountinued because it is being replaced by the MCX462+AT. The only diffrenc between +T and +AT is that +AT connects to socket lugs and not through motherboard. I personaly think thats bad becasue this heatsink weighs almost over 2lbs so lugs will break easily. Try and get your hands on an MCX462+T if you want air-cooled TEC.

the slk's base is WAY to small for a peltier. You need an alpha or a swifty along with a hefty chunk of copper for a cold plate to get that to work

Yeah, a cold plate is definatly neccesary. You dont want CPU and peltier to have direct contact. I dont know why but its supposed to be bad. TECs are awesome. I had a tester power supply and attached my TEC onto it. This test was done outside case. Within 2 minutes water started condensating on base. 1 minute later water starting freezing and turning into frost. 2 minutes later the whole base was full of frost. 2 minutes after that their was a base of ice under frost. 1 minute later their was kinda like snow falling off base becasue their was so much condensed water on it. This test was done with an ambient temp of like 90F(30C).

You do not want the peltier to directly contact the cpu because the coldside of the peltier is VERY inefficient at transfering heat across its surface - only a fraction of the TEC would be working to cool the CPU without a coldplate. A coldplate allows the entire coldside surface area of the TEC to contribute to cooling the CPU instead of just a small cpu die sized part of the TEC trying to cool the CPU - coldplates maximize the efficiency and cooling power of the TEC.

BTW steve, you need to have a clamping mechanism to hold the coldplate/TEC/heatsink assembly together. How you implement this is up to you to figure out, but a TEC requires an amount of clamping pressure that is far beyond that of the cpu core... so you can't clamp your heatsink down tight enough for the TEC without crushing your core. You need seperate clamps for HS/TEC/coldplate and to clamp that to the CPU.

You also need an appropriate power supply for your TEC. With more powerful TEC's this can be an expense problem if you want them to be run at Vmax - many PSU's cannot supply the amperage necessary and the ones that can supply enough cost some money.

There is a plethora of information about TEC's in the extreme cooling section. The forum search tool could be useful for finding information.

You also need to protect agaisnt condensation on CPU. You are gonna need a lot of dielectric grease and neoprene.

thanks alot for the response, I'm going to use my sk-7 instead the base is much larger than the slk-800. Heres how it is as i see it:

Core - Copper Plate - Peltier - Heatsink - Fan

is this correct? I have a fortron 530 watt power supply cant I just run it off that? For insulation where would I get neoprene or dieletric grease and how does one apply this? The clamping I can figure out

Well i doubt that powersupply can handel ur system and a 226 watt pelt unless it puts out like 40 amps on the 12v line. www.sidewindercomputers.com in the swiftech store under misc sells dielectric grease and spring retention mechanism to secure coldplate and pelt to heatsink. I dont know where you can buy neoprene gaskets but the gaskets are kinda like the old mouspad material this is kinda squishy and stuff.

I have to put the dielectric greas ein my pin grid array? that doesnt sound good... will that even stop condensation dont I want to stp condensation directly around the TEC

Yeah, just put grease all over the socket pin holes evenly, in a good 2mm thick coating or something. Then just smear it into the holes with your finger. It doesn't hurt anything, but prevents condensation inside the socket. Also put a light coating in the center of the socket, and then a tight square cutout of neoprene. I'd smear most of the board about an inch away from the socket too, just for security. And the back of the motherboard where the socket is too! You can use foam in that area. Becareful though, because most foams expand with a lot of force. It could break stuff. I bought some "non expanding" foam, which expanded at least a few mm after it dried. :\

rofl at the "non-expanding" foam does this dielectric grease stop condensation ? or just protect electrical parts against it? by the way the TEC i ordered is very very small only about 15mm by 30mm.. Where would I get neoprene? I found the dielctric grease thanks to constantinos but where is the neoprene? If I put a square of neoprene in the square in the middle of the socket wouldnt that stop my temperature diode from working?

would this work?

http://www.foamorder.com/products.cgi?cart=f2W0r0n11264189142n8B5b1V&item=neoprene

The grease just acts as an insulator to protect the electrical parts from moisture.

Yeah that neoprene should work They also make certain mousepads out of neoprene. I cut up my old allsop pad.

Is your pelt a 226 watt pelt? If so it will be running on the 12V line of the PSU, and if you look on the side of your PSU it should list it's amperage rating on that line. At 12 volts, a 226 watt TEC will draw something like ~20 amps. That means that you power supply needs to be rated for atleast ~20 amps to run your TEC, and yet a little more than that amount if it is going to be running your computer also. If you exceed the rated amperage for your PSU, you will not due so for long.

The kind of neoprene you want is closed cell neoprene - many mousepads are not made of this type of neoprene so they are not a good choice to use (You want an airtight material). I would guess that grade 2 neoprene is the type you want from that link you supplied. Check out www.directron.com if you want to also, they have neoprene.

The condensation proofing with dielectric grease may affect your thermal diode but that's not to worry - the thermal sensors in the socket are calibrated to account for airflow in the case so the temperature will already be messed up from everything else you are doing. If you want accurate temperature readings with this, it is a good idea to insert a thermal probe that touches the side of the die. You can post a thread, or do a search, about the best reasonable method of attaining somewhat accurate temperatures, and how to do this should be explained.

The dielectric grease is used to eliminate as much air as possible and limit air flow through your neoprene assembly. Air and air flow is the enemy within your condensation proofing. Air contains moisture, and if you have the right combination of air pressure, humidity, and temperature differential you will start to form surface condensation = bad.

Here is an old insulation guide by me:

Insulating your rig is really not a very large obstacle. Plan to insulate during a time when you have no other plans and will not be rushed for any reason, and getting it right will be easy.

Supplies:
You need neoprene, dielectric grease. and silicon.

Procedure: fill part of the inside of the socket with dielectric grease. insert a pad of neoprene on top of that to fill most of the area, top off inside socket with dielectric grease. cover back of die area on chip with dielectric grease. insulate the pins with dielectric grease (in holes of socket). cut neoprene to fit pretty snugly around socket. use silicone to adhere it to mobo. fill any gaps between neoprene and socket/mobo with dielectric grease, leave no air. put waterblock on and do the same around that as you did with the socket, leaving no air inside neoprene. leave no pathway, no matter how small, for air to travel from the case to inside the neoprene. take dielectric grease and cover the back of the mobo corresponding to the socket. take a pad of neoprene and adhere it to the mobo over the dielectric grease. leave no gaps for air. be VERY liberal with dielectric grease through the whole process. You shouldn't have any problems if you take your time. that's it.

Click to expand...

Here is some further information, these links are hella-useful:

Colin said:

Some links on pelts to help you learn.

Applying Peltiers

Peltiers in Practice

Picking a Pelt

Marlow Design Guide

Peltier Info

Overclockers Peltier Section

Kryotherm Pelt
Software

Click to expand...

And here's a little something about condensation that I wrote before:

IMOG said:

Condensation depends on surface temperature, air temperature, humidity, and air pressure. Air pressure is negligible if your region is relatively close to 1ATM/760torr. Condensation will/can only form on subambient surfaces.

This chart shows the surface temperatures (in black) that condesation will begin to form at:

Chart is by and from Bladerunner of these forums.

Click to expand...

Something that I didn't really mention, since you never answered my question about what wattage the TEC is that you will be using, is that you may not be able to cool a 226 watt TEC with a heatsink. Most heatsinks are designed to handle a heatload that is atleast below 120 watts with a reasonable to loud amount of airflow... with a 226 watt TEC, you will be trying to remove well over 200 to over 250 watts of thermal energy. That is a serious chore for ANY heatsink and will likely require a considerable amount of airflow. This is why watercooling is preferred for pelts - a large amount of high velocity airflow is not as necessary because of the larger heat exchange surface area of a heatercore.

Before investing more money into this, you may want to read some of these links and see if you can accomplish what you want to by doing things the way you have planned. You will likely be dissapointed with the results you will achieve.

I didn't want to be "that guy", and I hope this can work well enough to satisfy you, but I wouldn't want you to come back to the forums after you are done and be mad because no one warned you that you might end up with crappy temps.

Steve978 said:

ok I've given up phase change for the time being I have a refrigeration book on order and I'll read that before I attempt any serious phase change projects But I just ordered a peltier and from what I hear you can get some very good temps... Now would sitting the peltier between my slk-800a and core work? its a very very small peltier 15mm by 30mm is there something I need to know? Any good sites you can link me to?

Click to expand...

Bad idea.



what wattage is your pelt?

if your small pelt fits on the base of the slk800, then fine, otherwise no good..

you gotta use cold plate..

aircooled pelt is never a great solution. The heat will build up eventually and will lose overclock gradually. You won't be able to run at load at high clocks for long time.

supercooling the pelt is the way to go.

umm screw that chart.. the cold plate will get cold and create condensation but that is why you suppose to seal that up all the way. Other than that, his cpu is not gonna get cold enough to create condensation by aircooling the pelt. Ain't gonna happen. Take that lousy chart away, hahaha

I have an aircooled TEC and i am pretty sure it works fine. I dont know temps cause i dont have probe on CPU but i ordered one and should be here next week but i am thinking temp idle is at about 16C

Pikachu_Mommy said:

umm screw that chart.. the cold plate will get cold and create condensation but that is why you suppose to seal that up all the way. Other than that, his cpu is not gonna get cold enough to create condensation by aircooling the pelt. Ain't gonna happen. Take that lousy chart away, hahaha

Click to expand...

Your mostly right, the chart is pointless if he does a good job insulating... On the other hand however, it shows why he needs to go through the pain of insulating with conformal coating/dielectric grease and neoprene. It's a PITA but its wise to be careful, hence the chart for justification and general knowledge.

More importantly than the insulating issues however, it is highly unlikely that his setup will create a temperature differential that would cause condensation.

Supercooling a pelt is a "bad idea". If you are going to supercool, then you should directly supercool the CPU die. Peltier efficiency gets progressively worse at subzero temps; if you look at some of the equations in the links I presented earlier you can see why this is. If I remember correctly, I think it is the seebeck coefficient you need to examine to understand why what I say is true. I had wanted to look further into this around the time when I first joined the forum, but it gets far more complicated than is fun. If you don't like things that aren't a lot of fun, just take my word on this.

BTW, I forgot to mention that you should get a "potted" peltier... These peltiers are sealed so that no moisture gets between the hot and cold side of the TEC and shorts the device.

EDIT: The electrical power generated in a TEC is a result of the seebeck effect. The seebeck coefficient is adversely affected with a temperature drop below zero, and the further you go below zero, the less and less powerful your pelt becomes.