A TEC assisted CPU waterblock to prevent moisture build up and a TEC failure disaster.
A TEC assisted CPU waterblock to prevent moisture build up and a TEC failure disaster.
I think I see what you are going for here. Are you planning to use two loops?
This seems overcomplicated and inefficient to me. I do like it when people think up something original though.
Isn't condensation still going to be an issue since the chiller water is making contact with the metal exposed to the CPU which is exposed to the outside?
[ 3770K @ 4.2 : H100i : ASRock Z77E-ITX : GTX560 Ti : 16GB DDR3 1800 : +4TB : Bitfenix Prodigy : 2x Dell S2340M : Filco Majestouch-2 [Cherry Brown] : BX8a Deluxe]
I think the water at room temperature doesn't have the time to get chilled.
The CPU is hot and it only gets chilled at the center where there is immediate contact between the central copper piece of the 4th plate and the third plate.
Scarlet Infidel
No, One loop is enough. A splitter in and a join out.
This way the water never gets too cold.
Last edited by Marios; 03-01-2009 at 02:14 PM.
So the same water loop is in contact with both the cold and hot side of the TEC?
Yes. Room temperature water enters both chambers and gets out to enter the same loop. The CPU gets chilled from direct copper contact mainly.
The first part is delrin and other three parts are copper.
The last one (on the CPU) is only trimmed at three sides just to fit the 3rd plate and to create a path for the water.
This way it is easy to drill holes near the base of the plate.
Last edited by Marios; 03-01-2009 at 02:41 PM.
How will you make sure the water is at room temperature? The net effect of the peltier here it to increase the water temperature at the expense of electricity.
If you power a TEC with no cooling on the hotside within seconds it will burn out because the room temperature on the hotside is insufficient to cool it.
If you plan to cool a TEC with room temperature water you are pretty much doing a similar thing !!
It will cool it to a small degree due to the water movement but not enough to prevent it's demise.
The difference between the hot and cold sides will be too large.
The TEC will heat the water in the top of the unit you will not be able to maintain it at room temp so your loop will just get warmer and warmer. This happens in all single loops that pass both the cold and the hotside of a single TEC. This will also cause the TEC to burn out if you try to cool it with warm water. This is doomed to fail.
Result - dead TEC.
One hell of a lot of work for nothing......
To prevent condensation...simply insulate.
To prevent CPU loss due to TEC failure...simply set a temp limit in the bios or in a program such as realtemp or coretemp to turn off the machine when TEC fails...
Much easier than what your proposing...
Last edited by zipdogso; 03-01-2009 at 05:02 PM.
The water from the lower chamber (over the CPU) should keep its room temperature at the outlet.
Both the CPU and the TEC try to alter the water temp.
In case we give too much power on the TEC, the water in the lower chamber gets colder, but the water in the upper chamber gets hoter.
The water from both chambers' outlets mix together.
Then they enter the radiator.
Low noise fans on the radiator should be enough to keep the water in room temp.
Though I might insulate the middle of the CPU slot.
Just in case.
Water at room temp is enough to cool a TEC.Originally Posted by zipdogso
Last edited by Marios; 03-01-2009 at 04:20 PM.
You won't keep the water at room temperature (it's technically impossible with the setup you describe).
The only way the peltier could possibly reduce the CPU temperatures is if the heat conduction formt he peltier through the copper is greater than the heat conduction into/out of the water just above the CPU. This means when the peltier is off the block would need perform very badly to even get the design to work. Overall it will just give worse temperatures I'm afraid.
So you plan to underpower the TEC. (that's why room temp water will cool it yes ?)
Because your hotside cooling will be insufficient, if it doesn't eventually kill the TEC, IT WILL HAVE A HIGH TEMP. DELTA. ie the difference between the hot and coldsides. At a high temp delta your heatpump will be minimal and will not effect much cooling so the hot CPU (supposedly kept in check by the TEC.) will heat the water in the loop and your loop will get warmer and warmer.
Depending on you underpowering of the TEC (less than about 60%)with a high temp. delta the TEC will not even work.
Trust Scarlet and me your plan is a no-go.
Room temperature coolant will cool a TEC but insufficiently (that's the point I did'nt get across.) with your one loop it will not remain room temp it will get warmer.
You said "In case we give too much power on the TEC, the water in the lower chamber gets colder, but the water in the upper chamber gets hoter."
The problem is the water in the upper chamber will get very much warmer than the lower chamber will chill - in general the heat given off by the hotside will be at least 25% more than the amount of cooling from the coldside.
Your loop will get warmer how ever you look at it but really it is all a moot point because to achieve your planned result ....
To prevent condensation...simply insulate.
To prevent CPU loss due to TEC failure...simply set a shutdown temp in the bios or in a program such as realtemp or coretemp to turn off the machine when TEC fails...
Last edited by zipdogso; 03-01-2009 at 06:11 PM.
Scarlet Infidel
Even simple - unaided watercooling is enough to make a TEC work well, as we can see from other projects.
With a 45°C Delta, we need to keep the hot area at 50°C to get 5°C on the cold area.
Water is power.
With this waterblock halve the water flow through both CPU chambers is hot, halve is at room temp and they mix together.
Then we also have the refrigerator to further cool the water.
This helps to achieve much lower tmps at the CPU area with much less moisture build up and much less noise than regular designs.
And it even cools the CPU in case of a TEC failure.
No need to shut down immediately to avoid a disaster and restore the failed TEC.
Though in case of a watercooling failure (ie pump) we do have a big big problem.
The reason I wrote this post here is to avoid someone elso come and patent the same idea.
It is now an open project for anyone to try and build.
What is your avatar about?
It looks like a peltier sandwich with two waterblocks.
Last edited by Marios; 03-02-2009 at 12:54 AM.
The avatar is a model of a peltier water chiller I started making. The idea of it was to be cheap and low power and not to cool the water beyond the point where condensation occurs. The design had too many compromises really and I never finished it. Here is the thread if you want more detail.
I don't want to put you off innovating and coming up with new designs and I don't want to be rude. I will not pursue this any further but to say the design just will not work like you want it to. Feel free to make one and prove me wrong.
Last edited by Scarlet Infidel; 03-02-2009 at 02:48 AM.
OK I see...
My feelings also....but there are one or two points in your last post which i don't think are right
1.)
"With a 45°C Delta, we need to keep the hot area at 50°C to get 5°C on the cold area. " Here you have tailored the figures to fit your ideal, unfortunately it is rather fanciful, in real life it won't happen. without any apparent heat attached to the coldside (It looks to me like there is an intended gap between plates 3 - 4.) the coldside will be colder than you imagine and with only minimal cooling on the hotside (room temp coolant) I can see a temp delta larger than 45º. Even if you theory was correct (45º delta.) it would result in a heatpump of only 30% Qcmax or less ie very little cooling.
2.)
"Then we also have the refrigerator to further cool the water."
What refridgerator ?? The TEC is merely a heatpump taking heat from the lower chamber and dumping all that heat plus the heat generated internally by the TEC into the the top chamber it is a heat mover it does not actually do any heat lowering in fact as I already said it creates more heat in use. This is why using only one loop will just get warmer and warmer.. The heat will only be removed from the system if you use a dedicated hotside loop to do so.
I realize that this is your "baby" but you have blindfolded yourself in your willingness for this to work.
why not stick something like glycol between the CPU die or IHS and the TEC then cool the top with a normal water cooling loop. The problem with your current design is that TECs will also produce more heat than the amount of energy they can displace. So no matter what if you had a closed loop between hot and cold side it would always heat up. I realize that you have probably tried to think this through but I think you need to read up more on how TEC's operate and function.
I though it was something like that.
HAVE NO FEAR!
"AMD fallen angel"
You didn't get the memo? 1 hour 'Fugger time' is equal to 12 hours of regular time.
The hot side is the fridge of the watercooling loop. There we have fans to remove excess heat. There we not only have warm water going in from the upper chamber of the "TEC assisted CPU waterblock" but cold from the lower chamber too.
There are no gaps between the plates. I left them just for illustration purposes. Though we should use thicker rubber to the full extend between the second and third plates to prevent heat transfer to the lower chamber.
Last edited by Marios; 03-02-2009 at 10:03 PM.
if you're going to have 2 different heat signatures to cool and one of them is optional .. means you can switch on and off the TEC rite?
maybe you need to incorporate heatpipes, it's technically purchasable .. and it's quite easy to bend .. i broke a few .. lol
what do you think?
it's just a random brainstorming
but keep in mind .. watercool isnt bout 1 pass of water .. it requires .. lotsa loops
I read most of your project thread... shame you didn't finish it.It looks like it would work at least a little... were you so convinced that it would not work? You never say why in the linked thread.
Also has anyone else tried to make a tec waterchiller to get sub ambient temps... obviously Coolit has with their products... My dilema is... I am desperatly seeking to find or make a stand alone tec watercooler or tec assisted radiator system to cool an i7 below at or below ambient. The Boreas unit from Coolit breaks even on the i7 and water cooling surpasses it at a certain point of overclocking... I'll see if I can find the thread.
Seems to me... there is a market for a sub-abient cooler/motherboard combo. Couldn't the motherboard manufacturers make a MB that would be protected from condensation? There is a need here that is not being filled and I'm convinced that it is totally possible!
at least a cooler that takes the cpu down to or just a hair above ambient... Please!
The chiller in the thread was mainly a cheap and easy to build prototype so I could experiment without wasting too much money. The main reason I stopped work on it is simply that I never have time and I am rarely at home where my tools are (I'm a University student).
I have recently starting looking into parts for a more expensive but better chiller based in the same principles and using many of the same parts. I hope to have a few days free over Easter to get something going. I definitely something like an improved Boreas would keep many people happy. Unfortunately copper and TEC prices seem to have gone through the roof since I got mine and I believe that using lots of TECs is the key.
Getting sub ambient temps with a TEC is not a big problem....I read most of your project thread... shame you didn't finish it.
Also has anyone else tried to make a tec waterchiller to get sub ambient temps... obviously Coolit has with their products... My dilema is... I am desperatly seeking to find or make a stand alone tec watercooler or tec assisted radiator system to cool an i7 below at or below ambient. The Boreas unit from Coolit breaks even on the i7 and water cooling surpasses it at a certain point of overclocking... I'll see if I can find the thread.
Seems to me... there is a market for a sub-abient cooler/motherboard combo. Couldn't the motherboard manufacturers make a MB that would be protected from condensation? There is a need here that is not being filled and I'm convinced that it is totally possible!
at least a cooler that takes the cpu down to or just a hair above ambient... Please!
The problem is every Tom, Dik, Harry and his dog wants to stay above the dew point and fine control like that is a problem....not unsurmountable, but a problem none the least.
If you just want to cool and keep going sub ambient and depending on design etc even sub zero (no silly figures though - it's not phase !), within reason that is what you use TECs for. The coolit and boreas units use very small TECs undervolted....they have to because the current limit on a molex is for safety 7-8amps. To make their units available and easy to connect for all they have to use molex connection.
Build the chiller yourself and you no longer have this restriction you can undervolt high power TECs to get the efficiency and then multiply them to get back up to a sensible heat pump. So in general you would use at least 4. Theoretically you could build a coolant chiller for any heat load so i7 and high end cards are no longer a problem...HOWEVER you need to know a little about TECs, small electrical circuits, heatflow etc and be reasonably able on the DIY front.
This is a relatively "new avenue" so it helps if your a bit innovative we have yet to see designs that work well.
Once people stop being so concerned with the dew point and just do the sensible thing and insulate then perhaps we can get on and find some good designs for TEC water chillers.
I think you should separate cpu cool and tec cooling.if you're going to have 2 different heat signatures to cool and one of them is optional .. means you can switch on and off the TEC rite?
maybe you need to incorporate heatpipes, it's technically purchasable .. and it's quite easy to bend .. i broke a few .. lol
what do you think?
it's just a random brainstorming
HAVE NO FEAR!
"AMD fallen angel"
You didn't get the memo? 1 hour 'Fugger time' is equal to 12 hours of regular time.
In my system aiming for the water to be at ambient or just below (dew point) was for a couple of reasons. Firstly so I don't have to worry about condensations as you say. However the main reason was purely because I wanted a power efficient system and the cooler you go the more power you use and this seemed like the best trade-off.
In that case I feel your concerns were a little misguided.
Using multiple undervolted TECs (which achieves a good COP.) going sub zero doesn't affect things greatly. If you look at ferrotec performance charts and start by typing a reasonable hotside temp say 30ºC pick an input power (undervolt to 25% or less.)and average delta, then using the same charts enter -15 and recalculate and use the same input power and delta. The current draw rises slightly and obviously as a result of this the COP reduces slightly but it is fairly negligible. A COP of least 2 is still achievable.
Of course in the normal course of things you would have a job achieving a hotside of -15 i an really only saying this to illustrate a point that your concerns are not necessarily correct.
Last edited by zipdogso; 03-11-2009 at 03:00 AM.
Posting Permissions
Reply With Quote
Bookmarks