new idea, need suggestions

Mercury · 05-31-2003, 05:07 PM

so weve all seen direct die cooling, but what about direct-TEC cooling? heres my idea, u think it would work?

hara · 05-31-2003, 05:21 PM

It's been done before, by swiftech I think.

Mercury · 05-31-2003, 05:23 PM

you do understand, not WB cooled tec, but DIRECT TEC COOLING... as in the water actually hits the hot side of the tec..

if this is what your talking about, feel free to give me a link, i want to see what they did

Since87 · 05-31-2003, 05:24 PM

Swiftech is already doing it. I haven't seen any information on how well it works.

A lack of mechanical support pressing the TEC to the coldplate may be a problem.

Mercury · 05-31-2003, 05:28 PM

the picture doesnt quite reflect my idea, heres a better one

the pyramid opening for the water flow would only cover most of the TEC, the edge would be against the plexiglass. a lateral space would be all around the TEC, so that the space is wider than the copper cold plate. this way, i could mount bolts down the side, and pin the TEC in between the coldplate and plexi rim

surlyjoe · 05-31-2003, 06:04 PM

http://www.spodesabode.com/content/article/ppp

Mercury · 05-31-2003, 06:07 PM

all that is is a guy who messed up a direct peltier cooling, i want some detailed pics and results of someone whos done this :shrug:

winewood · 05-31-2003, 06:53 PM

The conclusions: Bigben2k was right on. A tec needs pressure to cool properly. Without having it sandwiched in between to hard surfaces, you loose efficiency AND the TEC would expand and contract. Much use would render this thing in pieces or just ruined. However, one way to proove that wrong it to make one. Why don't you?

Mercury · 05-31-2003, 07:43 PM

im not exactly the richest person, it cost money for the materials, since due to the strange design, everything has to be custom made :shrug:

RoboTech · 05-31-2003, 09:39 PM

Hey Mercury,

The Swiftech MCW50-T is almost exactly what you have drawn. It uses a water-cooled 80 watt Peltier for graphics processors. The hot side of the Pelt is in direct contact with the cooling water. The Swiftech water jacket is square instead of pyramid shaped. It works quite well...

I just finished a review of the MCW50-T which should be posted this weekend over at SystemCooling.com

winewood · 05-31-2003, 09:51 PM

I agree that it may work good for a graphics pelt, or something that could require little assistance. I understand that the heat given off by a cpu would greatly increase the heat variation, and therefore the expansion/contraction. Perhaps mercury could buy one of these stifty's and put a 226W pelt?

That may be less costly and provide the answer? I look forward to hearing the conclusions. I'm not rich enough to roll the dice on that guess though.

surlyjoe · 06-01-2003, 01:52 AM

Quote:

Originally posted by Mercury
all that is is a guy who messed up a direct peltier cooling, i want some detailed pics and results of someone whos done this :shrug:

well, its been tried and didnt work all that great is all I was saying

05-31-2003, 05:21 PM	#2
hara Cooling Savant Join Date: Oct 2002 Location: Malta, Mediterranean Posts: 662	It's been done before, by swiftech I think. __________________ - Every great HD crash day is the day before back-up day. - My Past System - "Better to reign in Hell than serve in Heaven." - Milton, Paradise Lost. - FMZ

05-31-2003, 06:53 PM	#8
winewood Cooling Savant Join Date: Nov 2002 Location: in my chair Posts: 574	The conclusions: Bigben2k was right on. A tec needs pressure to cool properly. Without having it sandwiched in between to hard surfaces, you loose efficiency AND the TEC would expand and contract. Much use would render this thing in pieces or just ruined. However, one way to proove that wrong it to make one. Why don't you? __________________ -winewood-

05-31-2003, 09:51 PM	#11
winewood Cooling Savant Join Date: Nov 2002 Location: in my chair Posts: 574	I agree that it may work good for a graphics pelt, or something that could require little assistance. I understand that the heat given off by a cpu would greatly increase the heat variation, and therefore the expansion/contraction. Perhaps mercury could buy one of these stifty's and put a 226W pelt? That may be less costly and provide the answer? I look forward to hearing the conclusions. I'm not rich enough to roll the dice on that guess though. __________________ -winewood-

05-31-2003, 05:23 PM	#3
Mercury Cooling Neophyte Join Date: May 2003 Location: Seattle, WA Posts: 20	you do understand, not WB cooled tec, but DIRECT TEC COOLING... as in the water actually hits the hot side of the tec.. if this is what your talking about, feel free to give me a link, i want to see what they did

05-31-2003, 05:24 PM	#4
Since87 Pro/Guru - Uber Mod Join Date: Sep 2002 Location: Indiana Posts: 834	Swiftech is already doing it. I haven't seen any information on how well it works. A lack of mechanical support pressing the TEC to the coldplate may be a problem.

05-31-2003, 06:04 PM	#6
surlyjoe Cooling Neophyte Join Date: Nov 2001 Posts: 38	http://www.spodesabode.com/content/article/ppp

05-31-2003, 06:07 PM	#7
Mercury Cooling Neophyte Join Date: May 2003 Location: Seattle, WA Posts: 20	all that is is a guy who messed up a direct peltier cooling, i want some detailed pics and results of someone whos done this :shrug:

05-31-2003, 07:43 PM	#9
Mercury Cooling Neophyte Join Date: May 2003 Location: Seattle, WA Posts: 20	im not exactly the richest person, it cost money for the materials, since due to the strange design, everything has to be custom made :shrug:

05-31-2003, 09:39 PM	#10
RoboTech Cooling Savant Join Date: Sep 2002 Location: Cincinnati, OH Posts: 229	Hey Mercury, The Swiftech MCW50-T is almost exactly what you have drawn. It uses a water-cooled 80 watt Peltier for graphics processors. The hot side of the Pelt is in direct contact with the cooling water. The Swiftech water jacket is square instead of pyramid shaped. It works quite well... I just finished a review of the MCW50-T which should be posted this weekend over at SystemCooling.com

Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)