• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

TEC chilled water. need some help.

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

orionlion82

Member
Joined
Jul 20, 2004
After deciding that I DID need Sub ambient cooling, but DIDDNT want water cooled Pelts, I've zeroed in on a part that will make this fantasy a reality.

The Swiftech MCW-CHILL 452.

And - Thats where things get Very Very Complicated, and i start getting lost.

For Starters: The MCW-CHILL 452 has TWO Hot side loops, EACH of them will be well in advance of 226W of heat EACH.

I was thinking about a BIXIII and a D5/655 for each hot side.
Will that even be enough? if not then what?
on the other hand - could I get away with combining the two hot loops (parelell, not series) into one? if so can i get away with just one pump, and BIXIII- or still need two of each?

Now watch closley as the plot thickens:

The cold side loop will be cooling my CPU, GFX and NB.
im thinking the DDA/350 is better for the cold side.
probably a storm WB and all that...


all the cold loop will have to be insulated, along with the other stuff that needs it too.

oh, and what sort of a case ?
I have doubts about the U2-UFO for this...
we might be talking 2 BIXIII's and maybe another rad or two if needed.

how to controll it?
lots of solenoids?
High Amp DC switches?
regular 12V switches?

Any questions that I Should have asked but diddnt?
(this is for upcoming Conroe BTW)
TIA, because im officially lost, folks... :attn:
 
orionlion82 said:
After deciding that I DID need Sub ambient cooling, but DIDDNT want water cooled Pelts, I've zeroed in on a part that will make this fantasy a reality.

The Swiftech MCW-CHILL 452.

And - Thats where things get Very Very Complicated, and i start getting lost.

For Starters: The MCW-CHILL 452 has TWO Hot side loops, EACH of them will be well in advance of 226W of heat EACH.

I was thinking about a BIXIII and a D5/655 for each hot side.
Will that even be enough? if not then what?
on the other hand - could I get away with combining the two hot loops (parelell, not series) into one? if so can i get away with just one pump, and BIXIII- or still need two of each?

Now watch closley as the plot thickens:

The cold side loop will be cooling my CPU, GFX and NB.
im thinking the DDA/350 is better for the cold side.
probably a storm WB and all that...


all the cold loop will have to be insulated, along with the other stuff that needs it too.

oh, and what sort of a case ?
I have doubts about the U2-UFO for this...
we might be talking 2 BIXIII's and maybe another rad or two if needed.

how to controll it?
lots of solenoids?
High Amp DC switches?
regular 12V switches?

Any questions that I Should have asked but diddnt?
(this is for upcoming Conroe BTW)
TIA, because im officially lost, folks... :attn:

Please do not dodge the swearfilter again.

nikhsub1
Forums Moderator


MCW-CHILL 452 has to be the most worthless and idiotic contraption you can imagine. It consumes ungodly amounts of power and will give you bad temperatures.

Pelts have to be the most inefficient way to move heat. But wait it doesn't end there, you will need to cool their other sides. With what water? Ok so add a pump, more heat dump more power draw... Ok so you have your pelts you have them cooled, but wait they cant cool the water they have to cool a plate the other side of which cools the water. Oh wait now you have to move this water, so another pump, more power drain, more heat dump to kill your temperatures? You loose SO much every time you transfer heat though a medium and pass it to the next. And you are dumping heat back in to your loop.

This contraption is the most inefficient and ludicrous system I have yet seen. below 0C while cooling cpu, gpu and nb? You have to be joking.


You want advice, buy a small compressor on eBay and some copper pipe and make yourself a chiller that will use half the power and will actually be able to get you below 0C.

"Another item to note is that it's certainly possible to hold the hot side water temp to 35C or less with heatercores and 120mm fans. That means that even a 150W CPU (such as Prescotts) should easily be held to no more than 10C under load." -procooling.com review


By the way intel conroe and the amd am2 socket cpu's are just toys put out as openers for the new technologies, the real preformers will be comming out in Q1/Q2 2007. Save your penies or you will be disapointed.
 
Last edited by a moderator:
Great input there. (except i hope you dont kiss that mouth with your mother.)

ive got more to consider now.

This in particular cought my attention:

"By the way intel conroe and the amd am2 socket cpu's are just toys put out as openers for the new technologies, the real preformers will be comming out in Q1/Q2 2007. Save your penies or you will be disapointed."

i would like to know exactly what is meant by that? the only things coming out in Q1/Q2 - 07 are two dual core conroe's stick together for a four core hybrid with some tweaks - as far as i know anyway. im still planning on conroe, but ill keep a sharp eye to the ground and see what i can listen.

i am aware there are better means to the end i want, but considering the WAY i want it, this is more like the only way ive thought of so far that works for me. not saying there arent unthought ways yet but i would be lucky to see them thought in my own eyes.

i dont want a pelt on each chip i desire to cool.
i dont want phase change. especially not a system for every chip.
i dont want any compressor noise.
i dont really want FREEZING TEMPERATURES, but the colder - the better. ... sub-ambient is the point of the excersise - and not too much noise.
ive allways liked the idea of pelts, and have toyed with buying them on several occasions.
plain old watercooling would be great, but i live in texas and dont use A/C, so i gotta do something about high ambient temps. high end air barely, barely cuts it.

and if worse comes to worse, theres a radiator shop Less than a block away. can allways sell off the BIX III'(s) if i (buy them at all) dont need them anymore, and see what they can do for me about more surface area.

i think ive decided on combining the hot loops in parellell as (if it were ) in series, well that would be a bit foolish i think. dumping hot water into a pelt - not for me thanks.
that means one pump, one pump circut, less tubing and less confusion. more room in the case as well, and more (possible) room for radiators (single 120MM size) if needed.
generally expecting that the U2 UFO would do me right. not sold on it for sure yet though,
far too tired for more posting tonight.
 
Last edited:
orionlion82 said:
i would like to know exactly what is meant by that? the only things coming out in Q1/Q2 - 07 are two dual core conroe's stick together for a four core hybrid with some tweaks - as far as i know anyway. im still planning on conroe, but ill keep a sharp eye to the ground and see what i can listen.

Well I wont lie here, I am not a fan of intel so I dont keep up that much with their road map but from what I thought the Tigerton is what they have in mind.

And this is speculation with a touch of fanboism, but I get the feeling AMD will regain the crown they are about to loose with the K8L also in Q1/Q2 2007.

orionlion82 said:
i am aware there are better means to the end i want, but considering the WAY i want it, this is more like the only way ive thought of so far that works for me. not saying there arent unthought ways yet but i would be lucky to see them thought in my own eyes.

Well you wont get what you want this way. I thinking realisticly your temeratures with the huge efficiency loss coupled with pump heat dump will make you unhappy.


orionlion82 said:
i dont want a pelt on each chip i desire to cool.

Understandable, they are huge power drain.


orionlion82 said:
i dont want phase change. especially not a system for every chip.

Thats why chillers exist. Thats what you want essentially, except your chiller uses pelts to cool the coolant, which is a horrible way to do it cost/power efficiency/preformance wise.

orionlion82 said:
i dont want any compressor noise.

I have a few 1/6hp embraco reciprocating compressors here, they run so quiet I cant hear them over the ambient noises from my computer and other things. Only way I can tell its on is either if it is dead silent in the room and the computer is off, or by putting my hand on it and feeling the slight vibration. Damn thing runs quieter then any water pump I own, and you will have two pumps with your setup.


orionlion82 said:
i dont really want FREEZING TEMPERATURES, but the colder - the better. ... sub-ambient is the point of the excersise - and not too much noise.

So you mean you want to run below ~75-85F (I live in CA, pretty close, and thats what ambient air temperature indoors is from night to day in summer, dought TX is far off, probbably hotter). Pretty high ambient temperatures. Like ProCooling review noted I dont know if you will be able to keep the hot side cool enough for proper preformance. I dont think you will get the temperatures you want.

And like I said as far as noise goes, its all a matter of the compressor. Sure if you yank a 3/4hp rotary out of a window ac unit it will be loud and hot, but of you get a nice small reciprocating piston compressor, it will run cool and quiet.


orionlion82 said:
i think ive decided on combining the hot loops in parellell as (if it were ) in series, well that would be a bit foolish i think. dumping hot water into a pelt - not for me thanks.

You might want to try both ways if you still want to go though with this and not convinsed of the chiller advantages in pretty much every way except maybe size but its not that much bigger if you do things right.

Form what I have seen the flow loss is so bad in parallel every instance I have seen preformed worce then in series, but I am no wc expert, with chillers you odnt feel that as much. Try both ways.
 
good stuff to think about, again, thanks for the input. might have come more questions when i have time to think (lunchbreak)
 
Aha! you were right about the chips.
Tigerton has the "Bus-Per-Chip" thing happening, but thats the server market. probably wont see them on desktops for a while after that even. love the idea though, but for me i think its time for conroe.
This Celly D (while nothing to sneeze at) was just a temporary thing to get me from northwood to conroe. not big on prescotts, even though this one is decent.

anyways, back to the hot loop and removing the heat with high ambient.

now - the water will be hotter than ambient, and the heat will flow to the cool (er) air.
i think all it is is a matter of surface area. too little, and its a disaster - we allready know that. were just not sure really (me?? calculations??? naw...) how much is Enough?

in all reality, if not enough, then more. thats easy.
now that ive thought all about it, and the point has been raised, it seems now that ive got it all planned out that the only issue is cooling the hot loop.

the only problem is, with that philosophy, it could become "more and more and more" and pretty soon ive got radiators Dangling from the celieng and tubing everywhere, and i suddenly have a heated waterbed, and the DEA starts flying over my house looking at the heat signature, and i get offered the "commercial" billing system by the power company.... thats a bit more than "more" and it could be understated a bit by saying that would be -too much.

i can live with a couple of pumps and a couple of radiators to get sub-ambient. thats easy. (now that ive worked out the pelt system)
 
I think getting sub-ambiant with pelting the in the loop (rather then the normal on die method) would be extreamly hard, I don't think Ive seen it done (Ive seen some try and fail miserably years ago)

It might be a good idea to ask who has tried it and succeded (Ive seen plenty of people start threads saying they were going to do it, then once they march off not beliving people that it's a bad idea they don't come back to give results ;) )
 
greenmaji said:
I think getting sub-ambiant with pelting the in the loop (rather then the normal on die method) would be extreamly hard, I don't think Ive seen it done (Ive seen some try and fail miserably years ago)

It might be a good idea to ask who has tried it and succeded (Ive seen plenty of people start threads saying they were going to do it, then once they march off not beliving people that it's a bad idea they don't come back to give results ;) )



hehe, okay, one vote for "put up or shut up" ?? :santa:
i cant see how it wouldnt work with enough surface area. as long as theres a temperature difference, heat will flow to cold, allways.
same thing as what makes the wind blow - high pressure going to low pressure - allways.
long as i can get heat off the hot side, i think ill be all set.
heres the key:
the hot side will allways be hotter than ambient, therefore heat will flow out. more surface area - the easier it will be to have better temps.

EDIT: and even if the TEC cooling is a total disaster - a complete bust- then what do i do? sell, smash, or bury the tec and have a rediculously powerful W/C setup. i dunno, i just dont see how no one has had good luck with it.

UPDATE: placed my order seconds ago for the 452.
i am officially taking the plunge.
look for a project log at OC-rev.com coming soon.
 
Last edited:
I dont agree with "for servers only" comment. If its for sale its for sale, no one is telling you what to use it on. Second for server just means ita a better chip, normally. Look at AMD opterons, they are basicaly cherry picked x2's. These are the better chips bar none.

If you only knew how much better a chiller is, but have it your way :rolleyes:

I still think the results wont be good, but I am also qurious to find out how this will play out. Be sure to keep us updated, and lots of pictures please :)
 
orionlion82 said:
look for a project log at OC-rev.com coming soon.

No no no no, post us a link right here in this thread.

You never know.. we might not like the idea but we just might know something about the subject that might help ;)

I didn't come to the conclusion that a chiller is better without looking at the alternitives, and this was one of them ;)
 
here's my 2 cents from a guy who has pelt cooled for 3.5 years: a waterchiller is going to rack up your electric bill using all the pumps, the multiple pelts, the extra fans, etc. not to mention cost a lot more out of pocket to begin with in that you need enough gear for 3 loops (oh but luckily only 2 rads). You could buy an AC unit for all of this and still save money with better temps not to mention more comfortable in the end.

Since you're in a very hot environment and you want sub-ambient temps you can either A. Direct peltier cool or B. phase change.

You said you wouldn't do phase change so you'd have to direct pelt. Keep in mind that the hotsides need to be cooled about 2x2.5 times as much as a cpu based on their wattage output so yes at least one big radiator is necessary and you better put it in the shade, basement, or dungeon!

WCing the NB is pretty much a waste as well, purchase a swifty or thermalright customized cooler and you'll be all set...

GPUs generally run really hot, even with ambient temps of 100-110, if you WC them, you'll probably still not get up to their idle temps (unless you used power save features). Slap on an mcw50 or so, buy a regular heatercore and a fan or two and you'll have just about as good as you're going to get.

Back to the cpu and direct pelting.... the days of pelting are virtually over because: its expensive, its tough to power and cool the pelt, it's difficult to keep from frying your stuff eventually, and it reaps not the best results especially with the latest chips.

my real advice: abandon the project and get a processor that isn't very temp sensitive like an Athlon64 or Opteron. If it gets to 110 in your house, you still got another 30 degrees of headroom and c/w count until your processor goes glitch, and then you can always turn down the voltage and speed a tad.
 
nicspits said:
here's my 2 cents from a guy who has pelt cooled for 3.5 years: a waterchiller is going to rack up your electric bill using all the pumps, the multiple pelts, the extra fans, etc. not to mention cost a lot more out of pocket to begin with in that you need enough gear for 3 loops (oh but luckily only 2 rads). You could buy an AC unit for all of this and still save money with better temps not to mention more comfortable in the end.

Since you're in a very hot environment and you want sub-ambient temps you can either A. Direct peltier cool or B. phase change.

You said you wouldn't do phase change so you'd have to direct pelt. Keep in mind that the hotsides need to be cooled about 2x2.5 times as much as a cpu based on their wattage output so yes at least one big radiator is necessary and you better put it in the shade, basement, or dungeon!

WCing the NB is pretty much a waste as well, purchase a swifty or thermalright customized cooler and you'll be all set...

GPUs generally run really hot, even with ambient temps of 100-110, if you WC them, you'll probably still not get up to their idle temps (unless you used power save features). Slap on an mcw50 or so, buy a regular heatercore and a fan or two and you'll have just about as good as you're going to get.

Back to the cpu and direct pelting.... the days of pelting are virtually over because: its expensive, its tough to power and cool the pelt, it's difficult to keep from frying your stuff eventually, and it reaps not the best results especially with the latest chips.

my real advice: abandon the project and get a processor that isn't very temp sensitive like an Athlon64 or Opteron. If it gets to 110 in your house, you still got another 30 degrees of headroom and c/w count until your processor goes glitch, and then you can always turn down the voltage and speed a tad.

Another vote for a chiller lol :santa:

Days of water and pelt cooling are numbered. Air will be the stock cooling for a while but phase change, DD and chiller SS's are becoming so simple, small, convenient and offer incredible performance that its going to become what water became to air not so long ago.
 
.... i believe thermosyphons will be the compromise of choice as anyone buying a retail box is not going to like the phrase "physical maintenance".

But AMD and Intel aren't even going for speed king as much anymore, their head hit hurt from hitting that brick wall, and now you see things like Vivo and low power, etc. Everyone's focusing on mobility and versatility, that means low power and lots of cache and crap.

Personally, the PS3 (first game console to be keyboard and mouse compliant), if you add in web browsing and office capability is more appealing then the typical PC... imagine one that outperforms a PC in games!!! omg!!! I might.... just might.... buy the PS4 and not build a new computer.... AAAHHHHH!!!!
 
I wanted to buy one of these tec chillers but then I saw the price its like 200$+ and takes something like 400w+ of power.
Good Luck with that.
 
Epox4life said:
I wanted to buy one of these tec chillers but then I saw the price its like 200$+ and takes something like 400w+ of power.
Good Luck with that.

my point exactly, plus for all that money and trouble the preformance is very poor, but if he has his mind set on it lets see how it turns out :)
 
well, ive been hard at work this weekend.
ive got all my wires and switches made up waiting...
and i have a secret thats going to really really help.
 
Cathar was doing some research on this and he's found that running several pelts running at 1/2 rated voltage or less would create a system far superior to current pelt chillers when coupled with an optimized chiller block. Of course, he was aiming for temps slightly above the Dew point. Subambient temps are better attempted with a compressor based system. There are a few threads on his pelt experiments at OC.au.
 
no doubt, cathar is the man.
ive toyed with the idea of something like that, but time will tell.

i gotta run to the supply house and get a few more parts i miscalculated on - tomarrow, and then i think ill have my secret weapon ready. the prototype was a smashing sucess before i took it too far.

my tests on it today told me i got a bit happy with a few things and its time to re-do it properly.
its gonna be interesting.
 
Moto7451 said:
Cathar was doing some research on this and he's found that running several pelts running at 1/2 rated voltage or less would create a system far superior to current pelt chillers when coupled with an optimized chiller block. Of course, he was aiming for temps slightly above the Dew point. Subambient temps are better attempted with a compressor based system. There are a few threads on his pelt experiments at OC.au.

That is completly wrong. Do you know what ambient means, its the temperature of the surrounding atmosphere, which is indoors pretty warm normaly, 75-85F. So shotting for sub-ambient temperatures is not that hard, you can get that with an air cooled pelt.

If you are refering to the dew point as the point when condensation occurs thats very vague. You can have condensation at +40C. It all depends on the ambient temperature, surface in question temperature and pressure. Depending on where you live it can be at a very hot or very cold end of the spectrum.

If you want to go below 0C then thats when you want to use a compressor. Although I have to say again, whatever you want, a compressor baised system is better every time in every way.
 
Back