Using TEC to chill after rad?

[silverphoenix]

I've been throwing the idea around for possibly using 1-3 low power TEC's to chill the liquid after the rad, mebe just a few degrees under ambient if possible or closer to ambient. Now I haven't done the math yet but wanted to get any opinions. A heatercore will be handling the majority of the heatload, the TEC's are just there to "chill" the water.

[[XC] riptide]

Whats gonna cool the TECS?

[silverphoenix]

I was hoping to use air to cool let's say 80Watt tecs.

[littleowl]

tec and air don't go good together.

[silverphoenix]

tec and air don't go good together.

well I can see that if ur directly cooling the cpu, but a good hsf should be able to handle about an 80watt TEC right?

[AliG]

I agree, stay away from tecs and air unless you plan to get the tecs heatsinks, and even still they better be good heatsinks as the tec's will burn themselves out. I would avoid trying a coolit type freezone/eliminator type thing, if you want subambient, go with a true chiller otherwise save your money

[littleowl]

well I can see that if ur directly cooling the cpu, but a good hsf should be able to handle about an 80watt TEC right?

no!

[Fraggle]

The idea is flawed anyhow; the radiator would be fighting the tecs even if they had the power to chill the water which they wouldn`t. The order of components is irrelevant in a closed system. If you must use a tec then put it in between the waterblock and cpu :up:

[littleowl]

The idea is flawed anyhow; the radiator would be fighting the tecs even if they had the power to chill the water which they wouldn`t. The order of components is irrelevant in a closed system. If you must use a tec then put it in between the waterblock and cpu :up:

you can make a tec chiller that will work better then directly on the cpu but your not going to get sub and it uses so much electric that you might as well make a chiller system or single stage.

[Fraggle]

To be sure but he`s talking one or two 80W tecs which ain`t enough as I`m sure you`ll agree.

[littleowl]

To be sure but he`s talking one or two 80W tecs which ain`t enough as I`m sure you`ll agree.

yes I do agree. That will not do! :yepp:

[emaniac]

hmm, i tried something similar and gained couple of degs C, but decided not to build it permanently in my rig :
http://www.xtremesystems.org/forums/showpost.php?p=2306332&postcount=13

TEC is rated 12V/5A and of ~45W usable power

[initialised]

The idea is flawed anyhow; the radiator would be fighting the tecs even if they had the power to chill the water which they wouldn`t. The order of components is irrelevant in a closed system. If you must use a tec then put it in between the waterblock and cpuThat's dogma, not strictly true. Yes the radiator has to shift all the heat but the water coming out of the radiator is colder than that going in otherwise what's the point of having a radiator?

I see it as Pump -> Radiator -> TEC (Cold) -> CPU (etc) -> TEC (Hot) -> Pump

The TEC cools the water, this cools the CPU, the TEC heats the water, the radiator dumps the heat into the air.

The problem you are likely to face is getting the flow rate right and fan cfm balanced. Basically your radiator/fan has to shift all the heat that the pump, TEC and CPU dump. which is also the case in direct die. However, while this method might be less efficient than direct die it does allow you to use a TEC with a Qc which is lower than the processor's TDP.

I believe that the DELL H2C system works like this.

[nubtrash]

That's dogma, not strictly true. Yes the radiator has to shift all the heat but the water coming out of the radiator is colder than that going in otherwise what's the point of having a radiator?

I see it as Pump -> Radiator -> TEC (Cold) -> CPU (etc) -> TEC (Hot) -> Pump

The TEC cools the water, this cools the CPU, the TEC heats the water, the radiator dumps the heat into the air.


I don't think it will work. Water has a high heat capacity so there is not much of a temp delta between any two points in the loop once it reaches thermo equilibrium. Check this link out:

http://www.xtremesystems.org/forums/showthread.php?t=77260


Look at the rad input and output temps there is only about .5 degee difference.

[[XC] riptide]

I don't think it will work. Water has a high heat capacity so there is not much of a temp delta between any two points in the loop once it reaches thermo equilibrium. Check this link out:

http://www.xtremesystems.org/forums/showthread.php?t=77260


Look at the rad input and output temps there is only about .5 degee difference.

Time for a measurement eh? Rough measurement... Inlet barb mount = 26.3C
Outlet Barb Mount = 25C. This was measured about a 1/2 inch back from the barb on the outside of the RAD with an IR gun. System E6850 @ 4.0Ghz, 120.2 RAD 4 120 fans (push Pull) + APogee GT Block, + DCC pump.

[initialised]

I don't think it will work. Water has a high heat capacity so there is not much of a temp delta between any two points in the loop once it reaches thermo equilibrium.http://marci.over-clock.com/tec/teclatest/tbanlcd.jpg What this read out shows is that the airflow through the radiator is not sufficient to cool the loop or that the loop is running at or close to it's minimum temperature.

dT(Rad) is determined by Water Flow, Air Flow, turbulence (pressure) and Thermal Resistance. Any closed loop liquid cooling system works by taking heated liquid from the cooled element, dissipating the heat from the liquid in a heat exchanger to another medium then sending more cooled liquid to the cooled element. If you are not seeing a big difference between inlet and outlet under load then the cooling system is inefficient or it is nowhere near its dissipation capacity.

Relatively inefficient cooling systems are perfectly adequate for cooling todays PCs, the industry is in it's infancy and often seems to be trying to reinvent the wheel.

I like Dell's approach, it's control system varies the TEC, pump and fan voltages simultaneously much like the cooling systems in engines where air and water flow are roughly proportional in a moving vehicle.

[NaeKuh]

omg how funny, i was just having this converstaion with owl not too long ago.

Well, a forum member, martin, iis going to building me the heat exchanger from scratch.

Basically it will have room to clamp 3x60mm tecs, have have 2 plates that need to be h2o cooled.

So i will be running 3 indpendant loops. Just adding 1 more to my exsiting rig.

Loop 1: Iwaki RD-30 -> PA120.3 -> TEC Prechiller -> CPU -> EK res.
Loop 2: D5 -> MCR220 -> EK FC -> EK FC -> EK res.
Loop 3: DDC 3.2 -> MCR120 or PA120 or PA160 -> TEC Prechiller -> DDC-3.2


The hot side will be on its own loop. I dont want to dump the heat back in to the main loop. The cold side will be in contact with loop1.

Im guessing 3 80W tec's should do it no?

[Serpentarius]

i tot the higher the watts on tec can divert more heat? 226w best?
i recall someone used a rheostat to adjust its voltage

[Duh]

tec and air don't go good together.

there are lots of devices that depend on the air cooling of a low wattage tec and have never failed..


The idea is flawed anyhow; the radiator would be fighting the tecs even if they had the power to chill the water which they wouldn`t. The order of components is irrelevant in a closed system. If you must use a tec then put it in between the waterblock and cpu :up:

wrong!

I see initialised pointed you the right way.. there seems to be a pre-negative thought in this section by some members..

[littleowl]

there are lots of devices that depend on the air cooling of a low wattage tec and have never failed..


only tecs below 80w and they are not worth the trouble when the amount of heat we are talking about!

Also the units your talking about most really good air coolers can do better then they do!


i tot the higher the watts on tec can divert more heat? 226w best?
i recall someone used a rheostat to adjust its voltage
yes and no. the bigger the tec the more heat it can pump out but also the more heat it will push into your water loop. The 226w tec is the biggest 50mm X 50mm but if you move up to the 62mm tec you get much better cooling.
if your using a meanwell power supply you can adjust the voltage through that. If your trying to run the tec on your pc psu your at your own risk because they can catch on fire!!

[Duh]

only tecs below 80w and they are not worth the trouble when the amount of heat we are talking about!

Also the units your talking about most really good air coolers can do better then they do!


yes and no. the bigger the tec the more heat it can pump out but also the more heat it will push into your water loop. The 226w tec is the biggest 50mm X 50mm but if you move up to the 62mm tec you get much better cooling.
if your using a meanwell power supply you can adjust the voltage through that. If your trying to run the tec on your pc psu your at your own risk because they can catch on fire!!
thats an opinion that does not lead members in the right direction. This is the TEC section. You seem to be against every single topic that has to do with TECS man. Not that i pretend that you lie but you are with the pre-negative piece of advice..


its not efficient
it wont chill as a single stage
it wont chill like a ln2 pot
it is expensive
it will raise your power bill
its not easy to deal with
if you are not careful you can light up your home


dont you get tired of posting the same stuff??

no offense intended but you should consider it.. :)

regards

J

[[XC] riptide]

Oooh..

Anyways.... Weren't there TEC coolers with delta fans out years ago?

[littleowl]

thats an opinion that does not lead members in the right direction. This is the TEC section. You seem to be against every single topic that has to do with TECS man. Not that i pretend that you lie but you are with the pre-negative piece of advice..


its not efficient
it wont chill as a single stage
it wont chill like a ln2 pot
it is expensive
it will raise your power bill
its not easy to deal with
if you are not careful you can light up your home


dont you get tired of posting the same stuff??

no offense intended but you should consider it.. :)

regards

J


yes I do get tired of posting the same stuff, and if that is a problem go ask nol why he don't post much in here now days! Tec is very dangerous and people that don't understand how they work need to know the bad things as well as the good things!

But for your info I love playing with tec's. If some one has a question I will answer to the best of my knowledge!!!

If you have a problem with my posts then please be sure to turn them all in to admins ASAP!! If I give bad advice or anything like that I would like to be in trouble for it!


riptide;2572438']Oooh..

Anyways.... Weren't there TEC coolers with delta fans out years ago?

yeah I think I remember them.. but I beleave they were only 50w tecs I cant be sure though

[jinu117]

I've tried exact concept years ago. Questionable thing would be, efficiency more than the concept. Basically what you are trying is chilling with radiator in loop. By taking some of work radiator can do through tec, you are decreasing amount of heat dump radiator can do (ambient is as close as it will get even in theory), reducing it's efficiency. It quickly becomes moot point unless you are really going full blown on tec using easily 200+w setup (and if you ever get below ambient temp on water line, radiator will act as heat source too... :P)
Honestly, too much trouble but no one is stopping you so spend your money and experiment :) Best way to learn.

PS) Oh... yes there was tec air cooler... :) Swiftech used to make them and it was 80w tec if I remember it right. Don't think it ever succeeded in market :P

[Duh]

@owl:sorry man for getting pissed off..should have handled the situation via PM

I can delete my posts if you want :)

Sorry again

J

[littleowl]

@owl:sorry man for getting pissed off..should have handled the situation via PM

I can delete my posts if you want :)

Sorry again

J

your post is fine! :D you are just stating your feelings! I do get tired of telling everyone the same thing over and over but if they ask a question I will answer to the best of knowledge... If I don't respond and someone want my opinion just send me a PM I will respond!

Sorry Duh if you don't like my posts but facts are facts and I don't want to miss lead anyone I want them to know everything and if I can answer I will that is the only reason I get on the forum. :D

[NaeKuh]

Duh, i know your a very informed member on this section, but i have to agree with OWL.

Heres my reason on why, because most of the post he brings in this section have been really acurate. I learned a lot about TEC's from NOL, SerialKiller, and everything owl has been telling me has been right on the ball with NOL's original statements.

However i always wished nol was sometimes wrong with sub ambient cooling. He pointed out all the wrong ways on my original TEC project and having a 17 or 18 yr old kill drill you the 101 on TEC's to a 30 year old is not something one likes to hear, but i take it as a golden paper and i also learned NOL usually knows the outcome to almost every scenario with sub ambient cooling. So i save money by not figuring the mistakes out myself and:

1. Risking a fire happening,
2. spending a ton of money only to have it sit in my unused parts box like my many other projects.


My project will have a special custom chiller right after the radiator that martin is going to build me. It will have room for 3x62mm TEC's however i will only use low wattage TEC's. Since having h2o temp @ ambient is impossible without a massive radiator, i wanted to use the chiller on a dial from my meanwell S320-24 to control set the 3 TEC's.

Im only aiming for ambient temps on h2o messured via temp probe. This is xtreme systems, so even if there is a 3-5C reduction in delta, i would be happy with the project. Only thing im waiting on now is for martin to free up some time so he can help me with this project.


But my general questions:
1. What is the h2o delta after it ran into a CPU + NB overclocked. NB voltage is 2 over factory settings on my P5K-DLX. CPU voltage is at 1.45Vcore Q6600 G0 loaded @3.73.
2. Im running a PA120.3 on a cpu/board loop. Only 2 blocks a ApogeeGTX + MCW30. This chiller a complete waste of my time?
3. I was originally considering serial killers low wattage/more tec concept so i wanted to limit it to 60Wx3 or 80Wx3 with room to scale later on if needed.

Idea still too loony to continue?

[Fraggle]

I see initialised pointed you the right way.. there seems to be a pre-negative thought in this section by some members..

Dude! I`m not `anti-tec` or anything lol :o) I`m just saying that if you`re trying to chill your loop then the last thing you want in there is a radiator.

[NaeKuh]

Dude! I`m not `anti-tec` or anything lol :o) I`m just saying that if you`re trying to chill your loop then the last thing you want in there is a radiator.

i dont think were trying to chill that much like a CoolIT product.

More like cheat ambients a little to make it lower then what it really is. :P

[Tony]

here you go...something i have been workin on:

http://www.ocztechnologyforum.com/staff/tony/TEC%20chiller/tn_CIMG2054.JPG

http://www.ocztechnologyforum.com/staff/tony/TEC%20chiller/tn_CIMG2055.JPG

Left cold side, right hotside. Cold side thermostat controls 1 psu which can have 1, 2 or 3 TECS off it.

Lowest temps see to date are -5C, issue is not enough cold side coolant capacity so more mods are coming.

[Fraggle]

Now that`s a tec chiller...

[littleowl]

here you go...something i have been workin on:

http://www.ocztechnologyforum.com/staff/tony/TEC%20chiller/tn_CIMG2054.JPG

http://www.ocztechnologyforum.com/staff/tony/TEC%20chiller/tn_CIMG2055.JPG

Left cold side, right hotside. Cold side thermostat controls 1 psu which can have 1, 2 or 3 TECS off it.

Lowest temps see to date are -5C, issue is not enough cold side coolant capacity so more mods are coming.

what tec's you running? What is this tec setup running on at that time or is that just open air temp? I would really like to see more pics of the case!?

[NaeKuh]

here you go...something i have been workin on:

Lowest temps see to date are -5C, issue is not enough cold side coolant capacity so more mods are coming.

OMG your almost as bad as Eddie EK. Showing off the bling, but not porn? :shakes:


Lets see the naked shots of this setup :up: :yepp:

[THE JEW (RaVeN)]

here you go...something i have been workin on:

htp://www.ocztechnologyforum.com/staff/tony/TEC%20chiller/tn_CIMG2054.JPG

http//www.ocztechnologyforum.com/staff/tony/TEC%20chiller/tn_CIMG2055.JPG

Left cold side, right hotside. Cold side thermostat controls 1 psu which can have 1, 2 or 3 TECS off it.

Lowest temps see to date are -5C, issue is not enough cold side coolant capacity so more mods are coming.

Rough specs with what hardware to achieve above? Thanks.

[Tony]

The case is home made and a little rough...i just wanted something i could house it all in.

TECS are 4X320W 14.5V units running either 12.5V or 5.5V using PC PSU's...i had next to no extra cash to make this happen using fancy power supplies.

TEC hotside is cooled with 2x120.3 rads and 12 xilence 120mm fans..DDC pump with Alphacool top and res....OK for hotside. 4 Swiftech blocks cooling the 4 TECS, 10mm Tube. OCZ's new blocks will cool the TECS better but its not released yet, hence i had to use some old GT's I had.

TEC coldside...custom double sided block/heat exchanger...I can not show pics as this is under NDA. 7/16th tube, DDC with Alpha cool top and res...not enough capacity hence mods coming soon.

Thermostats are LAE Electronic's brand, 240V but they do 12V AC/DC versions...it was all i had from old stock. One is read only, the other is fully programmable 150C to -45C and has hysteresis control of 1C to 10C either +- so i can average temps by 0.5C approx on the cold side water. IE I set 20..it hits 21 and its chilled back down to 20.

The right hand PSU supplies fans and pumps as well as 2 TECS at this time, the other PSU is thermostat controlled and supplies the other 2 TECS.
I actually think 1 TEC at 12.5V off the right hand PSU and 1 at 5.5V and the booster PSU running 2 TECS at 12.5V may be better but this depends on my tests on increased capacity on the cold side...
For now the booster PSU is running 2 TECS at 5.5V; running 12.5V pushes to much heat from the hotside and is heating the coldside due to it being over whelmed. Yes i can push 120.3's real hard with this setup to the point they just can not cope.

overall It all revolves around having enough cold side capacity and an efficient cooling system on the hotside...something i have not quite found yet but i hope to soon.

holding C2D CPU's at 4GHZ 10C idle is piss easy for this unit, infact it will hold then a lot lower but as mentioned before i start overloading the rads. I have been testing Phenom at 2.75GHZ with the CPU idle at 5C and load sub 25C...phenom has a high TDC so this was quite nice. Once i get it finer tuned I hope to be seeing sub zero idle's with sub 10C loads very soon.
Cold side coolant capacity is key...once I get a gallon or so being chilled it willl work a LOT better me thinks ;)

[Tony]

Ohh and i forgot, i will be adding a bypass loop for the hotside so i push the hotside cooling outside of the case you saw in my previous posts. I have a rain water storage tank with 220L of real cold rainwater cooling 2x20m long 11mmOD copper coils that are plumbed into my workshop watercooling setup. I have a feeling once i add the 3xtripple rads also to this monster loop I should have no problems cooling the TECS at 4x12.5V ;)

http://www.ocztechnologyforum.com/staff/tony/TEC%20chiller/tn_CIMG2180.JPG

With the cold weather coming in the UK i actually think i may stand a chance of freezing the water in the storage tank using the coil as the chiller and the rads to cool the coolant to sub zero...should be fun testing this :)

[NaeKuh]

Ohh and i forgot, i will be adding a bypass loop for the hotside so i push the hotside cooling outside of the case you saw in my previous posts. I have a rain water storage tank with 220L of real cold rainwater cooling 2x20m long 11mmOD copper coils that are plumbed into my workshop watercooling setup. I have a feeling once i add the 3xtripple rads also to this monster loop I should have no problems cooling the TECS at 4x12.5V ;)

With the cold weather coming in the UK i actually think i may stand a chance of freezing the water in the storage tank using the coil as the chiller and the rads to cool the coolant to sub zero...should be fun testing this :)

Very very nice Tony...

Omg NDA? so someone is coming out with a exchanger?

[Tony]

i did not say we were coming out with it...i just can not show you

[jinu117]

Yo Tony, coming to CES this year? :)

[THE JEW (RaVeN)]

Thank you for the followup post, Tony. It's appreciated.

[Scarlet Infidel]

Im working on a chiller myself, not as powerful as Tony's but hopefully a lot more efficient.

[Stigma]

Do yourself a favor, and get someone here on the forum to make you an economy water-chiller. Your idea is unfortunately doomed to be incredibly inefficient. An economylevel waterchiller, while probably a little more expensive will be leaps and bounds more powerful. If you don't want to go too low in the temps, just have the builder make you a small-scale system.

-Stigma

[Fujimitsu]

Do yourself a favor, and get someone here on the forum to make you an economy water-chiller. Your idea is unfortunately doomed to be incredibly inefficient. An economylevel waterchiller, while probably a little more expensive will be leaps and bounds more powerful. If you don't want to go too low in the temps, just have the builder make you a small-scale system.

-Stigma

He's right, phase-change waterchilling is simpler (relatively speaking) and more efficient, as well as probably being cheaper and more effective.

[Tony]

no where near as much fun testing and tweaking though ;)

Water capacity thats chilled is key...the more water you chill down the more efficient in becomes

[Stigma]

Well a TEC water-chiller DOES have the very nice advantage of being able to regulate the water temp to xactly what you want. You can't do that easily with a phase-cooler, since you can't start and stop it every 5 seconds. You could cycle it every few minutes probably, and if you had a really large chilltank that could possibly work without having too big differences in temp over time, but I don't think it would be a very healthy thing to do for the compressor.

But well... given the ineffectiveness of TECs to cool pretty much any of the really hot parts in a modern PC these days, TEC are just dead to me now after going phase :(

There is hope though. new TEC-like technology is supposedly under development, and should (according to the companies developing them atleast) be able to have an efficiency-level equal or even better than compressors. If and when that happens, the neo-TECs will be king in PC cooling for the simplicity of use and small size they offer.

Until then... phase is king! :)

-Stigma

[jinu117]

For the money, you could put in VFD compressor though... :) Ehehe.

[Stigma]

For the money, you could put in VFD compressor though... :) Ehehe.

Whats a VFD compressor, and how would that be able to regulate the cooling?

-Stigma

[Blauhung]

hehe, I've been thinking of creating my own version of what Tony has going on there for a few months (fittingly my name is Tony as well). I plan on milling my own blocks though.

[Tony]

I can dial in any temp i want within reason...IE test at 5C water...dead easy :)

Its due a rebuild and some mods...will be better then.

[Blauhung]

no where near as much fun testing and tweaking though ;)

Water capacity thats chilled is key...the more water you chill down the more efficient in becomes

Isn't that more of a Capacitor in the eyes of the system (or I guess more properly an inductor)? You are still pulling the same heat load out of the water, just there's more water so the overall temperature responds much slower to changes in the load or the sink.

[jinu117]

Whats a VFD compressor, and how would that be able to regulate the cooling?

-Stigma

Variable Frequency Drive equippped. Controls motor speed of copressor so what happens is you can regulate compression speed based on load condition :) It's newer technology and I believe only one company is producing it for hermatic reciprocating unit (there are scroll ones that has this in market already... copeland does scroll, embraco does recip).
Actually, such concept has been in use in larger scale industry where energy saving becomes huge issue, etc.
The additional cost of VFD drive puts some damper on cost side of story but I presume it will change over time as lot of newer technology does. (For machining motors, etc this is standard way of changing speed... especially in CNC)
BTW, while I like the TEC, I just made direct die unit that can handle total 550w with -4c evap temp drawing "only" 400w. This is the kind of efficiency refrigeration circuit has vs other technolgy for time being. (It could have gotten lower temp using different gas but I am going for lowest stress on system components)
Not saying TEC is a bad thing as conventional refrigeration circuit tend to be heavy, typically larger sized (not sure on this considering TEC need decent sized water loop really), and directionally challenged... (must be straight up).
In fact, it was developed for compensating the weaknesses of conventional refrigerations :P
I look forward to seeing you making some nice change Tony :P

[NaeKuh]

Jinu, martin is going to make me my heat exchangers soon.

I had a question, how should i design something like that to fit after the rad loop? i was thinkn of having it be 2 plates 62x3 plate for upgradeability.

Then i was wondering about the TEC's for it. Would it be better doing 3x60W @ 18V Or should i go higher @ lesser voltage? Im not looking at freezing the water, just want to shave off ~5-6C b4 it hits the cpu. Maybe a bit more if i get lucky.

Also would i require a meanwell for each tec? or can i get 1 large class meanwell and have it all wired to it?

[Cupcake]

One large PSU would cut it :)

a 300 watter would handle three 80s :)

You could use Mase 4 CPU Blocks to keep high flow

[Stigma]

Wow, a VDF sounds really handy! =)
I suppose they cost a ton more though, especially if its relativly new tech.

One of those would really be awesome for a chiller, so you could basically specify your desired temp and let it auto-regulate :)

Any idea what they cost compared to regular compressors?

-Stigma