DO OR DIE!!! TEC WaterChiller or bust!

[masher]

>> What's the reason behind using a non-water coolent?

>> Different evaporation/boiling points.

Which makes absolutely no difference in a system that isn't undergoing a phase change. In a single-phase system such as he's describing, your performance metric is the specific heat rating....which is higher for water than either alchohol or acetone.

Not to mention the fact that a high-wattage peltier might actually exceed the boiling point of those liquids, and the resultant steam cavitation would likely shut down the cooling entirely.

 

[masher]

>> I'm thinking I'll need to modify my plans and use two paralell blocks and four TEC's and a lower flow pump.

Err, why would you use parallel blocks (thus increasing your flow rate) then switch to a lower flow pump? Use a serial block setup and keep the same pump.

Hook up 600W of peltiers and you can possibly get your output temperature down 4-5C below ambient. That might take your core temp down almost a degree...at the cost of 450 KW-hours every month. Or, for one FOURTH that electric cost, you can hook up a single 150W peltier directly to your chip, watercool that, and see five TIMES the improvement in cooling.

Frankly, I don't think you're interested in a cooling project at all. You just want something loud, noisy, and impressive looking...no matter whether or not it actually performs

 

[sativa]

is there a specific reason that you aren't using the TEC directly below your processors water block?

and you can do the same for your video card.

You'll get much better temperatures and you won't have any increase in noise.

 

[RevelationX]

>> What's the reason behind using a non-water coolent?

>> Different evaporation/boiling points.

Which makes absolutely no difference in a system that isn't undergoing a phase change. In a single-phase system such as he's describing, your performance metric is the specific heat rating....which is higher for water than either alchohol or acetone.

Not to mention the fact that a high-wattage peltier might actually exceed the boiling point of those liquids, and the resultant steam cavitation would likely shut down the cooling entirely.

I agree with the first part, however that liability of a low boil point can take you here.

http://www.atcoamerica.com/how.htm

Use the peilters heat to pump the ammonium hydroxide or don't use a peilter at all and just use the heat from the cpu.If it is too cool you could adjust the amount of ammonium hydroxide. Hey..It could work...Or I could be all wet..Use the peiters to start pumping and use the coil to keep the res cool then pump that through the cpu block sans the radiator.But thats just crazy talk now...

As for why you would use alcohol as opposed to water in a pump and radiator system I've no clue..seems to me some like anti-freeze/anti-boil would be a good idea..but plain anti-freeze is bad as it neither cools or freeze protects as well as mix and I'm sure it can't be great for rings,seals,and plastic parts meant for water.

 

[Jonsey]

>> What's the reason behind using a non-water coolent?

>> Different evaporation/boiling points.

Which makes absolutely no difference in a system that isn't undergoing a phase change. In a single-phase system such as he's describing, your performance metric is the specific heat rating....which is higher for water than either alchohol or acetone.

Not to mention the fact that a high-wattage peltier might actually exceed the boiling point of those liquids, and the resultant steam cavitation would likely shut down the cooling entirely.

Exactly.

If you can use water, it's the best coolant. It cools much better than any kind of alchol due to it's very high specific heat. It takes much more energy to heat a gallon of water one degree than to heat a gallon of alchol one degree. Because of this, water can remove more heat than most any other substance on earth. There is no evaporation in a water cooling system, so that process doesn't even apply here. The only reason some people use straight alchol is because their chillers are doing -40c and that's about the only thing that won't freeze.

 

[Peach]

I have those big xeon fans on my radeon. 2 of em

 

[Quicksilver]

is there a specific reason that you aren't using the TEC directly below your processors water block?

Yeah, actually. It's because I was given these 2 72W TEC's and I figured that each one could be air cooled. The two together couldn't even keep up with by mobile 2400, let alone a 3 gig northwood. The previous owner used the two TECs to cool a celery 300a.

 

[Quicksilver]

I have those big xeon fans on my radeon. 2 of em

You are insane! I know it's fashionable to be "tha ole 'ardcore" around here, but JESUS MAN!! TWO? at full speed, the put out like 50 dbl

 

[masher]

>Yeah, actually. It's because I was given these 2 72W TEC's and I figured that each one could be air cooled. The two together couldn't even keep up with by mobile 2400...

You're still not making sense. Two of these can't handle a low wattage mobile cpu, yet you think 4 or even 8 will be able to chill a large water loop to the point it'll help cool a much higher power cpu?

A peltier is good at creating a temperature differential...dT. It's not good at removing heat. A 72W pelt is probably providing about 5 watts of actual cooling power....the rest is just waste heat. However, the dT created by the pelt can make OTHER cooling systems more effective. A watercooling system is very effective at removing heat--.if it has a good dT. It cannot create a temperature differential on its own, and in the absence of one, it can move no heat at all.

Use the pelt between waterblock and cpu to boost your dT, and let the water remove the heat. You'll have an effective cooling system. If you want something loud and impressive, put an air horn in your case.

 

[sativa]

>Yeah, actually. It's because I was given these 2 72W TEC's and I figured that each one could be air cooled. The two together couldn't even keep up with by mobile 2400...

You're still not making sense. Two of these can't handle a low wattage mobile cpu, yet you think 4 or even 8 will be able to chill a large water loop to the point it'll help cool a much higher power cpu.

thats exactly what i was thinking.. it doesn't make any sense that he thinks using them to cool the water would result in colder temperatures than using them directly on the processor, and using the water block to remove the heat from it (which, like you said, would create a higher difference in temperature)

 

[zer0signal667]

A peltier is good at creating a temperature differential...dT. It's not good at removing heat. A 72W pelt is probably providing about 5 watts of actual cooling power....the rest is just waste heat. However, the dT created by the pelt can make OTHER cooling systems more effective. A watercooling system is very effective at removing heat--.if it has a good dT. It cannot create a temperature differential on its own, and in the absence of one, it can move no heat at all.

How does the peltier create a temp differential effectively without removing heat? A peltier that has a Qmax of 72W will be able transfer 72W of power, I don't know where you got that 5W number from. Of course they are only able to achieve Qmax under ideal conditions, but to suggest a performance drop that huge is ridiculous.

 

[masher]

>> How does the peltier create a temp differential effectively without removing heat?

I didn't say it couldn't remove heat; I said it was very poor at doing so. It is one of the least efficient cooling mechanisms known.

>> A peltier that has a Qmax of 72W will be able transfer 72W of power...

No, it will consume 72W of power. And therefore create 72W of waste heat. The amount of heat it can actually remove is far smaller. A TEC operates off the Seebeck effect and is rated on a metric known as the coefficient of performance...similar to the carnot efficiency of a heat engine. When I was in grad school, COPs were on the order of 5-10%....though they may be somewhat higher now.

In contrast, a heat pump (aka a residential HVAC system) moves *more* heat energy than it consumes itself...substantially greater than 100% efficiency.

 

[Lazn_Work]

>> How does the peltier create a temp differential effectively without removing heat?

I didn't say it couldn't remove heat; I said it was very poor at doing so. It is one of the least efficient cooling mechanisms known.

>> A peltier that has a Qmax of 72W will be able transfer 72W of power...

No, it will consume 72W of power. And therefore create 72W of waste heat. The amount of heat it can actually remove is far smaller. A TEC operates off the Seebeck effect and is rated on a metric known as the coefficient of performance...similar to the carnot efficiency of a heat engine. When I was in grad school, COPs were on the order of 5-10%....though they may be somewhat higher now.

In contrast, a heat pump (aka a residential HVAC system) moves *more* heat energy than it consumes itself...substantially greater than 100% efficiency.

Atually not true. True TEC's are ineffeciant, but take at look at this 80W one here:
http://www.dangerdenstore.com/product.php?productid=87&cat=30&page=1

Imax = 8.0 Amps
Qmax = 80.0 Watts
Vmax = 16.1 Volts
Delta Tmax = 71 (C)
Size = 40mm X 40mm X 3.4mm

So that consumes a max of: 8 x 16.1 or 128.8 Watts, but it produces 80 Watts of cooling power. Qmax is the cooling power, not the power consumed by a TEC. (BTW that makes the efficency at 61%, so 39% of the power used is waste heat)

==>Lazn

 

[masher]

So that consumes a max of: 8 x 16.1 or 128.8 Watts, but it produces 80 Watts of cooling power. Qmax is the cooling power, not the power consumed by a TEC. (BTW that makes the efficency at 61%, so 39% of the power used is waste heat)

You're being misled by manufacturer statistics. QMax is indeed the maximum heat pumping ability (and if I misunderstood the original poster that his 72W figure was the QMax rating, I apologize). However, QMax exists only with a temperature differential of zero. Conversely, at the maximum rated temp differential, QMax = 0.

The real-world efficiency of a TEC is far lower....you can not exhaust heat fast enough to keep a zero differential....and indeed, even if you could, you wouldn't be "cooling", you'd simply be pumping heat from one infinite reservoir to another, without actually changing any temperatures.

 

[Peach]

You are insane! I know it's fashionable to be "tha ole 'ardcore" around here, but JESUS MAN!! TWO? at full speed, the put out like 50 dbl

ya man, on full blast it makes my desk vibrate

 

[Passive Anubis]

Sounds like you know what your talking about...
However, I would highly recommend you try it on another computer before you set yours up. Just any old comp will probably work. Remeber, jsut because somethign is good in theory, doesnt always mean it works out in reality. Good luck to you, i think you can pull it off.

 

[iddqd]

theres ALOT of problems with this plan first off two 72watt pelts isnt anywhere near enough power for todays cpus youll need at least two 226watt or even better two 300watt and youll never cool them with heat sinks the only way you will do it is with another seperate water cooling system, youll need three water blocks itll have to go waterblock TEC then waterblock then TEC then another water block with the cold side of both tecs facing inwards now youll need a seperate cooling loop with a seperate radiator and pump for the two outermost waterblocks. youre also going to need an industrial power supply a pc psu WILL NOT cut it i suggest meanwells there pretty good

I'm sure a 700W Zippy could power this setup.

By the way, I also suggest you try it without a working computer first. You can simulate the thermal load of a CPU by placing a propane burner under the waterblock. Tweak the flame until you get watertemps similar to your loop without the chillers. Turn on the chillers, and see if anything is different. I've seen this feat attempred before, and it did work, although not too well. But hey, you can try maybe you'll have more luck.

 

[cell_491]

no i doubt any pc psu can handle the amps reguired to power that many TEC's what he needs is a meanwell industrial psu.

 

[BallerX]

Go to walmart and buy a window ac unit... the cheapest one you can find. remove the fan that circulates air through the evaporator...remove the thermostat or relocate it to a really hot place like under the compressor so the unit never shut off unless its turned off...either build a case around the evaporator made of plexi(water tight please) or you can attemp to move it slightly to get it into a coleman 12 pack cooler. the wide flat kind!!! turn it on and circulate your water through that and you'll be doing a lot better than a gagillion TECs or whatever it is your planning... it has been tried a thousand times and the only one that made any difference was a funky tall carafe' looking thing that built up ice around a tec probe and then you flow your water through it...it reduced temps about 2deg. when placed inline after a radiator. 30 mins later when the ice melted around the probe it was over untill you turned off your comp and let it ice back up again...

 

[Lazn_Work]

Well everybody is telling me "don't do that, you won't see the kind of gains you're looking for"
and, it's not that I don't believe you guys, I just want to try this and use my gear. I'm thinking I'll need to modify my plans and use two paralell blocks and four TEC's and a lower flow pump.
Part of my reason for doing this also comes from my desire to build an external housing for a watercooler. I can't justify building a big, ugly enclosure for JUST a watercooler, I want a chiller too, even if it's just a room-temp one.

Any update? I am curious if you had any success.

==>Lazn

 

[gbomb944]

Apparently some of the peeps here smoke a lot of crack from some of the posts of I saw on this thread. Dude wants to put a couple of tecs in his loop and everybody starts going on about ammonia and building a phase change....rofl
Having open vats of alchohol and spraying r22 all over the place is not an option. I've seen the pics from those homemade compressor chillers and it's huge mess. Not to mention that you need to be hvac certified and know how to solder high pressure fittings. Me spilling a little glycol and water all the time is bad enough.

Anyway Here is my story....

I found a tec warmer/cooler in a thrift store with a psu for 20 bucks so I picked it up. I liberated the tec with a fairly giant heatsink from the little fridge and have been playing with it. The psu for this is a little switching type that gives 5.8a at 12v so the tecs getting 70w at best.

A 70w tec cannot cool a cpu unless it's <30w output so sticking it on the chip is out. I rigged a homemade block onto the cold side and put the hot side on the sink that came with the unit. The sink is about 120x110mm but it's aluminum and it is only about 3/4 of an inch thick so I think a better sink would help as it was pretty warm when I was running it.

I put it in my loop just like quicksilver was planning his. In front of the cpu and just after the rad so it would only get ambient temped water. I got about 2c lower temps by measuring with mobo monitor but I could tell that the water going into the chip was a pretty good bit colder than ambient, maybe 6 to 10 degrees by touching the exposed metal of the inlet to my cpu block. After a while even my rad felt a little cool to the touch.

I thought maybe the rad was holding temps upward so I removed it. I was able to run by removing the rad completely but only for web surfing so the 70w or so pelt could hold my winchester oc'd to 2550 and my 550 gph inline pump heat and maintain the same idle temps that I get with the rad on. If I had had a 170 watt tec or perhaps 2x70's instead of my one I wouldn't have been surprised at all to see sub ambient.

To say that a peltier will have NO effect whatsoever on the water in a loop is just retarded. My single 60ish watt one did give some chill so two 70's should definitely do something.

The main problem I could see with the two links to other guys that had tried was that they left the water they were using open so that the air could warm it. Go get a glass of something cold from the fridge with no ice and see how quick it warms back up. Also the guy with the big copper blocks didn't insulate his cold side. that huge copper block was just radiating cold back into the air. Cold water traveling through tubing is at least a little insulated.


So anyway I say rock on quicksilver and just build it with what you got. I'm planning on building something similar by adding some more pelts mysef. I was searching for info which is how I found this thread. I think in the end electricity would be a problem when you get a whole bunch of pelts running as well as room heat. But the great thing about a system like this is you don't have to run the pelts all the time unlike folks who have them strapped to the cpu.

I would build a custom hot plate that spreads the heat over two or three of those fans for each pelt because the cooling power is most efficient the colder you keep the hot side. At say 30% efficiency which is probably a reasonable expectation you'll be putting out 200 or so watts on the hot side and I doubt those zeon fans can pull that much through only one.
The closer you keep the hot side to ambient the more cold the cold side gets. My sink was aluminum and not very high tech but it did have a 120 cfm fan on it and it was probably 120-140 farenheit. Water on the hot side would be better but it would be too much hassle and noise to put in another loop.

Incase nobody believes you can do it just type peltier water chiller into google and you will see that people use pelts and water to cool lasers and special cameras and all kinds of stuff. This link is pretty cool.

http://www.tetech.com/assys/water.shtml

Notice the huge sinks compared to the watt ratings for each device.

Heres the industrial size model...

http://www.thermoelectric.com/2002/pr/lc/tlc/rlc-1400.html

Anyway don't get discouraged quick. If you've got water blocks just stick em in the loop and see what happens..

peace

 

[Gadfly]

I built something like this with some success, (10-15'C at the CPU), Here is the diffrent setups I used and the results:

water cooled starting temp: CPU 44'C (under full load), 75'F room temp

The water system was a custom DangerDen kit: TDX cpu block, DD 12V pump, black Ice 2x 80mm fan radiator.

The "Waterchilling block", was a MAZE4 water block, with a copper slab "cold plate" on the cold side of the TEC, with a massive (120mm x 135mm x 90mm) Copper HS I found at metal scrape yard on the hot side. I used a 165CFM fan to cool the heatsink. The TEC was a 150W with dedicated powersupply from Danger Den.

The water loops I tried went like this:

pump > waterchiller > CPU > radiator > tank > pump

Full load temp: 36'C
idle: 29'C
pump> CPU> Waterchiller> tank > pump

Full load temp: 39'C
idle: 31'C

pump > Waterchiller > CPU > tank > pump

Full load temp: 34'C
idle: 28'C

In order for this to work right, the chiller block needed about 10 min of "Pre-cool" to fully chill the coldplate, and the the water block before you power on the computer.

I used this system to cool an AMD64 3000+ (754) and ran it at 2.5ghz but it wasn't really stable. So I reverted to 2.4ghz with just the Danger Den water kit. The water chiller just didn't work well enough to justifiy the additonal noise for me.

I took the system all apart when I got my Vaporchill, so I don't have any pics to post, sorry

I do have the PC Mark though:
http://service.futuremark.com/compare?pcm04=2513243

Gadfly