Experimenting with Peltier device

[Delfistyaosani]

OK I'll try to explain better, if my ENG lets me :P

So the idea is to divide liquid stream coming from the pump into two parts, say LIQUID1 and LIQUID2. L1 is going to cool hot sides and return to radiator, lets call this LOOP1 (RADIATOR -> PUMP -> HOTSIDES -> RADIATOR). The second, L2 is chilled with cool sides, cools CPU and returns to radiator (RADIATOR -> PUMP -> COOLSIDES -> CPU -> RADIATOR). The tank is not directly connected to pump, but the radiator lies inside (I use a mix of oil and diesel for looping) outside area of tank is about 2-3 square meters (the tank is metallic) and that is pretty enough to transfer 1KW energy. However if water in tank heats us anyway I can simply let it out an fill with fresh cold water.

Get the idea now? I'll try to draw a variant in 3D MAX a bit later

[zipdogso]

OK I'll try to explain better, if my ENG lets me :P

So the idea is to divide liquid stream coming from the pump into two parts, say LIQUID1 and LIQUID2. L1 is going to cool hot sides and return to radiator, lets call this LOOP1 (RADIATOR -> PUMP -> HOTSIDES -> RADIATOR). The second, L2 is chilled with cool sides, cools CPU and returns to radiator (RADIATOR -> PUMP -> COOLSIDES -> CPU -> RADIATOR). The tank is not directly connected to pump, but the radiator lies inside (I use a mix of oil and diesel for looping) outside area of tank is about 2-3 square meters (the tank is metallic) and that is pretty enough to transfer 1KW energy. However if water in tank heats us anyway I can simply let it out an fill with fresh cold water.

Get the idea now? I'll try to draw a variant in 3D MAX a bit later

So you have 2 radiators....if you dont and both liquids go back to the same radiator you effectively have only one loop and you are running the same basic loop past both the hotsides and the coldsides. Athough you have shown this appears to work, your system WILL work MUCH BETTER if you disperse the hotside heat to a seperate radiator.
TEC's are heatpumps so when heat is applied to the coldside it is absorbed but all that happens to that heat is that it is passed to the hotside along with the heat a TEC naturally produces...so effectively you are just taking heat from liquid loop 2 and passing it onto liquid loop1 there is no chilling involved until that heat reaches your radiator by which time it is joined by the water warmed by your PC...that gives your radiator a lot of heat to get rid of. In a short run your setup might appear to be OK but if you ran it for any length of time I think you will find your coolant is just getting hotter and hotter.

To recap - [QUOTE=Delfistyaosani;3886530] "but the radiator lies inside (I use a mix of oil and diesel for looping) outside area of tank is about 2-3 square meters (the tank is metallic) and that is pretty enough to transfer 1KW energy." {/QUOTE]
You are forgetting all about the heat generated internally by the TEC's it could easily be 750-850w at 75% power cooling 1Kw your tank needs to deal with 1.8Kw not 1Kw.
I have told you this several times but you appear to just be ignoring me...what is it ? you don't understand ???

[Delfistyaosani]

I didn't say I'm gonna use 1KW effective heat pumping but POWER TECs WITH 1KW, don't think heat could appear from nothing, the heatsink will have to transfer TEC power + COMPUTER POWER right? and that is gonna be 1.2-1.5KW, not more. What about water, I can simply flush it out and fill with fresh periodically if it heats up. In winter there is no way to get even 10 degree in water cuz it's -10/-15 outside so the water is not even expected to melt, in summer it can be simply flushed/filled periodically or constant streaming for benching.

At this time I have three problems to solve

1. The motor heats up too much after 2 hours, think I'll have to get the second one and make them work one after another witch about 30min delta time.
2. Gotta get ~3mm copper plates with about 50x50 cm for TECs
3. Find a kind of metalized TECs and get a pair

If dividing one stream won't work I could separate sub and above zero ones, for example get a magicool system and connect to coolside "waterblock"

[zipdogso]

Well I am sure you obviously know best...good luck with project...

[Ultrasonic2]

zipdpgso is right... and it's obvious in your diagram. You electively have 1 loop that runs past the hot and cold sides.. there's is no possible benefit to a system like this since your just moving the heat from one part of the loop to another part of the same loop while adding more heat from the TEC.

The idea of using one loop to cool the hot and cold sides of the loop has been tried and tested in this forum before and it doesn't work. For this kind of idea to work it requires low flow rates. HOWEVER reducing flow rates results in worse cooling so the whole premise that this system is based on is wrong

[Delfistyaosani]

I'll add a second loop for CPU if one doesn't work

***

For chilling increasing "track" length will also increase the effect, not only slowing liquid down

[demonkevy666]

TEC's can be run perfectly well from batteries...what batteries you talking about...It all depends on the current they can deliver and for how long...if your thinking car/van/truck/lorry batteries i don't think there would be a problem it would just be a question how long the power would last.

actually aI was thinking lithium polymer batteries. you know like R/c car and planes, 7.4 volts 8,000mah 20C

but reading the ratting this is throwing me for a loop about it.

http://www.hooked-on-rc-airplanes.co...ery-packs.html

but then again an automotive or deep cycle is heavy and high amperage.

[Stigma]

zipdpgso is right... and it's obvious in your diagram. You electively have 1 loop that runs past the hot and cold sides.. there's is no possible benefit to a system like this since your just moving the heat from one part of the loop to another part of the same loop while adding more heat from the TEC.

The idea of using one loop to cool the hot and cold sides of the loop has been tried and tested in this forum before and it doesn't work. For this kind of idea to work it requires low flow rates. HOWEVER reducing flow rates results in worse cooling so the whole premise that this system is based on is wrong

I can only agree with this. You might think that the water will get cooled by the TEC and thus help keep your CPU temp cool before you dump it back into the line going to the radiator, but your really not looking at as a system. The problem is basically that the water running past the TEC will barely have any time to get cooled down on a single pass, so you won't really see any bnefit on the CPU temp. Meanwhile, all that juice that the TEC draws is generating a lot of heat, and you are just pumping that heat right back into the main loop. Basically if you run the system for more than a minute, you will start getting worse temps on your CPU than if you just didnt use any TEC at all.

If you do insist on using TEC cooling (wich I personally think is really inefficient for the kind of heat generated by modern computers), then what you want to build is basically this:

1 cold loop that has all the stuff you want to cool down attached to it.
1 hot loop, wich contains only the radiator to disappate the heat.
1 junction point where the two loops meet, and one or more TECs pump the heat from the cold loop to the hot loop.

This atleast works, but thats not to say it dosnt have issues like most TEC builds. Take care to have enough TEC cooling power to handle the heatload and, have a suficiently large radiator to handle the heatload of both the computer AND the added heat of the TECs. It also requires two pumps obviously, but atleast this system wuld not work against itself.

All in all I'd recommend looking into commisioning a phase liquid chiller instead. It will be a lot less complex, much MUCH more efficient and powerful, and probably not that much more expensive to be honest, when you take into account the price of the PSU(s) you need to feed the TECs with.

-Stigma

[Stigma]

actually aI was thinking lithium polymer batteries. you know like R/c car and planes, 7.4 volts 8,000mah 20C

but reading the ratting this is throwing me for a loop about it.

http://www.hooked-on-rc-airplanes.co...ery-packs.html

but then again an automotive or deep cycle is heavy and high amperage.

You obviously CAN power a TEC with a battery, but it seems to me like it would be really impractical. TECs draw oogles of current, and even if you got a high-performance battery that could handle that kind of load, then it would tap out the battery very quickly.

For example, that battery you gave as an example wuld be able to run a 300watt (power draw, not cooling) peltier for about 11-12 minutes. (assumingyou found the right voltage and all that obviously, im not saying hat particular battery is suitable, unless you atleast serial-connect it to 2x the voltage). That is to say, if that kind of power draw wouldn't just kill the battery - the battery itself would probvably need some kind of cooling to not take damage from overheating.

So basically why invest money into a complex system that can run your TECs for like 12 minutes at a time? It just seems so incredibly impractical... as if TECs weren't impractical in a lot of ways allready.

-Stigma

[Delfistyaosani]

Stigma

Phase u say... can u draw a diagram how to construct a 1-phase chiller from a common fridge and how to add that to existing one-loop liquid cooler?

[demonkevy666]

You obviously CAN power a TEC with a battery, but it seems to me like it would be really impractical. TECs draw oogles of current, and even if you got a high-performance battery that could handle that kind of load, then it would tap out the battery very quickly.

For example, that battery you gave as an example wuld be able to run a 300watt (power draw, not cooling) peltier for about 11-12 minutes. (assumingyou found the right voltage and all that obviously, im not saying hat particular battery is suitable, unless you atleast serial-connect it to 2x the voltage). That is to say, if that kind of power draw wouldn't just kill the battery - the battery itself would probvably need some kind of cooling to not take damage from overheating.

So basically why invest money into a complex system that can run your TECs for like 12 minutes at a time? It just seems so incredibly impractical... as if TECs weren't impractical in a lot of ways allready.

-Stigma

let me add in I would use the 40mm 225 watt tec which uses 24 volts max and 15.5 amps max

so it would be only be 30% of it's power extending the duration. you can add more batteries in parallel and series, parallel makes run time longer, series adds volts.

[Ultrasonic2]

If i was him i'd just build a standard 2 loop TEC chiller. i have one it works great. 0-c water temps at LOAD

[Delfistyaosani]

I'm getting more and more confused... which principle is more effective and less expensive overall? TEC or Vapor? Imagine i have a standard working fridge, how to use it as a chiller? As I see common fridges don't cool down fast enough to chill 20-30C` liquid to 0C`