peltier and phase change

[Kylian]

Hello I am going to build a new PC.
For the CPU I am going for a     AMD Threadripper 1950X.
For the cooling I am thinking of using a peltier stack and to cool the hot side of the peltier with a phase change (LittleDevil LD Phase Change) I don't really care about power usage at all.
The things I would like to know is.
Is this even possible?
Will a phase change be able to cool a peltier stack (ranging to get around -100/-150)
how bad is it for the CPU when having the PC on for 24/7 even with really good isolation.

[vanished]

There's no point in using peltier, this has been looked at many times before, they have limited cooling power, consume a ton of power to do what they can, and cost a lot for no benefit over a traditional system.  If you're imagining using the peltier to create a temp difference between the phase change plate and the other side so it gets even colder, that works when it's in the air under no load, but load it with the heat from a CPU and it will be no better than just sticking it directly on - maybe worse even.

 

There is a legitimate reason to use phase change but that doesn't necessarily mean it's the right choice for a normal 24/7 build either.  That's a power hungry chip but it's not that power hungry.

[Kylian]

1 minute ago, Ryan_Vickers said:

There's no point in using peltier, this has been looked at many times before, they have limited cooling power, consume a ton of power to do what they can, and cost a lot for no benefit over a traditional system.  There is a legitimate reason to use phase change but that doesn't necessarily mean it's the right choice for a normal 24/7 build either.  That's a power hungry chip but it's not that power hungry.

The reason I am looking at peltier is they can reach cryogenic temps for my understanding and phase change can get around -30 at best and I am looking for some thing to cool 24/7 to around -100 and below.
I have seen L&L cooling for ln2 how ever this is not a good choice for 24/7 (l&l cooling makes semi closed loop systems for ln2)

[vanished]

Just now, Kylian said:

The reason I am looking at peltier is they can reach cryogenic temps for my understanding and phase change can get around -30 at best and I am looking for some thing to cool 24/7 to around -100 and below.
I have seen L&L cooling for ln2 how ever this is not a good choice for 24/7 (l&l cooling makes semi closed loop systems for ln2)

I expanded on this in my edit above (sorry, posted too quickly I guess ) The issue with peltiers is, while they can develop a significant temperature differential between the hot and cold side while sitting in air, once you start dumping a bunch of heat into one side, they end up just shuttling it along while maintaining a completely unremarkable temperature, very much like a block of aluminium or any other good conductor.

[Kylian]

1 minute ago, Ryan_Vickers said:

I expanded on this in my edit above (sorry, posted too quickly I guess ) The issue with peltiers is, while they can develop a significant temperature differential between the hot and cold side while sitting in air, once you start dumping a bunch of heat into one side, they end up just shuttling it along while maintaining a completely unremarkable temperature, very much like a block of aluminium or any other good conductor.

are there any other options to get to cryogentic temps without having to keep refiling some thing like ln2

just give it a ton of power and it shut work. about power consumption i dont care if needed i can jst use a few kw if i want to

[Pixel5]

7 minutes ago, Kylian said:

The reason I am looking at peltier is they can reach cryogenic temps for my understanding and phase change can get around -30 at best and I am looking for some thing to cool 24/7 to around -100 and below.
I have seen L&L cooling for ln2 how ever this is not a good choice for 24/7 (l&l cooling makes semi closed loop systems for ln2)

peltier can not reach anything close to cryogenic temps.

they create a temperature differential so if you want the cool side to be at -100°C at all times you will need to cool the hot side to at least -70°C so you will need constant dry ice cooling on the hot side.

 

also what do you want to do with this why do you want -100C on a CPU that doesnt overclock very well?

[Kylian]

5 minutes ago, Pixel5 said:

peltier can not reach anything close to cryogenic temps.

they create a temperature differential so if you want the cool side to be at -100°C at all times you will need to cool the hot side to at least -70°C so you will need constant dry ice cooling on the hot side.

 

also what do you want to do with this why do you want -100C on a CPU that doesnt overclock very well?

Looking at this it seems they can https://www.quora.com/How-cold-can-a-Peltier-device-get

for overclocking if there is a other cpu with a ton of cores and that socket that can overclock realy well at those temps then i will choose that cpu

[Kylian]

2 minutes ago, Kylian said:

Looking at this it seems they can https://www.quora.com/How-cold-can-a-Peltier-device-get

for overclocking if there is a other cpu with a ton of cores and that socket that can overclock realy well at those temps then i will choose that cou

as far i know you can stack peltier to get them colder so to get -100 you use 3-4 peltier and the phase change on top.

well that was the idea

[Pixel5]

4 minutes ago, Kylian said:

Looking at this it seems they can https://www.quora.com/How-cold-can-a-Peltier-device-get

for overclocking if there is a other cpu with a ton of cores and that socket that can overclock realy well at those temps then i will choose that cou

yes it is possible to do with a stack like they use but how do you want to cool the hot side?

 

ifd you stack peltier elements you will lose efficiency more and more so to cool a 100W CPU to -100°C you will easily consume multiple kWh just for the peltier elements and ALL of that power plus the 100W of the CPU will need to go somewhere you will need an industrial grade chiller to cool that hot side so why not use that chiller on the CPU directly?

[Kylian]

1 minute ago, Pixel5 said:

yes it is possible to do with a stack like they use but how do you want to cool the hot side?

 

ifd you stack peltier elements you will lose efficiency more and more so to cool a 100W CPU to -100°C you will easily consume multiple kWh just for the peltier elements and ALL of that power plus the 100W of the CPU will need to go somewhere you will need an industrial grade chiller to cool that hot side so why not use that chiller on the CPU directly?

A chiller can not go below 0 it is using water. multiple KW well i am already consuming at least 60KW 24/7 so...

but if there are any good options to get to those temps with out geting to bulky like a ln2 compressor or a whole room with a cool instalation. the only other thing i know can get super cold is amonia cooling but i have no idea how it works

[Pixel5]

how do you think your freezer works? thats not different from an industrial grade chiller which absolutely can go below zero degrees.

 

take a look at this video to understand why doing what you want with peltier elements is a bad idea.

 

 

[Kylian]

2 minutes ago, Pixel5 said:

how do you think your freezer works? thats not different from an industrial grade chiller which absolutely can go below zero degrees.

 

take a look at this video to understand why doing what you want with peltier elements is a bad idea.

 

 

a freezer works exactly like phase change not a chiller as far i know?

a chiller is not using a compressor and gasses a phase change/freezer does. also the reason you cant put a freezer on its side.

 

[vanished]

23 minutes ago, Kylian said:

are there any other options to get to cryogentic temps without having to keep refiling some thing like ln2

just give it a ton of power and it shut work. about power consumption i dont care if needed i can jst use a few kw if i want to

I'm far from an expert on this but I'm not aware of any such system, and I have to imagine if one existed, extreme overclockers would use it rather than fuss around with LN2 and dry ice, etc.

[Kylian]

1 minute ago, Ryan_Vickers said:

I'm far from an expert on this but I not aware of any such system, and I have to imagine if one existed, extreme over lockers would use it rather than fuss around with one and dry ice, etc.

well there do exist semi closed systems only thing needed would be a compressor for the ln2 but that would get to bulky

[Kylian]

1 hour ago, Pixel5 said:

how do you think your freezer works? thats not different from an industrial grade chiller which absolutely can go below zero degrees.

 

take a look at this video to understand why doing what you want with peltier elements is a bad idea.

 

 

i found this

https://www.digitalstorm.com/cryo-tec.asp

[Pixel5]

29 minutes ago, Kylian said:

i found this

https://www.digitalstorm.com/cryo-tec.asp

yes systems like this do exist but not for -100°C and it still doesnt make any sense to use them for 24/7 operation.

 

the reason why nobody makes closed loop LN2 systems is thats its absolutely not practical and has no real world use for computers, if you want to overclock you do this to push the limits for short bursts and to get the highest scores, for normal overclocking you get a water cooler and be done with it.

[Kylian]

2 minutes ago, Pixel5 said:

yes systems like this do exist but not for -100°C and it still doesnt make any sense to use them for 24/7 operation.

 

the reason why nobody makes closed loop LN2 systems is thats its absolutely not practical and has no real world use for computers, if you want to overclock you do this to push the limits for short bursts and to get the highest scores, for normal overclocking you get a water cooler and be done with it.

a multi stage phase change might be able to go below -100 but i have not found them commercialy only single stage.

and it is not just for overclockign also just for fun and overclocking 

[SolarNova]

4 hours ago, Kylian said:

a multi stage phase change might be able to go below -100 but i have not found them commercialy only single stage.

and it is not just for overclockign also just for fun and overclocking 

if ur really interested in going into a 24/7 subzero build u can try my design. I have my project on hold atm.

 

You will need to build urself :

a phase change unit using R-1270 refrigerant (Propene) its the cheapest easily available, and coldest, refrigerant u can get for a single stage design. Which includes a plate heat exchanger.

a large (20-30litre) heavily insulated reservoir

a sealed, airtight PC case with 2 air hose attachments

 

You'll be using a dual loop. 1 to loop XT-1 coolant mix (58/42) between the reservoir and plate heat exchanger. And a 2nd 'normal' loop for the PC (from the reservoir)

 

The PC case u will need to fill with Nitrogen, hence the sealed airtight case, to ensure no condensation or icing occurs. I'd stil lsugest atleats some 'normal' subzero protection for the motherboard, GPU and CPU however, just incase.

 

If the chiller and reservoir are suitable insulted and build correctly u should be able to maintain -40c coolant.

 

You will want to run the 'chiller' system 24/7 using a temp controlled power unit so the reservoir is always at -40c or lower. The large reserve of chilled coolant will help stop the phase unit from being on all the time.

 

 

Note: You will need to take care with ur waterblock choices and will likely have to replace all o-rings with ones suitable for subzero temperatures. Alternatively solder the blocks and tubing.. Dont use acrylic topped blocks and use ether Copper tubing or glass tubing for the PC side loop if u want hard tubing,

 

[Nicklordzero]

You should actually go with a Chiller like LINUS did. I've tried doing such thing and funny thing is.... IT ACTUALLY ROCKS! I was able to cool a GTX Titan V & i9-9900K to 2c on major overclocks. Though you will need tons and tons of foam padding and insulator (vaseline) to keep it all from dying from liquid damage.

[bignaz]

If you have to ask this then sub ambiant cooling is water higher then your current skill level. 

 

Pelter? You list no info. 10w or something like a 300w? 

 

That little devil can hold 300w at -30c.   That 1950x will be around 220-300. I seen some clock pretty high and hit 300. You will need atleast a 300w tec and now you got all that heat that has to be dumped. -100c to -150c just won't happen your on a limited set up. 

 

Personally I would go with a 2 or 3 stage cascade and do it right. 

[Kylian]

9 hours ago, bignaz said:

If you have to ask this then sub ambiant cooling is water higher then your current skill level. 

 

Pelter? You list no info. 10w or something like a 300w? 

 

That little devil can hold 300w at -30c.   That 1950x will be around 220-300. I seen some clock pretty high and hit 300. You will need atleast a 300w tec and now you got all that heat that has to be dumped. -100c to -150c just won't happen your on a limited set up. 

 

Personally I would go with a 2 or 3 stage cascade and do it right. 

i have never build a Cascade unit before looking it up i find limid info and see that it is the same as phase change? corect me if i am wrong. Also it would be nice to fit in to the case (thinking of going with the thermaltake 900). A small unit next or below it would not be a problem but if i have to make a huge box of some sort it will be a bit mutch

[bignaz]

There's tons of info and there's no way around it. You want lower temps your going to have to go bigger.  Also note some cpu's have cold bugs.

 

 

And if your trying for a 24/7 5+ghz oc this is an absolute horrible idea.  Temps is the lest of the concern as the voltage your going to need is going to put way to much current on that cpu and that's what's going to kill it over time.  No way around this. 

 

 

Honestly your best bet is to just put the LD2 on it make sure everything is insolated front and back and done right and just bee happy around -30 to -50c.  Colder you need a bigger system and if I wanted that I would go for a minimum of a 2 stage. Back in the day on xs a guy did a custom 3 stage and it was pretty nice.