gabbax
05-17-2004, 04:44 AM
Air, water, compressors , dry ice and liquid nitrogen.
This is some well-known conceptions in cooling for processors today.
Todays processors are getting faster and thus producing more heat
and requires better cooling.
In this article/guide I will try to show some pictures and text that will describe
how to use dry ice or as the right name is CO 2.
I will point out that neither I will not take any responsibility
for any injuries that can occur to persons/hardware.
At -70 degrees Celsius the ice transforms to gas when it melts. As we all
know carbon dioxide is not healthy in vast amounts, so when you use this
method to cool down GPU/CPU make sure that ventilation is sufficent at all times.
The main purpose of this article is to give you overclockers out there some clue how to use dry ice, and what is called to be "extreme cooling" . In this category I will name a few: dry ice, cascade and liquid nitrogen. Using dry ice or as I call it DICE ( Dry Ice) requries some finess and timing.
What you will need first is a receptacle which is to be placed on the CPU/GPU. Preferable in copper because it conducts heat/cold quite good. I will not make any claims for measurements on the receptacle, but what you will be using is the bottom of the container. On that you hardsolder a copper-pipe in sufficent dimensons/length. It should have similar design as a waterblock , aka " fins" which gives broader contactarea which in turn gives better cooling.
Next step is giving right liquid because it will transfer cooling better than pouring DICE directly into the pipe.
At -78 degrees water is out of the question, the same goes for acetone and glycol. I have experience of two liquids, those can be easily purchased at closest gasstation. T-Red or Red ethanol freezes around -85c , the same goes for T-Blue. These shall be concentrated, otherwise they are useless.
Next are T -99.5 which is 99.5 ethanol. I have personally verified these temperatures directly at the manufacturer, sure it can be some minor differences but the temperatures should be correct. The first time I used T-Red which gave good temperatures but not as good as I thought. The reason could be the 5 % water in the T-Red which is reducing the performance, but probably they all are the same. In this particular test I used T- 99.5 which in use will need good ventilation.
If the receptacle has the dimension of 2 " or 50 mm you count 2 " of liquid in the pipe, it is a bit hard to know exactly but count 5-8 cm from the bottom to the edge of the liquid. These numbers are not stated and will need testing.
It will need some feelings when using dry ice, too much liquid leads to less cooling and the same goes for dry ice. Then the cold has nowhere to go, it shall all be measured so it will be a perfect combination of liquid and dry ice.
CPU-Container
http://w1.825.telia.com/~u82509108/DIcpu.JPG
http://w1.825.telia.com/~u82509108/DIcpu2.JPG
http://w1.825.telia.com/~u82509108/DIcpu3.JPG
GFX-Container
http://w1.825.telia.com/~u82509108/DIGfx2.JPG
http://w1.825.telia.com/~u82509108/DIGfx.JPG
Where do i buy dry ice ?
I do know that AGA in Sweden sells dry ice in " pellets" and in pulleys. The best are pellets because we want a big contact-area. Pulleys are ok but requires crossing.
So what is the price ?
I can not know what your receptacles/containers will cost to produce. If you are skilled then perhaps you will make them yourself and saving some cash. The dry ice here in Sweden is in my opinion cheap in comparison price/performance. I will point out that different stores have different prices so watch out. I personally puchasing no less than 10 kgs of dry ice pellets that is packed in a big box that contains 5kgs of pellets. An aprox price here in Sweden are 400 Sek inc VAT for 10kgs.
http://w1.825.telia.com/~u82509112/12.jpg
Performance/Temperature then ?
Because I doesn`t have any external thermometer I was forced to use the BIOS instead. After a while the temperature begins to decrease at the same time I pour down the DI pellets. I archived -35c and that is quite good, but I must point out that the BIOS figures is not to trust. With a compressor I archived -19c so the difference is quite big. The biggest advantage except from the extreme cooling dry ice gives is that it resist the heat from GPU/CPU.
I would say that Dry Ice cooling is way ahead of the compressor cooling in terms of overclocking/benchmarking. I lost just 1c in 32 secs of benching, a compressor looses 5 times of that. This could be the thing that gives you the edge in terms of benching/overclocking.
But what then ? I can`t run Dry Ice 24/7 ??
No that`s right, how would it look with a big truck standing outside your flat constantly ? In that the compressor/air/water cooling wins a big victory. Nah I see Dry Ice as a fun and different way of cooling and cost a fraction of a compressor. And is better for short times of benchmarking/overclocking. The advantage of a compressor is that you will have constant cooling power 24/7.
Consumption ?
Well it all depends, but we are not talking about a mere 10 mins and it is all over. At this try I had 10kgs of Dry Ice and a bottle of T-99.5 and I finished off 6-7kgs in aprox 7 hours of benchmarking and in effective benching ca 4-5 hours. So 10 kgs is enough for a weekend of benchmarking. The thing about Dry Ice is that it vapours by fuming. You will need to use it instantly because the volume is constantly decreasing. If it is cold it will last longer.
Condensation/Isolation
Hehe do I really need to say anything ? :-) If you are going to isolate do it properly. It is probably a no-ending job. I have only isolated the graphics-cards receptacle because of the frost wich in turn can drop onto the motherboard and then.... well do I really need to say ?? I used a black garbagebag, cut out holes for the AGP-slot and the CPU socket. It can " sputter " a bit if you pour too much of Dry Ice, so begin with little and increase slowly. After finished benching/overclocking turn it all off and dismantle the receptacles and canistors from the GPU/CPU, ( be careful because the CPU has a nasty habit of sitting too thight with the receptacle). Loosen all cables etc.... in with the stuff into the OVEN and turn it to 50c and wait ca 1-2hours to avoid any short circuiting.
Ex....of how u can do it
http://w1.825.telia.com/~u82509112/2.jpg
http://w1.825.telia.com/~u82509112/4.jpg
http://w1.825.telia.com/~u82509112/7.jpg
http://w1.825.telia.com/~u82509112/8.jpg
http://w1.825.telia.com/~u82509112/9.jpg
Final Words
Well what shall I say ? I´ve tried aircooling, watercooling and TEC/peltier. But Dry Ice is something extra because it requires something extra from you when using it and of-course it is very good cooling-method. It also gives the user a clue how to use the most xtreme ways of cooling todays CPU/GPU:s , i.e LN2/Liquid nitrogen etc...
I just say that if you have the possibility of trying, do it don`t hesitate for a second !!
Some Pics are coming up later this weekend, enjoy!
This is some well-known conceptions in cooling for processors today.
Todays processors are getting faster and thus producing more heat
and requires better cooling.
In this article/guide I will try to show some pictures and text that will describe
how to use dry ice or as the right name is CO 2.
I will point out that neither I will not take any responsibility
for any injuries that can occur to persons/hardware.
At -70 degrees Celsius the ice transforms to gas when it melts. As we all
know carbon dioxide is not healthy in vast amounts, so when you use this
method to cool down GPU/CPU make sure that ventilation is sufficent at all times.
The main purpose of this article is to give you overclockers out there some clue how to use dry ice, and what is called to be "extreme cooling" . In this category I will name a few: dry ice, cascade and liquid nitrogen. Using dry ice or as I call it DICE ( Dry Ice) requries some finess and timing.
What you will need first is a receptacle which is to be placed on the CPU/GPU. Preferable in copper because it conducts heat/cold quite good. I will not make any claims for measurements on the receptacle, but what you will be using is the bottom of the container. On that you hardsolder a copper-pipe in sufficent dimensons/length. It should have similar design as a waterblock , aka " fins" which gives broader contactarea which in turn gives better cooling.
Next step is giving right liquid because it will transfer cooling better than pouring DICE directly into the pipe.
At -78 degrees water is out of the question, the same goes for acetone and glycol. I have experience of two liquids, those can be easily purchased at closest gasstation. T-Red or Red ethanol freezes around -85c , the same goes for T-Blue. These shall be concentrated, otherwise they are useless.
Next are T -99.5 which is 99.5 ethanol. I have personally verified these temperatures directly at the manufacturer, sure it can be some minor differences but the temperatures should be correct. The first time I used T-Red which gave good temperatures but not as good as I thought. The reason could be the 5 % water in the T-Red which is reducing the performance, but probably they all are the same. In this particular test I used T- 99.5 which in use will need good ventilation.
If the receptacle has the dimension of 2 " or 50 mm you count 2 " of liquid in the pipe, it is a bit hard to know exactly but count 5-8 cm from the bottom to the edge of the liquid. These numbers are not stated and will need testing.
It will need some feelings when using dry ice, too much liquid leads to less cooling and the same goes for dry ice. Then the cold has nowhere to go, it shall all be measured so it will be a perfect combination of liquid and dry ice.
CPU-Container
http://w1.825.telia.com/~u82509108/DIcpu.JPG
http://w1.825.telia.com/~u82509108/DIcpu2.JPG
http://w1.825.telia.com/~u82509108/DIcpu3.JPG
GFX-Container
http://w1.825.telia.com/~u82509108/DIGfx2.JPG
http://w1.825.telia.com/~u82509108/DIGfx.JPG
Where do i buy dry ice ?
I do know that AGA in Sweden sells dry ice in " pellets" and in pulleys. The best are pellets because we want a big contact-area. Pulleys are ok but requires crossing.
So what is the price ?
I can not know what your receptacles/containers will cost to produce. If you are skilled then perhaps you will make them yourself and saving some cash. The dry ice here in Sweden is in my opinion cheap in comparison price/performance. I will point out that different stores have different prices so watch out. I personally puchasing no less than 10 kgs of dry ice pellets that is packed in a big box that contains 5kgs of pellets. An aprox price here in Sweden are 400 Sek inc VAT for 10kgs.
http://w1.825.telia.com/~u82509112/12.jpg
Performance/Temperature then ?
Because I doesn`t have any external thermometer I was forced to use the BIOS instead. After a while the temperature begins to decrease at the same time I pour down the DI pellets. I archived -35c and that is quite good, but I must point out that the BIOS figures is not to trust. With a compressor I archived -19c so the difference is quite big. The biggest advantage except from the extreme cooling dry ice gives is that it resist the heat from GPU/CPU.
I would say that Dry Ice cooling is way ahead of the compressor cooling in terms of overclocking/benchmarking. I lost just 1c in 32 secs of benching, a compressor looses 5 times of that. This could be the thing that gives you the edge in terms of benching/overclocking.
But what then ? I can`t run Dry Ice 24/7 ??
No that`s right, how would it look with a big truck standing outside your flat constantly ? In that the compressor/air/water cooling wins a big victory. Nah I see Dry Ice as a fun and different way of cooling and cost a fraction of a compressor. And is better for short times of benchmarking/overclocking. The advantage of a compressor is that you will have constant cooling power 24/7.
Consumption ?
Well it all depends, but we are not talking about a mere 10 mins and it is all over. At this try I had 10kgs of Dry Ice and a bottle of T-99.5 and I finished off 6-7kgs in aprox 7 hours of benchmarking and in effective benching ca 4-5 hours. So 10 kgs is enough for a weekend of benchmarking. The thing about Dry Ice is that it vapours by fuming. You will need to use it instantly because the volume is constantly decreasing. If it is cold it will last longer.
Condensation/Isolation
Hehe do I really need to say anything ? :-) If you are going to isolate do it properly. It is probably a no-ending job. I have only isolated the graphics-cards receptacle because of the frost wich in turn can drop onto the motherboard and then.... well do I really need to say ?? I used a black garbagebag, cut out holes for the AGP-slot and the CPU socket. It can " sputter " a bit if you pour too much of Dry Ice, so begin with little and increase slowly. After finished benching/overclocking turn it all off and dismantle the receptacles and canistors from the GPU/CPU, ( be careful because the CPU has a nasty habit of sitting too thight with the receptacle). Loosen all cables etc.... in with the stuff into the OVEN and turn it to 50c and wait ca 1-2hours to avoid any short circuiting.
Ex....of how u can do it
http://w1.825.telia.com/~u82509112/2.jpg
http://w1.825.telia.com/~u82509112/4.jpg
http://w1.825.telia.com/~u82509112/7.jpg
http://w1.825.telia.com/~u82509112/8.jpg
http://w1.825.telia.com/~u82509112/9.jpg
Final Words
Well what shall I say ? I´ve tried aircooling, watercooling and TEC/peltier. But Dry Ice is something extra because it requires something extra from you when using it and of-course it is very good cooling-method. It also gives the user a clue how to use the most xtreme ways of cooling todays CPU/GPU:s , i.e LN2/Liquid nitrogen etc...
I just say that if you have the possibility of trying, do it don`t hesitate for a second !!
Some Pics are coming up later this weekend, enjoy!