Making a CPU Pot

PeterPwned · May 28, 2011

Hi,

my benching rig is almost complete - all I need is a CPU pot. Now I don't feel like buying a pot from Koolance because 203,50 Euros (which amounts to 250-300 dollars) is just too much, but that is the pricing in Germany. I can't find any reasonably priced pots here - are there cheap European manufacturers that can be trusted?

I was thinking why not build it myself. Can't I just take a kitchen pot with a small diameter that fits on the socket and modify it? I could of course also make it out of a new piece of metal, but is there a point to that? Are there instructions for building one of these online? I actually really like challenges so I would almost prefer building it myself, if that is realistic.

Thanks,
Mark

Janus67 · May 28, 2011

From what I have noticed from looking at custom made CPU pots is that there is a lot of thought that goes into the creation of them.

To get an idea of what I mean check out:
http://www.xtremesystems.org/forums/forumdisplay.php?f=156

You may need to increase the filter of the last date to about a year to see a decent amount of them. Most of the pots are made from large blocks of copper and after being machined down weigh between 2-3.5kg. They also have different designs and features (slotted inside walls, bored holes in the bottom, screw-on tops, and then the mounting hardware as well).

I'm not sure if a home-made pot from 'around-the-house' items would be able to do what you want, although you may be able to get an OK result by buying some copper pipe+soldering on a pipe-end and so forth.

PeterPwned · May 30, 2011

What are the benefits of using a pot like this though versus using a copper pipe + pipe end. Technically the copper pipe should be just as capable of conducting heat.The only difference I have noticed is that the structure of these expensive copper block pots differs on the bottom of the inside of the pot. There seems to be more surface area there to enable better heat exchange. But will these you know 5-10 degrees make a difference? Isn't it the case that the colder you get the closer you get to the cold bug?

Voidn · May 30, 2011

The pot design on the inside is for more surface area (like you said) = responsiveness (quicker cool down, easier to heat up after cb). Copper is the best way to go, good thermal transfer. No kitchen stuff.

Seebs · May 30, 2011

PeterPwned said:
What are the benefits of using a pot like this though versus using a copper pipe + pipe end. Technically the copper pipe should be just as capable of conducting heat.The only difference I have noticed is that the structure of these expensive copper block pots differs on the bottom of the inside of the pot. There seems to be more surface area there to enable better heat exchange. But will these you know 5-10 degrees make a difference? Isn't it the case that the colder you get the closer you get to the cold bug?

Surface area and mass is what your copper pipe + pipe end would be lacking.

If you look at one of the commercially available pots you'll notice that the copper bases are thick walled and machined with multiple hole patterns and the such. I would guess that a copper pipe + pipe end would be just too light in mass to be able to hold any loads and the temperature swings would be so drastic that it would kind of defeat the purpose of subzero.

You could always look for Piotres or SF3D pots in Europe, but those are also going to be expensive. Or look for a used pot in different classifieds sections of all the benching forums; they are always popping up all over the place.

moocow · May 30, 2011

Duniek is a good guy to contact if you want an inexpensive euro-pot. He makes some pretty decent stuff for the money.

xsuperbgx · May 30, 2011

I made a home made one. I used a copper pipe and the copper insert from a intel c2d heatsink. The mass is low so it works best for lower power cpu's. It worked ok for short tests like cpuz and spi and wp32m. I used it for dice on amd and got some decent results.

ChanceCoats123 · May 30, 2011

The problem with making your own pipe-based pot is the HUGE lack of mass. Now, if you were to take a nice, solid piece of copper, (think: 1.5 to 2 inches thick by 2.5 inches in diameter) and add some nice surface area to it (think: drilling varying depths of holes) and then braze as copper pipe into that, you've got a more than ample pot. However, I wouldn't touch a regular pipe and cap "pot" with a ten foot pole.

xsuperbgx · May 30, 2011

You can make one and try it. Try it and you will get hooked and then you will have to buy a real pot.

moocow · May 30, 2011

Problem with that is that you can buy a real pot for not much more. A copper/ alu pot from Duniek is something like $70usd for people in europe.

ChanceCoats123 · May 30, 2011

moocow said:
Problem with that is that you can buy a real pot for not much more. A copper/ alu pot from Duniek is something like $70usd for people in europe.

Very good point. And another is that actual pots have much better resale value. Even if you don't like it, sell it to someone new for a minimal loss.

xsuperbgx · May 30, 2011

What cpu do have for benching?

PeterPwned · May 31, 2011

@Seba

Okay I get the surface area argument, but mass? Adding mass is about the easiest thing you can do. Get something heavy and attach it to the pot in a way that doesn't obstruct the cooling process.

Temperature swings? Can't find any cheap fluke or other thermometers for some reason, so its going to be a no thermometer thing the first few sessions anyhow. Technically I should be good though if I just make sure that there is a good amount of ice in the pot right?

I've been trying to look in different forum marketplace sections here in europe, but it turns out almost everywhere you need 100 posts to be able to even view the threads. No way am I going to take a day off to spam forums with posts.

@ChanceCoats

That would be crazy expensive. A 5cm by 5cm block of copper costs around 200 Euros lol

@moocow

Looks like you have a point there. Where do I find this guy?

@superbg

i7 930

hokiealumnus · May 31, 2011

Moved this to Extreme Cooling (with one week expiring redirect); might get a couple more views from people who don't frequent the benching section - though most of the two are one in the same around here.

Anyway, Duniek can be found at XS, though he doesn't seem too active these days. Piotres is more active and has his own web site selling his beautiful copper creations. He has photos of all of his containers on XS, just look click on 'find all threads started by piotres' in the 'stastics' tab of his profile and they could give you some ideas.

nzaneb · May 31, 2011

PeterPwned said:
@Seba

Okay I get the surface area argument, but mass? Adding mass is about the easiest thing you can do. Get something heavy and attach it to the pot in a way that doesn't obstruct the cooling process.

Temperature swings? Can't find any cheap fluke or other thermometers for some reason, so its going to be a no thermometer thing the first few sessions anyhow. Technically I should be good though if I just make sure that there is a good amount of ice in the pot right?

By mass, they are referring to cooling potential, not the overall mass of the pot. You will need cooling "mass" directly above the CPU, in order to prevent drastic temperature swings. If the copper pipe cap was 3/4" thick, it would perform MUCH better. Mass will allow the pot to maintain cold temperatures, surface area will allow you to cool the pot down faster (more contact with the cooling material). The correct balance is very dependent on MANY things (the cpu, the pot material, etc... another factor is what type of benchmark you plan on running)

NoL · May 31, 2011

And just to correct something I saw in here..

But will these you know 5-10 degrees make a difference?

The difference between a cap end pipe and a machined block pot will be more like 40-50C at load. A pipe cap will not be able to transfer the load to the dry ice. Remember that heat flows, not cold.

El<(')>Maxi · Jun 2, 2011

NoL said:
Remember that heat flows, not cold.

There is no distinction between hot/cold in conductance measurements, one will transfer at the same rate as the other.

NoL · Jun 2, 2011

Sorry Maxi, but regardless of the fact that it does not place direction upon flow, "cold" will never flow. Heat will always go from a higher temperature to a lower, it might not make a difference for many, but physics might get angry and squish us if we don't play to the rules

El<(')>Maxi · Jun 2, 2011

Go out in your shop and grab some copper pipe, then stick one end of it in a bucket of ice with your bare hand on the other end and wait about 5 sec...then come back here and tell me your response again

Bobnova · Jun 2, 2011

It all depends on how you look at it, physics says cold doesn't exist and it's all a matter of relative heat.

I say, it looks like it flows, feels like it flows, and acts like it flows

Regardless, mass is needed to store the cold (or store the lack of heat, either way) for sudden heavy heat loads.

Making a CPU Pot

Member

Benching Team Leader

Member

Member

Fronting as a Mod Member

Member

Benching Team Leader

d20 in a jacket

Benching Team Leader

Member

d20 in a jacket

Benching Team Leader

Member

Water Cooled Moderator

Senior Member

Member

Blank Senior Member

Member

Blank Senior Member

Senior Member

Similar threads