LN2 Evaporator Design and Reference Links

[nzaneb]

Reverse stepping, although at that height it will be impossible to machine.

 

[m0r7if3r]

Herm...well we're back to the post idea, which means it'll be slow to respond when poured into i think...that design also takes off a bunch of mass...I'd be interested to see analysis of both of them compared.

 

[Gautam]

I bet it'd perform fine, its just that the machining effort is unnecessary. You'd get a very similar feel with an unstepped base as well. You can get the same amount of mass with a flat base as a stepped base, obviously, and you can get the same amount of surface area simply by adding more holes.

btw

I agree that having a layout with a center hole probly better utilizes the space above the center of the die

Designs with a relatively low amount of mass in the dead center have historically not fared well.

 

[m0r7if3r]

Designs with a relatively low amount of mass in the dead center have historically not fared well.

This seems to almost directly contradict your earlier recommendation not to use a post...I know you know what you're talking about, so what option am I not seeing here?

 

[hokiealumnus]

I think he's saying to leave mass above the die by simply not drilling but a post is not necessary and may turn out to be excessively difficult to machine.

 

[EarthDog]

Reverse stepping, although at that height it will be impossible to machine.

Would making it a 2 piece allow the machining to happen?

What about holes on the EDGE of the pot as opposed to the center keeping more mass directly above the CPU for stable temps?

 

[nzaneb]

Would making it a 2 piece allow the machining to happen?

What about holes on the EDGE of the pot as opposed to the center keeping more mass directly above the CPU for stable temps?

Yes it would need to be two piece for that design to work. The length of the tooling required for that depth, would be much too large to do that intricate work. I believe this is probably why Vince went to Gemini style manufacturing. Tooling and time expense were cut dramatically.

The steps are quite exaggerated, and should really be cut in half.

Holes towards the edge would hold ALOT of potential, but the response would be slower... in theory.

The only reason for the steps G, was due to the fact that I don't like boring, lol, and am always a bit difficult. You are most likely right that you can get the same performance from a much more simple design. But then it would be just that, a simple design. A copper slug with a few holes drilled toward the bottom of it. And yes, I'm aware that most everything has been done before with regards to a Copper LN2 pot and this isn't even my pot

 

[ChanceCoats123]

Herm...well we're back to the post idea, which means it'll be slow to respond when poured into i think...that design also takes off a bunch of mass...I'd be interested to see analysis of both of them compared.

I fail to see how this design decreases mass... In the previous design, this shape was flipped over and cut out of the bottom, now it's the exact opposite, resulting in more mass, directly over the center.

Maybe I'm just crazy.

 

[Gautam]

I think he's saying to leave mass above the die by simply not drilling but a post is not necessary and may turn out to be excessively difficult to machine.

Pretty much, but furthermore what happens with a post is that there ends up being excessive mass in one place alone, but not much elsewhere, like in the Evo for example. The Evo overall does not have much mass nor surface area (when compared to the F1s). The post adds some mass but not enough to handle a modern CPU optimally. What ends up happening is that the container doesn't have the capacity to hold the CPU at a steady temp. The post is just a bandage fix, but it takes time to cool the post down, which adds a delay. So you end up with a slow pot that lacks the pull down power to keep a heavy heat load at a steady temp. It's a design that works well for a low heat load since a container even without a post would be able to hold the temp fine. The post would just add buffering and make the feel more comfortable, which is why it made sense a few years ago, though even then, I was frankly unsatisfied by the delay it caused.

On the other hand, adding a post to an already massive design like the F1EE ends up being vast overkill, and while the container easily would have the capacity to hold damn near any chip, it'd be a complete slug with the amount of time needed to pull down such an amount of mass, once again making it hard to control, albeit for a slightly different reason.

The bottom line is that you usually end up having the most responsive and easy to control containers when the mass is distributed roughly evenly across the base, with a *slight* bias towards the dead center. The one thing you want to absolutely avoid doing is hollowing out the region just above the die of the chip and leave large amounts of mass elsewhere. For this reason, the design in post #141 makes more sense than the one in #125. If you think about it, there is no good reason to leave more mass towards the edges of the container. It does no good there.

The only reason for the steps G, was due to the fact that I don't like boring, lol, and am always a bit difficult. You are most likely right that you can get the same performance from a much more simple design. But then it would be just that, a simple design. A copper slug with a few holes drilled toward the bottom of it. And yes, I'm aware that most everything has been done before with regards to a Copper LN2 pot and this isn't even my pot

Sometimes simple is all that's needed to get the job done. If you must make it complicated though, like I said, something inverted (though less exaggerated) than your design in post #141 would probably do well. I'd deem it mostly unnecessary, but it might help the overall feel of the container slightly.

 

[m0r7if3r]

I fail to see how this design decreases mass... In the previous design, this shape was flipped over and cut out of the bottom, now it's the exact opposite, resulting in more mass, directly over the center.

Maybe I'm just crazy.

we've gone from 2" of solid copper with holes in it and stepping above it to 1" of solid copper with side by side stepping and holes...looks like we've cut out 2 times the mass of the stepping to me.

 

[nzaneb]

looks like we've cut out 2 times the mass of the stepping to me.

^^^ precisely

 

[ChanceCoats123]

we've gone from 2" of solid copper with holes in it and stepping above it to 1" of solid copper with side by side stepping and holes...looks like we've cut out 2 times the mass of the stepping to me.

Ahhh.... I see it now.

Sorry Mort.

 

[dylskee]

I could machine nzaneb's concept but it would have to be in two pieces, I think that's why you see most of these in round stock. We could thread the end then mill the steps and make the second half in a lathe and screw the two pieces together. Other than that you wouldn't find a machine shop to make that design and if they did it would cost you $1500.!

 

[Bobnova]

Two pieces also means you can use far less copper. A copper slug to machine into a F1EE type pot is expensive. A copper slug to machine into a gemini type base costs far less, and the aluminum required to make the walls of the gemini costs a lot less then the copper for a F1EE would.
That, IMO, is the #1 goal. Vince probably cut his costs by 50% or more, at the price of a pot that leaks if you fill it.

 

[ChanceCoats123]

Which is why a circular base may work best. You could use the same threads as water coolers use (the ones that are meant to seal) and then the use of a seal would be greatly reduced.

 

[m0r7if3r]

what're you gonna use to seal it that's still good at -196c?

 

[I.M.O.G.]

Sealing isnt a problem if its circular, you just thread it and its good enough. If you machine the pieces with tight tolerances it isn't much of a problem with the square base either from what I've heard - I didn't think people were having much of a leakage problem with the gemini?

 

[Bobnova]

Fine threads+silicone grease works fine at dice temps, probably would at ln2 temps too.

 

[ChanceCoats123]

IIRC, it's called BSPT or British Standard Parallel Thread. The threads have a constant diameter as opposed to tapered threads which makes them seal much (much) better than NPT or National Pipe Tapered Thread.

Info.

 

[m0r7if3r]

http://en.wikipedia.org/wiki/British_standard_pipe_thread

bspt is tapered. you're thinking of BSPP, which is what we use for watercooling. It does not seal sans o-ring.