making my own ln2 pot, input appreciated

[Johnd0e]

alrite, i posted this over on ocn but i figured id double my chances of getting input by posting here too.

 

anyways, im making my own ln2 pot, as the title suggests, during breaks at work. so far ive taken measurements off my waterblock back plates to get things like diameter of the pot and were the holes need to be on the mounting bracket. only thing im at a loss for is placement of the thermal probe and how much surface area i should go for in the copper base? i believe its alot of surface area will make it a fast pot and too little will make it slow. but how much is to much and how little is to little?

 

anyways heres my cad drawing of what it will look like, pretty much just a round version of the gemini F1. consists of a copper base, aluminum canister, aluminum mounting bracket and aluminum extension.

 

 

overall diameter is 3.000" (can you tell im american), total height is 8.000". copper base is 2.000" thick.

 

EDIT:

i should mention to disregard all the holes in the copper base as thats just there as a visual aid for what ill be doing to gain the surface area(just like the kingpin pot i believe). depth, diameter and number of holes will change depending on surface area i decide on, hopefully with the help of others.

Edited April 5, 2016 by Johnd0e

[buildzoid]

The cool down speed is a combination of mass and surface area. The heat up is almost entirely tied to the mass of the pot. 1Kg of copper has a thermal capacity of 385J. So a 385W heat source would raise the temp by 1C in 1 second. Your pot's base with no holes will have a mass of about 2Kg. I'd just try to get as much surface area as possible while keeping the mass between 1Kg and 1.5Kg depending on what CPU you plan to use with the pot

[der8auer]

^^^^^this

 

Also try to avoid "round edges". You need sharp edges to break the leidenfrost effect. Just holes don't work that well.

[Johnd0e]

The cool down speed is a combination of mass and surface area. The heat up is almost entirely tied to the mass of the pot. 1Kg of copper has a thermal capacity of 385J. So a 385W heat source would raise the temp by 1C in 1 second. Your pot's base with no holes will have a mass of about 2Kg. I'd just try to get as much surface area as possible while keeping the mass between 1Kg and 1.5Kg depending on what CPU you plan to use with the pot

 

Thank you very much thats exactly the kind of help i was hoping for. Glad i posted here.

 

^^^^^this

 

Also try to avoid "round edges". You need sharp edges to break the leidenfrost effect. Just holes don't work that well.

 

Thank you for that info.

 

Cant say im educated on the leidenfrost effect, im aware its caused by very different temperatures coming into contact creating a barrier. Like cold water in a hot pan. Thats about as far as my knowledge goes so i guess ill have to do some more reading up on that.

 

By sharp edges are you speaking of the the corners where the horizontal and vertical faces meet?

 

Would machining circles into the block in a "bullseye" orrientation be more effective?

 

Thanks again.

[superpatodonaldo]

just to confirm what builzoid and der8auer said, to make a ln2 pot you have to use two rules:

 

In case of copper base

1) (°C/s) = [base weight (g)]*[0,385 (J/(g*°C))]/[load (W)]

2) delta temp (°C or K) = [base height (m)]*[load (W)]/( [390 (W/(m*K)]*[base area (m^2)] )

 

so you have to know on what CPU you'll use the pot, eventual CB

 

for example a not-so-height = (not-so-weight) base gives you a light base, fast pot

[Johnd0e]

so copper never showed up....apparently the guy who does all the ordering for the shop "forgot" to order my stock. anyways, i have enough copper to make 3 bases coming next week....i made sure he ordered it haha.

 

but in my spare time ive been looking at other bases on google and looking up more info on the leidenfrost effect.

 

as for the leidenfrost effect all i could find about "breaking" the barrier is an article from nasa or something explaining how they used nano particles to create a textured surface which resulted in the ability to make water boil over the leidenfrost effect point. Soooo since ill never have the means to do anything like that, i was thinking something along the lines of rough milling all the surfaces that will contact ln2, tapping all holes and possibly sand blasting the insides of the base to try to texture the finish.....at least ill try that on one base to see what happens.

 

heres what i figured for one of the bases, with weight compared to a complete solid base.

(i also changed my base height from 2.0" to 2.5" becuase the 2.0" base weighed ~1.5kg solid)

 

 

consists of an inner and outer ring milled into the base, outer ring has 16 x .25" tapped holes inside of it, and the inner ring has 8 x .25" tapped holes inside. center hole is .5" and also tapped.

 

i figure tapping will not only create more surface area then a straight hole but also give more sharp corners. not sure if that works, but i got the idea from the koolance base i think it was....i dont remember but somebody already did it is all i know.

 

thoughts, concerns, input?

[superpatodonaldo]

so many holes, don't know how it will work

 

I had a SF3D/Ryba pot, now I project a different base;

SF3D/Ryba is 50mm height with "holes" 25mm depth, my base will be about 50 mm without holes (holes just around the base);

 

when it will be ready I'll check the difference

 

that base You propose seems to be a fast base pot

[der8auer]

I did some research about the leidenfrost effect on my university a while ago. The leidenfrost temperature mainly depends on the temperature difference between the copper and LN2 but also the shape/surface of the copper.

 

This is an example (let's say Pot A). Section I is the part until the pot is completely filled with LN2. In Section II it turns into a linear cooldown which mainly depends on the surface area. In this case the cooldown was with 0.64 K/s. The linear cooldown (film boiling) will continue until you reach the leidenfrost temperature which was at about 50 K. Section III is called transition boiling and the LN2 will evaporate as soon as it hits the surface.

 

 

 

A lot of benchers know this behavior and we call it "glace the pot". At around -50 °C you put the heatgut or torch in the pot to create a thin ice layer on the surface of the pot. For some strange reason it immediately bypasses the leidenfrost point and goes to transition boiling. I tried to reproduce this by changing the surface structure of the copper. E.g. with grinding, polishing, etching and all that kind of stuff. The problem is that you will always have the ice structure in the end. So it doesn't really make sense to improve it that way.

 

 

The only thing you can change with surface area/structure is the linear area (Section II).

 

The structure has to be as sharp as possible with a lot of edges and not too fine. A thread works well but it has to be a thread with a certain diameter. M10 or above should work fine.

[superpatodonaldo]

Thank you Roman for your post, so, sadly, my poor english doesn't help me to understand that sentence

 

A thread works well but it has to be a thread with a certain diameter.

 

what do you mean?

Edited April 11, 2016 by superpatodonaldo
correction

[der8auer]

If you drill holes for surface area you should cut threads in there, too. Just to have more edges to break the film boiling.

 

[superpatodonaldo]

thank you

so you recommend to thread the holes, is it right?

[Johnd0e]

thanks der8auer. that helps tons, i dont think id have ever found that info searching google haha.

 

glad i was on the right track with threading the holes...if there is one thing im good at its google image search lol.

 

glad you pointed out the hole sizing and thread pitch before i started machining, i was originally planning on using 1/4-28 tapped holes around the perimeter and the inner circle, and then a 1/2-20 tap in the center....so equivalent to an m6 fine thread around the perimeter and inside holes and an m13 fine thread in the center. guess ill just redo it to be coarse thread and bump my outer holes to 3/8-16 thread (m10 equivalent)

 

thanks again, your a big help. material comes in tomorrow, so ill be starting on it asap.

[Johnd0e]

Finally got around to finishing up most of the work on the ln2 pot. next up is all the internal work on the copper, then sending out for nickle plating and anodizing.

 

 

 

[xxbassplayerxx]

Looks very nice!

[der8auer]

Really nice Reminds me of the days when I did the pots myself on the lathe Good fun.

[Johnd0e]

Thanks guys! Hoping it ends up performing good too, if not it can be a fancy looking paper wieght at least haha.

 

 

Rasparthe brought up a suggestion/critique over on ocn about moving the hold down lower to keep the threaded rods from moving around when tightening them causing the pot to move too. So im working on something to cure that from happening.

 

After that im going to go work on the copper block and order the springs, threaded rod and thumb nuts. Can anyone tell me how much force i should be aiming for? Id appreciate it alot.

 

 

Anyways thanks, im having fun making this haha.

[der8auer]

Total force between 100-400 N

[Johnd0e]

alrite, heres what im thinking for the base design.

 

 

green holes = 3/8-24 tapped

 

light blue = 3/8-24 tapped

 

blue = 1/2-20 tapped

 

red = .25" drill holes

 

 

 

and here is what im thinking of doing instead of moving my hold down lower. should definitely stop any kind of movement of the pot on the CPU when tightening down.

 

 

as always i appreciate any critique/input.

Edited April 27, 2016 by Johnd0e

[xxbassplayerxx]

As the pot gets cold, the rods get frosty and get stuck in the hold downs, which can make swapping processors or platforms difficult. Doubling the holes that the rods pass through is going to make this twice as difficult. I'd avoid it if possible.

 

This is more true with LN2 but can still be an issue with dry ice.

[Johnd0e]

i see, that makes complete sense. ill just move the hold down lower on the pot and be done with it. thanks bassplayer.

 

 

EDIT:

 

redid the drawing for the whole pot showing new hold down hieght. does this look better to you guys?

 

 

new height is 5 inches or 127mm.

 

Thats 1.375" (34.925mm) lower then what i currently have.

 

looking at pictures of pots on google, most seem to be roughly the same height as the Classified PCB or similiar graphics card, which is about 6 inches.

Edited April 27, 2016 by Johnd0e

[Johnd0e]

finished up the major machine work on the copper base today. didnt have time to debur and clean off all the oils/coolant. but here's what she looks like on the inside.

 

 

 

had to fight the copper a little bit to do those internal groves, broke a tool doing the bottom one, so now i have some "beautifying" to do...... but all in all went fairly easy.

[xxbassplayerxx]

Looks incredible! What's the current weight on it?

[buildzoid]

That looks like it will perform. You're gonna have to tell us a cool down to -196C time once it's done.

[der8auer]

nice work man Should work really well

[Johnd0e]

Thanks guys, cant wait to freeze this thing. Will definetly update with performance results.

 

Looks incredible! What's the current weight on it?

 

1.56kg