microfin wb made from akasa ak360/370

[leejsmith]

Ok, i ran the numbers, and there is indeed a problem: the ratio of the hydraulic equivalent of the outlet area, to the tube ID, is about 2:1, which means that the coolant is indeed being sucked out.

I'd recomend either using a smaller tube, or a bigger cup. If you follow Cathar's graph (above) you'll have some good guidelines, for those dimensions/proportions.

so my options are.

bigger cups with 2mm tube. but i can only go upto 3mm cups and that would make the number of cups smaller in the same area and also reduce the flow yes.

smaller tube with 2,5mm cups wich i can do as i have the 1/16 tube and now have a 1/16 drill bit. i did dome tests and the tube is very tight and will require some force but that should mean i dont need to glue them in place and there will be no chance of them being of center to the cups.

thanks Ben

[leejsmith]

after my visit to a local machine shop (in the back of some ones yard) I found the cost of cutting my 10mm copper down to 3mm was more than some new 3mm copper so i orded it and will get it later today. Its 50mm x 1m i think so i only have to make 1 cut each time.

the shop had a cnc machine (handy to know) and costed maching the copper base and acrylic top with just the cups , tube holes (no tubes they coudnt do anything that small) and screw holes. I would then finish the blocks my self But at £50 per unit i cant afford it at the moment after all the tools bought. besided i could buy a cascade at that price.

dont think it matters anyway i am getting better at drilling the cups and top anyway.

My idea of starting with a 1.5mm first works. I also found that if i make the 1.5mm tube holes in the acrylic first and then screw it to the copper to make the first marks into the copper i get very good results.

As i have bigger cups covering a larger area i dont want the outer corner cups to suffer from lower water velocity so i have changed the cup array to a rounder shaped with a width of 20mm.



the cups are still 2.5mm spaced 3mm apart. As i will have 3mm copper plan to make the cups 2.5mm deep wth flat bottom.

[leejsmith]

my joystick temp software is working very nice. I found the readings woudl fluctuate a lot so i added changed the program to take 30 readings every second and pass the average temps to the program. this way i get very stable readngs wich also helps with calibraton.

here is a sample graph i produced from one of my cup blocks againts the maze 3. both are delta temps cpu - air temps.
(note this is not actual core temps but a thermster stuck to the bottom of the cpu)



Orded some better thermisters that have have a tolerance between temps 0C and 70C of +-0.2C.

once i have removed all the little glitches if anyone wants it just drop me a PM. Also if there is anything anyone thinks could be a usefull feature i could add them too.

I have a 20x4 lcd in my pc and plan to display the 4 temp readings on that direct from the program. It could also pass them to MBM5 shared memory.

[bigben2k]

Seeing as we're going over that debate in another thread, I'd like to know how you put that joystick reader together! PM me, please?

Nice work on the bp.

[leejsmith]

Picked up the copper from the machine shop today plus some more 1/16 tube and some 3/64 x 0.006 i could now go smaller on the cup size wth the new tube.

the tubes are from a local model shop but they get it from a company in the us called special shapes.

http://www.specialshapes.com/brasstubing.asp

First i made a copper template for the base and jet layer so i can make them the same every time and i know the tops and bottom match too.



here you can see a finshed base with the 2.5mm cups with flat bottoms. Note the brass tubes used to align the top.




I got a little carried away wth the end mill trying to get the base as thin as i could without pushing the copper but failed on one cup at least you can see how well the tubes sit in the center of the cups.



will try again in the morning.

(edit: just remeberd can still use this base for a direct die version)

[bigben2k]

Nice.

[MadDogMe]

That looks so much better!. Very even for a homebrew . How small do you think you could go with the cups while maintaning 'order'?,2mm?, 1.75mm?...

[leejsmith]

Originally posted by MadDogMe
That looks so much better!. Very even for a homebrew . How small do you think you could go with the cups while maintaning 'order'?,2mm?, 1.75mm?...

do you mean the size of the cups or the space between them ?

the size of the cups will depend on what tubes i can get. As i am using the smaller tubes 1/16 OD i shouldnt have the problem as with the 2mm tubes that ben pointed out.

as for the spaceing i could try to go perhaps 4 or 3mm instead of 5mm. The 5mm is hard enough.

[leejsmith]

well here is the latest tub cup block.




the tubes are smaller than the ones i used before these are 1/16x0.014. I have the tubes 1.75mm inside the cups.

the cups are 2.5mm deep with almost flat bottoms i didnt want to push the copper out like before.

the jet plate is not perfect 1 row is off center but i should see some improvement from the last one i made. As i now have the template i can make the jet plate's until i get one i am happy with.

just need to seal it up and give it a test. cant wait to see how it performs as i have put a lot of time and effort into ths one not to mention the cost of a piller drill and cross vice.

only wish i could make a good groove for o-rings it would make life so much simpler

[bigben2k]

Excellent work!

[leejsmith]

thanks ben.

just did a leak test and visualy it looks very restrictive compared to my other blocks.

I just put it on my p4 test setup so will have a first test result later.

[MadDogMe]

Found them!...

Re: O~ring Grooves,
Could'nt you cut four grooves into the lexan top then finish them up with fine wet and dry. Or how about your router/dremmel attatchment?. The grooves don't HAVE to be in the copper, as long as there is room for them to expand/distort sideways, and the groove is abit 'shorter than the o~ring is high (so it compresses slightly) that's all that's needed. It does'nt need to be 'one continuous enclosed shape' either it can be a # cut like I said above, as long as the 'central square' has a good finish for the o~ring to mate against that's all that matters...

Looking very good as well. I'm wanting an X~Y Vice now! ...

[MadDogMe]

Lee!, what did you use to 'mill out' the lexan tops?. I remember you saying it was a Dremel but was it with the router attatchment?, and which/what kind of 'bit' did you use?...

[leejsmith]

i used the dremel with the router bench to mill the lexan.

re the o-ring it's not easy doing it by hand to make a nice round / oval shape for it to sit it. And doing it on the x-y vice would be just as hard.
i could practice a bit first.

[MadDogMe]

That's just it!, it does'nt have to be round. It seals just as well when the o~ring makes 90deg turns. Or do you mean the cross section?, cause that is usualy done 'square' as well like this: |_| not U shaped, a U shape would prevent the o~ring expanding/compressing properly...

Can you do turns on the X~Y table?. Do you turn both hadles at once or something?...

[ozzy7750]

i have read this thread from the beginning, and i am sure i have read it somewhere, but cant find it again.

where did you get the tubes from lee? and how did you cut them, with a pipe cutter (dunno if mine go that small) or just with a hacksaw?

any results yet? i think i will try to do one similar to this

[leejsmith]

Originally posted by ozzy7750
i have read this thread from the beginning, and i am sure i have read it somewhere, but cant find it again.

where did you get the tubes from lee? and how did you cut them, with a pipe cutter (dunno if mine go that small) or just with a hacksaw?

any results yet? i think i will try to do one similar to this

first welcome to the forums and many thanks for reading through this thread.

i get the tubes from a local model shop but they are made by a us company.

http://www.specialshapes.com/brasstubing.asp

Unlike the fin style blocks you just cant drill some holes into the base and point the tubes into them and hope it will work.
there is a balance between the size of the cups and the ID of the tubes. You need to have the total ID of all the tubes between 26mm2 and 30mm2.

the cups need to be perfectly spaced apart as the jet impingment only cools a small area and the cup base needs to be flat.

have at look at the

http://forums.procooling.com/vbb/sho...&threadid=6101

started by bigben2k that is where this idea started from and also

http://forums.procooling.com/vbb/sho...&threadid=6666

for an idea on what this design can achive when it's made by some one that knows what they are doing.

i used a diamond edge glass cutting wheel in a dremel sitting in a dremel router bench to cut the tubes and then hand file them.

unless you have access to a cnc machine your going to need a cross vice (or x-y vice) and a bench drill (or piller drill) for the base and tube holes doing it by hand is not posible. To make the first mark in the copper use a 1.5mm drill other wise the drill might slip giving uneven cups. Then use 2mm and 2.5mm followed by a 2.5mm end mill for the flat cup bottom.

on my current attempt compared to a maze 3 i have the following
on a p4 1.7 1.5C lower than the maze
on a athlon 1100 2C lower than the maze

good luck

[leejsmith]

ok made some more changes to the block.

first i think the last one didnt have enough jet orifice area with the smaller tubes.
please correct me if i am wrong but here are my calculations

the tubes have an OD of 1/16 or 1.5875mm

the tubes have walls 0.014 inches thick or 0.355mm

giving a radius of 0.43875mm

so each jet had an area of 3.14*0.43875*0.43875 =0.6044mm2

with 42 tubes total area is 25.38mm2

on this new version i have increased the number of tubes to 48 giving a jet orifice area of 29.0112mm2

The tubes are 0.5mm from the cup base or 2mm into the cups.




some jet action from a 1000l/h pump with 1.3m max head



will post some results later.

i might not be able to refine this design much more as these tubes are the smallest ID i can get so far. I can get smaller OD with thinner walls but the ID is about the same.

i could try the smaller OD tubes with 2mm cups.

[Gulp35]

I can get some 1/32" OD tubes with (i think i remember) .006 inch walls made out of copper here in the states. PM me for more info.

[ozzy7750]

wow, it is looking really good now.

the tubes come really small from Special shapes, but working with them at their smallest would be too hard in my opinion. I think you have picked the right size to use.

Now i just have to find some, I live in New Zealand, finding materials like this is impossible most of the time

[Cathar]

Originally posted by leejsmith
The tubes are 0.5mm from the cup base or 2mm into the cups.

Okay, this is your current biggest issue.

The tubes are WAY too close to the base of the cups.

Go back to google and do a little more reading on jet impingement...

Basically the way you have it setup now, you're probably losing around 30% of the jet's cooling potential due to its "too close" proximity to the cup base.

[leejsmith]

is the jet orifice area better now ?

i read through your thread over at ocau today and also the new deisng idea thread.

does the 4d rule apply when the jets are in the cups.

also is my 3mm base too thin ? should i have a thicker base allowing the tubes to be in the cups but far enough from the base for the jets form ?

i was almost at the point of giving up and just ordering a cascade from you but i spent too much on tools for this so i am going to carry on.

i found this block is 2C lower than the maze so about the same as the last one.

also i found the lower jets to be a problem my self today as i milled 0.5mm of the base of the last block that had the tubes 1.75mm into the base making them 2.25mm and the temps dropped back by 2.5C.

after some googling i found a study on jet arrays and there tests showed that the jet array provided the best heat transfer across a wide range of velocities when the height above the plate was 3d. this was for air would this also apply for water.

[leejsmith]

this morning i cut the tubes so they are only 0.5mm into the cups.

this now gives 2mm above the cup base.

after 3 tests this morning re-mounting each time the temps are almost the same as before i cut the tubes.

[leejsmith]

my block developement is sort of on hold for a while as i start a 5 week contract on a desktop rollout.

i did convert my main rig from 3/8 tube to 1/2 tube and installed the last block i made today.

Oh and i orderd a cascade too

BE I still plan on using the polycarb on a block and also make a cascade style gpu block at some point.

[ozzy7750]

I am going to try and make a gpu cooler the same way lee has doen his cpu.

Main body would be made from acrylic, with brass tubes as inserts.

A copper base would be attached to the bottom, and a 2mm piece of acrylic will be glued to the top. for barbs I will just use brass tube, with hose clamps done up really tight.

Water comes in central inlet in a 9mm tube, goes out two side in 7mm tube.

View from above:


view from below: