microfin wb made from akasa ak360/370

[leejsmith]

Originally posted by Blackeagle
Lee I like your idea of building custom built rigs. I of course have no idea what the market is like over there. MadDog is right about wait to see if the first couple sell before shopping for mills. (Unless this is just a prime excuse to buy the mill you want anyway! )

And thanks much for the pin sink, looking forwards to playing a bit with it.

it sounds like a good idea but i have to agree with MDM regarding the people that would buy one and when they make some mistake and all there components get wet who are they going to blame and expect to sort it out ? Me.

I have some other ideas but more on them later.

Anyway my new toys as the girl firend calls them arrived today.

one very nice looking cros vice and 3 endmils.

problem one the cross vice is alittle to big for my drill press.

problem 2 my drill that fits in the drill press has only 1 speed setting and the end mills just reck the copper at that speed.

what rpm do i need for the end mills. I need them to make the bottom of the cups flat after they have been drilled.

Plus i was thinking of using a 10mm endmill to make a chipset and vid cooler by milling a simple maze into the 10mm copper i have like the danger den z chipset blocks.

on a positive side the one turn of the cross vice x -y handels = 3mm so now i dont need to punch the pattern and can make a smaller design with some smaller tubes have.

what a shame i promised the GF i would spend this after noon with her Doh !!!

[Blackeagle]

TOYS ! !

I bought some new tools last night. Details in radom geek forum.

[leejsmith]

Nice Be. I like the sound of the drill with the laser marker I sure could use that.

i now have a piller drill (from argos thanks MdM)

after loads of messin with practice plastic and copper i thought i would be ready to make a near perfect jet array along with the cups in the base but i managed to miss align one row.
The piller drill with the cross vice is working very well it's just user error and lack of patients thats the problem.

on the latest version i used 2.5mm cups with 0.5mm between each of them. I tried to make the cups flat with a 2.5mm end mill but i think the chuck on the piller drill is not level as i am getting some wobbling and it makes a mess of the cups.

i have now started to use cpu burn to test the blocks it keeps a stable cpu temp unlike sandra the rises and falls in a cycle.

after some testing today i found the tube/cup block to be no better than the fin style block with a 1.5mm base and 27fins per inch.

i guess the one row i messed up is still causing parts of the base to get warm.

will try again later this week.

The lexan has not arrived BE has the pin cooler ?

[Blackeagle]

Hi Lee,

No sign of the pin cooler yet. And I'm real sorry to hear the Lexan is still MIA.

I talked to the guy that owns a PC shop I've done some buying at. The up & down security alerts here are really screwing up his maintaining his inventorys at normal levels. It's costing him some serious money.

All considered I have NO doubts that the problem is in the US postal service, not yours. From what I understand when they raise the security level from "yellow" which is one above "normal" to orange the security goes up a LOT.

I'll try shipping this time as I did the first time. Perhaps that will help. I sure hope so.

Remember when I sent the first time to you took less than a week. When I shipped some to MDM at a standard rate it only took 2 days longer. Now it's 3X as long and you still havn't gotten it. I shipped some to Puzdre the same day I sent yours,....... he hasn't gotten it yet. Flustrating.

God only knows what our postal people will do with a package coming into this country from overseas. Perhaps the pin sink will be here for my birthday (5 weeks). On the other hand I'll have my new work bench built and tools installed ready to go then. LOL

I'll cut some for you and have it ready to go. Should be into town on Thursday.

[MadDogMe]

How big is the I/D of the 2mm jets you are using lee?, Cathar used 0.8mm IIRC...

The cups don't need to be flat at the bottome do they?. I'd have thought a ball end would be best, but a normal drillbit V would 'deflect' the water up and out of the cups better than a flat bottom would...

How big a 'square' are you making with your cups lee?. For a P4 I'd try maybe a 1CM square...

How thich is the plate you're starting from?. I'd like to try 3mm with 2.5mm cups...

[leejsmith]

Originally posted by MadDogMe
How big is the I/D of the 2mm jets you are using lee?, Cathar used 0.8mm IIRC...

the bigger than 1mm smaller than 1.5mm based on what sized drill bit i can put inside.

The cups don't need to be flat at the bottome do they?. I'd have thought a ball end would be best, but a normal drillbit V would 'deflect' the water up and out of the cups better than a flat bottom would...

the flatter the better. rounded cups will just make the water flow out quicker. flat ones will create the impingment effect.
also lets say i have 0.5mm between each cup and 0.5mm base. If the cups are flat the the 0.5mm base will be the same thickenss across all of the cup but if it's round or v shape then the only point that is 0.5mm is the center and the rest of the cup will be getting thicker to the edge of the cup so the cooling effect wont work as well.

some anims i did for bens new design idea thread.

first rounded cups you get a nice flow out of the cup



second flat cups

How big a 'square' are you making with your cups lee?. For a P4 I'd try maybe a 1CM square...

i have a 20mm x 16mm area that the cups fit inside.
http://homepage.ntlworld.com/lee_smith/25mmtemplate.jpg

How thich is the plate you're starting from?. I'd like to try 3mm with 2.5mm cups...

the base plate is 4mm and the point of the drill is less than 1mm above the core. before i drilled the cups i used some plexi so i could see how close i was to the base.
I think the v shape is one problem and getting the cups spaced is another.

i need to check that the chuck on the piller drill is pushed on enough so i can stop the wobble and then i should be able to use the end mill.

[MadDogMe]

I'd remove the 'heatspreader' from the P4 if I was you and make the cup array the same size (well slightly bigger ) than the core. Are the ID of the holes to big?. It might make a good jet/spray when in the open air but what about under water/backpressure?...

Impingment is the force with which the water strikes the metal is'nt it?. like washing a path with a hosepipe, if you put your finger over it, it rocks! . therefore maintaning the waterspeed/velocity is important, would'nt getting it out of the cup after it has hit the BP hard&fast be better?. With the flat bottom it'll create more turbulance and have a bigger surface area of 'thin BP' but won't that turbulance create backpressure?. I don't know which is best :shrug: ...

It's strange the chuck does'nt wobble with the drillbits. Does the endmill have a perfect round shaft, or does it have grooves cut into it?...

[leejsmith]

Originally posted by MadDogMe
I'd remove the 'heatspreader' from the P4 if I was you and make the cup array the same size (well slightly bigger ) than the core. Are the ID of the holes to big?. It might make a good jet/spray when in the open air but what about under water/backpressure?...

i realy dont want to remove the heat spreader but i have lapped it down to show the copper in the center.
i could try the smaller tube i have but they are not metric so when i use the 2mm tube they have a very snug fit into the 2mm holes but the smaller tube is a fraction to big for a 1.5mm drill bit.

Impingment is the force with which the water strikes the metal is'nt it?. like washing a path with a hosepipe, if you put your finger over it, it rocks! . therefore maintaning the waterspeed/velocity is important, would'nt getting it out of the cup after it has hit the BP hard&fast be better?. With the flat bottom it'll create more turbulance and have a bigger surface area of 'thin BP' but won't that turbulance create backpressure?. I don't know which is best :shrug: ...

from what little i know the impingment is like you said a jet of water hitting a flat surface.
RE the cup shapes i am trusting cathar and bigben2k who both said the same abput flat cups being best.
if the cup is round then the impingment area is smaller and most of the jet is just directed out of the cup.

It's strange the chuck does'nt wobble with the drillbits. Does the endmill have a perfect round shaft, or does it have grooves cut into it?...

the endmill has a thread at one end it could be that. I am going to take the piller drill back anyway when i was assembling it the shelf that can be moved broke while i was pushing down onto some wood to install the chuck. That could be why the chuck is wobbling to.

here is a pick of my latest attemp. the top looks good and you can see the 2mm holes aligned with the cups.



and here the cups. i think i have a problem making the first mark into the copper. as i am drilling into the plexi and then into the copper. I might try a 1mm or 1.5mm drill first to make the marks then drill with 2mm after in the top.



i used the end mill on this one so the cups are flat and very thin but as you can see they made a mess of some of the cup walls. Not sure if i will finish this one but it might be worth while to see how well the flat cups work.

on the thermaltake block i removed some of the fins as cathar sugested and made a smaller jet plate. did some tests yesterday and after made the jet hole a little bigger and will test again today.

[bigben2k]

That's correct: the term "inpingement" describes an effect by which a jet of water suddenly strikes an object. The theory is that the striking creates a very small area where the water is thrown into turbulence (ref: Reynolds) and thereby reduces the thermal resistance, which allows the CPU heat to be transferred more efficiently.

The square cup is indeed better: the whole principle of Cathar's "Cascade" is based on what I described (here) as a double inpingement effect, where it first occurs at the bottom of the cup, then occurs again on the side wall. That gives you minimal thermal resistance, precisely where it is needed: right over the core. The round cup virtually eliminates the second inpingement effect.

It's going to be extremely difficult to beat this design, because it already maximizes all of the flow action of the water. Personally, I'll be trying it out with two Johnson pumps, and I'm hoping to throw the water into turbulence, as it enters the small tubes: something that I've never seen anyone do before, because the pressures involved are rather high.

Since the block will have brass microtubes, it should have a smooth inner surface, reducing the friction, and restriction of flow. Normally we're not concerned about that, because in non-turbulent flow, the surface smoothness of the tubing has little to no effect on the flow resistance.

It may sound crazy, but I'm going for the Chuck Yeager factor here

[leejsmith]

well the flat cup base even with some of the cup walls broke is now my best performng block beating the thermaltake block by 1C and the maze by 2.5C.

a strange thing happend when i was testing the maze 3 today.

i ran the first tests last night and the maze 3 was 7.7C above room temp (taken from a probe glued to the base of the cpu)

running the same test in the morning without remounting it and it was 8.4C above ambient. :shrug:

can the room temp effect how well the blocks or rads work ?


picked up a 1/16 dill bit today for the smaller tubes i have.

looking at the cascade thread gave me an idea to make a 2mm hole into the lower plex layer through to the copper and then glue one of the tubes into the plastic so when i need to align the jet tubes with the cups i can put the 2mm tube into a 2mm hole in the copper and get it a spot on every time.

also to solve the problem of drilling though the plex to make the drll mark into the copper without the drill bit moving i am going to start with 1.5mm drill. then use the 2mm in the plexi for the tubes after.
as the end mill makes the cups a lttle bigger when i make the cups flat i could make the space between each cup a little bigger to compensate.

bigben2k how does the impingement work on the cup walls ?

[Cathar]

Originally posted by leejsmith
bigben2k how does the impingement work on the cup walls ?

I answered this in the Cascade thread.

Water hits the base of the cup and the "shape" of the flow is a bit like an up-side down funnel. At the edges of this funnel-shape the velocity of the water is only a slightly less than the velocity of the water coming out of the jet (once you get past the outside of the funnal shape, the water velocity dramatically drops off).

The cup walls should be placed such that they are just outside of the funnel shape. The water travelling sideways smashes (impinges) on the cup walls giving a secondary impingement effect right where we want it.

Here's a quick image I found.

[bigben2k]

How did you measure your "room temp"?

Inpingement works by having water abruptly strike an obstruction. In this design, the first inpingement is quite visible: the water comes shooting straight out against the bottom of the cup. The second inpingement is a little harder to see: after the first inpingement, the water will spread out radially, and head straight for the cup's wall. That's your second inpingement.

As you can see from the animations, the square cup doesn't let water out so easily, and that's an indication that the inpingement works, because it's pretty restrictive.

In the round cup, there is no secondary inpingement: the coolant merely flows over the cup wall. You can see how quickly the water flows out of the cup.

Flow over a baseplate (or wall) is the old design, like any of the Maze series, from DangerDen. It's ok, but not high performance. In order for that design to perform well, the coolant would have to be in turbulence, but the pumping power required to do this is higher than what all of us have used so far. It also throws another critical variable in the block design: surface smoothness versus flow restriction.


Strapping on the helmet... where's my G suit?:shrug:

[leejsmith]

thanks guys,

at the moment i have the tubes 1.5mm into the cups and the cups are 3.5mm wth a 4mm base. i understand now that getting the depth of the tubes is just as important as the space of the cups and the shape of the cups.

if the tubes are to short then the funnel shape will not hit the tub/cup walls with full effect. I will take a guess that if they are too long this will have a negitive effect on performance too.

this is proving to be a challange but i wont give up.
My supply of copper is buzzbar and all the pieces i have are 10mm thick so i have to cut the copper in half each time. at least i can make 2 bases from it.

bb2k for my room temps i have a digital thermometer. It's never moved when i am testing and when i look at my 3 temp probes.

1. probe glued to the underside of the cpu.
2. aquarium water probe in the fill/bleed line
3. the digital therometer

they all show the same temps +- 0.2C in the morning.
it was strange as my temp readings are normaly very close.

i am currently calibrating a joystck probe for the rig i use for testing so at least i will have the same probes for cpu , water in , water out and room temps and they will all be recorded.

A friend made an smbus temp reader from an lm83 chip for me but MBM5 does not reconize it even in the smbus data dump.
I guess my mobo has a problem with it.

[bigben2k]

Yup. keep trying it, you'll find a good combination, just as Cathar has.

As for the temp, yes the room temp has an effect on the CPU temp.

[leejsmith]

i know it has an effect but dont we normaly measure how much above ambient is ? is this linear ? or is there a curve as the ambient temp increases.

[bigben2k]

I believe that it's almost linear.

If the rad was the single point where the heat escaped, the CPU temp would be perfectly proportional. But we know that the CPU itself will loose some heat around itself, by convection directly to air. I think it was Roscal that demontrated over at OC that the secondary losses were in the order of 1 or 2%.

But your temp swing is bigger than 1 or 2%: it's as if the CPU power jumped 5 to 10%.:shrug:

[leejsmith]

going back to the second impingement effect i have some more anims.

1. this is how i have the tubes at the moment about .5mm nto the cups. you can see there is no secondary impingment because the tubes are not deep enough into the cups



and here deeper into the cups. the arrows showing the water flow bounce of the cup walls and tube as it leaves the cup.



i now undestand how to make the cups aligned corectly while drilling through the plexi so i am confident my next attemp will be a good one.

A local engineering shop is going to cut my 10mm copper for me on monday.

[leejsmith]

almost forgot i have nearly finished a vb applicaton to read upto 4 thermisters connected to the game port and save to a csv file.
the main differance between my program and the one they used at benchtest is i have kept the decimal place on the temp.
I am attempting to include the calbration into the same program so there is no need to use the spread sheet. just enter the temps and click when the readngs are correct.

the only problem with the program is the joystick routing is a comercial addon to vb and shows 2 banners each time you start it.

[bigben2k]

Lee,

actually, i suspect that the flow is being sucked out of the cup, because of the very narrow outlet, when the jet tube sits in it. The water should flow straight out, and only turn once it hits the bottom of the cup.

Did you calculate the area of the opening, and/or the hydraulic equivalent?

[leejsmith]

Originally posted by bigben2k
Lee,

actually, i suspect that the flow is being sucked out of the cup, because of the very narrow outlet, when the jet tube sits in it. The water should flow straight out, and only turn once it hits the bottom of the cup.

Did you calculate the area of the opening, and/or the hydraulic equivalent?

if i increase the velocity of the water the arrows grow longer and make the diagrams very hard to understand. so the anims are using a very low water velocity.

as for the calculations no i didnt and wouldnt know how to. I am more of a build it and tweek it kind of guy. I would like to have a better understanding of how the blocks work but maths was never my strongest subject.

the models in the diagrams are not to any scale i was just playing with it to see what are the effects of the tube depth.

[bigben2k]

Ok, I understand.

There's an excellent website www.lmnoeng.com that has a bunch of formulaes and a few calcs, to do this.

If I have a chance, I could calculate and compare the hydraulic properties of the inlet and outlets involved here. PM me the dimensions

[MadDogMe]

Do you mean making sure the area around the jets OD is big enough not to cause too much restriction BB? (same size as jets ID or bigger?), or are you talking about something more elite? ...

I'd say the thread on the endmill is causing problems Lee. Can you get a thin threaded sleeve to thread over it?, to make it smooth again.. Or is it a weird reverse thread to prevent it untightening while turning?..

[bigben2k]

It would fall under "more elite".

In order for the coolant to jet properly, we have to avoid the effect by which the coolant is sucked out of the sides.

So it's a little calculation exercise. Now you'd think that it would simply involve making sure that the jet ID cross section is smaller than the outlet's cross section, but you actually have to calculate the hydraulic equivalent of the "ring" cross section.

In this link:
http://www.lmnoeng.com/PipeDuct.htm

you can see a formulae to calculate the hydraulic equivalent for both a square tube, and a ring-shaped tube.

[ed: sp]

[bigben2k]

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.

[leejsmith]

Originally posted by MadDogMe
Do you mean making sure the area around the jets OD is big enough not to cause too much restriction BB? (same size as jets ID or bigger?), or are you talking about something more elite? ...

I'd say the thread on the endmill is causing problems Lee. Can you get a thin threaded sleeve to thread over it?, to make it smooth again.. Or is it a weird reverse thread to prevent it untightening while turning?..

i removed the chuck ths weekend and made sure it's on correctly and so far no problems. as long as i dont put too much pressure on the end mll no more cup walls destroyed.