RFC: Impeller inside waterblock

[electrip_flip]

Hi,
this is my first posting so if this old news i'm sorry.

Did anybody tried following?
http://195.253.22.66/image/turbo2.jpg

O.K. it is a litle bit schematic but the point is the
brute force turbulence within the water chamber.

If needed i will supply some math which will show
the reasons behind this concept.

Pump is from an AquaClear Mini
http://www.arcatapet.com/item.cfm?cat=996
(Throw away the filter case ;-) )

Water in and out should be at different height
for having a slight pressure difference but
this assembly is not intended to replace your Eheim.

Sorry for my (de)nglish,
I'm from Germany,
Philipp

[packratbob]

u mean an inbuilt impeller?? to creat more turbulance in the water block. hmm that might actually work hmmmm but how to seel the thing and where does the water enter and exit??

[electrip_flip]

Originally posted by packratbob
u mean an inbuilt impeller?? to creat more turbulance in the water block. hmm that might actually work hmmmm but how to seel the thing and where does the water enter and exit??
Hi,
sealing should not cause any problem because the pump
has already an O-ring fitted which is ommitted in my little
drawing.
Later I will supply a second drawing with a view from the top.
If one imagine a 50x50mm baseplate the hose connections
are in the corners.

With respect to the turbulance, the numbers indicate
a factor of 2-4 for heat transfer (coefficient) in comparision
to conventional waterblock designs.

Bye,
Philipp

[JFettig]

YOUR A GENIOUS!!! now i wish i had acess to a mill this summer!!! id mill that out of acrylic and stick a copper bottom on it!, basicly make the same exact impeller housing for it and that would be it, and the block would be a waterblock and a pump at the same time!!

i hope someone has the stuff to try this...

[JFettig]

actually you could take the end off of a maxijet and glue that into it somehow, like take the impeller housing off, then cut the end suction side off and make the suction side rite on the waterblock...

[JFettig]

i hope you dont mind but i redrew your pic to show a better way to do this with an outlet... the inlet who knows were that should be....

[Moriquendi]

if its a centrefugal pump then the inlet must be around the drive shaft , or from below, right above the die.

for an axial impeller it would be easier but still pretty difficult.

hope this makes sense.

Idris

[electrip_flip]

Some technical data for the pump:
typical centrifual type pump for aquaristic use,
Weight: ca. 210 gramms, thats not much,
add maybe up to 200 gramms for copper and
50 gramms for parts and you are within (AMD)-spec.
Power consumption: 5 Watt (230 V)
Size: (as seen with impeller to bottom)
Motor housing:
Height: 44 mm
Width x depth: 55 x 37 mm
Flange (circular, shifted to one side):
Diameter: 28 mm (w/o O-ring)
Height: 14.5 mm (w/o impeller )
Impeller:
Diameter: 18mm
Height: 4.5 mm

Hope this is understandable,
maybe i try a drawing later.

Philipp

[electrip_flip]

my technical drawing ;-/
pump from side
impelller to bottom
http://195.253.22.66/image/pumpe.jpg
depending on your monitor
size is in real a little bit smaller.

Philipp

[gone_fishin]

You would need excellent mounting to get rid of vibration that would ruin the thermal interface material.

[parkan]

It's interesting that no one has said Welcome to the forums!

Anyway, that looks fairly interesting. I was thinking the stress on the seals might be an issue though.

[electrip_flip]

Originally posted by maskedgeek
actually you could take the end off of a maxijet and glue that into it somehow, like take the impeller housing off, then cut the end suction side off and make the suction side rite on the waterblock...
hi,
modded your moded pic ,-)
http://195.253.22.66/image/turbo5.jpg
Hope now it is clear where in and out should be located.
But this pump is not intended to drive water through the whole system.
The most important thing is turbulance
so my first drawing makes more sense for this purpose.

Philipp

[JFettig]

now thats looking good!

[electrip_flip]

Originally posted by gone_fishin
You would need excellent mounting to get rid of vibration that would ruin the thermal interface material.
Hi,
the vibration of this pump is pretty small,
call this a non-issue.
Using this pumps with aquarium for years,
leaking at the O-ring never occurred.

Philipp

P.S.: like your project very much
but thats another topic ;-)

[MajinSSJVegetto]

This is a cool idea!

The water below the out port wold become fairly stagnat, would it not?
You would need something to force water out to the sides and shrink the size of the outlet side a bit?

[gone_fishin]

Originally posted by electrip_flip

hi,
modded your moded pic ,-)
http://195.253.22.97/image/turbo5.jpg
Hope now it is clear where in and out should be located.
But this pump is not intended to drive water through the whole system.
The most important thing is turbulance
so my first drawing makes more sense for this purpose.

Philipp

WELCOME TO THE FORUMS!!!!!!

Ahh, I think you can cause turbulence by just using a small motor from a tiny model race car or something like that, It would only need to have the impeller go through the lexan top and maybe you could fit two or more on there for turbo power:) I've seen such small motors for sale on the shelf in radioshack at about an inch square and run off of 12vDC.
You're on to something here, great idea.

[electrip_flip]

About numbers:
A formula for calculating heattransfercoeffizient (alpha):
Dimension is Watt/(m*m*Kelvin)

350+2100*Squareroot(velocity (m/s))

Example1:
water standing still :350
normal WB (60liter/hour; 6mm diameter) :2000
Brute force ( @3000 rpm) : 3900

boundary layer is doomed to be eliminated ;-)

Philipp

[electrip_flip]

Originally posted by MajinSSJVegetto
This is a cool idea!

The water below the out port wold become fairly stagnat, would it not?
You would need something to force water out to the sides and shrink the size of the outlet side a bit?
No,
because the rotating impeller will introduce two effects:
First the water within the chamber will rotate in sync with the impeller.
Vmax is at the tip of the blades.
Second if impeller is mounted with some clearance to bottom or better
like in the drawing the whole assembly will act like an centifugal pump.
The center of impeller will suck water from the bottom and accelerate it to the tips of the blades.
This will cause a transverse rotation.
Both velocity vektors will add to something like a circular spiral/coil/helix.
Philipp

[futura2001]

I had this idea a while back, but then abandoned it for numerous reasons.
This would not be an effective way of cooling your CPU.
1: An impellor spins slower as you move closer to the center until you reach a dead spot or a vaccuum. The core of your CPU is in the center. If you have an impellor producing a dead spot right where you actually need all that cooling, your chip will burst into flames twenty seconds after you turn on the computer.
2: Mechanical failure. All that you have to prevent massive head loss is a motor. If that motor fails, your waterblock is screwed and the chip will ignite in twenty seconds.
3: As your waterblock becomes better at dissipating the heat, the efficiency decreases. At some point, the theoretical limit to what your block can accomplish will be reached. This is why at the upper end, blocks vary by a fraction of a degree in temps. Even if this worked properly, it would still only be marginally better than another high quality block. Might as well spend less time building a much simpler block that will work just as well...
Futura

[SaInTZeRo]

looks like someone has been reading the kaltmacher.de forum :D :D :D but who cares more people more opinions !!!:beer:

later

[Iron Hawk]

Here is my spin off of the Idea. Personaly I think mounting the pump on the block itself is a bad Idea, but if you were to make a block like this the Impellers would be driven by the water.

[The Spyder]

Exclent idea.... but how would the water flow to spin the propelors-
SPyder

[futura2001]

Impellors driven by water is an even worse idea than impellors driven by a motor. Do you realize how much headloss that would cause? I don't care how good of a pump you have, it is not going to make a difference. Again, your chip will have smoke emanating from it in 20 seconds.
Futura

[electrip_flip]

Originally posted by SaInTZeRo
looks like someone has been reading the kaltmacher.de forum :D :D :D but who cares more people more opinions !!!:beer:

later
That's me :D

philipp

[electrip_flip]

Originally posted by futura2001
I had this idea a while back, but then abandoned it for numerous reasons.
This would not be an effective way of cooling your CPU.
1: An impellor spins slower as you move closer to the center until you reach a dead spot or a vaccuum. The core of your CPU is in the center. If you have an impellor producing a dead spot right where you actually need all that cooling, your chip will burst into flames twenty seconds after you turn on the computer.
2: Mechanical failure. All that you have to prevent massive head loss is a motor. If that motor fails, your waterblock is screwed and the chip will ignite in twenty seconds.
3: As your waterblock becomes better at dissipating the heat, the efficiency decreases. At some point, the theoretical limit to what your block can accomplish will be reached. This is why at the upper end, blocks vary by a fraction of a degree in temps. Even if this worked properly, it would still only be marginally better than another high quality block. Might as well spend less time building a much simpler block that will work just as well...
Futura
My comments:

[1] This is not true, especialy if use with a design like shown above.
In reality there would be no dead spot even with a flat
baseplate
Never expect hydrodynamics act as mathematical simulation implies (much more turbulance).
Also you could shift the position of the impeller sideways.
Testing will show real world numbers.
If one never tests no answers will be given.

[2] Active surface of baseplate is in magnitude of medium
baseplate and there is no 20- second-BOOM.
Even without workin motor cooling is warranted
because in concept normal pump is still in place.
Counterquestion: what happens when your 'normal' pump dies?

[3] Heattransfer is a function of (active) surface,
temperatur difference and heattransfercoefficient (HTC).
Last one is dependent on water velocity.
If one can double HTC effiency is really better.
The problem of heattransfer in copper is dependent
on the geometry of the block.
This is also addressed in the schematics by having
large mass of copper around the cpu-die.

Hope this explanations clarify some issues,

Philipp

[packratbob]

i dont know much but it seems to me that its alot of work to do when u could probably inrease water flow with more powerfull pumps to increase water velocity. and achieve the same result i doubt that a small impeller the size that would fit in such an application would even mover the water at the increased volocities ur talking abt.

well if it works then great let us know an we will follow lol.

[electrip_flip]

Originally posted by packratbob
i dont know much but it seems to me that its alot of work to do when u could probably inrease water flow with more powerfull pumps to increase water velocity. and achieve the same result i doubt that a small impeller the size that would fit in such an application would even mover the water at the increased volocities ur talking abt.

well if it works then great let us know an we will follow lol.
This is how science works: think about it an DO IT.
To much guessing will lead to nothing usefull.
So this thread is about a maybe new way of increasing
cooling performance and we are looking for auxiliary
comments not for people who tell us not to try it.
Hope this was not offensive to you ;-).

Now technical comments:
Powerfull pump Eheim 1250 with 1.2 bar head or 1200 litres/hour.
Waterblock with 3/8"-connectors (ID=8mm)
and corresponding channel size (50 mm²).
Maybe Pump delivers amazing 300 litres per hour in setup.
Result : velocity is 1.657 m/s
Brute-Force-Pump: 2.827 m/s (@3000 rpm)

This is a too much theoretical approach but I think the potential
for a real gain is there.

Philipp

[Iron Hawk]

Originally posted by The Spyder
Exclent idea.... but how would the water flow to spin the propelors-
SPyder

Water would enter in one side, (at one of the small circles) and continue out to the other side. as it flows past the propellers they will start to turn, like a water wheel used in a flower mill.

[Iron Hawk]

Originally posted by futura2001
Impellors driven by water is an even worse idea than impellors driven by a motor. Do you realize how much headloss that would cause? I don't care how good of a pump you have, it is not going to make a difference. Again, your chip will have smoke emanating from it in 20 seconds.
Futura

Would a design more like this one reduce the headloss compared to the design above?

[MILLTHERM]

Why fool around with external pumps?

Ever see a Lab magnetic stirrer? They drop a teflon coated propeller in a beaker on a hot plate. A spinning magnet sealed below the hotplate surface causes the wheel to spin and and stir the fluid.

You have an infinite number of combo's for fluid movement here.
i.e You could scroll a cut out like a centrifugal fan and make a pretty good pump.

No internal parts to wear out and easier to seal.

[electrip_flip]

Originally posted by Iron Hawk


Water would enter in one side, (at one of the small circles) and continue out to the other side. as it flows past the propellers they will start to turn, like a water wheel used in a flower mill.
Those wheels would never spin faster than velocity
of the water accelerated by your pump so there is no win.
The impeller of a pump is spinning much faster.

Philipp

[electrip_flip]

Originally posted by MILLTHERM
Why fool around with external pumps?

Ever see a Lab magnetic stirrer? They drop a teflon coated propeller in a beaker on a hot plate. A spinning magnet sealed below the hotplate surface causes the wheel to spin and and stir the fluid.

You have an infinite number of combo's for fluid movement here.
i.e You could scroll a cut out like a centrifugal fan and make a pretty good pump.

No internal parts to wear out and easier to seal.
There are many pumps depending on this principle, often used
for salt water (marine?) fish tank.
But these are build for massive power consumption and flow (to big).

Wath is the problem using something that was build for this purpose?
The proposed pump is doing its job for years never leaking a drop water and without noteworthy vibration.
Even total weight of an assemly waterblock/pump would not exceed AMD recomendation.

Philipp

[xavierfg]

Edmund Scientific Miniature DC Pump (http://www.scientificsonline.com/Products/DisplayProduct.cfm?productid=1491 )

just a suggestion for a mini pump. Its a mini DC motor, so it has to have some pretty high rpms. Ive used them in projects, they're pretty quiet and dont vibrate noticabley

[Iron Hawk]

Originally posted by xavierfg
Edmund Scientific Miniature DC Pump (http://www.scientificsonline.com/Products/DisplayProduct.cfm?productid=1491 )

just a suggestion for a mini pump. Its a mini DC motor, so it has to have some pretty high rpms. Ive used them in projects, they're pretty quiet and dont vibrate noticabley

Cool thats definatly something to consider. Perhaps when I get the funds to build a block I will order some and see how they work.

[electrip_flip]

Originally posted by xavierfg
Edmund Scientific Miniature DC Pump (http://www.scientificsonline.com/Products/DisplayProduct.cfm?productid=1491 )

just a suggestion for a mini pump. Its a mini DC motor, so it has to have some pretty high rpms. Ive used them in projects, they're pretty quiet and dont vibrate noticabley
In my opinion this one is to small.
But if I had one I would try it anyway ;-).
What about this one:
http://forum.oc-forums.com/vb/showthread.php?s=&threadid=95448

Philipp

[Crazy Jayhawk]

I'd be worried about magnetic fields from the prop motor affecting the CPU.

[Szech]

What if the pump were on the side of the block, with the impellor right over the die? With this setup, the pump would not only be used to create turbulence in the block, it could also circulate the water in the system.

Diagram

[MajinSSJVegetto]

Originally posted by Crazy Jayhawk
I'd be worried about magnetic fields from the prop motor affecting the CPU.

Thats what I was thinking. Think what could go wrong like this!

You could extend the pole though, having the pump 6"+ inches away.

[OKCFUNKY]

you could take ironhawks diagram take those mini dc motors and have them power those impellers. then when the blades spin, they would neutralize the dead zones as the water in the block would continually move, cooling more, but you would definitely have to have a high gph pump run the water in the water in there, as there would be a very significant drop in the output push after it goes through the blades, but since the water would build up in there, it would cool more, and the blades would increase in speed, further creating more turbulence. Like a watermill sorta. Over all, it should work very nicely, if it works as great as it possibly could, it could cool the thoroughbreds good enough. It could also have a potential for making a much bigger block for a hard drive cooler for scsi drives at 15k rpm. That is a great idea. good luck