My new waterblock - Page 2

nicozeg · 10-31-2002, 10:47 AM

Couldn’t explain it better Ben! Besides, it would be more flow restrictive I think.

Gmat: Spirals in the face would require a CNC lathe, you need to coordinate rotational speed, angular position and tool advance. Almost impossible, only an ultra expensive industrial machine could do that.

Anyway, the water is moving radialy, so spirals or rings do the same.

MDM: cyanoacrilate glue does wonders! Also fresh resin over cured one adheres very well.

I thought about a channel system to provide only one exit, but that would complicate the holding clip (socket type). Opted for my “external flow integrator” (almost done).

bigben2k · 10-31-2002, 11:23 AM

It might be a little more restrictive, but not significantly. The main issue is that reversing the steps greatly increases the boundary layer, the water closest to the baseplate, and that's bad.

Note: because of the type of turbulence induced, there is an optimal flow rate: too fast and the vortex will negate the next step. too slow and there won't be a vortex. This is where us block makers have to rely on experimentation, because there's no simple calculation or computer modeling that will allow us to calculate any of this easily.

It's all similar to aerodynamics, or plane makers, where they use turbulators to increase the resulting lift of a wing. The major difference is that air is compressible, and a liquid is (mostly) not. Also, plane designers only have to deal with the resulting noise, but otherwise don't care about the effect of the induced turbulence (think about how much more noise a plane makes when the flaps are down). In block design, turbulence is critical to improving cooling, so the geometry of the turbulence is important.

Nicozeg, what's your baseplate thickness? FYI, I may "borrow" some of that turbulence idea to incorporate into my block

Essentially the same design, but with fins.

nicozeg · 11-07-2002, 12:30 PM

Sorry Ben for the delay in my reply, I’m having a lot of work these days, with almost no time left for anything else.

About bp thickness, it varies between 3 and 4 mm. I found that a good tradeoff between thin bp and heat spreading over the base.

I agree that the turbulence and “aerodynamics” part is the more intuitive in my block design; I want to do a similar base without turbulators to see how much they impact.

See you’re burning your brain with numbers in your design

, I opted for a theorist approach in mine and then test run different base configurations, it’s very easy to exchange the copper discs.

I’m almost done with my block, hope this week end have some time to make the clip

. In the meantime here’s how I’m going to merge the four outlets:

Puzzdre · 11-07-2002, 01:44 PM

Errrr...Sputnik? Nonononono......Atlas? Nononono....

Man! This looks sooo professional! Great work!

bigben2k · 11-07-2002, 01:53 PM

Nice, I suppose that you cast that too?

Yeah, I've got a cold today, and I think it came from the calculations! I'm not going to have a testbed for this, so I'm trying to get at least some kind of reference point about how it's going to perform.

Puzzdre: I believe you're thinking about the Mercury capsule:

Puzzdre · 11-07-2002, 02:02 PM

bb2k: Bingo!!!

sorry for your cold...I just hate that sore nosethrills feeling...

nicozeg · 11-07-2002, 08:14 PM

Thanks Puzzdre; So this is "The mercury flow integrator"

For all the ones interested in my plastics process here’s a quick step by step. (After some tweaking in photoshop to make it fit in a 75 kb attachment)

1- The main channel shape
2- The outer shape mold in place
3- Curing the resin with a lamp and aluminum paper (this time it was epoxy with some pigment in the mix)
4- Rough shape in lathe after peeling off the cardboard
5- Final shape done
6- Drilling the smaller channels
7- Polished
8- Final assembly.

MadDogMe · 11-08-2002, 04:15 AM

Luverley!

. Can you dye this stuff nicco?...

BB2K, what is making the water 'curl under' in your pic on the last page?. the way I see it there would/will be a dead spot there(that will curl, but on itself). imagine driving a car over a surface like that both ways, the way it is now would be smoothest, the other way will be more 'turbulant' (and slightly more 'restrictive'

)...

I'd try a smooth one and beadblast it. maybe try one with lines 'spoking' out from the centre with some fancy dremel work, or scored with a sharp blade?, to increase surface area...

8-Ball · 11-08-2002, 07:47 AM

MadDogMe,

fluid flow is very unlike drinving a car over some bumps.

Small eddy currents would be created over the lip. These eddy currents are the perfect example of turbulent flow, and are the reason planes stall when the angle of attack of the wings becomes to large.

having the vertical sections facing flow would be more likely to induce dead spots as well as greatly reducing the flow.

Hope this helps

nicozeg · 11-08-2002, 08:38 AM

Yup! Any color you like

You’re right about the water curling over itself, that’s a vortex. It effectively reduces the boundary layer, because it’s mixing water close to the surface with water in the flow.

Lots of tests to do with base shape, only need time

.
Also have in mind some "cathar clone" ideas

bigben2k · 11-08-2002, 08:45 AM

Pretty sharp 8-ball!

You really have to try to visualize the flow, in order to understand it. If the lip is facing the flow then, with the right flow rate and lip height, you'd induce a vortex above the copper.

If the lip is not facing the flow, then the water "sees" a sudden drop in its travel, which causes it to curl, right at the bottom of the channel.

Inducing vortices in the middle of the channel is (mostly) useless. Inducing vortices at the channel surface increases the heat transfer.

If it's any help, try to picture it all on a larger scale, where there's only a thin film of water moving over it.

Nicozeg: nice work! Thanks for the info, and the pics!

nicozeg · 11-11-2002, 12:52 PM

YIIIIIIEEHHAAAAAAAA!!!!

YYYAAAAAAAAAAAAABBBBADABADUUUUUUUUUUUUUUUU!!!!!

I'ts alive!!!

The clip gived a big fight, but I was decided to finish it on sunday, so after some design simplifications It's done!

The material for the clip I used was carbon fiber bar I had; it's elastic properties made it an ideal spring material and is easy to work with. For the holds to the socket lugs I had in mind some molded plastic pieces, but that required more time than available. I ended up with some uglier aluminium chunks that did the work ok.

Can't show finished pics cause my dad goes on travel for a couple of weeks and take out the digicam.

This is the last pic I could take.

mattah · 11-12-2002, 03:28 AM

Very interesting and original block.

I just got myself one of Cathar's Little River: White Water blocks, and this looks like it uses some of the same ideas; high water velocity, and multiple exits.

Let us know how it goes..

nicozeg · 11-12-2002, 12:14 PM

Well, after some testing I can say that this block performs about 3 - 4 deg. C better than my previous block. I'm a bit dissapointed, expected a bigger improvement cause I know the other one don't was a great performer.

Anyway, this is only the first prototype, there’s a lot of tweaking to do. I found that the exhaust system ended up more restrictive than expected; the id. area of the four brass barbs is equal to the id. of the main barb, without compensating for the increased drag of the smaller hoses.

In the flow tests I made my first block take 32 seconds to fill one gallon, the new one outputting water directly to container takes 35 seconds, and when assembled with the “flow integrator” separated by 15 cm of hoses takes 45 seconds. Ouch! My pump is an intermediate model rated at 12w, roughly similar in specs with an eheim 1048.

What I’m sure right now is that my asus mobo is an absolute crap at reporting temps. It’s an A7S333, bought it because was cheap and promised diode reading; but in fact it’s only used for security shutdown and uses in socket for temperature report. Besides, the ambient reading can’t be trusted at all. The only temp I’m sure comes from a trusty old analog (mercury) thermometer I have in my res. I know that motherboard testing at all (as BillA would say) is flawed, but this are more like blind tests.

So I ask for suggestions about some mobo in the cheap range that really reads the diode, I’m going to upgrade as soon as I found someone to sell this one.

MadDogMe · 11-13-2002, 03:01 AM

Assus COP out eh?!

. Unfortunatly sockets are all we have. I agree with Billa in regards to testing/comparison, but when he shoots down an individual who's only trying to gauge his temps I don't see the point. not unless you can offer a better way...

I'm interseted in the results, Ghetto or not. as long as you explain how and what they were taken with...

MiusenHausen · 08-20-2003, 10:48 AM

WOWW !!!

nicozeg · 08-20-2003, 12:06 PM

Thanks for the bump

, I have almost forgoten this block, as I dismantled it after a few days of use to improve some design flaws, and never finished it. I replaced it with another one that looks very ugly, but is a great performer.

This is the current status of this block:

starbuck3733t · 08-20-2003, 01:06 PM

beautiful work!!! I take it your AMD board doesn't have the 4 mounting holes in the PCB?

nicozeg · 08-20-2003, 02:04 PM

Yes it have the holes, but a good implementation of the socket clip is a lot more convenient and appropiate for a light block like this.

Goksly · 08-20-2003, 04:29 PM

rofl - thats some pro work going on there. looks like you have the kit and the knowledge - cant wait to see the results

10-31-2002, 11:23 AM	#27
bigben2k Responsible for 2% of all the posts here. Join Date: May 2002 Location: Texas, U.S.A. Posts: 8,302	It might be a little more restrictive, but not significantly. The main issue is that reversing the steps greatly increases the boundary layer, the water closest to the baseplate, and that's bad. Note: because of the type of turbulence induced, there is an optimal flow rate: too fast and the vortex will negate the next step. too slow and there won't be a vortex. This is where us block makers have to rely on experimentation, because there's no simple calculation or computer modeling that will allow us to calculate any of this easily. It's all similar to aerodynamics, or plane makers, where they use turbulators to increase the resulting lift of a wing. The major difference is that air is compressible, and a liquid is (mostly) not. Also, plane designers only have to deal with the resulting noise, but otherwise don't care about the effect of the induced turbulence (think about how much more noise a plane makes when the flaps are down). In block design, turbulence is critical to improving cooling, so the geometry of the turbulence is important. Nicozeg, what's your baseplate thickness? FYI, I may "borrow" some of that turbulence idea to incorporate into my block Essentially the same design, but with fins. __________________ WBTA (Water Block Testing Alliance) Nordic Hardware WC 101

11-07-2002, 01:44 PM	#29
Puzzdre Cooling Savant Join Date: Aug 2002 Location: Croatia Posts: 969	Errrr...Sputnik? Nonononono......Atlas? Nononono.... Man! This looks sooo professional! Great work! __________________ 'Out of cheese error... ...please reboot the universe (press the GBL to continue)'

11-07-2002, 01:53 PM	#30
bigben2k Responsible for 2% of all the posts here. Join Date: May 2002 Location: Texas, U.S.A. Posts: 8,302	Nice, I suppose that you cast that too? Yeah, I've got a cold today, and I think it came from the calculations! I'm not going to have a testbed for this, so I'm trying to get at least some kind of reference point about how it's going to perform. Puzzdre: I believe you're thinking about the Mercury capsule: __________________ WBTA (Water Block Testing Alliance) Nordic Hardware WC 101

11-07-2002, 02:02 PM	#31
Puzzdre Cooling Savant Join Date: Aug 2002 Location: Croatia Posts: 969	bb2k: Bingo!!! sorry for your cold...I just hate that sore nosethrills feeling... __________________ 'Out of cheese error... ...please reboot the universe (press the GBL to continue)' Last edited by Puzzdre; 11-07-2002 at 02:33 PM.

11-08-2002, 04:15 AM	#33
MadDogMe Thermophile Join Date: Aug 2002 Location: Just shut up ;) ... Posts: 1,068	Luverley!. Can you dye this stuff nicco?... BB2K, what is making the water 'curl under' in your pic on the last page?. the way I see it there would/will be a dead spot there(that will curl, but on itself). imagine driving a car over a surface like that both ways, the way it is now would be smoothest, the other way will be more 'turbulant' (and slightly more 'restrictive' )... I'd try a smooth one and beadblast it. maybe try one with lines 'spoking' out from the centre with some fancy dremel work, or scored with a sharp blade?, to increase surface area...

10-31-2002, 10:47 AM	#26
nicozeg Cooling Savant Join Date: Sep 2002 Location: Santiago, Chile Posts: 403	Couldn’t explain it better Ben! Besides, it would be more flow restrictive I think. Gmat: Spirals in the face would require a CNC lathe, you need to coordinate rotational speed, angular position and tool advance. Almost impossible, only an ultra expensive industrial machine could do that. Anyway, the water is moving radialy, so spirals or rings do the same. MDM: cyanoacrilate glue does wonders! Also fresh resin over cured one adheres very well. I thought about a channel system to provide only one exit, but that would complicate the holding clip (socket type). Opted for my “external flow integrator” (almost done).

11-08-2002, 07:47 AM	#34
8-Ball Cooling Savant Join Date: Feb 2002 Location: Oxford University, UK Posts: 452	MadDogMe, fluid flow is very unlike drinving a car over some bumps. Small eddy currents would be created over the lip. These eddy currents are the perfect example of turbulent flow, and are the reason planes stall when the angle of attack of the wings becomes to large. having the vertical sections facing flow would be more likely to induce dead spots as well as greatly reducing the flow. Hope this helps

11-08-2002, 08:38 AM	#35
nicozeg Cooling Savant Join Date: Sep 2002 Location: Santiago, Chile Posts: 403	Yup! Any color you like You’re right about the water curling over itself, that’s a vortex. It effectively reduces the boundary layer, because it’s mixing water close to the surface with water in the flow. Lots of tests to do with base shape, only need time. Also have in mind some "cathar clone" ideas

11-08-2002, 08:45 AM	#36
bigben2k Responsible for 2% of all the posts here. Join Date: May 2002 Location: Texas, U.S.A. Posts: 8,302	Pretty sharp 8-ball! You really have to try to visualize the flow, in order to understand it. If the lip is facing the flow then, with the right flow rate and lip height, you'd induce a vortex above the copper. If the lip is not facing the flow, then the water "sees" a sudden drop in its travel, which causes it to curl, right at the bottom of the channel. Inducing vortices in the middle of the channel is (mostly) useless. Inducing vortices at the channel surface increases the heat transfer. If it's any help, try to picture it all on a larger scale, where there's only a thin film of water moving over it. Nicozeg: nice work! Thanks for the info, and the pics! __________________ WBTA (Water Block Testing Alliance) Nordic Hardware WC 101

11-12-2002, 03:28 AM	#38
mattah Cooling Neophyte Join Date: Nov 2002 Location: Perth, Western Australia Posts: 1	Very interesting and original block. I just got myself one of Cathar's Little River: White Water blocks, and this looks like it uses some of the same ideas; high water velocity, and multiple exits. Let us know how it goes.. __________________ "I had to hit him, he was starting to make sense."

11-12-2002, 12:14 PM	#39
nicozeg Cooling Savant Join Date: Sep 2002 Location: Santiago, Chile Posts: 403	Well, after some testing I can say that this block performs about 3 - 4 deg. C better than my previous block. I'm a bit dissapointed, expected a bigger improvement cause I know the other one don't was a great performer. Anyway, this is only the first prototype, there’s a lot of tweaking to do. I found that the exhaust system ended up more restrictive than expected; the id. area of the four brass barbs is equal to the id. of the main barb, without compensating for the increased drag of the smaller hoses. In the flow tests I made my first block take 32 seconds to fill one gallon, the new one outputting water directly to container takes 35 seconds, and when assembled with the “flow integrator” separated by 15 cm of hoses takes 45 seconds. Ouch! My pump is an intermediate model rated at 12w, roughly similar in specs with an eheim 1048. What I’m sure right now is that my asus mobo is an absolute crap at reporting temps. It’s an A7S333, bought it because was cheap and promised diode reading; but in fact it’s only used for security shutdown and uses in socket for temperature report. Besides, the ambient reading can’t be trusted at all. The only temp I’m sure comes from a trusty old analog (mercury) thermometer I have in my res. I know that motherboard testing at all (as BillA would say) is flawed, but this are more like blind tests. So I ask for suggestions about some mobo in the cheap range that really reads the diode, I’m going to upgrade as soon as I found someone to sell this one. Last edited by nicozeg; 08-20-2003 at 12:08 PM.

11-13-2002, 03:01 AM	#40
MadDogMe Thermophile Join Date: Aug 2002 Location: Just shut up ;) ... Posts: 1,068	Assus COP out eh?! . Unfortunatly sockets are all we have. I agree with Billa in regards to testing/comparison, but when he shoots down an individual who's only trying to gauge his temps I don't see the point. not unless you can offer a better way... I'm interseted in the results, Ghetto or not. as long as you explain how and what they were taken with...

08-20-2003, 10:48 AM	#41
MiusenHausen Cooling Neophyte Join Date: Aug 2003 Location: Santiago, Chile Posts: 1	WOWW !!!

08-20-2003, 01:06 PM	#43
starbuck3733t Cooling Savant Join Date: Nov 2002 Location: State College, PA Posts: 338	beautiful work!!! I take it your AMD board doesn't have the 4 mounting holes in the PCB? __________________ Goliath: 3.4E@3.91/Abit IC7, Maze4 (temporarily) + custom splitter to crazy 4-way watercooling parallel loop: X800XT @ 520/1280 + AC Twinplex, AC Twinplex Northbridge, Silenstar Dual HDD Cooler, Eheim1250, '85 econoline van HC + 2x120, 1x120 exhaust - polished aluminum frame panaflo L1As, 2x18GB 10K RPM U160 SCSI, 4GB PC4000. I wanna be BladeRunner when I grow up! Project Goliath - nearing completion.

08-20-2003, 02:04 PM	#44
nicozeg Cooling Savant Join Date: Sep 2002 Location: Santiago, Chile Posts: 403	Yes it have the holes, but a good implementation of the socket clip is a lot more convenient and appropiate for a light block like this.

08-20-2003, 04:29 PM	#45
Goksly Cooling Neophyte Join Date: Aug 2003 Location: Nottingham, UK Posts: 24	rofl - thats some pro work going on there. looks like you have the kit and the knowledge - cant wait to see the results

Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)