Originally Posted by teamtd11 Great Article! Its good to see inside the D-Tek without taking mine apart;) I always woundered how the water would go across the whole block when the output is only in 1 corner, but i did move from the simple TDX to the fusion :o
Actually, unless you bought yours very recently It'd probably be worth taking it apart. I read an article on the FuZion a little while back in a forum and loads of people were finding bit's of black "gunk" floating around in there loop/s.
Turns out that one of the rubber seals is actually injected into the unit once it has been assembled, which unfortunately leaves a lot of weak rubber material loose where it squeezes out inside the unit. When I took mine apart, a lot of it had got stuck in the pins and was clearly impeeding it's cooling potential.
Can't think where I read about it now, I'll see if I can dig it up.
Most people seem to take their FuZions apart and cut the plastic seal. Seems to be the only thing wrong with the block. Having said that, I'd go for an Acrylic topped block anyday.
Nice article, informative, unbiased and encouraging.
The Aquacomputer block in the article was the DI version, don't think that was mentioned.
For my loops I modded an Aquacomputer Cuplex XT from 1_in_1_out to 1_in_2_out. It suited my loop, reduced restriction (I reckon) and meant that the input was directly over the countersunk jet holes as the author recommends.
Definitely an improvement with my temps. I've changed my mobo and cpu since that build and haven't put the watercooling back in. I wonder how it would cool my Q6600 compared to the X2 X4400+ previously.
Originally Posted by Da Dego Erk! Hey Sam, I thought I left the word 'five' in there. Originally I intended to add a look into an 'up and coming' in there - but after digging through the best of the best, the fifth block really didn't make the cut. I'd prefer not to slander it without a proper review, so I withheld judgement on it. :)
And Whiskey, you are correct that this is a standard TDX - I didn't get my MC version in time and forgot to change the references :) Sorry bout that! The article is corrected now. Dan wouldn't have been happy for me to feature his "old" block as the new - and I've not tested the MC's performance.
I'm glad you guys like it so far! This has been in my head to do for a while - I've been looking around for watercooling resources and found things particularly lacking. Not so much in "this block good, this block bad," but the why's and the wherefor's.
Quote:
The MC-TDX by Danger Den is a popular
and very effective high-flow block.
^^ Picture comment, first page mate :)
Edit : Spotted another errant MC :
Quote:
The channels empty out into the smooth edges of the block, where they reach an outlet barb on one side. This helps restore laminar flow to create an orderly exit though it's not quite as well designed in this respect as the Danger Den MC-TDX and certainly not the D-Tek FuZion.
I was surprised to find you didn't really mention if/how multicore blocks have an effect on the design. Surely the 'hot spot' in a quadcore block is different to a single core block? Also the picture of a block not mentioned anywhere in the article on the final page is a bit strange.
Well, i loved the article, i own 2 machines filled with Aqua-computer blocks and radiators. But my next machine probably wont get any WC in it. One of the main reasons of this, was because of this article from Joe at procooling.com. It opend my eyes...
when reading it, at first i was like O.o... i don't get it, but then it hit me and i did, much thanks to your onion :P
I now know alot more on water cooling blocks. <3
I've had a few ideas during the course of reading it though that I thought I'd pose to you all.
First, solving (or at least alleviating) the boundary layer issue.
Would it be possible to use some sort of heat transfer paste on the inside of a water block?
Maybe some sort of paste with very small molecules, that would form a much smoother layer between itself and the coolant. Much smoother than possible with copper at any rate. And if it was something with good thermal properties it'd pass the heat on very well from the block to the coolant.
My second thought was rifling.
Could you use rifling on the intake of the waterblock to increase the turbulence to help break the boundary layer?
An idea I had in my head for a block design was to have a right-hand rifling on the entry to a block then having a left-hand rifling to the block's baseplate tapering out to a smooth exit near the edge. The way I'm picturing it in my head is the water would hit hard onto the reversed direction rifling of the base plate but after that, laminar flow would be restored very quickly. Should keep the block's overall obstruction quite low while giving great turbulence at the heat source. :)
well, vortices (rifling) turbulence does not make.
the best way to create a turbulent flow in my opinion is to use something called a hydro-shear (US & International patents belonging to John Melbourne)
the way a hydro-shear works is using angled jets to create a vortex flow and then using a set of jets angled in the opposite direction to destroy that vortex flow.
this principle works on anything from 10ft in diameter right down to very small scales (1/2 inch or less)
my father did a good deal of work on the concept and design, and the waterblock I designed actually uses this technology to remove all the worries about boundary layers.
I would put up the design for you all to see, but there are some patent and design issues that prevent me from doing so.
Originally Posted by [USRF]Obiwan Well, i loved the article, i own 2 machines filled with Aqua-computer blocks and radiators. But my next machine probably wont get any WC in it. One of the main reasons of this, was because of this article from Joe at procooling.com. It opend my eyes...
I know deep down that water cooling really does not improve my overclock that much. Point in fact, my 3800+ AMD X2 got to 2.45ghz on air (from 2.0 stock). On water I was able to get it to 2.7 stable and 2.8 unstable. So for about $200-250 I got an extra 150mhz. I know it isn't worth it. I also know my 3870 would OC about the same on air as it does on water. In fact, I could probably push the core on it another 150mhz if I just flashed the bios. It sure as hell wasnt worth the $120 waterblock I put a $250 video card.
Originally Posted by Bluephoenix the way a hydro-shear works is using angled jets to create a vortex flow and then using a set of jets angled in the opposite direction to destroy that vortex flow.
Would that not have a similar effect to this...
Quote:
Originally Posted by me The way I'm picturing it in my head is the water would hit hard onto the reversed direction rifling of the base plate but after that, laminar flow would be restored very quickly.
not really, with the rifilng, you have the flow edges disrupted, but as he said it would be quickly restored and still have issues with anything the center flow went over.
the hydroshear destroys the flow down to the molecular level, resulting in a much greater degree of turbulence and thus the flow takes longer to reassemble itself (loop goes from cpu to GPU, and the flow only goes back to smooth after the GPU)
I never thought I'd see transport mechanism tackled on bit-tech. I'm not sure that the laminar flow diagram was used properly, but that's okay. Great article!
I am currently building a water cooled gaming beast called Exterbis, http://exterbis.parahelix.org/
Could you tell me, currently the radiator is mounted at the front, behind the 3 140mm door fans (Case: LianLi PC-P80)
I am worried about the hard drives heating as the air flows through the radiator, any suggestions?
Comments 26 to 43 of 43
ReplyActually, unless you bought yours very recently It'd probably be worth taking it apart. I read an article on the FuZion a little while back in a forum and loads of people were finding bit's of black "gunk" floating around in there loop/s.
Turns out that one of the rubber seals is actually injected into the unit once it has been assembled, which unfortunately leaves a lot of weak rubber material loose where it squeezes out inside the unit. When I took mine apart, a lot of it had got stuck in the pins and was clearly impeeding it's cooling potential.
Can't think where I read about it now, I'll see if I can dig it up.
EDIT: link
Nice job though, be looking forward to the next bit. Take a look in my latest project log and you'll see my version of a TDX...
The Aquacomputer block in the article was the DI version, don't think that was mentioned.
For my loops I modded an Aquacomputer Cuplex XT from 1_in_1_out to 1_in_2_out. It suited my loop, reduced restriction (I reckon) and meant that the input was directly over the countersunk jet holes as the author recommends.
Original assembled:
http://homepage.ntlworld.com/mysterae/parallel_hex/cuplex_xt_mod_01.jpg
Original internal water path:
http://homepage.ntlworld.com/mysterae/parallel_hex/cuplex_xt_mod_03.jpg
The plan to extend the return paths to 2:
http://homepage.ntlworld.com/mysterae/parallel_hex/cuplex_xt_mod_08.jpg
The plan to block off the original inlet and make it an outlet (before drill through the block):
http://homepage.ntlworld.com/mysterae/parallel_hex/cuplex_xt_mod_09.jpg
The middle block drilled, and the plexi top drilled and tapped:
http://homepage.ntlworld.com/mysterae/parallel_hex/cuplex_xt_mod_13.jpg
The previously inlet blocked off with 'quick metal' or other type of filler:
http://homepage.ntlworld.com/mysterae/parallel_hex/cuplex_xt_mod_16.jpg
The middle part of the block with an additional return path milled clear:
http://homepage.ntlworld.com/mysterae/parallel_hex/cuplex_xt_mod_17.jpg
Assembled and leak testing (sucessfully!). Reminder - The middle port is the inlet, the other two outlets (RBX style):
http://homepage.ntlworld.com/mysterae/parallel_hex/cuplex_xt_mod_19.jpg
Definitely an improvement with my temps. I've changed my mobo and cpu since that build and haven't put the watercooling back in. I wonder how it would cool my Q6600 compared to the X2 X4400+ previously.
it's not as if CPU watercooling is complicated. it's a copper block and you inject water in the middle for highest effectiveness. woo.
^^ Picture comment, first page mate :)
Edit : Spotted another errant MC :
From the EK section.
I was surprised to find you didn't really mention if/how multicore blocks have an effect on the design. Surely the 'hot spot' in a quadcore block is different to a single core block? Also the picture of a block not mentioned anywhere in the article on the final page is a bit strange.
when reading it, at first i was like O.o... i don't get it, but then it hit me and i did, much thanks to your onion :P
I now know alot more on water cooling blocks. <3
I've had a few ideas during the course of reading it though that I thought I'd pose to you all.
First, solving (or at least alleviating) the boundary layer issue.
Would it be possible to use some sort of heat transfer paste on the inside of a water block?
Maybe some sort of paste with very small molecules, that would form a much smoother layer between itself and the coolant. Much smoother than possible with copper at any rate. And if it was something with good thermal properties it'd pass the heat on very well from the block to the coolant.
My second thought was rifling.
Could you use rifling on the intake of the waterblock to increase the turbulence to help break the boundary layer?
An idea I had in my head for a block design was to have a right-hand rifling on the entry to a block then having a left-hand rifling to the block's baseplate tapering out to a smooth exit near the edge. The way I'm picturing it in my head is the water would hit hard onto the reversed direction rifling of the base plate but after that, laminar flow would be restored very quickly. Should keep the block's overall obstruction quite low while giving great turbulence at the heat source. :)
the best way to create a turbulent flow in my opinion is to use something called a hydro-shear (US & International patents belonging to John Melbourne)
the way a hydro-shear works is using angled jets to create a vortex flow and then using a set of jets angled in the opposite direction to destroy that vortex flow.
this principle works on anything from 10ft in diameter right down to very small scales (1/2 inch or less)
my father did a good deal of work on the concept and design, and the waterblock I designed actually uses this technology to remove all the worries about boundary layers.
I would put up the design for you all to see, but there are some patent and design issues that prevent me from doing so.
hope this is some food for thought.
I know deep down that water cooling really does not improve my overclock that much. Point in fact, my 3800+ AMD X2 got to 2.45ghz on air (from 2.0 stock). On water I was able to get it to 2.7 stable and 2.8 unstable. So for about $200-250 I got an extra 150mhz. I know it isn't worth it. I also know my 3870 would OC about the same on air as it does on water. In fact, I could probably push the core on it another 150mhz if I just flashed the bios. It sure as hell wasnt worth the $120 waterblock I put a $250 video card.
But damn it sure is fun and it does look nice.
Would that not have a similar effect to this...
the hydroshear destroys the flow down to the molecular level, resulting in a much greater degree of turbulence and thus the flow takes longer to reassemble itself (loop goes from cpu to GPU, and the flow only goes back to smooth after the GPU)
Could you tell me, currently the radiator is mounted at the front, behind the 3 140mm door fans (Case: LianLi PC-P80)
I am worried about the hard drives heating as the air flows through the radiator, any suggestions?
-
« Previous
-
1
-
2
-
Next »
Discuss in the forums