|
|
Water Block Design / Construction Building your own block? Need info on designing one? Heres where to do it |
Thread Tools |
10-04-2002, 05:02 PM | #1 |
Cooling Neophyte
Join Date: Sep 2002
Location: Sweden
Posts: 20
|
Ultimate waterblock
In theory, what is the most important factor of a waterblock, its way of handling water, its amount of surface to the water, its warmth capacity characteristics, or non of above.
Lets have a normal waterflow from the pump, normal water temp, and a normal rate of watts. |
10-04-2002, 05:48 PM | #2 |
Cooling Neophyte
Join Date: Jan 2002
Posts: 29
|
Mathematically, they are all equally important. This is one reason why water cooling as efficiently as possible is a balancing act and can be difficult.
If you mean what will cool better, a straight fitting flat on the die or a maze of fins and crap, that would depend on flow rate and pressure. Trying to make two blocks with diametrically opposed methodologies would simply prove that they are equal according to Fourier's equation. I will add that the easiest thing to do to help yourself out is to use copper get the highest (practical) thermal conductivity for your block. In my mind, that's a given. |
10-04-2002, 05:57 PM | #3 |
Cooling Savant
Join Date: Aug 2001
Location: SLO, CA
Posts: 837
|
Have you tried reading throught our on-going discussion thread about block efficency?
It might have the answer to your question: http://forums.procooling.com/vbb/sho...&threadid=3792 Good luck reading throught all of it but it does have QUITE a bit of good info.
__________________
Athlon64 X2 4200+ @ 2.5Ghz (250FSB x 10) OCZ VX 1GB 4000 @ 250FSB (6-2-2-2 timmings) DFI LANParty nForce4 Ultra-D SCSI Raid 5 x (3) Cheetah 15K HDDs LSI Express 500 (128MB cache) OCZ PowerStream 520W PSU ATI X850XT PE (Stock) DTEK WhiteWater + DTEK Custom Radiator Eheim 1250 |
10-04-2002, 07:17 PM | #4 | |
Cooling Neophyte
Join Date: Sep 2002
Location: Sweden
Posts: 20
|
Quote:
I have red it and there where plenty of interesting stuff there! |
|
10-05-2002, 03:28 AM | #5 |
Thermophile
Join Date: Aug 2002
Location: Just shut up ;) ...
Posts: 1,068
|
I wonder too,(more so after reading that thread!! ), seeing as the 1/2inch Swiftech with it's 'hollow' inside performs dead on par(or better) than all the wierd and wonderful 'maze type' designs.
Seems H20 has 'hit the wall' so to speak,???... PS, this has'nt stopped me coming up with a wierd and wonderful 'Triple decker' design! ... |
10-05-2002, 04:33 AM | #6 | |
Thermophile
Join Date: Sep 2002
Location: Melbourne, Australia
Posts: 2,538
|
Quote:
It's possible to do a fair bit better than a Swiftech (I am right now) Just gotta push the boundaries more than current designs do. Last edited by Cathar; 10-05-2002 at 04:56 AM. |
|
10-06-2002, 05:18 AM | #7 |
Thermophile
Join Date: Aug 2002
Location: Just shut up ;) ...
Posts: 1,068
|
I was'nt saying the swiftech was the pinnacle, they are one of my least favorite manufacturers, but they are a perfect example of a straight through block performing on par with maze/channel type blocks...
I do think H20's hit a wall as far as the current medium goes though, any improvements are small and seem to result from fine tuning the whole system, rather than block inovations IMO. I don't really think there is an 'Ultimate Block' design out there, it's more of an 'ultimate block for a given system'. perhaps it should be the 'Ultimate System' we're looking for?... That said it would in no way stop me playing around in design if I had a mill, Utabintarbo has kindly offered to make a solidworks model for me to show my basic design, hopefuly if it looks good to people someone will offer to make one up [nudge ~nudge ] [/wink~wink] ... Last edited by MadDogMe; 10-06-2002 at 05:23 AM. |
10-06-2002, 06:03 AM | #8 | |
Thermophile
Join Date: Sep 2002
Location: Melbourne, Australia
Posts: 2,538
|
Quote:
Seeing at least a 4C gain here (at the CPU diode) with my custom block over a "Swiftech" on a hot CPU (AthlonXP @ 1925MHz/2.15v), using the same pump/radiator for both. It's probably possible to wring another 1-2C or so out of it, but to do so would involve fairly expensive manufacturng. I know what you're saying about Swiftech not being the pinnacle, I'm saying it's not even close. There's a lot more that can be done, I have the results to prove it, but it doesn't involve milling through blocks with 4+mm wide mill bits (read - cheap manufacturing) or using flat plate blocks (read - cheap manufacturing) or merely using the inlet nozzle in a sub-optimal way (read - cheap manufacturing). By cheap manufacturing, I'm talking about the $20USD/block to make range, which is where most manufacturers are sitting at. This is probably something that should be added to Phaestus's "Myths" page. "Myth #6 - Waterblock technology is currently at about as far as it can be taken" |
|
10-06-2002, 10:37 AM | #9 |
Responsible for 2%
of all the posts here. Join Date: May 2002
Location: Texas, U.S.A.
Posts: 8,302
|
It's probably a combination, but on the top of my list would be:
-flow speed (not flow rate) -fin design over the core area -baseplate thickness (not necessarily in that order) |
10-06-2002, 07:54 PM | #10 | |
Thermophile
Join Date: Aug 2002
Location: Just shut up ;) ...
Posts: 1,068
|
Quote:
I think more could be gained just from using 5/8 barbs with 12mm inner diametre tubing, to gain the fullest from the flow we have. matching block~channels to bore diametres of tube ect. ************************************************** **************** BigBen, can you explain flow rate and flow speed?, I think I'd rather have as much 'mass' of water moving through as quickly as possible . |
|
10-06-2002, 10:15 PM | #11 | |
Thermophile
Join Date: Sep 2002
Location: Melbourne, Australia
Posts: 2,538
|
Quote:
People often think flow rate is best, but really it's water velocity that counts more than flow rate. Extra flow rate does help, but only because it increases the water velocity. It is possible to boost water velocity by about 4x easily by adding restrictions are important sections of the block and still not have flow rates tumble much. Having high flow rates and high velocity can be done, but only by having a very large and powerful pump which will in itself add so much heat to the water that it simply stops being beneficial. As with most things in life, there's a balance and that balance (in my mind) would be a ~30-40W pump rated for around 15lpm (~4gpm) but with as great of a pressure head as is feasible given the motor power output and a well designed impeller. Stick that onto a block that forced the water to go through ~20mm^2 of openings boosting water velocity to about 4-5x what is typical in most of today's blocks, and you'd probably still see around 2.5-3gpm flow rates with 1/2" tubing/barbs feeding the water. ie. it hinges on the pump. |
|
10-07-2002, 02:03 AM | #12 |
Cooling Savant
Join Date: May 2002
Location: Da UP
Posts: 517
|
I agree. spending a little more on a pump can get you better performance with the right kind of design. People are already willing to throw away $60 on pumps, what's another $40?
I just put in another one of my prototype blocks. It has many restrictive passageways but I am still able to get high water velocities in the block because I have a pump that will handle the pressure. One of them puny pumps with the impellar free floating by a magnet will just rotate less under back pressure, thus you will not get the benefit of higher water velocity when it is forced through a restrictive opening in the block design. |
10-07-2002, 04:37 AM | #13 |
Thermophile
Join Date: Aug 2002
Location: Just shut up ;) ...
Posts: 1,068
|
If that's the case why do you use 1/2inch barbs?, why not 3/8?.
Like I said I'd rather have the most water at the most speed. to me, with what I have, it seems having no restrictions and letting the pump flow freely are the way to go, unless you'd call a 1250 a high pressure pump ... And why have a restrictive outlet? what can that serve?. I still think that block design is'nt the way forward. I think tuning the system to what you have is the all important thing. ******************************************** Last edited by MadDogMe; 10-07-2002 at 04:58 AM. |
10-07-2002, 05:16 AM | #14 | |||
Thermophile
Join Date: Sep 2002
Location: Melbourne, Australia
Posts: 2,538
|
Quote:
As hard as it may be to believe, it's possible to get better performance by causing the flow rates to be lowered through adding restrictions in important places in the block (didn't I say this above already). Quote:
Quote:
If you belong to the "good enough" crowd, then why are you here side-lining the efforts of people who want to talk about pushing waterblock technology further? Last I heard, no one ever complained about getting 4C (or more with a better block than mine) lower temperatures just by changing one element in their system. I'll let you ride your hobby horse. It's clear you're only really here to talk down efforts towards raising the bar, and pass them off as "unimportant" or "trivial". |
|||
10-07-2002, 05:33 AM | #15 | |
Thermophile
Join Date: Sep 2002
Location: Melbourne, Australia
Posts: 2,538
|
Quote:
My CPU is hot. Putting out around ~80W under full load (real watts - ~140 according to "Radiate" which is a joke). Depending on what water-block I'm using, the CPU will run anywhere between 16C and 22C above the water temperature. I have a large dual fan radiator. Running the two fans on it at a quiet 7V, the water climbs to a mere 2.8C above the radiator air intake temperature. If I run the fans at a noisy full blast 12V, it holds the water at around 2.0C above the radiator air intake temperature. Not a whole lot to gain there. Double the radiator size. Get down to a 1C above radiator air intake water temperature. Now I have 3 pumps here. The Eheim 1048, 1250 and a Pondmaster 4200, rated for 10W/600lph, 28W/1200lph, 90W/4000lph respectively. What's the best pump to use? Currently the Eheim 1250. Why? Because although the Pondmaster pushes nearly double the flow rate through a waterblock that the 1250 can push, and improves water-block efficiency by around 0.75C on average due to the extra flow rate, the extra pump heat added into the water with my radiator fans at 7V heats up the water by about an extra 0.7C over the Eheim 1250. ie. next to no net gain whatsoever for sucking down an extra 62W of power and a what's now an audible pump (the Eheim 1250 is silent). So what does this leave us with? A radiator that has nowhere to go. Can't get the water temperature much closer to ambient unless we go to a stupidly large radiator, and even then, we only have at best 2C to gain with a "perfect" radiator, but realistically, more like a 1C gain. Pushing up the pump size doesn't help us any. At best you'll see maybe a 0.5C gain over a 1250 with an optimally balanced pump for wattage and pressure which I described above after taking extra pump heat into account. What does that add up to? 1.5C (1.0C for the radiator, 0.5C for the pump) What did a water-block redesign buy me? 4C Water-block design IS the way forwards. Pump and radiator technology while still possible to improve in small ways, are well known technologies that are pretty much at their limit. Water-blocks are still in their infancy with respect to what we can get out of them performance wise. Last edited by Cathar; 10-07-2002 at 06:06 AM. |
|
10-07-2002, 06:34 AM | #16 | ||||
Thermophile
Join Date: Aug 2002
Location: Just shut up ;) ...
Posts: 1,068
|
Quote:
Quote:
Quote:
Quote:
Please don't get snidey Cathar, please take note of the smileys I put in my posts, mainly the lack of 's . and notice the fact that I said regardless of everything I said it would'nt stop me doing exactly the same as you if I had the tools and space and money. ************************************** I still think that block design is'nt the way forward. I think tuning the system to what you have is the all important thing. -------------------------------------------------------------------------------- What I meant was I don't think there's a revolutinary independant design thats going to blow everything else away, I think it's more of design/tuning the block to GIVEN pump and rad carichteristics. not just 'generic' waterblocks. creating the block to be part of a GIVEN whole is where I think it's at :shrug: . can you not see where I'm coming from?. I think you've misunderstood what I meant, maybe I did'nt articulate myself well enough?. whatever, please don't get . Last edited by MadDogMe; 10-07-2002 at 06:46 AM. |
||||
10-07-2002, 07:41 AM | #17 |
Thermophile
Join Date: Sep 2002
Location: Melbourne, Australia
Posts: 2,538
|
Okay, I have some more hard data from the results I've been gathering for a water-block review.
For the Eheim 1048 & 1250, my block is seeing indistinguishable performance difference. The 1048 pushes 4.2lpm through the block, the 1250 6.3lpm. With a high pressure high-flow Pondmaster 4200, there's 10lpm being pushed through the block. For all of those pumps, my block is at least a 4C clear lead over any other block I'm using (Maze 3, Cyclone 3 [Swiftech copy which performs exactly like a Swiftech], Cyclone 4, Cyclone 5) for bare CPUs (AthlonXP, Duron), and 2C for P4's with the IHS, for a CPU pushing out >75W of actual heat, in relative terms, that's a >20% improvement. I'm sorry if 20%+ doesn't seem like a big jump to you. To me it is. ie. the same benefit is seen across a wide variation of flow rates and pumps. The radiator just sets the water temperature that enters the block. Changing the radiator has the same effect (for all intents and purposes) on all blocks. The only effect changing a radiator can have on block performance is if it's radically restrictive. ie. yes - there can, is and does exist ways to design blocks for better performance in almost all commonly used circumstances (I haven't yet tested for <0.5gpm, or >3gpm). I'm just getting frustrated at your continuing attempts to play down the importance of water-block design when there's clearly a long way to go ahead of what the current market blocks give. It may seem I'm trying to boast here about my achievements, but unless I come up with something factual to counter your statements, then any contradiction to your statements would be merely speculation, as yours are. I have solid proof that you can do better. Do you have solid proof that you can not? This is what's frustrating me. Last edited by Cathar; 10-07-2002 at 08:26 AM. |
10-07-2002, 08:44 AM | #18 |
Thermophile
Join Date: Aug 2002
Location: Just shut up ;) ...
Posts: 1,068
|
Cathar I'm not in the least interested in bandying bullshit with you, can't you see that?. have you a problem with someone having a different opinion than you?, I do not. how many 'I thinks or IMO's do I have to tag on the end?. I'm not saying there is'nt room for improvment. My opinion is that there is no ground breaking revelation/improvement to be made purely in block design. does'nt mean I'm not all for sqeezing those last 6 or 7 DegC out of them For JC's sake!.
A block alone does not make or break a system 'regardless of pump and rad and tube', never has never will, that is my opinion. if you choose to take that as a statement of fact, thats your perogative, but remember I'd love to be proved wrong!! .I'd be a first to buy the block!. now how about we lets the thread be used for other peeples opinions like it was made for?. No problems MadDogMe... |
10-07-2002, 11:51 AM | #19 |
Responsible for 2%
of all the posts here. Join Date: May 2002
Location: Texas, U.S.A.
Posts: 8,302
|
To answer your question MadDog, flow speed is the speed (velocity) at which the water flows. Flow rate is the total amount of water that passes through, over a given period of time.
for 1/2 in tubing versus 1 inch tubing, for example, given the same flow rate, the coolant will flow faster in the 1/2 inch tubing. This is relevant because the flow speed will dictate how turbulent the water is, and the more turbulent it is, the better it will be at taking the heat away from the block. The disadvantage of course, is that it can be awfully restrictive, so a better pump is needed (more head), but that in turn adds a greater heat source to the loop, so the whole waterblock design has to take this into consideration. As for how much better it can be... that's where the baseplate comes in. The thinner the base plate, the more heat is available to be taken away, but if the flow rate isn't high enough, then the heat just sits there, and performance suffers. As for what it's all worth, what it comes down to is 2 factors: -overclock -longer life of the CPU Putting it in terms that can be measured, i.e. money, we can calculate what it's worth to overclock a system. if your (basic) PC costs $1'000, and you can overclock it by 20% with the right equipment, then you must not spend more than $200 in this venture, otherwise you're not getting your ROI (return on investment). Longevity is unfortunately one of those factors that's very hard to measure, since there's little data. Some people will say that an overclocked CPU will last an average of one year, so that might be a factor too, or it might just cancel itself out. The point was that a CPU that's cooled better, will last longer. If a little tweaking gets me 4C, I'll take it, any day. |
10-07-2002, 11:54 AM | #20 |
Responsible for 2%
of all the posts here. Join Date: May 2002
Location: Texas, U.S.A.
Posts: 8,302
|
Cathar, IMO, there might be an advantage in tilting the barbs to your block, down to 45 degrees. It's not easily done, but given your configuration, it might be worth a shot. Note: it might affect the flow dynamics though, where there would be more turbulence at the bottom, instead of the top.
|
10-07-2002, 02:35 PM | #21 |
Cooling Savant
Join Date: May 2002
Location: Da UP
Posts: 517
|
By MadDogMe
"A block alone does not make or break a system 'regardless of pump and rad and tube', never has never will, that is my opinion" end quote Strongly vague opinion if you ask me. Regardless of pump, rad, tube. So given any combo of these, the waterblock cannot break the system? Go get yourself the BlueCooling's BTMS block and live with it And the reverse is true also, take an open design 1/2" swiftech and choke it off with 1/4" tubing and a grab bag of 90 degree elbows, throw in a tiny rad with 1/4" barbs and a little 12volt 1/2lpm squirt pistol of a pump, that will most surely effect the system. Now just because you can screw up more drastically with the combination of all the other parts in the loop does not mean that a block is insigificant in the equation. by bigben2k "The thinner the base plate, the more heat is available to be taken away, but if the flow rate isn't high enough, then the heat just sits there, and performance suffers" end quote Technically I don't follow the first part. Increasing surface area makes more heat available to the coolant and the heat transfer rate (by velocity and properties of coolant) determines how much is stored up in the bp, higher velocity means higher transfer rate and lower block temp which results in lower die temp. Thinner bp means less stored heat energy to begin with. The second part yes, higher flow rate brings higher coolant velocity. I think what Cathar is trying to point out, is that he is using a restriction as a nozzle directly over strategic areas in his block. This lowers the overall flow of the loop but increases heat transfer because of the extreme velocity directed at various points of heat transfer, for a beneficial tradeoff. He is not using a restrictive barb nor do I. I in fact do use a 5/8" OD barb as the inlet on my design. The inner geometry of my block creates a velocity change throughout the short flow paths of the block and directly over the center of the core but also does not sacrifice as much overall flow volume. Both are radically different than what is available in the various mazes and open designs for an improvement,in my opinion Edit: Observation, Did the other "ultimate" thread max out or something?:shrug: |
10-07-2002, 03:23 PM | #22 | |
Responsible for 2%
of all the posts here. Join Date: May 2002
Location: Texas, U.S.A.
Posts: 8,302
|
Quote:
So if the heat is available at a higher transfer rate, then there is an opportunity to achieve better cooling. This is where flow speed, and fin design comes in: because a flat baseplate alone can be improved so easily with fins, it adds potential for better cooling. Add more flow, for more turbulence, and those make up, IMO, the three factors that are the most important. Energy stored in the bp is irrelevant, once the rig achieves a balance point (my mistake). A restriction, wether it's part of the barb or not, will achieve better cooling, because it creates a highly turbulent flow, which is great for improving the heat transfer to the water, as long as the accelerated coolant is located in a critical area. The problem with this, again, is that it's awfully restrictive, and given the same pump, will reduce the flow rate. The reduced flow rate will deteriorate the performance. As I remember, G_F, you and I went over this, some time back, but in your particular case, removing the nozzle gave you better performance. So again, the fin design is critical, because there may be some instances (Cathar for example) where an impingement may yield better results. The "Ultimate waterblock? Theory" thread is very much alive, and is now (as of today) the most viewed thread in all the forums. In fact, my design (#2) suggests a similar arrangement, but with fins in a radial pattern, because what's the point of having a center inlet if the fins are just going to go across it? I say, let's optimize the block for a radial, center inlet, set of fins that's less restrictive, impinge the thing, and watch in awe... My design isn't tweaked out yet, but it's there. |
|
10-07-2002, 03:48 PM | #23 | |
Cooling Savant
Join Date: May 2002
Location: Da UP
Posts: 517
|
Quote:
.................. Energy stored in the baseplate is what determines the heat of the die, if it balances out at a lower temp in the baseplate then the die will be cooler. .................. Agreed .................. That was in a different block and the tradeoff was in the negative direction for that design, higher overall flow was more beneficial. A restriction is built into the base of my EQ design which increases the velocity at the center of the bp. |
|
10-07-2002, 04:06 PM | #24 | |
Responsible for 2%
of all the posts here. Join Date: May 2002
Location: Texas, U.S.A.
Posts: 8,302
|
Quote:
................. Agreed, see above. ................. ................. So you don't have any flexibility, in case there's room for improvement. That's OK, I'm sure your block performs nicely. I really need to take the time to finish my #2 design, then I'll need someone to mill it for me... |
|
10-07-2002, 04:20 PM | #25 | |
Cooling Savant
Join Date: May 2002
Location: Da UP
Posts: 517
|
Quote:
:shrug: Can't wait to see it when it is done |
|
Currently Active Users Viewing This Thread: 1 (0 members and 1 guests) | |
|
|