View Full Version : Preparing for Water
Rotarian_SC
08-06-05, 10:50 PM
Right now my setup is good for all intents and purposes, and I don't plan on changing anything for two years. I hope to wait about two years, and then water cool my CPU and GPUs and buy a new PSU, and have it perform at least decently for future games. I'll use that watercooled system for another one and a half to two years.
Now at the end of those two years of water cooling, I know my system won't be very good, but at that point I'll be able to buy a new computer. If I can, I want to reuse my old water cooling setup on a new, but not very expensive system straight from the start. Is there much turnover in water cooling parts though, will a good water block/pump now still be a good water block/pump in two years?
My second question is what do I need to do to prepare for a water cooled system. I've read many things, but I know there's nothing like hands on experience. Right now, I have practically none. I built my computer, but I've never modified a component like put an aftermarket cooler on anything. These parts are also relatively new and expensive, so I don't want to mess them up. What should I do between now and two years so I'll be able to bolt a WC setup onto my computer and not screw it up?
Well first off, if you're going to cool your machine with water in a few years, see you then. Seriously though, if you're going to ask the questions now, why not DO it now? It's not all that bad; I just finished my first water-cooled setup and I'm mighty proud of it. Now, to answer your questions:
1) I don't see any 'radical' new block technology coming up in the future, unless they find a cheaper, more conductive metal than copper or a better way to transfer heat from the CPU die. If you go with a universal block like the Swiftech Storm, you'll be safe for a while to come.
2) Research, ask questions, and DO IT. We can give you all the knowledge in the world but if you can't apply it, then it's useless. Take the necesary precautions, ask questions where necesary, and take the time to be thorough and careful. Simple as that.
Good luck to ya.
Seriously, I would come back in two years. There is a LOT changing in watercooling. Just in the last six months, the quiet, yet powerful D5 came out. The PA160 from thermochill is the first radiator really designed just for PC watercooling to come out, and a lot more is going to be done in this area in the days ahead. If you do some reading on procooling, you will see that there are a lot of ideas and refinements "in the pipe" so to speak.
The Swiftech Storm just came out and was licensed from Little River Waterblocks, aka Cathar. This block is the best thing developed yet to date for mass production (keyword mass- not including G5). But give it a few years and it will be old news as watercooling continues to be refined. CPU's and GPU's continually are changing and putting out more and more heat. At some point, PC manufacturers may be forced to start watercooling due to increased heat loads, which is already becoming an issue with dual-core processors. This would dramatically alter the PC watercooling landscape. An "excellent, top-of-the-line" watercooling setup that you install today will be a "good" setup in a year or two, and then will be only a decent setup sometime thereafter.
But back to the other question. Why wait? What are you waiting for? The only real "preparation" is to read around on this and other good forums like XS and procooling to get an idea of how things work and how people install and troubleshoot setups. But there is nothing like reading and planning your WC setup, and then putting it together and making it happen! Why not go with a simple WC setup now? Get yourself a basic pump like a via aqua 1800 ($30), a cpu block used or something from the classifieds ($30?), and go get a heatercore for $20, build some kind of shroud, get some tubing from mcmaster ($10), and do it? You really learn watercooling by DOING watercooling.
Rotarian_SC
08-07-05, 10:50 AM
Well, there are a couple reasons I can't do it now. First, well I don't really need to do it because I can play everything will all the bells and whistles and it plays fine. Second, I would have to go and buy a new and larger case, which would probably run me another $150 for a good one. I would need a better PSU too, except I think that might be around $200. So add that on a cheap WC system and the overall upgrade is around $450. I also want these components to last me four years, so I don't want any uncalled for stress on them now.
I'm not asking questions here like what model block should I get, or what pump because that advice will have changed in two years I understand. I'm more asking what I should do between now and then to prepare for having a WC system, which is something I can't really ask later.
I'm more asking what I should do between now and then to prepare for having a WC system, which is something I can't really ask later.
In that case I would say don't put any more money into your current rig, and save save and save for later. If you can play most games now, you should be able to play most future games on lower settings untill you get a new computer all together.
rogerdugans
08-07-05, 11:12 AM
My suggestion:
build another computer. :)
Seriously: pick up old, cheap parts and you should be able to build something that you are less nervous about screwing with for pretty short money.
Pick up used water cooling components for the above system- again, this stuff can be found for pretty low money.
The system you can build for less than $200 total for everything (including water!) is NOT likely to break any speed records or anything, but it will do a few good things for you:
1) give you practice changing heatsinks and blocks
2) give you practice modding a case for water cooling components
3) give you a spare system in case (knock on wood) something happens to your current computer
4) Saves you a LOT of money if you DO screw up. ;)
To really build a complete, water cooled computer for that kind of money IS possible, but it can take quite a bit of time to find parts cheap enough.
Your 2 year time limit should be enough. :)
You have top-of-the-line components now from your sig. A 500w Fortron Blue Storm is PLENTY for watercooling. Most DC pumps only use like 30w max (if that), which is a drop in the bucket for that PS. I have the same PS, and if you add up the wattage requirements of components, that PS has plenty of juice. So you have no reason to upgrade your PS.
I don't know what case you have, but there are very few cases that you can't watercool. Any kind of midtower can be watercooled.
The only thing you can do if you plan on waiting is to simply read around on forums like here on OC, Xtreme systems, maybe procooling, etc. and see what people are doing and learn from their questions/problems/solutions. But you still will have a learning curve once you actually DO watercooling. Reading about it and figuring out what you want and your goals with watercooling are very important, but nothing replaces hands-on.
Rotarian_SC
08-07-05, 01:00 PM
http://www.mustangmods.com/data/10014/dscn0429_01.jpg
It looks a bit crowded for one to me, although I may come to learn otherwise. I have a midtower case. Having as much of the WC system external as possible would help I suppose. I have a Lian Li 65B with 80mm fans so I would imagine it would take a bit of case modding to make it work if I chose to use it.
Hopefully I can get a bunch of cheap parts from my school (or people at my school) for free to build a system I can WC and not worry about, to work on my learning curve with. Hopefully I won't have to spend any money on it at all, except for some WC equipment.
When I do get a pump for my computer I would imagine it would have to be fairly strong as pumps go because it will have to handle 2 blocks for gpus and one for the cpu, but that would still take only about 30W to run?
I have a feeling if I save and save I still won't have have the money for a system as nice as mine would be if I WC, although that may change in two years, but I doubt it.
Check out the link in my sig: I watercooled my system for the first time (also in a mid-tower) with near top-of-the-line components and everything went flawlessly. I don't see why you can't do it either. :)
Rotarian_SC
08-07-05, 01:25 PM
Nice job, and I admit it is tempting (and I would plan on adding 2 more waterblocks than you have now, which might be a bit tighter for space). But right now I don't have the money, experience, or knowledge to risk it. I'm trying to save every scrap I can. I didn't get the ability to pay for my current system by going out and buying everything I wanted :). Also the potential gain isn't much because I don't even need to OC this system right now (blasphemy on this board ;) ) but don't worry because that will definately change. From now on when I do computers I'm going to buy less expensive systems and hopefully WC them from the start, starting over from almost scratch every 2 years instead of every 4.
I was considering WCing my 6800 GT as you are, but I figured the stock cooler is fine (for now), but those (sorry to say) ugly-ass Ledteks of yours definetly need a new cooler/water blocks. :p
I guess I can see where you're coming from on the issue of money, although I haven't had the issue myself. I'm lucky enough to have had an extremely high paying job through high school and now into college which fuels my addiction. :)
As for the gains: I not only wanted a more efficient way of cooling my hot-ass Prescott, but also desired a quiter system. Not having a CPU fan is a huge thing, and being able to undervolt my 120x38mm (120CFM@45dB) fan to 5V allows me t o operate in any but the most demanding circumstances, almost silently.
Seriously, I would come back in two years. There is a LOT changing in watercooling. Just in the last six months, the quiet, yet powerful D5 came out. The PA160 from thermochill is the first radiator really designed just for PC watercooling to come out, and a lot more is going to be done in this area in the days ahead. If you do some reading on procooling, you will see that there are a lot of ideas and refinements "in the pipe" so to speak.
The Swiftech Storm just came out and was licensed from Little River Waterblocks, aka Cathar. This block is the best thing developed yet to date for mass production (keyword mass- not including G5). But give it a few years and it will be old news as watercooling continues to be refined. CPU's and GPU's continually are changing and putting out more and more heat. At some point, PC manufacturers may be forced to start watercooling due to increased heat loads, which is already becoming an issue with dual-core processors. This would dramatically alter the PC watercooling landscape. An "excellent, top-of-the-line" watercooling setup that you install today will be a "good" setup in a year or two, and then will be only a decent setup sometime thereafter.
Hmmm. Interesting points. I personally feel we are getting pretty close to as good as water-cooling can offer today. Still some small ways to go, but we're down to ekeing out the last few tenths of a degree C here and there.
Assuming a true 100W heat load on a 12.5x12.5mm CPU die-size.
For waterblocks we have at absolute best ~2.5C or so to gain over something like a G5, and that gain is not going to come easy at all. To get more than 1.5C better than a G5 we'd have to start cracking some pretty major physical concepts on liquid-metal heat transfer. So let's postulate that there's still 2.0C realistically left to gain with waterblock advances.
For pumps, incorporating their effective pumping power and heat dump. As far as ideal single pumps go we do have two existing choices in the AQX-50Z @ 15.0v (if you can even get it to work at that voltage), or the Iwaki RD-30 @ 17-18v. Can still do better than that though. If we instead compare against the best of the available stock voltage pumps on the market, we have about a 0.7C improvement to gain on our 100W heatload by building a truly optimised water-cooling pump. If people are using the AQX-50Z @ 15.0v or the RD-30 @ 17-18v, then there's only about 0.2-0.3C left to gain from pump improvements.
For radiators, let's assume we're talking about a single 12cm 30dBA fan on a 12x12cm radiator performance. For our 100W heat-load, once we factor in existing radiator efficiency and the thermal capacity limit of air we can conclude that at best we'll do about 1.5C better than what's on the market right now in a 12x12cm form factor. This is not to slight existing radiator manufacturer's efforts, I'm dealing in pure theoretical conjecture here and such may not even be realistically achievable in the slightest.
Okay, add all that up and an existing top-of-the-line single 12cm fan radiator based watercooling kit which for our example CPU size would have a total system C/W of around 0.14 (or a 14C rise above ambient-air of the 12.5x12.5mm CPU die at 100W heat load), that at absolute best that'll come down to around a 0.010C/W assuming everything was at the pinnacle of what water-cooling can do. i.e. at absolute best, 4C better.
Realistically though, in the next 2 years I don't foresee water-cooling in terms of our theoretical reference sized kit improving by much more than 2-2.5C. That last 1.5-2C above that is going to be a mongrel to achieve and will probably take a bit longer than 2 years to achieve.
So Cathar, you don't think that any major improvements in watercooling are on the horizon? Maybe I am niave, but although the MCP655 for instance may not break any new records performance wise over the MCP650, it sure is a lot quieter! Just recently the HWLabs rads dramatically dropped in price also. And the PA160 is the first alternative to really offer decent cooling with just one low CFM fan. These may not be performance improvements per say, but I would think of them as notable improvements that are very recent. Not to mention your Storm being mass-produced.
Do you see anything else on the horizon in terms of cooling besides the current water setups? Do you think we will see more PC makers going water?
Oh, those have all been very significant improvements. It's going further that's the trick. I'm looking ahead to the next two years, given everything you just mentioned as already being what's available today.
We will see radiators slowly improve and get cheaper. From a performance viewpoint though, like I said above.
The "ideal" computer water-cooling pump that is realistically achievable is one that is:
20' or 6.5mH2O of peak pressure head
3.5gpm or 13.5lpm of peak flow
11.5' or 3.5mH2O of pressure head at 2gpm or 7.5lpm (**)
12vdc
25dBA noise level at most
20W peak power consumption at (**) flow/pressure noted above (preferably less)
Maximum of 10x10x10cm bounding box size, including barbs, preferably even smaller
Such a pump would hold near universal appeal for high, moderate, and low flow setups alike.
We will start to see the rise of cheaper/better 20-25dBA fannage optimised radiators too.
The next step for water-cooling beyond this will be active dewpoint-level water-chilling, which is something that's been worked on by myself and others.
ls7corvete
08-07-05, 11:20 PM
Oh, those have all been very significant improvements. It's going further that's the trick. I'm looking ahead to the next two years, given everything you just mentioned as already being what's available today.
We will see radiators slowly improve and get cheaper. From a performance viewpoint though, like I said above.
The "ideal" computer water-cooling pump that is realistically achievable is one that is:
20' or 6.5mH2O of peak pressure head
3.5gpm or 13.5lpm of peak flow
11.5' or 3.5mH2O of pressure head at 2gpm or 7.5lpm (**)
12vdc
25dBA noise level at most
20W peak power consumption at (**) flow/pressure noted above (preferably less)
Maximum of 10x10x10cm bounding box size, including barbs, preferably even smaller
Such a pump would hold near universal appeal for high, moderate, and low flow setups alike.
We will start to see the rise of cheaper/better 20-25dBA fannage optimised radiators too.
The next step for water-cooling beyond this will be active dewpoint-level water-chilling, which is something that's been worked on by myself and others.
Evaporative cooling? Is there something in the works or just something you see as the next step?
First you say that system pricing is justified and them you go and start promoting evap, sigh, I dont get it sometimes.
Anyways, I see some basturd cooling coming in the future, semi-evap, semi-pelt(something like mine...), small phase change, non-standurd heat pipe applications and any combination of the above. All the above is being developed. I think that evaporative cooling combined with a low volt pelt and put into a 5.25 package would be very sexy.
Evaporative cooling eh? Is there something in the works or just something you see as the next step?
First you say that system pricing is justified and them you go and start promoting evap, sigh, I dont get it sometimes.
No, no. Not evap at all. I never said anything about evap.
I mean a TEC-based water-chiller with a current controller that monitors ambient, relative-humidity, and the cold-side water temp, and holds the water temperature at ~0.5C above the dew-point (or otherwise as cool as it can get if dew point drops too low).
I implemented a low power draw unit (~90W power draw max) capable of cooling about 200W of heat load thusly, and have been playing with it since. Right now the project is awaiting me to find some time to take it to the next stage with the help of an electronics power supply whiz-kid I know.
More details here:
http://forums.overclockers.com.au/showthread.php?t=334206
Been playing with this stuff off and on for the last 3 years. Proven it to a point where it can address all heat-load scenarios, but I just need to find the time to get it finalised. As it stands, it bridges the gap between high-end water and effective single-stage phase-change, but retains the flexibility of a regular water-cooling setup.
ls7corvete
08-08-05, 12:20 AM
No, no. Not evap at all. I never said anything about evap.
I mean a TEC-based water-chiller with a current controller that monitors ambient, relative-humidity, and the cold-side water temp, and holds the water temperature at ~0.5C above the dew-point (or otherwise as cool as it can get if dew point drops too low).
I implemented a low power draw unit (~90W power draw max) capable of cooling about 200W of heat load thusly, and have been playing with it since. Right now the project is awaiting me to find some time to take it to the next stage with the help of an electronics power supply whiz-kid I know.
More details here:
http://forums.overclockers.com.au/showthread.php?t=334206
Been playing with this stuff off and on for the last 3 years. Proven it to a point where it can address all heat-load scenarios, but I just need to find the time to get it finalised. As it stands, it bridges the gap between high-end water and effective single-stage phase-change, but retains the flexibility of a regular water-cooling setup.
OIC, I was aware of that setup but it got below dew-point.
http://mcshaneinc.com/html/5C7-350.html
http://forums.procooling.com/vbb/showthread.php?t=6600&highlight=variable
Meant to put that link in one of the procooling threads But I guess I messed it up. That system seems a bit complex, I guess it all comes down to how cheaply the peltier blocks can be made for. I would be interested in seeming how 320W units would work compared to the 226 ones you are currently using.
Rotarian_SC
08-08-05, 10:00 AM
For my junk setup, what would you say is the bare minimum that I should have? I would assume an old graphics card (not onboard graphics), but is a P3 ok, etc?
for a junk setup, it can be anything that works. people were overclocking 486DXs up to a whopping 66MHz back in the day. If you want to get into watercooling, the p3 would be a good start. a new p3 system can be had for less than a hundred bucks these days. onboard graphics is ok, unless you are going to OC that too. might actually get better overclocks without the extra current drain from the video card.
Rotarian_SC
08-08-05, 02:05 PM
Well I planning to WC my GPUs as well, so I want some practice on a card. I'm not really worried that much about how far it will OC either, because it's not like I'll be using it to play games or anything :p.
I was wondering if the mechanisms for fitting a water block had changed a good deal because of the chipset or something. I would still like my junk system to be good practice and somewhat similar to what I'll have to do on my current computer.
most blocks (swiftech dangerden) have mounting kits for a wide range of graphics/cpu (chipset can be more complicated (especialy nf4's)) so all that you would likely need to move the system from the jnk to the final would be a different mounting kit
Different hardware require different mounting solutions. Not everything works out. Sometimes components interfere or just don't fit. Modification is a word you will learn to use. Its a process to say the least. The one thing the test rig will do, is give you confidence to run water inside a computer, a step a few are hesitant to take. The main thing is to learn the principles and techniques, and sweat the small stuff later.
rogerdugans
08-08-05, 03:02 PM
The main thing is to learn the principles and techniques, and sweat the small stuff later.
Wise words right there, and can be applied to anything in life. :)
Moto7451
08-08-05, 03:28 PM
Chipset holes for AMD setups are the same. Same goes for all the radeon cards since the 8500. The P4's retention mechanism for its chipsets are the same also i believe (not sure about the S423 stuff). NVidia cards have been kind of all over the place.
A good junk system would consist of any cheap Athlon XP mobo with mounting holes (the ASRock KT400, KT600, & KT880 boards comes to mind) would work great along with a 1700+ TBredB or something similar. That way you can test under overclocked conditions since that setup will overclock easily. Another option would be a Sempron 2200+ if you can't find a 1700+ for a lot price. Add in an 8500, 9000, 9100, or 9200 with mounting holes (SE, Pro, w/e - its all good) & then you're ready to test out a water cooling system with modern components. A P3 doesn't have the same heat load as a modern CPU but an AXP is much closer so I think it'll be a better choice for a sub $100 test rig.
Rotarian_SC
08-08-05, 04:10 PM
Ok, I'll try to get an Athlon XP board if I can, but there aren't any cheap gpus you know of that have similar mounting to a 6800gt? If not I think my brother had a 9000 (not mobile) in his Alienware, which he broke the backlight on. I'll see if I can get that out of his if I can, and if it's a mobile or not.
Thanks for the info guys.
rogerdugans
08-08-05, 05:05 PM
I don't know much about the newest/bestest video cards (or anything else, either, lol) but I do know that for quite a while now some manufacturers change mountings quite a bit-
I have had a few GF4 Ti cards and at least two different mount spacings.
The FX5900 has another dimension, and my 6600gt has yet another one.
Keeping a chipset block from card to card will most likely be more of a problem than a cpu block...
But if/when you get the hang of this stuff, almost nothing is impossible. :)
I wouldn't say you necesarily need something with the same dimensions as a 6800GT, as practice on any card should suffice to give you some hands on experience when installing a block on a different card.
I still say you just charge it on your new setup! ;)
rcillig
08-08-05, 07:05 PM
1 month ago I decided to try WCing, took me about a week or so to decided on a setup the next 3 weeks finding my parts most of them used..... Im up and going for under $200 and Im very happy... learning as I go most of the time.... but ive read most every post in the WCing section over the last month... its all there... if you got a problem someone has posted it. Just got for it as soon as you can.... its a blast.
Rotarian_SC
08-09-05, 10:36 AM
Do you have any suggestions where to buy the junk system cheap. I think I should be able to get a better deal than the egg is offering here. Is it safe to buy off of Ebay? I'm not able to access the deals part of this forum yet, but is this sort of stuff reliably and cheaply sold there as well?
I think the most expensive part of the system will be my heatercore. Heatercore technology doesn't seem to be one of the things that is changing much if the best and cheapest ones out there are the kind off 70s and 80s cars. I'm figuring that I'll need a decently strong heatercore to run two blocks, so I might as well prepare it to be my future radiator for three blocks. I'll probably go for a '77 Bonneville one, but I have a couple questions. To save room in my case, would it be detrimental in any way to have it as an external one? What would you reccomend I should have on it to handle 3 blocks, 3x120mm? And also, what sort of cheap case should I get for my cheap setup.
rcillig
08-09-05, 10:27 PM
Ive bought lots of stuff from ebay, just check out the selller before you bid. Also when you get access here in the forums you can get great deals.
Rotarian_SC
08-12-05, 10:04 AM
So far, I have gotten free an old Compaq Presario 7470 with an AMD K6 3D Processor. How's that? I'm not too familiar with archaic technology. I don't think it's too good because it sounds like it is having a heart attack every time it comes under any load.
rogerdugans
08-13-05, 12:35 PM
So far, I have gotten free an old Compaq Presario 7470 with an AMD K6 3D Processor. How's that? I'm not too familiar with archaic technology. I don't think it's too good because it sounds like it is having a heart attack every time it comes under any load.
;)
Learn that system by getting it straightened out a bit- maybe it has a dead or dying fan?
Most water cooling components can easily move from one system to another- pumps, fans, tubing, radiators.
Water blocks are a bit trickier: sink/block mountings are sometimes very different from one architecture to another... A bit of creativity and modding can rectify this, but it may be worrisome for those who are new at this stuff.
There are however, two socket types that are pretty much compatible: Socket 7/Super Socket 7 (as in old AMD k6 chips) and Socket A. :)
Most blocks that use the socket-tab mounting method should work on an old socket 7 board, so you may be in luck, my friend.
Rotarian_SC
08-24-05, 12:24 PM
Ok, I've got a video card lined up in place, but now I'm looking to find a 1700+ TBred. Couple questions, what should I be looking for. I know that it is the 130nm core as opposed to the Palimino, but not all Ebay sellers list that, so does it have more L2 cache also? I also found out through a thread on here that they started locking them around number 0343, so I should get one with a number lower than that. If a seller has trouble finding the number on his chip where would it be? I did a search here, but it didn't turn up much easily accessible info.
vBulletin® v3.8.7, Copyright ©2000-2012, vBulletin Solutions, Inc.