I'm a complete newbie when it comes to water cooling, so I've been reading everything I can find on the subject. I have a tentative design for a quiet water cooled system, and I'd like comments/suggestion/whatever.

I have a YY Cube case. This is a two-chambered case with the motherboard in the left chamber and the drives and PSU (mounted at the bottom!) in the right chamber. Photos of the case are available at the manufacturers site and at Case Outlet for your viewing pleasure.

My plan is to:

1. Remove the card guides in the front of the MB chamber and the 3.5" drive brackets in the rear of the drive chamber.

2. Enlarge the 120mm intake in the MB chamber and put a tall radiator just inside that intake. The radiator will be a single-pass design. Do some cutting and drilling in the front bezel to let more air into the intake.

3. Use a pair of 120mm L1A or Papst fans to suck air through that radiator into the MB chamber. Use a shroud to space the fans away from the radiator. Run fans at lowest speed that will still work for cooling.

4. Block the side 92mm intake in the MB chamber and the small vent holes at the rear of that chamber to force all air outlet through the drive chamber.

5. Install a Seasonic PSU in the stock location.

6. Above the PSU, install a 2-271 radiator (see Leaky Car) with a shroud to adapt it to the pair of 92mm outlet vents in the rear of the drive chamber (where the 3.5" bays used to be.) NO FAN ON THIS RADIATOR. The airflow through this radiator will be due to overpressuring the case with the 2x120mm intake and only an 80mm output fan on the Seasonic.

7. Put the pump and reservoir in the drive chamber.

8. Set up the water cooling loop to look like this:

pump
| (1/2"ID)
2x120mm radiator
| (1/2"ID)
CPU (White Water block)
| | (2 x 1/2"ID)
2-271 radiator
| (1/2"ID)
manifold
| | | (3 x 3/8"ID)
GPU/Northbridge/disk drives
| | | (3 x 3/8"ID)
manifold
| (1/2"ID)
reservoir
| (1/2"ID)
pump

The White Water block has a single inlet and two outlets. Most people seem to use a Y connector to combine the two outlets, but the 2-271 radiator has two inlets and a single outlet, so I want to use that as my "Y connector." The 2-271 is a bit of an unknown quantity. I figure it must be at least a double-pass radiator, and I don't know how much flow restriction it imposes. The White Water block likes a lot of flow.

Since most of the smaller water blocks for GPUs, NB, etc. tend to have 3/8" barbs and need to fit in tighter areas, I thought that using a manifold to split the loop into parallel 3/8"ID lines for the GPU, NB, and disks (and maybe a SB and/or rectifier loop) will give me a little more flexibility for positioning tubing without restricting water flow. Unless the splitter and combiner manifolds create too much restriction. Unknown territory here.

I'm not sure about cooling the vidram, SB, and rectifiers. With a pair of 120mm fans blowing (slowly) across that MB, maybe I could just stick heat sinks on everything that gets hot and be happy with it. I'm planning on using an Asus A7N8X Deluxe MB and an ATI All-In-Wonder 9800 Pro.

I have absolutely no metalworking skills, other than soldering plumbing together. I'm planning on using commercial water blocks for the CPU, NB, and GPU and soldering together a "plumbers delight" setup for a pair of hard drives in a sound isolation box. The hard drive box will sit at the bottom of the 5.25" drive bays in the drive chamber.

So what do you think? Will I need some huge wombat of a pump to push the water through all of these things? Is the manifold idea a good plan? How about that second radiator acting as an "intercooler" of sorts? Has the government secretly been slipping hallucinogens into my drinking water?

Thanks for reading,
Scott The Long Winded

Wow, thats a lot, but I'll tell you what I can. I'm watercooling and planning on getting a YY cube for my set up, so I've thought a bit about this myself.

First of all, fitting the radiator on the left front will be difficult- between the left edge of the case and the door is just barely large enough for a 92mm fan.

I don't think the second radiator would work very well passive, since the high restriction will leave most air escaping elsewhere, like the vent holes on the left side. However, another low powered 120mm won't add noise over the front 2 fans, and would result in excellent water temps.

You might want heatsinks on the vid ram, a heatsink for the southbridge would be wasted, but I have one anyways; I took off my northbridge heatsink and stuck it on my southbridge. I'm gonna WC my MOSFETs just for fun too. I have no metalworking skills either; my plan is to get soft copper tubing and a pipe bending spring (I don't even know what that is, but it lets you bend soft pipe without kinking) and use it to make a HD cooler, MOSFET cooler, pump cooler (more on that next), and a PSU cooler.

The manifold is an excellent idea; small blocks like a GPU block or a NB block can be very restrictive; my DD z-chipset is more restrictive than my radiator and maze3 combined. but since I don't need much flow through it, running it in parallel with the other coolers will reduce restriction.

And the pump... with an Eheim 1250 you'll probably end up with pretty low flow rates. For a little less ($55) I bought a Little Giant PCL-020 at my hardware store that has 12.2' shutoff head, 300 GPH at 1'. But it consumes 47 watts. I'd like to replace it with an Iwaki MD-15Z. Its a bit more expensive ($120) but its got pretty much the same flow/head curve, but only consumes 31 watts. This implies to me that its a much superior motor which will not only be cooler but quieter too... I'd love to hear one for myself.

Thanks for the input, Zhentar.

I think that Yeong Yang makes several different versions of the YY-0221. The one I have looks like the one in the Virtual Hideout Review and the front looks like photo in that review. It looked like a 120mm would fit there, but I didn't measure. Looks like I need to measure this weekend. I don't think I've ever seen a core that would fit well with a pair of 92mm fans. That could be because I've never looked for one, though Maybe I can do something strange with an angled radiator...

I was planning on sealing up all the holes so that the air was forced out though the PSU and right-rear fan holes. I'd probably get some leakage around the front drive bays, though, so I may take your advice about putting another 120mm on the rear rad, if it doesn't make things too loud.

A pipe bending spring is a close-wound spring (like the old screen door springs) that fits snugly over the tubing you're bending. The coils of the spring keep the tubing from kinking. The coils don't let the "sides" of your bend push outward, so the kink can't get started. I was thinking of a similar idea for the VRs and the disk drives. I've done plumbing, but never milling and drilling.

For a disk block I'd like to have copper plate tall enough to mount two drives on each side of the drives. I'd do a "U" of tubing running down one plate, across the end of the drives, then back up the other plate. I've been trying to figure out how to take 1/2" copper tubing and keep it round on both ends for soldering to plumbing fittings (barbs and elbows) while flattening it in the middle for better contact with plates. Maybe something ugly involving dowels and a vise would work...

Thanks for the info on the pumps. I was looking toward an L30 or something similar.

Scott

I've been trying to figure out a way to flatten the tubing some too, the best I've come up with is beat it with a hammer a bit and then sand that flat. Fortunatly, the drives and MOSFETs don't put out alot of heat so you don't need a great contact.

Thanks for the info on the pipe bending spring, I had some guesses but I wasn't sure.

Give an inch or two between the rad and the front of the case and you could make a 120mm to 92 mm shroud; the radiator itself is so restrictive the space between the door and the side wouldn't restrict the air much at all.

Good place to start... If you haven't found them yet, I would suggest the articles by 'BillA' over on Pro-Cooling and Overclockers. A great deal of my information has come from there, and the various watercooling threads on Pro-Cooling.

OK

Seems reasonable, the single pass rad is a good idea for lower water flow resistance, and possibly more flexibility in arranging airflows. However remember that the number of SP rads is very small compared to the number of DP rads, so you may find yourself limited as to the sizes you can choose from. FWIW, I reccomend the 2-342 as nicely sized for 2 x 120mm's - it is a little wider, but a perfect length.
All good ideas. Note that you should use 38mm thick fans to get maximum suction pressure, and space them at least 1" off the rad face. Also check the PQ curves on any prospective fans and make sure they have good static pressure, and maintain a good flow volume as resistance increases. A reasonable working Assumption is that a good fan will deliver about 50% of its free air rated CFM when sucking through a rad.

Good for ensuring airflow through the case, but not good for cooling any hot components on the mobo - essentially this will be putting the mobo in a stagnant air pocket... I would reccomend giving at least some exhaust airflow over the mobo.

This is probably not the best idea... You will be adding a great deal of airflow resistance, considerable water flow resistance, and not gain much in terms of cooling, in fact I wouldn's be suprised to find it a net loss!

Bear in mind that the CPU will only increase the temp of the water going through it by about a degree or less.

You might get some additional cooling of the overall temp by going through the second rad, but I doubt it would be very much. However you will have added a great deal of restriction to the airflow, giving you LESS flow through both rads than you might have gotten from the single rad up front. Further, you will be restricting the water flow by adding two manifolds plus the radiator itself (which looks to be a very restrictive unit - each side appears to be dual pass, with one side only having 5 tubes, and the other having 7 judging from the picture... Assume 1/2 that count is used for each pass, and you have serious flow restriction)

I would reccomend getting rid of the second rad entirely. Instead sort the secondary coolers (all of those devices are either less heat sensitive than the CPU and/or need less cooling BTW) into 2 groups w/ approx equal flow restrictions and route an output of the WW to each group. Let each group dump into the res w/ a seperate inlet. This will give max flow through all components, and give the optimal cooling. On the airflow side, try to split the flow so that about half the air goes out each chamber.

Reasonable. The Ehiem 1250 should give plenty of volume for this setup. If you want a different pump, look for head pressure more than volume.

Set up the water cooling loop to look like this:

pump
| (1/2"ID)
2x120mm radiator
| (1/2"ID)
CPU (White Water block)
| | (2 x 1/2"ID)
GPU/Northbridge/disk drives
| |
res
| (Biggest hose you can make fit!)
pump

It is also best to arrange the plumbing so that the water flows upwards from the pump to the res as much as you can. The res should ideally be the highest point in the loop. This will minimize trapped air in the system (Which is deadly to it's cooling ability) and make getting rid of it as easy as possible.

As I mentioned above, this looks like a very restrictive rad for water flow. It *might* also be hard to come by - the application is for a few Ford / Mercury cars of late 60's vintage. Leakycar shows it, but the website I use for cross references doesn't show it, and when I try to look it up by application I keep getting the wrong core.

If you really want to, you might try splitting the flow further among the secondary blocks so that they are in parallel rather than in series. (Remember that to a first approximation plumbing works much like wiring, flow resistance goes up in series, down in parallel)


Reasonable again, but probably no real need to go to extremes.
Assuming you have halfway decent airflow in the mobo chamber, you shouldn't need extra cooling on anything you wouldn't worry about in an AC system. Experimentation will be needed. If you do need more water cooling, get a copper plate the appropriate size, and solder a length of copper tubing to it. WB's made that way will be enough to cool just about anything except the CPU. (It is how I'm making my drive blocks btw)

[/quote]I have absolutely no metalworking skills, other than soldering plumbing together. I'm planning on using commercial water blocks for the CPU, NB, and GPU and soldering together a "plumbers delight" setup for a pair of hard drives in a sound isolation box. The hard drive box will sit at the bottom of the 5.25" drive bays in the drive chamber.[/quote]
Not a bad plan, however I would advise that you look closely at what each block is made from, and avoid the dreaded 'mixing of metals' problem - since you are going to be using a copper / brass rad, I wouldn't allow any aluminum parts in the system. Alot of the commercial blocks are made out of aluminum because it's easier to machine and less expensive, but they won't play nicely with copper in the long term.
Mixed bag; Not if you pick a pump w/ a good head pressure rating; no; definitely no; perhaps, send me a sample for testing

[/quote]Thanks for reading,
Scott The Long Winded[/quote]
Gooserider The Even Longer Winded

Good place to start... If you haven't found them yet, I would suggest the articles by 'BillA' over on Pro-Cooling and Overclockers. A great deal of my information has come from there, and the various watercooling threads on Pro-Cooling.

OK

Seems reasonable, the single pass rad is a good idea for lower water flow resistance, and possibly more flexibility in arranging airflows. However remember that the number of SP rads is very small compared to the number of DP rads, so you may find yourself limited as to the sizes you can choose from. FWIW, I reccomend the 2-342 as nicely sized for 2 x 120mm's - it is a little wider, but a perfect length.
All good ideas. Note that you should use 38mm thick fans to get maximum suction pressure, and space them at least 1" off the rad face. Also check the PQ curves on any prospective fans and make sure they have good static pressure, and maintain a good flow volume as resistance increases. A reasonable working Assumption is that a good fan will deliver about 50% of its free air rated CFM when sucking through a rad.

Good for ensuring airflow through the case, but not good for cooling any hot components on the mobo - essentially this will be putting the mobo in a stagnant air pocket... I would reccomend giving at least some exhaust airflow over the mobo.

Probably not the best idea... You will be adding huge amounts of airflow resistance, considerable water flow resistance, and not gain much in terms of cooling. I wouldn't be suprised to find it a net loss!

Bear in mind that the CPU will only increase the temp of the water going through it by about a degree or less.

You might get some additional cooling of the overall temp by going through the second rad, but I doubt it would be very much. However you will have added a great deal of restriction to the airflow, giving you LESS flow through both rads than you might have gotten from the single rad up front. Further, you will be restricting the water flow by adding two manifolds plus the radiator itself (which looks to be a very restrictive unit - each side appears to be dual pass, with one side only having 5 tubes, and the other having 7 judging from the picture... Assume 1/2 that count is used for each pass, and you have serious flow restriction)

I would reccomend getting rid of the second rad entirely. Instead sort the secondary coolers (all of those devices are either less heat sensitive than the CPU and/or need less cooling BTW) into 2 groups w/ approx equal flow restrictions and route an output of the WW to each group. Let each group dump into the res w/ a seperate inlet. This will give max flow through all components, and give the optimal cooling. On the airflow side, try to split the flow so that about half the air goes out each chamber.

Reasonable. The Ehiem 1250 should give plenty of volume for this setup. If you want a different pump, look for head pressure more than volume.

Set up the water cooling loop to look like this:

pump
| (1/2"ID)
2x120mm radiator
| (1/2"ID)
CPU (White Water block)
| | (2 x 1/2"ID)
GPU/Northbridge/disk drives
| |
res
| (Biggest hose you can make fit!)
pump

It is also best to arrange the plumbing so that the water flows upwards from the pump to the res as much as you can. The res should ideally be the highest point in the loop. This will minimize trapped air in the system (Which is deadly to it's cooling ability) and make getting rid of it as easy as possible.

As I mentioned above, this looks like a very restrictive rad for water flow. It *might* also be hard to come by - the application is for a few Ford / Mercury cars of late 60's vintage. Leakycar shows it, but the website I use for cross references doesn't show it, and when I try to look it up by application I keep getting the wrong core.

If you really want to, you might try splitting the flow further among the secondary blocks so that they are in parallel rather than in series. (Remember that to a first approximation plumbing works much like wiring, flow resistance goes up in series, down in parallel)


Reasonable again, but probably no real need to go to extremes.
Assuming you have halfway decent airflow in the mobo chamber, you shouldn't need extra cooling on anything you wouldn't worry about in an AC system. Experimentation will be needed. If you do need more water cooling, get a copper plate the appropriate size, and solder a length of copper tubing to it. WB's made that way will be enough to cool just about anything except the CPU. (It is how I'm making my drive blocks btw)


Not a bad plan, however I would advise that you look closely at what each block is made from, and avoid the dreaded 'mixing of metals' problem - since you are going to be using a copper / brass rad, I wouldn't allow any aluminum parts in the system. Alot of the commercial blocks are made out of aluminum because it's easier to machine and less expensive, but they won't play nicely with copper in the long term.
Mixed bag; Not if you pick a pump w/ a good head pressure rating; no; definitely no; perhaps, send me a sample for testing


Gooserider The Even Longer Winded

Thanks for all the info, Gooserider. Looks like I have a lot more reading to do.

I didn't realize that the CPU block doesn't really raise the coolant temp that much. If that's the case, then the second radiator is definitely not needed, and I save the noise of another fan on it. I guess that explains how small diameter tubing systems like Innovatek and Koolance still manage to be reasonably effective without a lot of water flow. They're not particularly quiet from what I've read, though.

The YY has an opening between the chambers all along the top of the MB tray. I figured that most of the air from the intake 120mm fans would take that route to the drive chamber rather than stagnating over the MB. Maybe not. I'll have to experiment. Maybe a single 120mm drawing through a rad on the intake with a 92mm intake fan on the side would work better. Gotta run them slow, though, and soft mount the 92mm.

I may also try splitting the WW outlets into a pair of 3/8"ID lines from each to run to the other blocks. Wrestling 3/4"OD lines all around the inside of the case doesn't sound like much fun. Definitely copper/brass throughout, though. I did pick up that mixing metals is evil. Makes me wonder why the cu/al Swiftech blocks are so popular. Sure, it's anodized aluminum, but I'd still worry. I had a guy from Swiftech tell me he's running their standard GPU block on an AIW 9800 Pro card. The folks at Danger Den told me that theirs doesn't fit and they're working on one that does. I made noises of doubt to the Swiftech guy, asking if the TV tuner doesn't get in the way, but he insisted that he's been running it at home just fine. Dunno.

Thanks for the name at ProCooling. I've read many of the articles there, but I haven't dug through the forums much. More reading for me!

Scott

I wouldn't worry too much about stagnant air in the mobo side, if your watercooling the GPU/NB/CPU/MOSFETs theres not anything left that produces appreciable amounts of heat. You might have 10 or 15 watts of heat dissipation there if you have PCI cards, but it shouldn't be a problem.

If you've got enough flow theres not much of an increase of coolant temp going through blocks, though if you get the flow too low it will matter.

Fitting 3/4" OD tubing won't be much of a problem in such a large case, even 1" OD tubing might fit, but after splitting the flow 3/8" is fine; BladeRunner splits his loop 4 ways from the WB and goes even smaller than that.

Anodized aluminum won't be as much of a corrosion problem, unless something gets through the thin anodized layer; then you've got a problem. For example, brass barbs can potentially cut through the anodization. Anodized aluminum is safer, but I still would avoid it.

ProCooling doesn't have many up to date articles, but thanks to the waterblock design forum there are quite a few watercooling experts; you can find info on whatever you need in them. Unfortunatly some of these experts are real assholes, and aren't too easy on newbies, but they still provide a lot of valuable information.

Two quick points...

1. Over on ProCooling, the earlier mentioned 'BillA' goes by the handle "Unregistered", with the accurate tagline 'AKA BillA - he's grumpy'

2. I agree with the earlier assesment of some of the pro's over on the ProCooling forums. They can be a$$#0!e$ and rough on the newbies, but my observation was that much depended on what you asked, and even more importantly, HOW you asked it... (And I think this applies to almost any area BTW!) Pro's are human just like the rest of us, and they get tired of answering 'dumb newbie questions' The biggest single thing I would advise is to do your homework before posting - there are lots of articles explaining the basic concepts and how things work. READ THEM!
then think about them, and READ THEM AGAIN! Then if you still don't understand, post a question, but make sure your question shows that you tried to figure it out for yourself first.

I know I look far more favorably upon the person who says 'I read this article / FAQ, and don't understand' than I do the person who asks the question the article was written to answer... (it is worth noting that the person who tried but doesn't understand is usually more polite about it as well.

Don't tell a pro that he's wrong because you don't like his answer - he might be, but it's not likely. Instead say 'I've observed that...' or 'My understanding of that is...', WHERE IS MY ERROR? - this will usually get you a better explanation than a challenge.

I did the above, and have had very little problem with any of the folks on that board.

Gooserider