advice needed for ambitious WC project (dual Opt, RAID, more)

Velda

Limp Gawd
Joined
Aug 10, 2001
Messages
410
I am currently planning out the build of my next system, and need some help working out the water cooling aspect of things. Im not really new to water cooling, but i am on cooling at this level. I probably wont get to building it till around winter, but i want to plan everything out now so i can mod my case the few weeks im home over the summer and fall.

The system is a dual 2.4GHz Opteron on a Tyan Thunder K8W. 2 gig of ram total, and four Seagate Cheetah 15K.3 73G Ultra320 in a RAID 10 with an adaptec 2120S.

Even with my very nice ingramMicro discount this system hits the $5,000 line.

The question is how (or perhaps IF) to water cool it all?

My main concern is cooling the RAID. Ive checked out Koolance's HDC1-A01 Hard Drive cooler that lets you cool two drives at once, but i really question how well this works with such fast and active drives. Does anyone have any experience with this? I have searched around with little information on it in actual practice.

The second problem with the Koolance drive cooler is that i can only find it in 1/4" size. While i can find 1/4" blocks for the chips, i really question cooling this entire system on such small ID tubing.

The way i see it i will need two CPU, one (perhaps two) Chipset, one GPU, and two HDD. Wouldnt it be a bad idea to try and cool all of that with 1/4" or does it make a difference?

Could i just size down from 1/2" at the HDDs? Sizing down would drastically impair the water flow wouldnt it?

I thought of another solution where i would run everything in 1/2" and then once it get to the HDDs just Y the 1/2" into two 1/4" and feed them both off of that, then take the return from the HDD and Y it back from the 1/4" to 1/2" before it hits the rad. In basic theory i should keep the same pressure and speed with a slight decrease from the resistance of the split.

One last question is monitoring the system. I know Innovatek sells a "Flow-O-Matic" which can report the speed of the flow in the system. How is this displayed? Looks like its just a 3 pin connector that could be reported like a fan's speed. Would it be a good idea to have one at the beginning and one at the end of the system? Another downside is they seem to only sell it in 1/4" sizes.

Id also like to add temperature probes at the beginning and end of the system to monitor water temp. I know ive seen this done before but i cant remember any source of information on it.

This all seems rather daunting, would it even be worth it to try and water cool such a complex rig?

thanks in advance for any assistance.
 

Tigerbiten

[H]ard|DCer of the Month - February 2007/January 2
Joined
Sep 24, 2002
Messages
5,028
Your new setup sounds like an upgraded version of mine.
I only cool my cpu's ( 2x2400+'s ) and hard drives ( 5x36 Gig 10k scsi ) with water.
The rest is cooled by air.

The question is how portable do you want it.
Mine is not.
Two Lian-Li PC70 cases. one each side of my Sony FW900.
One side has pump, Rad, etc.
Other side has everything else.
I used large bore ( 1.5" ) to join the two halfs togther and in first case.
Then manifold block to split it to two 1/2" pipes.
These run to my cpu's.
Used a design similar to the Whitewater block, one in two out.
Ran the four tubes to my homemade HDD coolers.
Then back to a manifold block to join together before going back to the pump.

If you are going to split a single 1/2" pipe then its 2x3/8" or 4x1/4".
You have to work by area not diameter.

Its very [H]ard not to buy a new system like that.

Luck........ :D
 

kronchev

[H]F Junkie
Joined
Feb 23, 2001
Messages
12,051
if you bought all of that, then used a koolance to try to cool it, i would have no choice but to slash my wrists; i would lose all hope in humanity.


you can make harddrive coolers, just take some copper pipe and have it snake around on top up and between the drives. if you want to get [h]ardcore you can get copper stock and carve out blocks but that requires more time, money, and equipment than just attaching copper pipes together

please, please please dont get koolance for such a great setup.

1/4" vs 1/2" is a whole world of cooling difference. stay at 1/2" the whole way or youll KILL your flow.
 

Velda

Limp Gawd
Joined
Aug 10, 2001
Messages
410
Hah, i cant believe i didnt notice the 1/2 to 1/4 conversion. I spend so much time working in a 3D environment that i really should get that quicker.

Mobility isnt an issue at all. This thing will likely sit in my apartment for the next four years and never see sunlight once i settle down.

That rig sounds very [H]ard. From the way you describe it every block gets fresh water from the pump correct? That would set me at ease a little more, but i bet that is one crazy tubing job.

I really dont know much about one brand over another, i just read the reviews and try to figure it out on my own. Koolance was just the only company selling a HDD cooler that would cool two at a time.

Have any examples of home made HDD coolers? I get the gist of just making your own pipes, but i dont see how they could achieve the same effect on a HDD. Most of the time i have seen HDD coolers they have been on the circuit board. This seems slightly counterintuitive, but i guess it makes sense that you are keeping the chips cool and not just the metal case of the drive. If i really can cool them all simply by making a snake out of some copper pipe that would rock, i just dont want to trust $1,500 in HDDs and all my data to it until ive read some figures and heard some real world experience.
 

AggieMEEN

Gawd
Joined
Mar 2, 2003
Messages
756
kronchev said:
if you bought all of that, then used a koolance to try to cool it, i would have no choice but to slash my wrists; i would lose all hope in humanity.
....
please, please please dont get koolance for such a great setup.
...

While I don't necessarily share Kronchev's rather passionate dislike for the Koolance product line, I totally agree with him in that you do not want to use Koolance on your system.

Perhaps what you should be looking for are passive HDD coolers. I don't know how much space you have to work with, or how much airflow you've designed into your system, but I would imagine that four HDD coolers might work. Browsing through Xoxide, I found this unit, which, although it is not 1/2", is bigger than 1/4".
 

kronchev

[H]F Junkie
Joined
Feb 23, 2001
Messages
12,051
Velda said:
Hah, i cant believe i didnt notice the 1/2 to 1/4 conversion. I spend so much time working in a 3D environment that i really should get that quicker.

Mobility isnt an issue at all. This thing will likely sit in my apartment for the next four years and never see sunlight once i settle down.

That rig sounds very [H]ard. From the way you describe it every block gets fresh water from the pump correct? That would set me at ease a little more, but i bet that is one crazy tubing job.

I really dont know much about one brand over another, i just read the reviews and try to figure it out on my own. Koolance was just the only company selling a HDD cooler that would cool two at a time.

Have any examples of home made HDD coolers? I get the gist of just making your own pipes, but i dont see how they could achieve the same effect on a HDD. Most of the time i have seen HDD coolers they have been on the circuit board. This seems slightly counterintuitive, but i guess it makes sense that you are keeping the chips cool and not just the metal case of the drive. If i really can cool them all simply by making a snake out of some copper pipe that would rock, i just dont want to trust $1,500 in HDDs and all my data to it until ive read some figures and heard some real world experience.

well youll want pipes all around it. on top of the drive casing, since that is the major source of radiating heat from the motor. on the bottom, hopefully touching the hottest few chips with some thermal tape as insulator, if thats needed. the thing with SCSI drives is that the motors get hot hot hot, and that can be the cause of failure. usually, the chips getting hot isnt a problem.

heres a few links from the guys who REALLY know cooling:

http://forums.procooling.com/vbb/showthread.php?t=9228

http://www.digital-explosion.co.uk/index.php?articleID=31
 

Velda

Limp Gawd
Joined
Aug 10, 2001
Messages
410
i have seen those hdd coolers that just pipe some water across the sides... doesnt really seem like they would do much.

Case realestate is about as much of a non-issue as the price of this project is, within rational need. The desk im building can accommodate just about anything. I have been considering the Lian-Li V2000B or perhaps a cube. Id like the possibility of racking the entire thing in a 4U to join the others but im not sure how probable that is with so much water cooling.

The guide on building your own HDD water cooler is very educational and within my scope of possibilities. I guess i can start drafting that up and get it built before i start putting the system together. I think il just adapt the same design and make two of them.

Tigerbiten's note about his system running in parallel now has me thinking about trying to do the same. With the possibility of as many as 7 blocks to cool it seems like a better idea, but an even more daunting challenge. especially to monitor. couldnt that cause current variations, or is that even an issue? It just seems like a bad idea to reuse the same water through so many blocks.

As far as monitoring goes does anyone know of a flow meter in the 1/2" variety? I have searched for quite a while with no luck. Temperature shouldnt be an issue, mod some probes into the tubes, im just worried about keeping tabs on water flow.

thanks again for all the assistance.
 

AggieMEEN

Gawd
Joined
Mar 2, 2003
Messages
756
Velda said:
....As far as monitoring goes does anyone know of a flow meter in the 1/2" variety? I have searched for quite a while with no luck. Temperature shouldnt be an issue, mod some probes into the tubes, im just worried about keeping tabs on water flow.
...


Flow meters come in different varieties, but the majority of these units are pretty expensive. You could probably obtain one from McMaster-Carr. Is it your intention to be able to read precisely the rate of flow, or just confirm that water is flowing? In the case of the former, you'll be shelling out some money. In the case of the latter, you could design a reservoir such that the incoming flow creates visible turbulence on the water's surface.

I use the turbulence method. It isn't precise, but it does work, and does not lead to the formation of bubbles.
 
Top