water cooling temps...problem?

The alternative to direct air cooling

Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee

Post Reply
gotensan01
Posts: 16
Joined: Sat Aug 14, 2004 10:13 pm
Location: Chicago

water cooling temps...problem?

Post by gotensan01 » Thu Aug 26, 2004 8:25 pm

I just finished my water cooling setup but was very discouraged to discover that my idle and load temps are only about 2 C cooler than with my hsf. Here are pictures of my setup and a possible source of error. Do you guys have any suggestions as to why my water cooled temps blow? One reason I came up with is that it may not have enough air exhausting from the system.

1911user
Posts: 109
Joined: Sat Mar 13, 2004 12:08 pm
Location: Oklahoma

Post by 1911user » Thu Aug 26, 2004 9:16 pm

Do the air bubbles show up only when the pump is running? If so, I'd guess you have something not sealed well around the chipset block or the way that tube attaches to the barb.

Is the tubing pulling on the cpu block? The bend from cpu to chipset looks tight especially if that is 1/2ID 3/4OD clearflex tubing. Tubing pressure will affect mounting pressure and that affects temps. The tubing from the gpu block to the pump inlet looks like a tight bend also.

I also noticed no fan shrouds. They make a nice difference in sound level and rad performance. With only one of the fans pulling (not pushing) through a 1-2 inch shroud, you'd probably get about the same performance as both push/pull with no shroud. It might even be better since a shroud would pull air through a larger area of the rad core for better cooling and probably quieter also.

gotensan01
Posts: 16
Joined: Sat Aug 14, 2004 10:13 pm
Location: Chicago

Post by gotensan01 » Thu Aug 26, 2004 9:49 pm

1911user wrote:Do the air bubbles show up only when the pump is running? If so, I'd guess you have something not sealed well around the chipset block or the way that tube attaches to the barb.

Is the tubing pulling on the cpu block? The bend from cpu to chipset looks tight especially if that is 1/2ID 3/4OD clearflex tubing. Tubing pressure will affect mounting pressure and that affects temps. The tubing from the gpu block to the pump inlet looks like a tight bend also.

I also noticed no fan shrouds. They make a nice difference in sound level and rad performance. With only one of the fans pulling (not pushing) through a 1-2 inch shroud, you'd probably get about the same performance as both push/pull with no shroud. It might even be better since a shroud would pull air through a larger area of the rad core for better cooling and probably quieter also.
I fixed the air bubble problem. I just put a sponge into the res. I guess small amounts of air bubbles were being sucked into the system from the res. The sponge catches all the bubbles and releases them to the top slowly.

There are actually three tight bend areas. From the cpu to chipset, chipset to gpu, and gpu to pump. LOL, that's three in a row!!! I ordered coolsleeves but I think they sent me the wrong size because it would be extremely difficult to fit it onto my clearflex tubing. The bends are tight enough to flaten the tubing a little. However it doesn't create a concaved bending so I figured the same amount of volume is passing through. I don't think it is causing the blocks to make bad connections with the chips though. The blocks are on there pretty tight.

I completely agree about the fan shrouds. The reason why I didn't get any is because I didn't think I can fit two fan/shroud combos for a push/pull system. Now that I think about it though, I agree with you that having one fan/shroud combo pulling air across the rad would be equal if not better.

So you think flatening of tubes and no shrouds could account for such lousy temps?

1911user
Posts: 109
Joined: Sat Mar 13, 2004 12:08 pm
Location: Oklahoma

Post by 1911user » Fri Aug 27, 2004 4:25 am

Coolsleeves doesn't reduce the bending pressure, it just helps keep the tubing from collapsing. There are 2 sizes of coolsleeves; one for 1/2OD tubing (500 model) and the other 5/8OD tubing (625 model).

Having used 1/2ID 3/4OD clearflex, I know how much it doesn't like to bend tight. I use strips from large diameter PVC fittings to help hold the bend pressure. I also use soldered copper fittings for even tighter bends. The tubing could easily be affecting the block mounting. Tightening the block down even tighter is not a good solution IMO. The tubing pressure needs to be contained so the block isn't pulled or twisted. I looked at your system last night and mentally worked out a different tube routing that would eliminate the tight bends and could eliminate the reservoir if you wanted to use a T instead.

What I thought of is to turn the pump around and move it closer to the fan on the rad. Connect the pump intake to the gpu block barb pointing out of the case(can't tell if this would be tight or not). Connect the gpu block to the radiator barb inside the case. Connect the outside rad barb to cpu block barb closest to the back of the case. Connect the other cpu barb to the reservoir then the res to the chipset then chipset to the pump outlet.

I am replacing the 1/2ID clearflex tubing in my system with 3/8ID 1/2OD tubing. I'll strech it over 1/2 barbs or use swiftech quick connects (if available). The routing hassle just isn't worth it IMO and my blocks don't benefit much from higher flow rates. I may keep 1/2ID tubing for the pump inlet since that is the one place where the larger tubing may noticably have a benefit. Choking the pump inlet hurts pressure (and flow rates) noticably.

gotensan01
Posts: 16
Joined: Sat Aug 14, 2004 10:13 pm
Location: Chicago

Post by gotensan01 » Fri Aug 27, 2004 5:42 am

1911user wrote:Having used 1/2ID 3/4OD clearflex, I know how much it doesn't like to bend tight. I use strips from large diameter PVC fittings to help hold the bend pressure. I also use soldered copper fittings for even tighter bends. The tubing could easily be affecting the block mounting. Tightening the block down even tighter is not a good solution IMO. The tubing pressure needs to be contained so the block isn't pulled or twisted. I looked at your system last night and mentally worked out a different tube routing that would eliminate the tight bends and could eliminate the reservoir if you wanted to use a T instead.
I don't understand how you are using the pvc or copper to reduce flatening. Also by using a T, do you mean a T valve that routes one way or another? If so where can you find these?

1911user wrote:What I thought of is to turn the pump around and move it closer to the fan on the rad. Connect the pump intake to the gpu block barb pointing out of the case(can't tell if this would be tight or not). Connect the gpu block to the radiator barb inside the case. Connect the outside rad barb to cpu block barb closest to the back of the case. Connect the other cpu barb to the reservoir then the res to the chipset then chipset to the pump outlet.
If I did this though, wouldn't the the cpu be receiving the warmest water? Also, does the rad have designated intake, output barbs? Or does it work either way?
1911user wrote:I am replacing the 1/2ID clearflex tubing in my system with 3/8ID 1/2OD tubing. I'll strech it over 1/2 barbs or use swiftech quick connects (if available). The routing hassle just isn't worth it IMO and my blocks don't benefit much from higher flow rates. I may keep 1/2ID tubing for the pump inlet since that is the one place where the larger tubing may noticably have a benefit. Choking the pump inlet hurts pressure (and flow rates) noticably.
Are you sure you can stretch it over the 1/2" barbs?

1911user
Posts: 109
Joined: Sat Mar 13, 2004 12:08 pm
Location: Oklahoma

Post by 1911user » Fri Aug 27, 2004 7:22 am

gotensan01 wrote: I don't understand how you are using the pvc or copper to reduce flatening. Also by using a T, do you mean a T valve that routes one way or another? If so where can you find these?


If I did this though, wouldn't the the cpu be receiving the warmest water? Also, does the rad have designated intake, output barbs? Or does it work either way?

Are you sure you can stretch it over the 1/2" barbs?
There is a limit of about 2.5 inches for the bend radius of 1/2ID 3/4OD clearflex tubing. Any tighter than that causes the tubing to flaten. See this thread for more detail: http://www.ocforums.com/showthread.php?t=281094

There is a description of t-lines in the first 1/3 of this thread: http://forums.silentpcreview.com/viewto ... 68&start=0

Waterblock order does not matter. The temperature of the water varies less than 1 degree C as it goes around the loop.

It does not matter on a radiator which barb is used.

I haven't done it with 3/8ID tubing, but others have. By heating the tubes with near-boiling water, I've streched 1/2ID tubing over 5/8OD barbs and even a little larger. It did take some heat and pressure, but it went on. Clamps are still needed even when streching the tubing over larger barbs.

Another tubing option is Tygon (or silicone) tubing which bends easier and a little tighter, but it is expensive. Also, I'm not convinced 1/2ID tubing is the best choice for everyone. 3/8ID is much more practical and easy to use. Worst case has my temps going up a degree or 2, but it probably won't be noticable with the components I use. For pure performance, 1/2ID is a clear winner unless you go to 3/4ID or larger (don't laugh, people have done it).

gotensan01
Posts: 16
Joined: Sat Aug 14, 2004 10:13 pm
Location: Chicago

Post by gotensan01 » Fri Aug 27, 2004 7:37 am

Okay, now I see how the pvc and t line work. Unfortunately that pvc idea will restrict me from fitting the bends inside the case unless I remap all the connections. I will take your idea into account when remapping. So the T line opening just needs to be higher than any other component of the water cooling system? I think my res does the same job. I'm off to make a shroud right now. I haven't decided whether or not I'm going to remove the chipset block. Doing this will require I remove the mb from the tray and I don't feel like doing that right now. Unless of course you strongely recommend it?

1911user
Posts: 109
Joined: Sat Mar 13, 2004 12:08 pm
Location: Oklahoma

Post by 1911user » Fri Aug 27, 2004 8:23 am

gotensan01 wrote:Okay, now I see how the pvc and t line work. Unfortunately that pvc idea will restrict me from fitting the bends inside the case unless I remap all the connections. I will take your idea into account when remapping. So the T line opening just needs to be higher than any other component of the water cooling system? I think my res does the same job. I'm off to make a shroud right now. I haven't decided whether or not I'm going to remove the chipset block. Doing this will require I remove the mb from the tray and I don't feel like doing that right now. Unless of course you strongely recommend it?
What I'm posting are my observations and experience. They are suggestions at most.

A t-line would be a replacement IF you wanted to remove the res, they take up more case room and can make routing the tubing harder. If you're happy with it, I'd probably leave it. The T-line opening does not have to be the highest point in the system; the vacuum that would form if the water tried to escape holds it in place. If there was a way for air to enter at the top of the water system, then water would flow from a low T-line because no vacuum would form.

Chipset blocks are a personal thing. I use a passive Zalman NB cooler on my systems which have a decent amount of airflow (not silent). I don't think the NB benefits much from water cooling, but it doesn't hurt much either unless it causes routing difficulties with the tubing or is a serious flow restriction. For extreme FSB overclocks or a case with very minimal airflow, a NB block can be useful.

HammerSandwich
*Lifetime Patron*
Posts: 1288
Joined: Sat Oct 25, 2003 3:21 pm
Location: 15143, USA
Contact:

Post by HammerSandwich » Fri Aug 27, 2004 2:02 pm

Is that a Hydor L20? Such a pump might not give much flow with a 3-block system.

1911user
Posts: 109
Joined: Sat Mar 13, 2004 12:08 pm
Location: Oklahoma

Post by 1911user » Fri Aug 27, 2004 2:30 pm

He had a different pic up last night showing the back of the pump. I think it is an L30. He posted a thread on procooling and IMO has serious case airflow problems. The front bezel looks very restrictive and taking off the side case cover lowers cpu temps by 4C. The airflow needs to be cleaned up before loop improvements are going to show.

I think all of the blocks are low restriction so the pump and flow should be fine. The cpu block is a swiftech mcw-5000 (pin fin). The gpu block is most likely an older dangerden model that has an impingement wall in the middle but is an open design. I don't recognize the NB block, but it can't be that bad flow-wise.

This setup should work good once the airflow, fan shroud, and tubing pressure issues are resolved. JMHO

gotensan01
Posts: 16
Joined: Sat Aug 14, 2004 10:13 pm
Location: Chicago

Post by gotensan01 » Sat Aug 28, 2004 8:18 am

1911user, thanks for all your help first off. Yes it is a Hydor L30. Right now I think the biggest problem is the lack of fresh air that can be drawn from the outside of the case like 1911user has stated. However I have the wavemaster case and I am really at a loss as to how to get more holes for fresh air. I already hacked up the inside and I didn't really want to mess up the outside. Do any of you guys have suggestions as to maybe how to do it cleanly?

EDIT: It is a mcw5000-64 cpu block, danger den old style z chipset block, and old style danger den gpu block with 45 degree angled barbs.

I also update the pics on my website.

1911user
Posts: 109
Joined: Sat Mar 13, 2004 12:08 pm
Location: Oklahoma

Post by 1911user » Sat Aug 28, 2004 9:55 am

This is the part where hardcore watercoolers will want to shoot me:

You don't have to do anything to your setup. You can decide the current setup is good enough and change nothing. You also know that more fresh air to the rad will lower your temps, but the price is opening (cutting, drilling, etc.) your nice (expensive?) case. Is it worth it? Maybe you could compromise and open it up some without changing how it looks from the outside.

Aesthetics vs. Performance vs. Noise --------------A tricky balancing act (if all 3 are important)
Last edited by 1911user on Sat Aug 28, 2004 6:36 pm, edited 1 time in total.

gotensan01
Posts: 16
Joined: Sat Aug 14, 2004 10:13 pm
Location: Chicago

Post by gotensan01 » Sat Aug 28, 2004 11:54 am

1911user wrote:Here's the part where hardcore watercoolers will want to shoot me:

You don't have to do anything to your setup. You can decide the current setup is good enough and change nothing. You also know that more fresh air to the rad will lower your temps, but the price is opening (cutting, drilling, etc.) your nice (expensive?) case. Is it worth it? Maybe you could compromise and open it up some without changing how it looks from the outside.

Aesthetics vs. Performance vs. Noise --------------A tricky balancing act (if all 3 are important)
I really like this plan. I totally agree with you. I know I can increase performance by messing up my case but it isn't worth the time, money, or case. Or at least not right now. I am water cooling the cpu, nb, and gpu with kinda better temps and a lot less noise. All in all I guess I am satisfied.

Post Reply