It is currently Tue Sep 19, 2017 11:32 pm

All times are UTC - 8 hours




Post new topic Reply to topic  [ 145 posts ]  Go to page 1, 2, 3, 4, 5  Next
Author Message
 Post subject: Water Cooling Circuits Analysis (A liquid cooling work log)
PostPosted: Fri Aug 20, 2004 7:23 pm 
Offline
SPCR Reviewer

Joined: Thu Dec 11, 2003 9:53 pm
Posts: 2696
Location: Scarsdale, NY
As some of you know, I will be conducting a comparison test of three different water cooling cirtcuit arrangements.:

1: Single pump, one CPU block, one GPU block, one northbridge block, one 120mm heatercore and a dual-bay reservoir

2: Same setup, but two pumps, in serial

3: Dual pumps; both pumps pull water from the reservoir via a y-splitter, but one pump pushes through the CPU block and the second pump pushes through the northbridge block and GPU block, then the two paths combine through a y-splitter before entering the heatercore

Now, as some of you probably also have read by now, I had some serious gunk and leakage woes with my initial setup. With some help from a few forum members, but in particular Hammer Sandwich, GlassMan and senior editor Russ Kinder, I've been able to make some very good improvements to the system, and now have it prepped up; the reservoir is still coming in, so the system has not yet been filled and run yet, but once it comes in, I should be ready to test the first setup. These changes were made:

1] moved the entire heatercore and fan assembly into the case and mounted it out front; the PCI-slot passthrough was leaking and it was virtually impossible to stop it

2] bleeding the heatercore completely in the original location was virtually impossible due to the inlet and outlet being at the bottom; they are now at the top, to promote proper bleeding

3] the fan was pushing through; now it pulls through the radiator and blows it out the front of the case

4] MOSFETs on the mainboard were exceeding 100C due to there being no exhaust fan out back and being completely unsinked; Microcool MOSFET ChipSinks have been installed

5] Microcool SouthSink installed for good measure

6] arrangement of flow through the blocks has been redone in order to eliminate 180-degree U-bends; new arrangement only has 90-degree L-bends

7] shorter tube lengths

8] all anti-kink coil has been eliminated due to its detrimental affects on flow through tubing; to fix flattened tubing in bends, 1/2"ID tubing is clamped loosely around the 1/2"OD tubing of the system where necessary (only one spot at the current time, from the northbridge block to the GPU block)

9] the exhaust fan has been reinstalled, now that the heatercore assembly does not prevent it; this should help a great deal with the MOSFET situation

10] fill n' bleed is being swapped out for a reservoir; the filling and bleeding issues were driving me insane

11] Fluid XP is being replaced by distilled water with Swiftech HydrX additive

12] external 3.5" drive tray was snipped out; the removable cage doesn't need it, since it secures itself to the bottom of the half-height bays, anyway; this provides that last bit of extra room I needed for the heatercore

13] Swiftech RadBox had the center of the mounting plate cut away with my rotary tool to dramatically free up airflow, and the remaining edges were deburred using a sanding disc

14] all old tubing was thrown out due to contamination and replaced using fresh Tygon

15] pump's 1/2" fittings were replaced using 3/8" fittings from McMaster-Carr, as the 1/2"-3/8" adapters were leaking like a bullet wound to the heart; thread now sealed with teflon tape

16] all blocks flushed with hydrogen peroxide and then blown out with compressed air

17] all tubing carefully redone, proper clamping principles applied, proper clamps utilized, tubing geometry carefully arranged, tube lengths more accurately measured, flow optimized, difficult spots double-clamped

Because I have three fans exhausting (front/heatercore fan, rear/exhaust and PSU), I obviously will have to have some sort of intake; the 92mm AcoustiFan mounted to the case side panel window will handle this responsibilty. Its nearly centralized location compared to all the exhaust fan will definitely help in that respect.

Here's how it looks so far (I wanted to make this post more dial-up friendly, so I used very heavy compression and didn't apply my patented soft-edges; sorry!):

Per usual, click on the small image here for a much larger version with superior definition.

[URL=HTTP://WWW.NgTechnik.Com/Scenario0/Whole-Rig-Big.jpg]Image
Here's the system as a whole as it stands right now.[/URL] Reservoir isn't here yet, so its tubing has not yet been run.

[URL=HTTP://WWW.NgTechnik.Com/Scenario0/BayRes-Space-Big.jpg]Image
Here's where the Bay-Res will go.[/URL] I may have to bend up some of the tabs in there (the ones designed to support the drives from beneath), because it's a full-height model.

[URL=HTTP://WWW.NgTechnik.Com/Scenario0/Pump-Big.jpg]Image
The pump, much more handsome with its new 3/8" fittings in black, installed with teflon tape for an even more secure seal.[/URL] The 1/2"ID clamps have been precisely gapped to ensure the ultimate tube seal. That dirty stuff around the inlet is leftover from removing the stupid sticker that C-Systems places there from the factory, with a message to visit the web site for the user's guide; talk about one way to make life harder for the installer! :evil: I have not yet tried to remove this stuff because I do not wish to risk scratching the finish on the pump.

[URL=HTTP://WWW.NgTechnik.Com/Scenario0/Scenario-0-Big.jpg]Image
Here's how the tubing is currently arranged, to test the baseline arrangement.[/URL] Flow goes reservoir->pump->northbridge->GPU->CPU->heatercore->reservoir.

[URL=HTTP://WWW.NgTechnik.Com/Scenario0/CPU-Block-Big.jpg]Image
Swiftech MCW-6000A CPU block, optimized for low-flow, yet still extremely competent at high flows; a solid all around block, and the socket mounting system is also extremely handy.[/URL] Those two standoffs belong to the SP-97 that was once there. The block mounts fine, even with those standoffs in the way. Dual 1/2"ID snap clamps on each fitting to ensure zero leakage; there was some leakage here with the original worm-drive clamps that came loose over time; while single snap clamps eventually solved that problem, I decided that dual snap clamps this time around would be absolutely foolproof.

[URL=HTTP://WWW.NgTechnik.Com/Scenario0/Northbridge-Block-Big.jpg]Image
Danger Den Maze4 chipset block--1/2"ID snap clamps just as in the original arrangement and spaced properly to ensure absolutely perfect seal.[/URL] There was no leak here, originally, anyway.

[URL=HTTP://WWW.NgTechnik.Com/Scenario0/GPU-Block-Big.jpg]Image
Danger Den NV-68 block for nVIDIA 6800-series cards; cools the memory as well as the core.[/URL] 1/2"ID snap clamps were utilized here, and gapping carefully adjusted to ensure zero leakage; there was never a leak here, either, with the original setup.

[URL=HTTP://WWW.NgTechnik.Com/Scenario0/Smooth-Kink-Fix-Big.jpg]Image
Here's where I surrounded the tight bend in the 1/2"OD tubing with a piece of 1/2"ID tubing and then lightly clamped, to fix the kink without introducing ripples into the tube walls, which the anti-kink coiling would have done.[/URL] This is similar to what Russ did in his new Reserator-based K7 rig, only with two different diameter tubes.

[URL=HTTP://WWW.NgTechnik.Com/Scenario0/Heatercore-Big.jpg]Image
Here's the heatercore. Dual 1/2"ID snap clamps were utilized on each fitting, even though the original arrangement didn't leak, even with worm-drive clamps; I want to be extra secure, nonetheless.[/URL] This is set up so that the fan pulls air through the radiator and pushed it out the front of the case. I may or may not seal up the sides of the RadBox. The RadBox normally has a filled in mountplate with many slats to fit an almost infinite possible number of case installs; since I only need the outermost slats, I used a rotary to cut out the center and deburr the remaining edges, to promote proper airflow out of the case.

[URL=HTTP://WWW.NgTechnik.Com/Scenario0/MOSFETsinks-Big.jpg]Image
Microcool MOSFET ChipSinks.[/URL] The last I ran the system, the MOSFETs were exceeding 100C under full CPU load; I do not wish to take the risk of MOSFETs holding back my overclock.

[URL=HTTP://WWW.NgTechnik.Com/Scenario0/Southsink-Big.jpg]Image
Microcool SouthSink, just for good measure. Some K7 overclockers have reported the southbridge as holding their overclocks back.[/URL] Hey; even if it doesn't do anything to help, at least it looks nice!

-Ed

EDIT: Added pump images and high-resolution linked images.

EDIT2: Fixed an error in formatting.

EDIT3: A typo I found.

EDIT4: Another typo and some really ugly grammatical errors.

EDIT5: Corrected the spelling of the Microcool sink names.

EDIT6: Updated thread title to bring in those interested in CSP-750's effectiveness and un-fuglied my image description links.

EDIT7: Changed the title of the thread slightly, now that CSP-750s are no longer involved (see page five)...

_________________
Contributing Writer, SPCR
NgTechnik

Want something reviewed? Help us get samples!
Donate for Patron or Friend Status!


Last edited by Edward Ng on Tue Dec 21, 2004 4:59 pm, edited 7 times in total.

Top
 Profile  
 
 Post subject:
PostPosted: Fri Aug 20, 2004 8:40 pm 
Offline

Joined: Sat Mar 13, 2004 12:08 pm
Posts: 109
Location: Oklahoma
Your setup looks much cleaner and should perform better. Nice!

I've used 1/2ID 3/4OD clearflex tubing for the 2 water setups I've done previously and it's a royal pain for bends. I'm changing my setup around and switching to 3/8ID 1/2OD tubing soon. I was planing to order the coolsleeve material.

Did the coolsleeve actually cause flow turbulence or was it just a potential problem?

Did you consider using a T-line? You've got the perfect location at the outside radiator inlet. A T-line would also lower the amount of tubing in the system. This ties into something I noticed.

I think the tubing from the rad to the cpu block may kink once it has the weight of water added. That tube looks a little tight and could easily cause the block to have less contact pressure with the cpu die. One thing I found important is to have minimal tubing pressure pulling on the water blocks. You'd be amazed how just a little pulling pressure can affect contact between the block and chip which also affects temperatures.

_________________
rig #1 Watercooled XP2500 @ 2.5GHz, Nforce2-400, watercooled 9700 Pro (370/330), Seagate 120GB, Antec SLK3700 & PSU
rig #2 Aeroflow cooled XP2500 @ 2.3GHz, Nforce2 IGP, 9700 Pro (405/370), Seagate 80Mb, Enlight mid-tower & PSU
6.5 GHz Folding


Top
 Profile  
 
 Post subject:
PostPosted: Fri Aug 20, 2004 8:49 pm 
Offline

Joined: Fri Aug 01, 2003 10:45 pm
Posts: 587
Location: North Billerica, MA, USA
Nice job overall, but a couple of points...

1. Opinions vary - I wouldn't trust those nylon snap clamps as far as I could throw them, and would feel far safer with stainless worm-drive clamps. Important thing though is that you don't leak.

2. Fan mounting is VERY BAD! Sucking air through the rad is good, that usually works much better than trying to blow through it. OTOH, you have the fan right up against the rad fins, this greatly reduces your cooling power because of no airflow through the fan dead spot and the close coupling between the fan and the rad is going to increase your noise. (plus bad airflow making you run the fan faster...) I've also been under the impression that standard SPCR practice is to avoid air blowing out the front of cases because that increases noise.

Recomendation - move the rad up against the case front, and put that shroud (with all air passages except through the rad fins sealed) on the back side of it and then put the fan so that it is sucking air in through the case front, through the rad and blowing into the case.

This will warm the case interior a small amount (2-3*C) but reduce your noise by getting the fan into the inside of the case with the rad between you and it. The shroud will reduce the fan beat noise from the blade / rad distance and improve your cooling by enabling airflow through the 'dead spot'. The improved cooling may also allow you to slow the fan. It will also balance your airflow better so that you may be able to slow down (or eliminate) the other fans.

Gooserider

_________________
Building Dual Athlon MP system, Tyan mobo, U160 drives, Server Cube case, Linux ONLY, lots of other goodies. Will water cool, attempting to make as silent as possible.


Top
 Profile  
 
 Post subject:
PostPosted: Fri Aug 20, 2004 8:50 pm 
Offline
SPCR Reviewer

Joined: Thu Dec 11, 2003 9:53 pm
Posts: 2696
Location: Scarsdale, NY
Thanks!

Several people noted to me that the coolsleeve reduces performance, particularly if you use it in vast amounts like I did; I had no time to compare with and without, but I'll take the word of majority for what it's worth in that case...

I'm not sure exactly what a T-line would be used for; please elaborate.

If it should kink, I can apply the same fix I did for the connection between those two blocks; I take 1/2"ID tubing, cut it to sufficient length to cover the kink, slice it open, wrap it around the 1/2"OD tubing and lightly snap clamp it down. The snap clamps pull far enough apart to allow this, and I tried to slide the 1/2"ID over the 1/2"OD but there's way too much friction, so even if the tubing weren't fitted yet, I'd have to slice the 1/2"ID no matter what to get it onto the 1/2"OD.

I did check the amount of pressure on that line and I find that it's fine; it's hardly tugging. If I made it any longer, it would've been pushing on the block towards the back side; if I made it shorter, it would pull as you mentioned.

-Ed

_________________
Contributing Writer, SPCR
NgTechnik

Want something reviewed? Help us get samples!
Donate for Patron or Friend Status!


Top
 Profile  
 
 Post subject:
PostPosted: Fri Aug 20, 2004 9:13 pm 
Offline
SPCR Reviewer

Joined: Thu Dec 11, 2003 9:53 pm
Posts: 2696
Location: Scarsdale, NY
Crossposted, Gooserider...

1. All my leaks that sprang up from my last install were at worm drive clamp points; they come loose over time, even after I retightened them. The only remedy were the snap clamps. This is from my personal experience so far.

EDIT: Btw the snap clamps are useless if you just try to close them with your bare hands; they must be clicked to the last notch using pliers to work right.

2. The fan is not directly against the fins; there is a 1/2" deep shroud between the fan and the heatercore. The gap is wide enough that there is no blade proximity noise generated, as well. The noise is bound to be there; my noise standard is lower for this build because it need only be quiet enough that I cannot hear it over the noise of gaming with my Sennheiser HD497 headphones. Maximum overclock is the prime focus point once the noise level is at that certain level, and that it is.

The system will be overclocked extremely heavily; the CPU puts out over 140 watts of heat and the 6800 is known to produce at least 110watts of heat as well (GPU and memory combined); when I first had this system running, the back wall behind the heatercore got quite hot, as does the heatercore itself at the inlet; I am not going to blow such hot air back into the system.

This is not water cooled to near dead silence; this is water cooled for acceptable acoustics while attaining an extremely high overclock for gaming performance while maintaining portability (thus the seemingly poor case choice, acoustics-wise; this case is lightweight and compact for portability) without resorting to compressor cooling (loud plus risk of condensation) or peltiers (risk of condensation as well).

Slowing down or eliminating the other fans is not an option. I thought the 100C MOSFET temps would have been enough of a sign that this system is running harder than normal, but let me outline clearly how this system will be operating when all the messing around is done with:

Barton at >2560MHz with 2.1volts driving it
two 512MB pieces of OCZ EL PC3200 at >DDR446 and 11-3-2 2.5 timings with minimum 2.8volts driving it
6800GT at >430MHz core and >1200MHz memory speed (this exceeds stock speeds for Ultra, of 425/1100)

Those numbers are based on the best numbers I got when the water system was previously running in its highly shabby state. I expect better numbers once this thing works right.

I'm doing the circuit analysis at stock speeds, so don't worry, it won't be a bunch of worthless crap numbers that you guys will be looking at but will have no use for and would bare no relevance whatsoever to your own configurations, but at the same time I must and will test the system in a noisy configuration. I can easily turn down the fans to a low-noise configuration for the stock speed testing if you guys want.

Please don't take my response in a negative way; I highly value your input. The issue is there are certain variables that you probably did not know of that have to be dealt with, and I have outlined them here. As I said, I will try my best to accomodate the needs of your guys, but I am doing this completely for my own benefit, and sharing what I learn with you guys as a fringe benefit. This will be one of my in-forum, unofficial write-ups/articles, as far as I can tell, which is why I'm not going all-out on silence. I still believe that the information that will be presented bares relevance to the setups that everyone else has and/or is considering, now that C-Systems makes dual pumps so easy to do.

The focus here is the thermal performance difference between different circuit arrangements, not how quiet I can/am willing to run my system with these pumps. Pro Cooling's article on the pumps should be more than enough proof that these pumps can perform. Unfortunately, I do not have proper acoustics measuring equipment to give you guys solid numbers on the acoustics of these pumps, and I also have not had any experience with any of the other pumps on the market, so unfortunately I cannot make any subjective comparisons to other pumps that you may use as a reference. The only thing that's definite is that I am highly satisfied with the acoustics of these pumps, and if I were to build a machine for silence I would not hesitate, for a minute, to choose these pumps to do the job. They can and will do a great job of pushing water quietly when softmounted inside a dampened case, and it doesn't take an in-depth analysis for me to tell you that much.

It does take an in-depth analysis to compare the effects of different circuit arrangements, alternating only the placement of pumps and blocks within the system, leaving the reservoir and heatercore arrangements static as a control measure.

Cheers! :)

-Ed

EDIT: Shroud is 1/2" deep, not 3/8".

_________________
Contributing Writer, SPCR
NgTechnik

Want something reviewed? Help us get samples!
Donate for Patron or Friend Status!


Last edited by Edward Ng on Sat Aug 21, 2004 6:26 pm, edited 2 times in total.

Top
 Profile  
 
 Post subject:
PostPosted: Fri Aug 20, 2004 9:27 pm 
Offline

Joined: Sat Mar 13, 2004 12:08 pm
Posts: 109
Location: Oklahoma
I missed the fan against the rad. That isn't good for 2 reasons. The first is you get more air turbulence hence noise. The second is not using the dead spot in the center of the fan and the area not covered by the fan. Having the fan an inch or so away from the rad surface solves both problems. People use a shroud to hold the fan away from the surface and seal any air leaks so all air through the fan goes through the radiator. In many instances, the radiator surface is larger than the fan and the shroud encloses some or all of the larger area to cool the water more effectively. Shrouds are quieter and efficient. The shroud in this pic is a tubberware dish but it worked well. $11 buys a commercial model that looks better.
Image

A T-line is like a small reservoir and air trap combined. It is also where fluid is added to the system. Draining the system can be a challenge. In the 2nd pic, you can see a white cap threaded onto a grey fitting which goes down to a white barbed, upside down "T" fitting. That is the T-line; cheap and space efficient. The picture doesn't show clamps on the T-fitting, but they were added later. I also only use metal, worm-drive clamps. The line needs to be capped to prevent evaporation. The first pic shows a copper T-fitting (good clamps are an absolute must with smooth copper fittings!) while the second shows a plastic T.

Image

_________________
rig #1 Watercooled XP2500 @ 2.5GHz, Nforce2-400, watercooled 9700 Pro (370/330), Seagate 120GB, Antec SLK3700 & PSU
rig #2 Aeroflow cooled XP2500 @ 2.3GHz, Nforce2 IGP, 9700 Pro (405/370), Seagate 80Mb, Enlight mid-tower & PSU
6.5 GHz Folding


Top
 Profile  
 
 Post subject:
PostPosted: Fri Aug 20, 2004 9:31 pm 
Offline

Joined: Fri Aug 01, 2003 10:45 pm
Posts: 587
Location: North Billerica, MA, USA
T-lines act as bubble traps and fill points, essentially the same job that a res does so normally you would have one or the other, but not both. IMHO a regular res does the job better, I'm not so sure about a bay-res. Your Swifty fill + bleed kit was essentially a fancy version of a T-line, FWIW.

Glad to see you making sure the lines aren't putting pressure on the block, but as 1911user said, you should recheck it when the system is full to make sure that this doesn't change.

One last point, is that you should make sure the line between the res and the pump inlet imposes as little flow restriction as possible the centrifugal pumps that we use for WC stuff react very badly to negative pressure on the suction side, so you need to make sure they have NO resistance or even slight positive pressure at the intake. This increases flow efficiency and reduces noise due to less cavitation.

Gooserider

_________________
Building Dual Athlon MP system, Tyan mobo, U160 drives, Server Cube case, Linux ONLY, lots of other goodies. Will water cool, attempting to make as silent as possible.


Top
 Profile  
 
 Post subject:
PostPosted: Fri Aug 20, 2004 9:36 pm 
Offline
SPCR Reviewer

Joined: Thu Dec 11, 2003 9:53 pm
Posts: 2696
Location: Scarsdale, NY
So my guess is what you guys really mean is that the shroud I'm using isn't deep enough? In that case a simple remedy is for me to find a deeper one; one that's as deep as the fan center hub is wide in diameter.

I'll let you guys know when I find one that's deeper than what I have. I'm definitely not up to task to build my own from scratch.

As I said, the snap clamps have worked flawlessly for me once I figured out that they must be sealed using pliers. The worm drive clamps always came loose on my sooner or later and the only solution was plier-shut snap clamps.

Gooserider I will definitely recheck all lines once it's filled; that was another mistake I made before that I will not repeat.

I'll see if I can add a tiny bit of length to the line from the res to the pump once I have the res, to relax the curve so as to further ease the intake of the pump; this was the primary concern I had when I decided to switch from the fill & bleed to the reservoir, as I knew that the dead body of water in the res would only be counteracted by making optimal use of gravity in the line from the res to the pump.

-Ed

EDIT: I'd like to add that I distrust the full bleeding ability of this heatercore, so I will no longer rely on fill n' bleed or t-lines to do the job on my system. I will have to use a reservoir so long as I'm using this heatercore that loves to hog air in the system to itself. Thanks for the t-line suggestion, however, as technically it should provide easier flow to the pump; that I do agree with.

_________________
Contributing Writer, SPCR
NgTechnik

Want something reviewed? Help us get samples!
Donate for Patron or Friend Status!


Top
 Profile  
 
 Post subject:
PostPosted: Fri Aug 20, 2004 9:52 pm 
Offline
SPCR Reviewer

Joined: Thu Dec 11, 2003 9:53 pm
Posts: 2696
Location: Scarsdale, NY
Okay, I've found this:

http://www.svc.com/cw-mini-coolshroud-black.html

But the problem is that it appears to only have one set of holes, on one side. I don't understand exactly how the gaskets would be used. Furthermore, I cannot picture how I can use this in my current setup the way I have the current shroud, because the current shrouds has one set of holes on both sides, one side for the bolts through the heatercore and the other side for attachment of the fan.

I rely on both sets of holes on my shroud in order for the entire assembly to support the weight of the heatercore. Without both sets of holes on the shroud, it can only sustain the weight of the fan alone, which means the only arrangement would be to blow through the radiator to the front of the case, rather than pulling air through the radiator and then blowing it out, as it is currently set up. Again, there's no way I'm going to pull the air in through the heatercore and then blow it into my case; that will raise my ambient temperature like crazy due to the amount of heat I'm goign to be working with.

Anyone have any other ideas or explainations as to how I could implement this shroud such that it can support the weight of the heatercore?

In the meantime, I will continue seeking other shrouds...

-Ed

EDIT: Alrighty, I've found the Mini-Coolshroud's big bro:

http://www.cooltechnica.com/Merchant2/m ... Coolshroud

From the looks of it, I can probably use it the way I am using my current shroud. The primary worry I have is whether or not it's so big that I cannot utilize my through-heatercore bolts that I have now to attach this new shroud. I suppose for that price, it's worth a try.

EDIT 2: Further research indicates that this probably won't allow me to use bolts on both sides; the heatercore I'm using is only 5.5" wide and that full size CoolShroud is 6" wide with what appear to be 1/4" lips, which places it such that it allows boltless fan mounting, but doesn't allow bolt fan mounting...

DTek carries a 1 3/8" deep shroud that, by the measurements, should fit. I do not know why D-Tek Customs indicates that it does not fit, so I better e-mail them and ask what's going on; this is getting quite annoying...

_________________
Contributing Writer, SPCR
NgTechnik

Want something reviewed? Help us get samples!
Donate for Patron or Friend Status!


Top
 Profile  
 
 Post subject:
PostPosted: Fri Aug 20, 2004 10:32 pm 
Offline

Joined: Sat Mar 13, 2004 12:08 pm
Posts: 109
Location: Oklahoma
Please don't shoot the messenger!

After looking at the heat load you envision, I think you will need more radiator capacity. A heatercore tall enough for twin 120's would be a good start, but I don't think it'll fit in the lower intake part of the case even with all bays but the 5 1/4 removed. The option that makes the most sense to me is mount another single 120mm radiator on the back except maybe a chevette size instead of the pro-120.

Anyway, the rear 120mm case fan could be moved outside the case and mounted over the exhaust hole. That would still pull air over the MOSFETs. Deliverying 2.1 volts means you NEED good case airflow to cool the motherboard which is what mainly cools the MOSFETs and the rest of the power conversion circuit. A shroud and second radiator could be attached to the back of the 120mm fan. That would leave room for the PSU exhaust; it's going to be working overtime to supply power for the monster gaming rig. I'd put the rad inlet tubes at the bottom as before because it would make the external tubing shorter and better protected.

About the first shroud kit you saw, the bolts go through the rad, the shroud, and to the fan mounting ears. The wide part of the shroud is pressed against the rad surface and doesn't have seperate bolts. The neoprene gasket has a cutout that only leaves a border about 1/4 wide. That border piece goes between the rad surface and shroud to seal any air leaks. The pics don't show it, but my current shroud uses weatherstripping for the same sealing effect.

The second shroud is for a chevette heatercore (6x6 rad area) and won't work for your smaller pro-120. I guess the pro-120 is narrower as well as shorter than a chevette core.

_________________
rig #1 Watercooled XP2500 @ 2.5GHz, Nforce2-400, watercooled 9700 Pro (370/330), Seagate 120GB, Antec SLK3700 & PSU
rig #2 Aeroflow cooled XP2500 @ 2.3GHz, Nforce2 IGP, 9700 Pro (405/370), Seagate 80Mb, Enlight mid-tower & PSU
6.5 GHz Folding


Top
 Profile  
 
 Post subject:
PostPosted: Fri Aug 20, 2004 10:45 pm 
Offline
SPCR Reviewer

Joined: Thu Dec 11, 2003 9:53 pm
Posts: 2696
Location: Scarsdale, NY
I cannot do the arrangement you said out back; the same issue of the holes is back there as well. I can't directly mount anything using the standard 120mm fan hole setup. The only way to attach something is either a fan via the included plastic clips or anything using 120mm fan hole config once a Swiftech RadBox is installed...

I realize I'm giving my poor heatercore excessive work to do; I was considering switching to a Thermochill 120.2 heatercore back when I had the thing set up hanging off out back, but I've come to realize that fittings on top would work far better than fittings on bottom (for any heatercore/radiator), and if I were to install a radiator that way out back, it would add a vast amount of additional tubing to the system, which is extremely bad...

Which is why I'm going to just sacrifice acoustics for better flow by running my fan harder with the core I have now. I know there's always diminishing returns this way compared to a bigger core, but this really is easier.

Thanks for the suggestion, I think I'll leave larger heatercores to future setups...

-Ed

PS Since when have I been known to shoot the messenger (not counting XG2004)? :wink:

_________________
Contributing Writer, SPCR
NgTechnik

Want something reviewed? Help us get samples!
Donate for Patron or Friend Status!


Top
 Profile  
 
 Post subject:
PostPosted: Fri Aug 20, 2004 11:02 pm 
Offline

Joined: Sat Mar 13, 2004 12:08 pm
Posts: 109
Location: Oklahoma
The solution for the 120mm rear fan mount is to drill 4 holes. Isn't that standard practice on SPCR? especially for slk3700 cases.

For bleeding purposes, the limiting factor for the rear mounted rad is the tubes in the slot cover. You have to get the entire rad and tubing below the pass-thru for effective bleeding. With all of the tubing and blocks everywhere, completely bleeding the system will be a real challenge. My last system took 2-3 months to completely bleed. I had to keep adding an inch or so of coolant to the T-line every two weeks for over 2 months. At first I thought it was a leak, but could never find one. It hasn't needed any topping off in several months. I think it was air trapped in the cpu block that slowly dissolved into the water then was trapped in the T.

I wonder if there is a way to attach a vacuum pump somehow to speed the bleeding process especially on multi-block, multi-rad systems?

_________________
rig #1 Watercooled XP2500 @ 2.5GHz, Nforce2-400, watercooled 9700 Pro (370/330), Seagate 120GB, Antec SLK3700 & PSU
rig #2 Aeroflow cooled XP2500 @ 2.3GHz, Nforce2 IGP, 9700 Pro (405/370), Seagate 80Mb, Enlight mid-tower & PSU
6.5 GHz Folding


Top
 Profile  
 
 Post subject:
PostPosted: Fri Aug 20, 2004 11:12 pm 
Offline
SPCR Reviewer

Joined: Thu Dec 11, 2003 9:53 pm
Posts: 2696
Location: Scarsdale, NY
If I drill four holes where the standard holes belong, it will notch in between the two sets of holes that Tt already has there; it will open up the case so there are four diagonal slats and the result might be quite an unstable mountpoint for a heavy heatercore. I really can't elaborate any more on the issue with the mountpoints on this case; if I were to cut a set of holes completely unaligned to the original two sets, it would then involve cutting out the 120mm hole for the air to pass through, and I can guarantee serious, "uglification," of my case if I tried that, myself.

This is Thermaltake Tsunami Dream, not an SLK-3700. I'm not entirely sure drilling holes is standard practice, either; I have yet to ever drill a hole into anything computer related (cut out grilles, yes; deburr, yes; cut off parts that get in the way, yes; drill precision placed holes for mounting of heavy objects to improve airflow, no).

The bleed issue you mentioned is exactly correct, and I will not be able to maintain a position where the radiator is below the PCI slot line; for me to cut the precisely right-sized, nearly perfectly circular holes into a higher portion of the case myself successfully is like me trying to hit a 3-point shot from the opposite end of an official-spec NBA court; it's not impossible, it's just not ever going to happen.

Again, thanks for the suggestions, but once again, they're not particularly good options given my current situation and who you're suggesting them to (I really am NOT a very handy person; you ought to see how unclean my cuts to the RadBox came out! :oops:).

-Ed

_________________
Contributing Writer, SPCR
NgTechnik

Want something reviewed? Help us get samples!
Donate for Patron or Friend Status!


Top
 Profile  
 
 Post subject:
PostPosted: Sat Aug 21, 2004 7:38 am 
Offline

Joined: Sat Mar 13, 2004 12:08 pm
Posts: 109
Location: Oklahoma
I know your case is not an slk3700, but if you look at the pictures closely, it is very similar. Similar enough to say they are based on the same basic chassis including the rear fan mounting. When I drilled the back of my case, yes the holes became diagonal slots. As far as support, I don't know the best method.

Another option would to place another radiator inside the case up against the top behind the PSU. That is somewhat common, but adds its own bleed challenges. It also requires a blowhole or 2 in the top of the case.

I just have a very strong hunch that you will not be able to do what you want to do without excessive component temperatures or the lack of stability to get where you want to be. I hope to be wrong.

_________________
rig #1 Watercooled XP2500 @ 2.5GHz, Nforce2-400, watercooled 9700 Pro (370/330), Seagate 120GB, Antec SLK3700 & PSU
rig #2 Aeroflow cooled XP2500 @ 2.3GHz, Nforce2 IGP, 9700 Pro (405/370), Seagate 80Mb, Enlight mid-tower & PSU
6.5 GHz Folding


Top
 Profile  
 
 Post subject:
PostPosted: Sat Aug 21, 2004 10:29 am 
Offline
SPCR Reviewer

Joined: Thu Dec 11, 2003 9:53 pm
Posts: 2696
Location: Scarsdale, NY
Moving the radiator internally provided a massive improvement to the portability of the system; remounting something externally as before would recomplicate the transportability of the machine. Having made such a great improvement to this aspect of the machine, I'd hate to take away from it, so I'd prefer to make the best of what I can with the current heatercore. Now that I know I don't have to have heavy things hanging out the back hampering tranposrtation, I will avoid messing it up again at nearly any cost.

I would've done this from the beginning had it actually occurred to me that I could fit it inside this way; it just didn't pop into my head until the second build.

I mean if portability weren't such an important issue, I'd have utilized a steel case and at minimum a Thermochill 120.2 heatercore, if not a 120.3, and a larger case at that, which would give me more room to work.

-Ed

EDIT: Btw the design is extremely similar, but Tsunami is aluminum, not steal, so mounting a heavy heatercore->shroud->fan combo there would be less stable than on the steel SLK3700.

EDIT2: That is, if I turn the two sets of holes into four diagonal slats; it would hurt the integrity of the aluminum backwall of the case much more than on a steel case.

EDIT3: Maybe you don't own an SLK3700...

_________________
Contributing Writer, SPCR
NgTechnik

Want something reviewed? Help us get samples!
Donate for Patron or Friend Status!


Last edited by Edward Ng on Sat Aug 21, 2004 6:28 pm, edited 3 times in total.

Top
 Profile  
 
 Post subject:
PostPosted: Sat Aug 21, 2004 5:49 pm 
Offline

Joined: Fri Aug 01, 2003 10:45 pm
Posts: 587
Location: North Billerica, MA, USA
Ed, The issue with your fan mount is where the fan is, not so much the shroud. Currently you have the fan smack against the face of the rad then the shroud between the fan and the case. I would put the shroud against the rad (and tape up the holes in the side of the shroud) and the fan against the case. This would get rid of the fan turbulence noise from the fan blades beating next to the rad face, and allow airflow to come through the entire rad face (and it sounds like you need all the help you can get in that regard)

The downside is that you might get some interference on the fan output side from air against the case. If this is a problem, it looks to me like you might have room to slide a second narrow shroud on the other side of the fan. Note that making basic shrouds is relatively easy, I suspect that you could probably get something appropriate from the plastics department at your favorite Wal-mart, or your S.O.'s Tupperware collection

As to the question of blowing air through the rad into the case, the folks at ProCooling all seem to say it's a non-issue - that the maximum temp increase for air going through a rad in a PC WC configuration is 2-3*C, probably less. This heat is distributed, not point concentrated like it is in an AC setup, so there won't be as much transfer back to the PC innards. The temp increase won't be enough to significantly reduce the cooling on those parts dependent on air cooling. The parts w/ waterblocks aren't going to be heated at all since they are effectively insulated by the blocks which are inherently going to be hotter than the incoming air.

At the same time, you will be cooling the rad with ambient air, probably the coolest available air source, certainly cooler than air that's gone through the case. Since the rad's efficiency depends on the Delta-T between it and the cooling air, (again sounds like you need all the help you can get) that cooler air could make a difference. It also solves some of the shroud issues, since the rad would be up against the front of the case (which might need to be cut out under the bezel) and the shroud would only need to support the fan.

Now given the lower airflow numbers that are commonly used on SPCR, you might get moved back towards the high end of that temp increase, but IMHO you'd still come out ahead. Consider the Delta-T you would consider acceptable between the intake and exhaust of an AC system. 90%+ of that increase comes from the components you're watercooling in this setup, so they won't add anything significant to the temps inside the case in a WC box. Thus the only temp increases will come from the few remaining air cooled components. So why would it be a problem to blow slightly rad warmed air into the case - you still end up with less of a Delta-T between the intake and exhaust...


Gooserider

_________________
Building Dual Athlon MP system, Tyan mobo, U160 drives, Server Cube case, Linux ONLY, lots of other goodies. Will water cool, attempting to make as silent as possible.


Top
 Profile  
 
 Post subject:
PostPosted: Sat Aug 21, 2004 6:14 pm 
Offline
SPCR Reviewer

Joined: Thu Dec 11, 2003 9:53 pm
Posts: 2696
Location: Scarsdale, NY
Gooserider, the shroud is 1/2" deep and it's between the heatercore and the fan (black material); you're talking about the RadBox used to mount the assembly (painted blue).

Come on I know it's a shallow shroud but why doesn't anyone see it?!? The RadBox is no shroud and as stupid as I may look here I'm not THAT stupid that I can't tell a RadBox from a shroud!!! Like I said if it's too shallow, fine; that's valid, but I say one...more...time..."The fan is not directly up against the heatercore (unless you consider a 1/2" gap to be directly against it)!!!" I'm also smart enough to realize that a shroud goes between the fan and the heatercore; what good at all would it be in any other arrangement if it's designed to deal with proximity noise and central dead spot? Come on, give me more credit than that!

That RadBox has the slats in it that allow me to even mount that assembly there; the holes in the front intake don't line up with the holes in the fan; look closely at the RadBox and you see that the bolts going through it from the front do not even line up with the bolts that pass through the fan into the RadBox.

I can flip the fan around so that it will will blow through the radiator, in which case the heatercore will run cooler but the case ambient will be higher, but I cannot change the order of the components as they extend from the front of the case. In other words, it must go RadBox->fan->shroud->heatercore, from front to back, due to the method I must use to mount it all. I cannot have it going RadBox->heatercore->shroud->fan because the RadBox does not have holes where the bolts extend through the heatercore to attach the shroud. If I run a second set of bolts through the heatercore to mount it directly against the RadBox (assuming I were to go out and buy some), they would interfere with the shroud.

-Ed

EDIT: If it helps at all, notice that the bolts going through the heatercore do not line up with the bolts going through the fan; this is because the holes for the fan on the shroud are in the normal position for 120mm fans, but the holes for the heatercore are farther out to the sides and closer together, rather than at the corners. Can you see the shroud now? As I said, the shroud is where it belongs, but if you guys think it's too shallow (even I think so by now--you guys can't even see the damn thing!), I will continue seeking a replacement.

EDIT2: A typo.

_________________
Contributing Writer, SPCR
NgTechnik

Want something reviewed? Help us get samples!
Donate for Patron or Friend Status!


Top
 Profile  
 
 Post subject:
PostPosted: Sat Aug 21, 2004 8:34 pm 
Offline

Joined: Sat Mar 13, 2004 12:08 pm
Posts: 109
Location: Oklahoma
I see the existing shroud; it's much better than no shroud at all.

The challenge seems to be the radbox because it is limiting good options. To secure the radiator, dump the radbox completely and use several zip ties through the bottom of the radiator fin area around to the lower drive cage mount. So long as there is a 120mm hole in the front, air will find a way to get to the rad; it doesn't have to line up perfectly. The tubing will also help hold the top of the rad in place.

The rad needs the coolest air and that is directly from the outside. Good case airflow will also help your system overclock. A performance rig needs more than minimal case airflow.

_________________
rig #1 Watercooled XP2500 @ 2.5GHz, Nforce2-400, watercooled 9700 Pro (370/330), Seagate 120GB, Antec SLK3700 & PSU
rig #2 Aeroflow cooled XP2500 @ 2.3GHz, Nforce2 IGP, 9700 Pro (405/370), Seagate 80Mb, Enlight mid-tower & PSU
6.5 GHz Folding


Top
 Profile  
 
 Post subject:
PostPosted: Sat Aug 21, 2004 8:50 pm 
Offline
SPCR Reviewer

Joined: Thu Dec 11, 2003 9:53 pm
Posts: 2696
Location: Scarsdale, NY
There is more than minimal case flow; the 120mm exhaust fan and 92mm intake fan wil both be running 12volts, as will the rather high speed 120mmfan attached to the shroud on the heatercore (likely the fastest fan of all). Okay, it's not MONSTER case flow (I used to run a Xaser II with 2 80mm LED fans up front, one 92mm LED fan up front, one 80mm LED fan on a top blowhole, two 80mm LED fans out back and one 80mm LED fan in the side panel, all running 12volts), but it's still more than a major majority of SPCR users; as I said, I'm not going for any seriously low level of noise. Come to think of it, the combination of the PSU fan and exhaust fan should be nearly as much as the fan pulling through the radiator.

The heatercore isn't lined up squarely as it is; by using the RadBox, I have the heatercore assembly slid over to one side, in order to clear the side panel. That's another reason why I rely on the RadBox; the slats in the mounting plate allow horizontal adjustment.

Anyway, as I said, if I feel it's necessary, I can always flip the fan later; that's just not a major problem. I'm trying to figure out a decent looking way to seal the sides of the RadBox so that none of the hot exhausted air gets deflected back into the case that would be easy to remove and reinstall should I find the need to make adjustments to the fan/heatercore assembly. Tape would be really, really unprofessional, and collect dust, as well as leaving crap on my RadBox, and it won't be reusable in any way. I was thinking something like aluminum strips cut to size, but I'm not sure how to secure it.

I'm trying to see how I could use zipties to secure the heatercore to that bracket without interfering with the shroud; the only way I can see would involve cutting into the shroud, which would allow for a slight, but minor leakage of air...

-Ed

EDIT: I will likely place the thermistor from the AF92CT intake fan right into one of the MOSFET sinks, which may get hotter than anything else in the system at this rate. If that doesn't promote sufficient spin, I can always do the thermistor snip n' short mod, which would allow the AF92CT to spin at full blast.

_________________
Contributing Writer, SPCR
NgTechnik

Want something reviewed? Help us get samples!
Donate for Patron or Friend Status!


Top
 Profile  
 
 Post subject:
PostPosted: Sun Aug 22, 2004 4:23 pm 
Offline

Joined: Fri Aug 01, 2003 10:45 pm
Posts: 587
Location: North Billerica, MA, USA
Me bad... (unless you snuck in and changed the pix on us) as I can see that shroud now, I think that I was seeing it as part of the fan before. Yes, a 1/2" shroud is better than nothing, but it is way skinny, it would be better if it were at least 1-1.5" I was assuming that you were using the radbox as a shroud also - I define shroud as any item (with or without a fancy brand name) that directs the flow of air between a fan and a target object, and the radbox fits that definition. (BTW, even with the current setup I'd tape up the side slots in the rad box to force the air out the front and keep air from escaping and getting recirculated through the rad.)

I agree with 1911user that the radbox looks like a big part of your problem, and I don't see any reason why you are married to it. What structure is there in or around the sheetmetal in the front of the case under the bezel? I can't see if there is any issue from the picture, so I would consider seeing if it were possible to get the rad right up against the front case metal wall. If I could, then I would cut out an opening in the case wall leaving just 4 corners to pass securing bolts through the rad in order to anchor it. Use some weatherstripping tape to seal the rad / case joint and supply vibration isolation (and consider using zip ties instead of bolts)

Then fasten a deep,(possibly home made or tupperware) shroud to the rear face of the rad and the fan to the shroud, with more weatherstrip to seal and isolate as needed. (The fan sucking through the rad and blowing into the case) This will give you a cleaner install that is lighter weight, probably better looking, and much better flowing instead of the current setup.

Again, I repeat, the increased heat in the case from having it blowing air through the rad is trivial. OTOH, if you have the rad fan blowing in the same direction as the PSU and exhaust fans (so that the rad is the primary air intake) then you will have in effect a serial fan configuration which is highly desireable in a WC setup. Serial fans don't move more volume than a single fan in free air conditions, but DO produce the higher static pressures which are needed to get maximal airflow through your (relatively high airflow restriction) rad.

If I'm understanding your current fan setup, you have the rad fan and PSU fans fighting over the airflow coming in through a side vent fan, which is a poor to negative fan config. Your cooling would be better if you had the rad as sole intake with all the other fans doing exhaust to help the rad fan deal with getting the max flow into the box.

Gooserider

_________________
Building Dual Athlon MP system, Tyan mobo, U160 drives, Server Cube case, Linux ONLY, lots of other goodies. Will water cool, attempting to make as silent as possible.


Top
 Profile  
 
 Post subject:
PostPosted: Sun Aug 22, 2004 4:58 pm 
Offline

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

_________________
rig #1 Watercooled XP2500 @ 2.5GHz, Nforce2-400, watercooled 9700 Pro (370/330), Seagate 120GB, Antec SLK3700 & PSU
rig #2 Aeroflow cooled XP2500 @ 2.3GHz, Nforce2 IGP, 9700 Pro (405/370), Seagate 80Mb, Enlight mid-tower & PSU
6.5 GHz Folding


Last edited by 1911user on Sun Aug 22, 2004 7:29 pm, edited 1 time in total.

Top
 Profile  
 
 Post subject:
PostPosted: Sun Aug 22, 2004 5:31 pm 
Offline

Joined: Fri Jul 23, 2004 11:10 pm
Posts: 657
Location: AB, Canada
i wonder what kind of case exhaust you'd need to use to create enough negative pressure to get decent airflow through a rad with no fans on it; you'd have to seal the rest of the case really well but it'd be neat to have 2-3" of extra space on the inside of the rad. :) just in general, obviously doesn't apply here.

and i don't mean for this to even sound like a suggestion since it would mean a ton of work, but i was thinking if i used a single 120mm rad and didn't mind having the front of case open, it'd be nice to sit the radiator in the bottom of the case and duct it to the bottom front bezel (would actually be more like a long/curved shroud). would let you get cool air and not exhaust it into the case, and not mess with "normal" airflow.

Can't tell what rad you're using, but there's a very nice article from last year at http://www.overclockers.com/articles778/ . Looks like those single 120mm rads get rid of a surprisingly decent amount of heat.


Top
 Profile  
 
 Post subject:
PostPosted: Sun Aug 22, 2004 8:49 pm 
Offline
SPCR Reviewer

Joined: Thu Dec 11, 2003 9:53 pm
Posts: 2696
Location: Scarsdale, NY
1911user wrote:
deleted


:?


Re: Gooserider...
I'll take a look at the possibility of mounting the heatercore directly to the front wall of the case tomorrow; if I can mount it with cableties without punching new holes, I would be somewhat more inclined to do it, but that would imply that it won't shift so far over to the side that it interferes with the side panel...

You have to understand that if I attempt to fabricate my own duct, even with some fitted Tupperware, you're talking about an almost 100% guarantee of absolute, total non-professional looking results. Edward Ng trying to hand-modify a Tupperware container into a fan duct will result in some seriously home-school looking rigging. I'm no Bob Villa--as far from it as one can possibly imagine.

Flipping the fan around so as to pull air in from outside and exhausting into the case seems like the only way to get you guys to shut up about it (that's a joke, fellas :wink:), so I'll go ahead and do that, but I may end up simply using my current arrangement, flipping the fan so it blows through the heatercore (find some clean, professional looking way to seal the RadBox, rather than mere tape which will collect dust).

The problem with going directly against the front wall with the heatercore is that there's no way for me to use the existing bolts to attach the stock shroud back onto the heatercore without drilling holes out to allow the bolts to pass through (although admittedly, if I do this, it would allow me to use the stock shroud bolts not only to install the shroud, but also to mount the heatercore itself to the front wall). It would be an insane pain in my butt to take everything out of this case so that I can drill four holes without worrying about metal dust/shavings. I can't even guarantee that this would work at all, since it may force the heatercore back over so far to the side panel that I can't close the panel; if I try to slide the heater core more inward from the panel and mount it in this fashion, there may not be any metal on the side closer to the panel for fastening of the bolts at all! It may also require me to cut out more from the front wall just to free up more air flow, since it would be pushed way over past the edge of the current hole...

Come to think of it, I think I'll shoot some pics of the front wall for you guys to get an idea what's going on.

-Ed

EDIT: The photos (click for the all-important high-res, low-compression view):

[URL=HTTP://WWW.NgTechnik.com/HCassembly/Current-Arrangement-Big.jpg]Image
Here's a direct profile view so you can get a proper idea of the scale and vertical spacing available to me.[/URL]

[URL=HTTP://WWW.NgTechnik.com/HCassembly/Side-Panel-Gapping-Big.jpg]Image
This is my attempt to show you how much space I have to work with from the side panel as it is already--it's about 3/8"-1/2" from the side panel, but this angle, which is just about the best I can manage, isn't 100% clear.[/URL]

[URL=HTTP://WWW.NgTechnik.com/HCassembly/See-Through-Big.jpg]Image
This is the clearest shot I have of the shroud; this also gives you an idea of the depth and shape of the RadBox. Also perceivable is exactly where the bolt holes are on the shroud for both, the heatercore and the fan.[/URL]

Image

[URL=HTTP://WWW.NgTechnik.com/HCassembly/Front-Wall-Straight-Big.jpg]Image
Here's a direct shot of the assembly from the front, including the wall. The first thing I did with this case before I ever installed anything was to snip out the front and rear grilles, leaving this opening. You can also get an idea how I cut out the center portion of the RadBox to free up flow (actually, there's only one more slat from the top and from the bottom, and then one really big solid wall of metal, had it not been cut away using the rotary). Notice that my cuts weren't all that straight at all. :oops: I pushed the RadBox as far over as I could on its slats to clear the side panel, which is why it's not centered in the opening. I wasn't willing to cut out any more of the mounting plate, as it may compromise rigidity of the thing excessively for the weight it has to handle, what with the system expected to travel quite a deal, it needs to remain extra rigid.[/URL]

_________________
Contributing Writer, SPCR
NgTechnik

Want something reviewed? Help us get samples!
Donate for Patron or Friend Status!


Last edited by Edward Ng on Sun Aug 22, 2004 9:41 pm, edited 1 time in total.

Top
 Profile  
 
 Post subject:
PostPosted: Sun Aug 22, 2004 9:40 pm 
Offline

Joined: Sat Mar 13, 2004 12:08 pm
Posts: 109
Location: Oklahoma
The deleted part was:
Edward Ng wrote:
EDIT: Btw the design is extremely similar, but Tsunami is aluminum, not steal, so mounting a heavy heatercore->shroud->fan combo there would be less stable than on the steel SLK3700.

EDIT2: That is, if I turn the two sets of holes into four diagonal slats; it would hurt the integrity of the aluminum backwall of the case much more than on a steel case.

EDIT3: Maybe you don't own an SLK3700...


That level of frustration means it's time for you see how the current setup works then start planning for the next revision.

Gooserider has been doing serious watercooling longer than I have, but I read and research subjects thoroughly plus I have a technical background. We are used to seeing certain characteristics in good water systems and to not see that takes some adapting. Doing watercooling without case mods is unusual especially for performance systems. I'm really curious how well your current setup will work if you bring air in the front intake. That's about it for the easy mods to the current setup except res/T selection and that's mostly personal choice; doesn't affect temps if done correctly. You'll have to decide if the results are good enough.

To get an idea of our mindset, look seriously at this radiator for your next revision. It looks like it'll just fit inside the lower front of the case if you remove ALL hardware for the lower and mid drive bays (that means drilling rivits). http://www.ocforums.com/showthread.php?t=117055
Make a 2 inch deep shroud that covers the entire rad surface and uses a 120mm fan. I'll predict your loaded cpu temps will easily drop several degrees compared to the current setup. It won't be plug-n-play though.

_________________
rig #1 Watercooled XP2500 @ 2.5GHz, Nforce2-400, watercooled 9700 Pro (370/330), Seagate 120GB, Antec SLK3700 & PSU
rig #2 Aeroflow cooled XP2500 @ 2.3GHz, Nforce2 IGP, 9700 Pro (405/370), Seagate 80Mb, Enlight mid-tower & PSU
6.5 GHz Folding


Top
 Profile  
 
 Post subject:
PostPosted: Sun Aug 22, 2004 9:43 pm 
Offline
SPCR Reviewer

Joined: Thu Dec 11, 2003 9:53 pm
Posts: 2696
Location: Scarsdale, NY
Crosspost, 1911; check the photos in my EDIT.

You guys must remember this is my first ever water cooling setup! I hope I'm not the only one who's amazed as it is that I got this much done with it...

Can we please leave some of the much harder-core modding to a later setup? :cry: I mean come on as it is this system comprises a good 50% of all the serious mods I've ever done for myself already! Do you guys have any idea how many systems I've already built in my time? It's hard to suddenly start doing such advanced stuff (advanced to me, at least) out of the blue having been accustomed to a certain standard level of labor.

It's not like I'm not trying, here! This is going WAY above and beyond anything I've done before. Some people, like myself, need to scale a 10' wall before tackling the 200' mountain...

-Ed

EDIT: Btw by me saying, "maybe you don't own an SLK3700," I was not taking a crack at your knowledge or understanding of that case. I had made a prior comment saying, "your case," rather than, "the SLK3700," which I subsquently modified to simply state, "the SLK3700." I denoted this change with that statement, "Maybe you don't own..." I don't like making EDITs without stating why I EDITed, it's like going back and changing something on you so I don't look wrong when you pointed out something I said that's wrong, and I don't do that. When I'm wrong, I'm wrong, and I admit to it, time after time., just as I am now (and always have in the past in these forums). Sorry for the mix up, and no hard feelings. I never doubted yours or Gooserider's experience; no one with such lack of experience such as myself on this topic has a right to do so.

On the same note, the same lack of experience with serious modding is the reason why I am doing my best to avoid tackling a mod that is beyond my current skill level and possibly botching the living crap out of it. My first failure at water cooling in general made me wary enough and I was ready to quit. That I'm taking it up and trying once more is hard enough, let alone going all-out and trying things like drilling rivets and finding ways to mount things into places that have no preset, easy way to do so.

My frustration is not with you or Gooserider. It's with the fact that from the sound of it, this is going to be a complete failure unless I go more than all-out. If that's what it takes then maybe I am in over my head and should just leave this analysis to someone else that is more qualified.

_________________
Contributing Writer, SPCR
NgTechnik

Want something reviewed? Help us get samples!
Donate for Patron or Friend Status!


Top
 Profile  
 
 Post subject:
PostPosted: Sun Aug 22, 2004 10:19 pm 
Offline

Joined: Sat Mar 13, 2004 12:08 pm
Posts: 109
Location: Oklahoma
Your frustration (especially with radiator size, selection, and placement) is an example of why watercooling isn't and won't be mainstream for awhile. It isn't as easy as air cooling even if you do assemble your own computer. The first setup is always the hardest though. You are also placing importance on how it looks, that makes the job harder.

To finish this one up, just turn the front fan around to draw air from the intake, seal up the sides of the radbox, and see how it works. If you decide to make it better, then the mods get more advanced.

You'll want to plan to lay the PC on it's side when traveling. That will keep the video waterblock from trying to tear the video card out of its slot or cracking the video card circuit board. The GPU block is very heavy and could do damage with a big shock during transit. Make sure the reservoir is OK with the case on its side; it needs airtight seals. This applies to PCs with heavy heatsinks either air or watercooled.

_________________
rig #1 Watercooled XP2500 @ 2.5GHz, Nforce2-400, watercooled 9700 Pro (370/330), Seagate 120GB, Antec SLK3700 & PSU
rig #2 Aeroflow cooled XP2500 @ 2.3GHz, Nforce2 IGP, 9700 Pro (405/370), Seagate 80Mb, Enlight mid-tower & PSU
6.5 GHz Folding


Top
 Profile  
 
 Post subject:
PostPosted: Sun Aug 22, 2004 10:32 pm 
Offline

Joined: Sat Mar 13, 2004 12:08 pm
Posts: 109
Location: Oklahoma
Your setup will not be a complete failure. Your system is a good start except airflow through the radiator. For every degree you can lower your air temperature to the rad, that is a degree cooler for your CPU and everything else watercooled. If you can save 5 degrees on air temp, that is an easy 5C temp drop for your CPU. There are very few cheap changes that affect temperatures so much.

Realistic worst case is your radiator can't get rid of enough heat for the massive overclock you have planned at acceptable (and stable) temperatures. Try it, but you MAY have to back off the overclock somewhat; it isn't a certainty. The same is true for the GPU.

_________________
rig #1 Watercooled XP2500 @ 2.5GHz, Nforce2-400, watercooled 9700 Pro (370/330), Seagate 120GB, Antec SLK3700 & PSU
rig #2 Aeroflow cooled XP2500 @ 2.3GHz, Nforce2 IGP, 9700 Pro (405/370), Seagate 80Mb, Enlight mid-tower & PSU
6.5 GHz Folding


Last edited by 1911user on Sun Aug 22, 2004 10:37 pm, edited 1 time in total.

Top
 Profile  
 
 Post subject:
PostPosted: Sun Aug 22, 2004 10:37 pm 
Offline
SPCR Reviewer

Joined: Thu Dec 11, 2003 9:53 pm
Posts: 2696
Location: Scarsdale, NY
I always transport on its side; tall things like to fall over in moving vehicles...

There is no particular frustration on my part with the radiator size, selection or placement now that I have it rotated to allow proper bleeding. I'm happy with this arrangement (again, I'm assuming it bleeds properly this way, and I don't see any reason why the other heatercores that have been recommended to me would be any easier to bleed, with the exception of heatercores with bleed valves). You guys are the ones that aren't. I'm not trying to single you two out; it's just a fact. I was actually quite satisfied as it was with the setup when it wasn't even working right until I saw mother nature crashing the party, and I am confident that this arrangement will perform better still. Am I not the one among the three of us that's trying to KEEP it like it already is, rather than changing it? Were I the one that's frustrated with it, I'd be out to change it or fix it, not keep it as is.

I don't have insane expectations here. Considering I felt the results of the mediocre setup to be more than satisfying, this second setup should be even more stellar to me. For you guys, it already seems disappointing, and I wouldn't be surprised if the results, once this system is running, will also disappoint the two of you. Of COURSE it will; you've probably seen more than your fair shares of uber fluid cooling systems that will put the whoopass on mine. On the converse, if this system works, period, I'll already be more than happy with it, assuming it works at least as well as the first setup.

I already said I will turn the fan around. I need to figure out a way to seal the RadBox that won't collect dust like crazy (tape likes to pick up dust; I think we all know that much), and then that will be done as well.

In the end I'm not here to go all-freakin'-out. I'm here to learn how the arrangement of the two pumps will affect the performance of the system. I think that's the same thing the rest you are here for, in regards to this topic. Does not having the absolutely, "mostest bestest," heatercore arrangement run the risk of completely rendering my results worthless to you guys? I sure hope not.

In other words, so long as this system works, that's all that matters, because the focus is the three circuit arrangements. If my heatercore setup is crappy, fine--so long as it remains equally crappy with all three pump/block scenarios, it should not come into play, am I not correct? I think we need to refocus on the priorities of this circuit comparison.

-Ed

EDIT: Crosspost, 1911...

That massive overclock you stated ran fine already with the busted setup. Unless you have an inkling that my new arrangement could be worse than before, there is no need to worry about my ability to run the former overclock figures, at all.

I ran 2533MHz at 2.1volts on the CPU and 418/1200 on the GPU/GDDR3 on air cooling. What are the chances that this arrangement could possibly perform worse than an SP-97 with 12V 92mm M1B and the stock 6800GT cooling?

EDIT2: Do either of you guys use AIM, per chance? :lol: :lol: :lol:

_________________
Contributing Writer, SPCR
NgTechnik

Want something reviewed? Help us get samples!
Donate for Patron or Friend Status!


Top
 Profile  
 
 Post subject:
PostPosted: Sun Aug 22, 2004 11:04 pm 
Offline

Joined: Sat Mar 13, 2004 12:08 pm
Posts: 109
Location: Oklahoma
Honestly, I think you'll get similar temps compared to air cooling, maybe a few C better. The reason is the surface area of an SP-97 heatsink and the stock video cooling are probably similar to the surface area of the radiator. The big performance advantage to watercooling is being able to efficiently move the heat to a much larger surface for cooling. Only the rad you have isn't that much of a larger surface.

Part of the reason we keep subconciously picking on the rad is it is small by performance watercooling standards. Radiator size is one of THE primary factors that determine how well a water cooling setup works; period.

It is hard work to use big rads. You either need a large case (that probably needs mods) or build an external cooling box that contains the large fans, large heatercore(s), and possibly the pump. The external solution works well, but isn't very portable. Most "ubber-water" setups have large radiator capacity where cool air can easily flow through them; it's not a huge secret.

The advice I normally give for people wanting to seriously watercool is to stuff the largest heatercore possible into your case, use a shroud for the fan(s), then worry about the rest. The radiator should be the first priority, not the last. As you might expect, this advice isn't well recieved because it's hard and not as much fun as discussing which is the best waterblock or pump to purchase.

OK, I've killed this subject. Just finish the setup and have fun.

_________________
rig #1 Watercooled XP2500 @ 2.5GHz, Nforce2-400, watercooled 9700 Pro (370/330), Seagate 120GB, Antec SLK3700 & PSU
rig #2 Aeroflow cooled XP2500 @ 2.3GHz, Nforce2 IGP, 9700 Pro (405/370), Seagate 80Mb, Enlight mid-tower & PSU
6.5 GHz Folding


Last edited by 1911user on Mon Aug 23, 2004 9:07 am, edited 1 time in total.

Top
 Profile  
 
 Post subject:
PostPosted: Mon Aug 23, 2004 8:12 am 
Offline

Joined: Thu Jul 01, 2004 4:55 am
Posts: 168
HEY ED!!!!!
AWESOME!!!!

_________________
The GlassMan
Silicon, at the bleeding edge of technology for 20,000 years


Top
 Profile  
 
Display posts from previous:  Sort by  
Post new topic Reply to topic  [ 145 posts ]  Go to page 1, 2, 3, 4, 5  Next

All times are UTC - 8 hours


Who is online

Users browsing this forum: No registered users and 0 guests


You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot post attachments in this forum

Search for:
Jump to:  
Powered by phpBB © 2000, 2002, 2005, 2007 phpBB Group