TJ07: Black n Watercooled Project

[alphakry]

Alright, so I figure I start the official log on this project. As previously indicated in my other thread, I'm looking to build out a TJ07 housing a dual-loop watercooled system. This thread is also being mirrored on [H].

Project Goal: Maximum Performance with Minimum Sound

------------------------ Table Of Contents: ------------------------

Post #2 - Unofficial Sponsors
Post #31 - 8800GTX single bracket installation and first paste test
Post #40 - 8800GTX EK waterblock modification - sunken screws
Post #46 - Motherboard Tray Drilling & Dual Res Mounting
Post #52 - ATX Removal
Post #53 - Powder Coating
Post #66 - Leak Test #1

------------------------ Primary Components: ------------------------

• CPU: Intel Quad-Core Q6700 2.66GHz
• MOBO: EVGA 122-CK-NF68-A1 680i SLI
• RAM: G.Skill PC2-6400 4GB (2x2GB) CL5-5-5-15
• PSU: Cooler Master Real Power Pro 1000W
• GPU: Dual EVGA 8800GTX 768MB (SLI Configuration)
• HDD: Dual Western Digital 80GB Raptor (Raid 0 Array)
• SOUND: Soundblaster X-Fi Fatal1ty w/ Front Panel
• NIC: Killer-NIC Offload Card
• DVD: Lite-on DH-4B1S Blu-Ray/DVD/CD Rewriter

------------------------ Water Cooling Components: ------------------------

• CPU: D-Tek FuZion CPU Block
• MOBO: Northbridge: EK-NB EVGA 680i Acetal;
• MOBO: Southbridge: Swiftech MCW30
• RAM: None yet.
• HDD: Danger Den Aqua-Drive Hard Drive Block
• GPU: Dual EK FC8800 GTX Delrin
• PUMP1: Swiftech MCP350 w/ DDCT-01s PetraTech Top
• PUMP2: Swiftech MCP350 w/ DDCT-01s PetraTech Top
• PUMP3: Swiftech MCP350 w/ EK X-RES 100 Top
• RADIATOR1: Thermochill PA120.2
• RADIATOR2: Thermochill PA120.2
• RADIATOR3: Swiftech MCR220-QP Res
• RESERVOIR: Dual EK-150 rev.2

Loop Configurations:

• Loop 1: PUMP1===>RAD1===>CPU==>RES1
• Loop 2: PUMP2===>RAD2==>GPU1===>GPU2===>RES2
• Loop 3: PUMP3===>RAD3===>MOSFETS==>N.BRIDGE==>S.BRIDGE==>H DDS
  
  
  the placement of the remaining components is yet to be determined. I'm a believer in shortest path, with the exception that a pump really should always be fed input from a reservoir and output should first hit the loops most critical block.
  
  Tubing Options (3/8", brand to be determined):
  Tygon R-3400 3/8" ID - 1/2" OD - black and sexy (most likely choice)
  Primoflex 3/8" ID - 5/8" OD (1/8" wall) - amazing kink resistance
  Tygon R-3603 3/8" - thick and strong

------------------------ Items No Longer Used and Now For Sale: ------------------------

• Lian-Li VCOOL PC-V1000BPLUS Aluminum ATX Mid-Tower, Modified for Water Cooling
• Koolance HD-50-L06 Hard Drive Cooling Block
  
• Swiftech MCRES-525 5 1/4" Bay Reservoir from the H20-Apex kit, revision 1
• Cooler Master GeminII LGA CPU Cooler w/ brackets

[alphakry]

It may be a bit early in this projects life to be thanking people, but I feel I must!

a BIG THANKS goes out to John at Jab-Tech. He's been an incredibly informative and helpful resource in this and many of my other projects and I'm sure he'll be continuously hearing from me before this one is done! Between quick shipments, friendly email responses and the utmost competitive pricing - he's the very first stop any time I ever need a part. 95% of parts you see in this build, from the blocks to the yate-loons, were acquired with his help.

I also must thank Gary Stofer at Sidewinder. He's a valuable asset to this community and his willingness to help us (See virtualrain's post on Tubing Comparisons) really doesn't go unnoticed. He provided me with some tubing that looks to be the decided upon component in this build.

On that same note, Eddy_EK also needs a quick thank you for providing this forum and the watercooling community in general with amazing products that are really designed with OUR thoughts and suggestions in mind! Guys like him and Cathar are really gods amongst men who keep many of us addicted to this world of water! Plus, I am hoping to really put pressure on him to get a Mosfet block designed for the eVGA 680i motherboard before this year is out - and hopefully it will see it's prototyping here in this project! ::Crossed fingers::

And lastly Alex and Quoc at Petra's Tech Shop - for being one of, if not THE best shops out there dedicated to Xtreme systems and keeping them cool!



anyhow, now that I got that out of the way - lets get a few teaser shots in here to keep everyone on page with what's lying on the work bench...

[RADCOM]

fantastic project I shall follow with this to the end, keep posting m8 and good luck!

[alphakry]

As you can see in the pics, I experimented with the idea of inverting the motherboard. However, because of this - the stock windowed panel becomes useless. Now considering a big secondary goal of this project is to look good, we can't have that - and I don't really want to cut into the other panel to add dual-windows - so it looks like I'll be nixing the 750W Silencer for another PSU that provided longer cables - namely the 8PIN PSU and Main ATX power connections. Any recommendations?


I also think I'll be selling off the MCP655 - as the TJ07 is too narrow and the Thermochill radiators are far too thick - so that bottom compartment is QUITE stuffed - leaving no room for a big pump like the 655. I think i'll be selling it off (see my [h] profile for such business) and replacing it with another Petra modified MCP350. For this same reason, the 120.3 will be replaced with a 120.2 - to give a bit more room to work with.

I prefer the 350 over the 355 for the fact that it's quieter. I don't think I'll need the extra flow and don't want the extra heat dump of the 355 - so the quieter yet still amazing 350 should do the trick!

Stage 1

• Sell off MCP655 for MCP350+Petra's Top
• Sell off PC Power&Cooling 750W for a undecided replacement
• Swap 120.3 for a 120.2 for more room
• Return Motherboard tray to original configuration

[Senater_Cache]

exciting, please take lots of pics of the blocks being mounted, and all little innovations you will do.

I live for that stuff

[RADCOM]

I can do swap for pa.120.2 to 3 but guess you live in da US

[Nate P.]

Very nice! I'll be sure to follow this one closely!

[alphakry]

Radcom - yea sorry, I'm in NY. Plus Petra's is allowing me to return it in exchange for their last 120.2 in stock!

Any TJ07 owners out there that have a PSU recommendation with long enough wires for a clean install in this case?

I've always been a huge fan of Seasonic but don't remember them making very long wires. I really wanted to give my first PC Power&Cooling PSU a shot - but between the suggestions that it's not by any means "silent" - and the fact that the main power wires just aren't long enough - it's got to go!


I'm waiting on some right angle radiator fittings (yuck, i know. but it's the only way those 2 thick as brick Thermochills are going to route from the bottom compartment!) plus some tubing to arrive before the next big update! I'll be back shortly!

[AndyM]

I've always been a huge fan of Seasonic but don't remember them making very long wires. I really wanted to give my first PC Power&Cooling PSU a shot - but between the suggestions that it's not by any means "silent" - and the fact that the main power wires just aren't long enough - it's got to go!

Corsair HX620. Made by Seasonic also. Nice modular cables so no extras floating around, and long enough as well.

[kimandsally]

Corsair HX620. Made by Seasonic also. Nice modular cables so no extras floating around, and long enough as well.

Yes agreed I have the TJ07 with the Corsair and the cables are plenty long enough, don't need to add more I think everyone knows about the HX620.

[alphakry]

Yikes @ Modular. Was always skeptical of these types of PSU's due to the chance of poor connections, causing all types of variance... but i'll do some research! Thanks!

[Jigsaw]

And I thought MY water cooling was expensive... I do have a few questions though. You have a PA120.2 and a PA120.3 listed and yet a single PA120.3 would be sufficient to cool your loop. It would have saved you a fair amount of money to go with just the one rad and a Laing DDC 3.2 pump (like me ).

Why did you decide on two separate loops?
Why didn't you spend the extra ~$30 and get a second PA120.3 to match the first one?

[DarthBeavis]

And I thought MY water cooling was expensive... I do have a few questions though. You have a PA120.2 and a PA120.3 listed and yet a single PA120.3 would be sufficient to cool your loop. It would have saved you a fair amount of money to go with just the one rad and a Laing DDC 3.2 pump (like me ).

Why did you decide on two separate loops?
Why didn't you spend the extra ~$30 and get a second PA120.3 to match the first one?

Love the rig in ur quote . . .but want pix . . .

U wanna be my new friend

[alphakry]

Why did you decide on two separate loops?
Why didn't you spend the extra ~$30 and get a second PA120.3 to match the first one?

As previously indicated, i'm interested in maximum performance here. Throwing all of these parts into 1 loop will CRIPPLE the flow... not to mention add tons of heat... 2 loops will allow me to break up the primary heat dumpers (namely keeping the SLI rig from adding heat to the CPU loop) plus segment off any blocks that will impede flow...

I've ran a dual loop setup on my last 3 projects and have always been a fan of the flexibility and performance. Hell, I originally contemplated making this project a THREE loop!

Allow me to quote ranker in a previous conversation we had:

C2D 6850's run very cool with the new G0 stepping. At 4.0 GHz, which is easily obtainable with 1.5v, it'll have a heat dump of about 150W. Throw in a few chipsets a piece at 25W + 25W respectively and you can throw all of that in your PA120.2.

Each GTX puts out about 130W. Two in SLI gives you 260W worth of heat to dissipate. If you throw in your RAM, mosfets, HD's, we're safely estimate them to be 100W total. At 360W, this will all go onto you're PA120.3.

And to answer your second question, though I swore I did that on my last update -Thermochil radiators are BRICK thick... and there is only so much room in any case, even the TJ07... dual 120.3's would NOT fit in this case without serious modification or external mounting.

[Jigsaw]

Love the rig in ur quote . . .but want pix . . .

U wanna be my new friend

I will link my worklog when it gets started. Still waiting for parts to arrive fr it. It is a mod for CeBIT.

[Jigsaw]

You inspired me to go dual in my project as well lol!

[Origin_Unknown]

Very nice! I'll be sure to follow this one closely!

Me too, I've just had a glance over the thread right now. looks very promising.
Ill have to read it properly tomorrow as i really should be going to bed, work tomorrow

[NicePants42]

As previously indicated, i'm interested in maximum performance here. Throwing all of these parts into 1 loop will CRIPPLE the flow...

That's why people use two pumps in series.

not to mention add tons of heat...

And that's why people use more than one radiator.

I've ran a dual loop setup on my last 3 projects and have always been a fan of the flexibility and performance.

Flexibility - can you explain what makes dual loops more "flexible" than a single loop? With a single serial loop, the entire cooling capacity of the radiators can be used most effectively to cool any component being loaded, so I'd give the "flexibility" nod to the serial loop.
Performance - Have you ever used a single loop setup? Considering that the temperature differences created by loop design alone are very small, I would argue that the impressive performance you see with dual loops is virtually no different than you'd see with a single loop.

Allow me to quote ranker in a previous conversation we had:

C2D 6850's run very cool with the new G0 stepping. At 4.0 GHz, which is easily obtainable with 1.5v, it'll have a heat dump of about 150W. Throw in a few chipsets a piece at 25W + 25W respectively and you can throw all of that in your PA120.2.

Each GTX puts out about 130W. Two in SLI gives you 260W worth of heat to dissipate. If you throw in your RAM, mosfets, HD's, we're safely estimate them to be 100W total. At 360W, this will all go onto you're PA120.3.

Ranker's comments indicate that your setup will work because your radiators will have enough capacity to dissipate the heat that your components generate - Ranker does NOT state that dual loops are the best way to go.

I'm not trying to say that you shouldn't use two loops - I used two loops until a month ago - the reason I'm arguing with you is that you are justifying two loops with unsound or vague arguments, and possibly misunderstanding Ranker's comments. I agree with Ranker that your setup will work just fine, but I don't believe that dual loops are ever a better performing solution than a single serial loop if the same components are used.

As for a quiet PSU, I would recommend the Thermaltake Toughpower 1200W and Corsair HX620 based on hear-say, and the Cooler-Master Real Power Pro 1000W (available in modular and non-modular configs) based on Jonnyguru's review and my own personal experience.

Good luck with the project!

You inspired me to go dual in my project as well lol!

May I suggest looking through this thread? Especially Marci's comments.

[alphakry]

NicePants42 - without picking through every statement you made with a counter statement, being this probably isn't the best place for such an in depth discussion (a thread like this would be)- let me sum up my dual-loop preference like this:

I prefer the dual-loop setup for the flexibility of options that allow me to isolate different types of blocks.

For example: The FuZion CPU block has much better flow characteristics then the HDD or RAM cooling blocks... using these in the same loop will hinder the performance of the CPU block. By using multiple loops, I can segment out flow-reducing blocks from much more critical blocks and still achieve an overall silent, well performing system. Remember, I considered using THREE before deciding it was overkill - just to keep said blocks away from the two main beasts in this system - the SLI setup and the CPU.

Plus, performance proof aside, it's peace of mind to have my CPU isolated from my GPU's, considering the differences in heat output - as per the examples put forth by ranker.

Additionally, logistically speaking, having 2 radiators on the bottom of the case - it would probably be too much up and down routing of tubes to have two radiators configured into a single loop while avoiding succession.
To avoid having the immediate dissipated heat going from one component to another, I wouldn't want GPU going to CPU, nor radiator going to radiator. A solution to this would be to have a radiator between each of these components, however I expect it would be a lot of tubing without the radiators being in different places.

So, based on my pre-planning and past experience - a dual-loop setup allows me the flexibility to achieve all of these goals.

Make more sense?

[alphakry]

Also - to address your two-pump solution. I admittedly have not done a ton of research on this design idea. I've never run 2 pumps in one loop for more then failure redudancy.

However, consider a loop with only two components - a high-flow CPU block like a Storm or FuZion - and an incredibly restrictive chipset block. I can't see how adding a successive pump will help improve performance in this loop as the restriction of that chipset block is going to affect the overall flow of the loop - as it acts a bottleneck.

Unless I need to update my knowledge of viscosity and fluids. Again, I'm quite open to this discussion - just not sure if it has a home on my build log.

[NicePants42]

I prefer the dual-loop setup for the flexibility of options that allow me to isolate different types of blocks.

For example: The FuZion CPU block has much better flow characteristics then the HDD or RAM cooling blocks... so by using multiple loops, I can segment out flow-reducing blocks from much more critical blocks and still achieve an overall silent, well performing system.

Well stated. Grouping important components into a single, high-flow loop and the unimportant components in a secondary loop makes sense.

Performance proof aside, it's peace of mind to have my CPU isolated from my GPU's, considering the differences in heat output.

This is where I don't understand your logic - if you're going to use two loops, and plan to separate the CPU and GPU, then one or both of your main component loops will have to suffer the addition of unimportant blocks like RAM, NB, SB, etc. which will reduce performance. This flies in the face of your initial reasoning.

Logistically speaking, having 2 radiators on the bottom of the case - it would probably be too much up and down routing of tubes to have two radiators configured into a single loop while avoiding succession. To avoid having the immediate dissipated heat going from one component to another, I wouldn't want GPU going to CPU, nor radiator going to radiator.

You are imagining problems where there are none. Allow me to quote Marci:

Order of components in a loop makes no difference.

You're missing the fundamentals of watercooling. http://www.overclockers.com/articles1088/

So dual-loop allows me the flexibility to achieve all of these goals.

Make more sense?

All I understood was that with your current plan, you're using two loops because of inconsequential tube routing preferences rather than your perfectly reasonable initial argument of separating high flow from low flow. Not that there's anything wrong with that.

Your plan of separating important components with low restriction blocks from the rest of the crap is a great idea, but it would suggest that separating the CPU and GPU blocks from the rest of the system is more important than separating the CPU from the GPUs. I suggest either using three loops like you originally intended (1 for CPU only, 1 for GPUs only, 1 for everything else) or changing your dual loop setup (1 for CPU + GPUs, 1 for everything else) and just make sure you use the PA120.3 for the CPU+GPUs.

Also - to address your two-pump solution. I admittedly have not done a ton of research on this design idea. I've never run 2 pumps in one loop for more then failure redudancy.

Nothing wrong with redundancy.

However, consider a loop with only two components - a high-flow CPU block like a Storm or FuZion - and an incredibly restrictive chipset block. I can't see how adding a successive pump will help improve performance in this loop as the restriction of that chipset block is going to affect the overall flow of the loop - as it acts a bottleneck.

A bottleneck that the head pressure of a second pump can help to overcome. Have you, by any chance, played around with MartinM210's flow rate estimator? I heartily recommend taking it for a spin.

As for theoretical discussion and justification of your build decisions, I can think of no more appropriate place than this thread. If you'd prefer to restrict discussion to posts like "Looks great" and "Good luck", that's fine - but please be careful not to make blanket statements about best performance and dual loops that may not apply to all situations, since some people may be misinformed.

[nibble]

From an aesthetics point of view get the EK NB block, they are far nicer looking than MCW30's and performance is a non issue really, water will spank air no matter how simplistic the block design. Pressure drop is also really low on both.

[alphakry]

nibble - i agree. I have every intent of running the EK NB block - just have to make sure it will fit correctly. I got the MCW30 just to give it a try since i've heard such great things about it.

NicePants - I agree with your statement about adding the components later in the loop will just be counter intuitive to my opinions of flow restriction. As you can see, I have not come to a decision regarding the placement of these blocks. It's still in planning and for all I know, i may go back to my plan of 3 loops.

My concerns of GPU vs CPU are still valid in my opinion because I want to try and make the heat dissapation of each loop equal or at least more balanced. Having the CPU and CPU in the same loop would sky rocket the amount of heat in that loop.

[NicePants42]

My concerns of GPU vs CPU are still valid in my opinion because I want to try and make the heat dissapation of each loop equal or at least more balanced. Having the CPU and CPU in the same loop would sky rocket the amount of heat in that loop.

So the question then becomes one of flow rates vs. overall heat in the loop affecting performance. A complex question, but in this case, given that you are using a PA120.3 and a PA120.2, I would argue that 410W (maximum load) is not too much for the PA120.3, and the increased flow rates across the key components would be a good thing. Balancing the heat load is not important - what is important is knowing that you have enough dissipation capacity for a given loop.

I don't have much to back that recommendation up with, though. Maybe you could send Ranker a PM.

[BNDMOD]

Suscribed!!!! impressived...