Worklog: Project Thief - Fully Watercooled Dual PC

[stren]

Been slow on the updates because work has been kicking my behind. Anyway the 580 3gb cards came in. I know some of you may say why not get 680's (or even 670's), well these were a lot cheaper (awesome 2nd hand deal) and will perform plenty well enough to max out 2560x1600 @ 60Hz. With my 480's I never had a problem running out of processing power, but I did run out of vram, so I think these will be plenty until I upgrade the monitors Once I upgrade then I'm sure I can justify a few GK110s

As you can see two of the backplates are 480 ones:

So I had to switch the cards around so that the 580 one was at the bottom of the stack. I had to take them apart anyway to check for nickel problems and gunk. I'd be more concerned if I wasn't suspecting that I'll change GPUs again before the build is finished. Here they are rearranged and plugged in:

The 120 radiator is for the motherboard only loop:

I'm still working out how to route the tubing for optimal performance without making it look like a mess - should I come out the side or go out the top?

Although the board has two loops on it, it will really look like it has three as there will be a 560 rad in between the cpu and gpu:

Motherboard loop routed with temporary fittings - I thought this would look better. I don't really like the way it looks - too busy and crowded, I need to work out a way to keep the tube out the way more. Maybe a memory block would help:

 

[PCTC2]

Yeah. Work can definitely take away from the fun time.

Glad to see an update. Looking good. I think we all wait around for an update to live vicariously through you stren, since this is the dream build we all wish we could build.

(Though I guess I can't complain with all my computers either.)

 

[stren]

Yeah. Work can definitely take away from the fun time.

Glad to see an update. Looking good. I think we all wait around for an update to live vicariously through you stren, since this is the dream build we all wish we could build.

(Though I guess I can't complain with all my computers either.)

Haha yeah you certainly have a collection!

 

[OVerLoRDI]

stren works on his amazing computer and I work on my car

Parallel GPU links can make tubing so much easier. Like in your project you have in and out on the bottom of the cards. I did something similar, had in and out on the top, saved my bacon when I was lacking a 90 degree fitting for clearing my PSU.

Great work and pics. As ever, excited to see the final outcome.

 

[stren]

Ok so a quick recap -not much had been done the last few weeks as I was busy with work and then went on a brief vacation. Now I'm back the build will continue, however I'll also be doing various things in between including a review of Rich Chomiczewski aka Spotswood's new tech station. I'll also be doing a CPU waterblock round up which will include:

DT Sniper
DT 5Noz
EK Supremacy
EK Supreme HF
XSPC Raystorm
XSPC Rasa
MIPS Iceforce HF

Thanks to DT, EK and Indigo Xtreme for sponsoring the blocks and TIM, thanks to Rich for sponsoring the tech station which will be used as the test bench :thumb::thumb:

So without further ado, let's get back to some pictures

The tech station arrived from Rich in a fairly small box:

However there was still plenty of room for packing:

The case comes "mostly" assembled, here's the motherboard tray:

that mounts above the base which includes mounts for DVD drives, HDs and SSDs:

Here's the rest of the frame:

And all the screws and accessories, interestingly Rich includes an allen driver, as well as an allen key, and a torx key also

The frame makes use of extruded aluminum with a pattern that makes it easy to use screws and bolts/nuts to easily customize the setup. Here's one vertical strut getting slid onto the base:

There is then a hole allowing that screw to be tightened. Before you know it all 4 are up:

 

[stren]

You can then add some feet:

Then attach the PCI card support to the motherboard tray and then attach both to the frame:

Then the PSU can be attached with two small plates:

Although this is solid enough for benching (and the PSU does not move), a 4 screw mount would be more solid if you were moving the case as there is potential for the PSU to bend those small plates. The hard drives, dvd and SSDs use small rubber grommets that get screwed into the base of each device:

You can then slide these on the same extruded frames which have pre drilled access holes to make it easy to swap them in and out:

More to come tomorrow!

 

[stren]

I also had ordered a crystalfontz setup so that I could monitor temps accurately, although the aquaero is more of a system, it doesn't support the dallas temp probes that can be calibrated to <0.1C accuracy. This setup should let me do that. The package arrived while I was away:

More packaging:

Taking it out of the bag - it's the 635 module with the 4 line LCD:

They'd pre fitted the SCAB module which enables the temp sensors to be hooked up:

Not quite sure what was with the stray spray paint:

Decided to mount it on the top of the test bench:

However the right side couldn't be screwed down so I added some support so I could at least push the buttons if needed:

The test bench came with extra extrusion pieces so you can add on extra devices, I decided to use my RD30 pump for the cpu block test:

 

[stren]

I'd also ordered a King Instruments flow meter (similar to the one Martin uses):

It's more accurate and less restrictive than the impeller types. However it's large - here it is zip tied to the case next to the 400mm EK res:

And now the final setup waiting for the 2nd R4E/3930K to arrive:

 

[stren]

Time for a belated update:

I used the nice quick release thumb screws to remove the EX 560 radiator from the TX10. I used two extra pieces from the tech station to build a support beam for it:

Then used some of the accessories to grip it either side so it was stable, but yet also quick to remove:

Then the 2nd R4E board came in:

And a 2nd cpu:

 

[stren]

Installed the motherboard - I learned I should loosen the standoffs from the tray a little as they holes have enough movement that the standoffs may be mispositioned. After screwing everything in I want back and tightened it. Then added the cpu:

Then some ram:

Then removed the ram in order to fit the DT 5Noz using the Sniper thumbnuts:

Then put the ram back in:

Added a gpu (9800 gt is enough for the cpu test) and that's it for now!

 

[OVerLoRDI]

I love those motherboards. Keep up the good work!

 

[stren]

I love those motherboards. Keep up the good work!

Time for more:


Photo update:

Box was huge, almost as wide as a 560 rad! The dac/amp is large too, much bigger than it seemed from photos:

Deep too:

Time to get the test bench up and running:

The flow meter needed some barbs, I got these 1/2" brass barbs from home depot:

I forgot that with no o-ring, you need teflon tape, that would bite me later. Clamped it down:

Matching one for the bottom input port:

Hooked up the Iwaki RD30 pump fed directly from the 400mm reservoir:

 

[stren]

I had the return come in the bottom and used the longest internal tube to direct the flow to the top of the res in order to help bleeding. Then connected the CPU:

Hooked up the rad, and ran out of clear tube, had to use some of the leftover pink:

Time to fill it up:

2.55GPM while bleeding with the Iwaki at 19V, at 29V it was pushing a smidge over 3.5GPM:

Ready to start overclocking:

 

[Aikouka]

As a bit of a side note, that CaseLabs store is rather interesting. I've been using my old Lian Li PC V2000B case as a server case for awhile, but given my desire to space my hard drives with a buffer between each, I don't have a terribly high hard drive capacity. I think I can hold ~6 right now.

I was looking at their cheaper/smaller options, and their site says the M8 can hold up to 24 with optional bays!

I may be in heaven!

But my wallet is going to be in hell. :sneaky:

 

[Schmide]

That vertical tube reminds me of the old school bong. Wonder if anyone else here remembers?

There was a guy. Pepsi over at overclockers.com that built vortex watercooling bongs similar to this.

But 6-7' tall 3-4' diameter.

Sadly the chances of finding any picture about it are null and void as he was contracted to build one for one of the guys at OCers and failed to produce. He was subsequently labeled a troll and thief and erased from history.

 

[stren]

As a bit of a side note, that CaseLabs store is rather interesting. I've been using my old Lian Li PC V2000B case as a server case for awhile, but given my desire to space my hard drives with a buffer between each, I don't have a terribly high hard drive capacity. I think I can hold ~6 right now.

I was looking at their cheaper/smaller options, and their site says the M8 can hold up to 24 with optional bays!

I may be in heaven!

But my wallet is going to be in hell. :sneaky:

Haha yeah CaseLabs really know how to support hard drives. My case can fit something like 128 if you really want to. The M8 is a great case though, heavily recommended :thumbsup:

That vertical tube reminds me of the old school bong. Wonder if anyone else here remembers?

There was a guy. Pepsi over at overclockers.com that built vortex watercooling bongs similar to this.

But 6-7' tall 3-4' diameter.

Sadly the chances of finding any picture about it are null and void as he was contracted to build one for one of the guys at OCers and failed to produce. He was subsequently labeled a troll and thief and erased from history.

Yikes! Yeah I've seen some bong coolers around the past few years, not a vortex style one. I thought about a bong, but it's dry enough I feel like I'd spend too much buying distilled water haha.

 

[ViviTheMage]

awesome build, wish I had sponsors so I could build something like this up

 

[shenoavox]

i must thank you for this thread. I just read through this and I leanred a whole new plethora of information

 

[stren]

i must thank you for this thread. I just read through this and I leanred a whole new plethora of information

Alright - while I haven't made progress on the project, I have made progress on the CPU water block review. Here's how I did the temperature sensors (some of these photos are repeats but are included for completeness as a how-to guide):

For this I was following Rubidium's excellent guide with some modifications. His guide is a must read if you're planning to copy this.

Dallas one wire temperature sensors are much more accurate than the standard 2 wire thermistor type sensors that are used in current water temperature probes. However they need to be water proofed and mounted in the loop which means some work on my part.

So here's what I ordered - the temperature sensors and the crystalfontz USB interface- there are many options here, the one I chose was not the cheapest but gives a good amount of screen space:

The package came in:

In addition to the main screen display, you'll need a SCAB module that enables the sensor monitoring, plus a USB cable, plus the temperature sensors too.

Here's the break down of how to plug it in

We also needed some other parts to build the temperature sensors into, the bitspower rotary q fitting is very flexible but has more ports than you'll probably need so you'll need to order some stop fittings or a different t adapter:

Here's what I ordered from mcmaster-carr at the top and at the bottom is a sample order for enough parts for one temp sensor from sidewinder (I had some parts already so only ordered what I needed). If you clean the stainless steel syringe well you may be able to reuse it, I ordered two just in case. The parts arrived:

 

[stren]

We also needed some other parts to build the temperature sensors into, the bitspower rotary q fitting is very flexible but has more ports than you'll probably need so you'll need to order some stop fittings or a different t adapter:

Here's what I ordered from mcmaster-carr at the top and at the bottom is a sample order for enough parts for one temp sensor from sidewinder (I had some parts already so only ordered what I needed). If you clean the stainless steel syringe well you may be able to reuse it, I ordered two just in case. The parts arrived:

I ordered the brass metal brush, I probably should have ordered something else though - maybe a steel one? Eitherway it seems to have worked for me. Brass barbs make life easier because otherwise you have to remove the nickel plating from the fitting.

First off rough up the surface of the temperature sensor, I used 220 grit regular sand paper:

Then rough up the interior surface of the barb using the metal brush (fit it in a drill and spend a few minutes doing the old in and out). I also sanded the end of the barb and it's chamfer with 220 grit. I then cut some small neoprene washers to fit around the temp sensors and hold it centered in the barb:

I bought the smallest I could find from home depot. If you can find something that's a perfect fit (any rubber grommet or washer that fits would be fine) then it would be better. You want to make the fit as perfect as possible otherwise the glue will leak past and it could ruin all your efforts. Fit a 2nd washer to keep the wire centered at the back of the barb. Now we're going to mix up the glue. Get your glue mats together:

I found about a quarter of the 1.7oz size would fill 2-3 3cc syringes. One 3cc syringe's worth is probably enough for 2 fittings. I mixed it in a disposable plastic shot glass that I had laying around.

 

[stren]

Then take the plastic syringe, dip the end in and start sucking up the glue:

Stop when you have enough and wipe off the tip:

Attach the needle by screwing it on while pushing it in. Be careful to catch the glue that comes out while you do this:

Fill up around the temperature sensor slowly avoiding making any bubbles. You will want to fill up just above the edge of the fitting as the glue will shrink down a bit. As the glue will run you need to secure it so that it doesn't run out. I put mine loosely in a vice:

Here they are after drying for 24 hours:

You can see the glue is lower than the edge of the fitting. If your washer/grommet wasn't a good fit, a good amount of glue may have leaked past. If it's too low you can always top off later.

 

[stren]

Now you're going to remove the grommet/washer and fill the other side of the fitting all the way to the top. Again take it slowly and avoid bubbles, keep going around and around the wires and try and keep the wires centered. Leave to dry in an upright position so that it doesn't run out the side:

All done:

You can then mount them in the bitspower q fitting:

I then added on my VL4N QDC's (along with g1/4 g3/8 adapters):

You can then add stop fittings and mount to your favorite waterblock and then leak test:

Looks kinda frankenstein, but hopefully the performance will be worthwhile :thumb:

I also made a video of the test setup here:

http://youtu.be/eO0ec6NLfa8

 

[Rubycon]

Wouldn't have been easier to buy this instead?

http://www.sidewindercomputers.com/big1tesestfi.html

That TO-92 package (sensor) along with the epoxy seal has un-tested durability. Honestly it would probably be fine, but it should be tested rigorously to be sure.

 

[stren]

Wouldn't have been easier to buy this instead?

http://www.sidewindercomputers.com/big1tesestfi.html

That TO-220 package (sensor) along with the epoxy seal has un-tested durability. Honestly it would probably be fine, but it should be tested rigorously to be sure.

Yes it would but those are very inaccurate. For normal loop purposes those would be fine, but for a detailed block performance review the dallas ones are way superior and considered the standard. Plus you can daisy chain up to 32 of them or something. Hopefully the seal will prove strong. Rubidium made his years ago and never posted to say there was an issue (though that's not proof of anything). Skinnee used the same sensors with an even shadier water proofing method for years, so we'll see.

Fingers crossed!

 

[Rubycon]

Themistor sensor is plenty accurate the problem is what most people connect them to.

If you need accuracy there's Cole Parmer, Omega, and Agilent instruments with 8 significant digit reporting but that gets expensive.

Just make sure you have real proof of cooling and not just depend on pump RPM sensing to protect your system.