So, thanks to my work rota totally shafting me and me having to work like every day, and basically going to be sacked if I turn up late again, I've not been going out and have been playing with hardware and trying to sort my main rig out before the new year...

Haven't really got too far, I'm gonna have to work on it through the day before I go to work meh...

Anyway, I sized up where the PA120.3 can go: Chieftec Mesh Bigtower PA120.3 test fit

and then cut the holes for the tubing inlets above the PSU (each hole takes about 3 or 4 minutes to cut)




the hole below the PSU is where the second IEC mains socket was for the Meanwell 320W PSU for the Swiftech TEC/Water block

I guess New Years Day is my next day off... so I can do a decent amount of work then...

I never realized just how big that case is until you put the PA120.3 next to it.

Too bad my sound doesn't work. It looks like there was a fair bit of explanation in that video.

hey noob, good job, finish it already.

Get to work, slag! Everyone's been waiting for this build for like a year now (more? didn't you start this when I started mine almost two years ago?)

After seeing your crazy water system layout a while ago, I'm very eager to see you start working on this again!

yep I'llhave to check it out again when my main rig is back running, right now I'm using my second rig.

Finally Mr. Slacker gets going again?

What's with just putting the hardware on the carpet? I wouldn't do that...

its in the Kitchen it looks like Vinyl flooring... But a nice mess off water blocks lol

You put carpets down on the kitchen floor in Germany? You are so weird lol

you shouldnt have carpets in the kitchen at all.

I see you found a nice spot for your reference 680i board

Meh, it looks like carpet, talk about crappy video quality

Fillports

Ok so with the last few builds in this case I've used a Swiftech MCRes-Micro reservoir, but it's kind of a pain to fill (need to swing it out and use a funnel) and not the easiest thing to drain either.

This time I have two loops, one for the CPU, NorthBridge, and RAM; and a second separate loop just for the graphics cards. I didn't really fancy the idea of two reservoirs and don't really see the point of them to be honest.

I decided to use Dangerden Fillports instead, one blue one and one black one. Blue goes with Intel, but also goes with the blue Swiftech MCR-220 radiator. Black doesn't really go with nVidia, but I'm not making a Christmas tree here lol, and black goes with the Thermochill PA120.3.




I forgot how laborious cutting holes in this case was. It really takes forever, I took a pizza out the oven before starting the second hole and it was almost cold by the time I'd finished!




So here are the fillports in place, and in the close-up you can see they will be firmly attached when the foam/rubber is squeezed tight by the locking collar.






The next logical components in the watercooling loops are the pumps. I decided to go with this orientation/position to maximise clearance from the motherboard and expansion cards.




If you didn't think the Swiftech MCP-655 and Laing D5 are the same, guess again lol. The pumps are proximally very close, but have a few mm of clearance between each other and the case chassis.




The fillports will be connected to T-pieces at the bottom of the case.




That's all I had time for this evening, but I've got all the case cutting done, so next is to line the rear entry holes with rubber C-section edge molding for the tubing to the PA120.3, and then take my main rig offline so that I can test fit the motherboard, two graphics cards with blocks, CPU with block, NB with block, and RAM with blocks, and then cut up some tubing for the two loops.

I like the fillports, cool doohickies! I still would've gone with green for the heck of it though (To accentuate it from the case/ each other). Anyways, judging by the pics, they screw on and off via hex wrench? Huh, I would've imagined that they would've been easier to access via a handled screw (car gas cap?) or something. I suppose you wouldn't velcro a hex wrench to the back of your case would you

Question, do you plan on touching up your case with some paint? Or even repainting the thing entirely? Now that I look at it, I can see there are quite a few scratches/blemishes - probably even more, as you've been working on this thing for a year... and judging by the state of your room/dwelling...

Also, now that I look at the pumps... That case seems really small now lol

Looking good though!

I got a reservoir from TribalOVERKILL that is setup for 2 separate systems but one res. His stuff is through ACRyan, maybe you might be able to find one from there. Looking good so far and I'll keep my eye open on this mod. Happy Modding to ya.

I like the almost flush appearance of the fillports... I wouldn't want a protruding cap... the "fillport" of the MCRes-Micro needs a flat screwdriver... and I very rarely change the water, maybe once or twice a year, I don't think it would kill me to find an allen key (hex wrench).

Last time I tried to touch up one of the side panels, the ensuing sanding created more patches of primer from cutting through the paint... it took forever... I need to learn how to fix scratches without sanding down a massive area. I mean, I'd *like* for the paint work to be perfect, but I don't live in a garage (sometimes I wish I did lol) and the thought of doing touch-up paint work makes me feel slightly ill.

I might just leave the paint as is, get all the watercooling done, clearcoat the panels, mount the acryclic sheet (Plexi), mount the UV CCFLs, run wiring for the 13+ fans and CCFL invertors, and call it a day lol.

If anyone wants to step in with any pearls of wisdom regarding fixing scratches on cellulose based spraypaint + cellulose clearcoat, feel free to divulge the information.

Spray paint scratches way too easily, even with plenty of clear coat. If I had known that when I repainted my Stacker, I wouldn't have bothered at all. I swear, I can put scratches in the side panels just by looking at it the wrong way.

i wonder where u got that idea of the color coded filler cap's/loop's! i was guna do something very similar bule for intel and green for Nv. As for the paint i think powder coating is the way to go! but as always nice work. enjoy the cold pizza

Tubing inlets

Ok, so the inlets holes were cut, but they weren't ready to feed the actual tubing through. You need some sort of grommet or edging to protect the tubing from the metal in order to stop it being cut by the case.

Due to the limited space, I chose U-type (C-type) rubber edge molding as you can cut it to fit and odd shape a lot easier than a plastic grommet.




First you cut at a diagonal, then cut off a piece slightly larger than the circumference of the hole so that you have some extra overlapping. It may be difficult to get the rubber molding to slip over the metal, so run it under a hot tap to make it more flexible. I ran a small allen key along the slit to encourage it to open up. When you've got it all the way around, mark the end, then cut a diagonal (the same orientation as last time) to finish the custom grommet. Make sure you don't cut too much off and you end up with this:





And repeat for the other side.




Then a little trimming on the inside was necessary for clearance of the PSU mounting plate. Note that this is highly undesirable as the molding is not holding on from both sides now, and pushing tubing outwards pushes the molding out of the hole unless you hold it while feeding tubing through. Due to the location of the radiator however, this was the only option.

Graphics cards

Well after the highly exciting tubing holes, it was time to move onto something else. I already had a BFG 8800GTS 640MB OC with a DangerDen NV-88GTS block that I fitted myself. On deciding to go for SLI, I bought a BFG 8800GTS 640MB OC WC (Watercooled) which comes prefitted with the same DangerDen block, but branded with the BFG logo.




Peering between the waterblock and the graphics card memory modules, I saw what looked like white thermal paste. I wasn't really too thrilled about this, could it be Ceramique? I didn't want to guess, I had to know.




What the hell is this? Looks like dried up cheap standard silicone paste! Just look at the state of it and the appalling coverage.




You can see the GPU is an A2 revision. My first card is an A3 which is the later, cooler running, and often highly overclockable type (I've reached 702 / 2088 with it). The A2 revisions run hotter and generally don't OC as high, but maybe that is because of the heat difference, so I'm hoping the water will bring the A2 at least near the A3 OC. I went for Arctic Silver, this is how it looked before installing the block.

Looks pretty good, that stock paste looked nasty though...

CPU and Northbridge blocks

The Dangerden 975x chipset block (Maze4) fits the odd hook mount points of the DFI 680i LT. So that was easy.




Installed the Scythe Universal Backplate and the DTek mounting screws ready for the DTek Fuzion block.




But first I had a little surprise for the Fuzion, a washer and an accelerator nozzle. The washer seals the slight gap inside the block that causes some of the inlet flow to channel straight out to the outlet. So as well as this problem being corrected, the washer bows the base slightly. Finally the accelerator nozzle (the Quad type) splits the flow to each core and increases pressure.

Here's the block apart, the washer (with peelable adhesive backing) resting up against the mid-section, and the various accelerator nozzles.




The quad nozzle, and in place.






Make sure you reassemble the block so that the flow divider of the quad nozzle accelerator is oriented to split to each core.




Looking at the CPU with the lever at the bottom, the two parts of the quad core are up and down, not side to side.


(photo from nfm @ XS)


CPU ready to go. RAM ready too.




CPU to Northbridge tubing cut to size.




At this point, I had to mount the motherboard in the case to get an idea of the position of the pumps in relation to the graphics cards, motherboard, etc.








Room enough to connect the monitor and audio.

CMOS reset switch

Well, this is hardly an original idea, but something I've wanted to make for a long time. I opted for a toggle switch that has two positions, and is spring loaded so it returns to the original position. So, to reset the CMOS, I can turn off the PSU, hold the switch in the other position, then let go. It's just impossible to get to the CMOS reset jumper behind two pumps, a soundcard, tubing, etc etc.

I adapted a USB internal case cable and trimmed the black short. So I had white as the common (pin 2), green as normal (pin 1), and red as reset (pin 3). This is how the connector was already wired up, so I didn't have to rearrange them.








I drilled a hole for the switch on one of the cross-brace parts of the chassis that runs above the motherboard. Easy enough for me to reach, but out of the way of inquisitive noobs (so they don't reset the CMOS while the PC is running!)


More tubing!

After that little diversion, it was time to finish cutting all the tubing.








Time for some water.




I realised that reservoirs help a lot to bleed the air out of the system, and T-lines really don't do much at all. With the Swiftech MCRes-Micro reservoir, I could pretty much just run the system for 10 minutes and it would bleed the air by itself.

After about an hour of stop-starting the pumps, I still had air in the system.




Fortunately, it was only a fine stream of bubbles by this point. I left it running over night and it had bled itself of air when I woke up.

All aboard

With the waterloops bled and leak tested, it was time to hook everything up and install all the hardware.






Obviously I'm not going to use fan power adapters and splitters eventually, but for now, it works. And yes, I do have 9 harddrives lol.

My case weighs an absolute ton now... about as heavy as my 22" CRT. Hopefully one of my housemates has bathroom scales so I can weigh it when it's done haha.

Turns out that using a fan as a radiator mount doesn't work! The fan housing warps under the loaded radiator weight and the fan blades touch the sides and so the fan can't turn. HMM!

One of the things I never used from the Swiftech H20-Apex (Storm) kit is the Radbox. Right now it seemed like the option I should have originally taken. I mounted the open part on the case so that the exhaust fan can push the air out of the case. The partially open part I put against the radiator as the fins aren't flush with the radiator surround, so the Radbox isn't really stopping airflow.

This is with the Panaflo M1A serving as the radiator mount. Note the 80mm fan inside the case, stood on a small box. This is extracting air from a Thermalright HR-09S Type 3 mosfet sink.




I propped the radiator up with a couple of boxes on top of a speaker stand. Also, if you are thinking about the Thermalright mosfet sink, I had to bend the heatpipe towards it being a HR-09U so that it could clear the EPS-12V cable enough so that the mosfet sink wouldn't get tilted from being pushed up from underneath.




So I took the fan off and mounting the plate part to the radiator using some rubber washers to substitute the thickness of a fan housing. This is important so that the radiator screws don't puncture the fins or water channels.




Then I mounted the fan internally and spaced the Radbox from the case with some more rubber washers for noise isolation and also to match the 38mm thickness of the fan.




So how are the temps?

CPU: 34.5°C idle, 47.3°C load
GPU (A3): 37°C idle
GPU (A2): 46°C idle


A nice drop in temps for the CPU compared to air. I've got to redo the OC as I was using Geil Ultra before, so now I've got to rework the settings on the OCZ FlexXLC RAM and have a go at upping the CPU/FSB too. And also OCing the graphics cards in SLI...

But first, I'm going to break the seal on Crysis and Unreal Tournament 3 and enjoy the fruits of my labour!

Whoa!!!!! :bows many times: you are the king! I've said it once and I'll say it again! You are the king! You should make that cmos switch thing into and occ guide! Like your volt mods etc. I think I'm going to try it out I have one of those wires kicking around so I'll just have to get a switch and wire it.

Impressive as usual Mr. nrg.

So, are we EVER gonna see that case closed with all parts attached?

Good stuff, but how are you gonna mount the rad for the final version?

the rad is mounted using a Swiftech Radbox... I can't see any problems with continuing to use it

ahh, didn't see that bit. It's too early in the morning

Not bad Nrg. i can See that Cmos switch Being used a lot for older mobo's. And i just so happen to have the same switch

But again Nice job

:Edit

But i guess you didnt no that Dfi already fixed that Problem Of reseting the Cmos

Screen shot of the DFI 680 manual

Nice! It's good to see it all coming together.

I didn't know when I was making the switch, I read the manual cover to cover about 10 minutes after lol

I might use it for the Safe Mode boot (allows you to enter bios without clearing values)

Beats playing hide and seak with the jumper after you droped it tryin to pull it out with Tweazers lol


that safe mode boot is quite handy. after it fails to boot about 10 times it safe boots so your little switch might speed the Ocing up a bit.

Looking good, and wow. I didn't know there was such a difference between in the A2 and A3 in terms of temps - that's rather astonishing

Oh and just a suggestion on the CMOS reset toggle. Have you considered getting a micro or even surface mount 12v or 5v relay (SPDT) and sticking the "off" connections to the center pin (Pin 2) of the CMOS cable and hooking up the solenoid to a spare molex? That way when the power is on, the relay is on and the connection on the pin 2 is broken - protecting from accidental CMOS resetting whilst powered on?

This would have been so useful with my old 680i before it went and pulled a --/C1 on me...

@firky, I don't know about you, but I hate taking my case apart, getting down on my stomach every 5 minutes to try and reset the CMOS without cutting myself while overclocking. I'd much rather have a switch on or in my optical drive bay or something.

don't pin1 and pin2 have to be connected when the board is running?

anyway, I've taken the switch off the CMOS reset header... DFI has that "EZ clear" that allows a CMOS reset by doing a cunning Power and Reset combo which you can do with the regular power/reset buttons on the case...

so the switch is now hooked up to the Safe Mode header which kinda does a similar thing to a CMOS reset in that it gets the board to boot at stock, but the values in the BIOS are the same as they were when you last saved it...

I'm making progress towards 3826MHz (previous 3400 MHz)... I think everyone's going to be impressed with the temps at 1.616v quad core load

Ah well, if the safe mode header is the same way as the CMOS header.

You can get a DPDT relay instead and on the other side, have pin 2 (pulled before the relay connection) and pin 1 together on the "on" side. So when you turn the thing on, it'll disconnect the switch, and connect pins 1 and 2.

ah I see... that might be something I'd do when I have kids my housemates don't mess with my computers lol

so, some OC progress and temps (especially for you Frank, RESULTS! )

yeah so times suck cos I'm running the RAM unlinked and slow... just CPU stability for now:

SPi 1M @ 3826 MHz, 1.600v x 102%


SPi 4M @ 3826 MHz, 1.600v x 103%


SPi 8M @ 3826 MHz, 1.600v x 104.5%


SPi 16M @ 3826 MHz, 1.600v x 104.5%


Load temps during SPi 4M run, 1.600v x 104.5%


Ambient temps, I dunno, I'd guess 25-26°C because it's warm enough for me to not be wearing any pants

SPi 32M is running now, if that passes then next is OCCT 1/2 hour, then all 3DMarks I guess, in case there is some 3D stability thing for the CPU somehow (like Micron RAM in the past)

Bloody hell, I should get some WCing VERY nice clocks, keep us posted! I am very interested in your BIOS voltage settings though (chipset, FSB etc), as I haven't quite figured out the voltages on my Blood Iron yet

Ha ha how ture is that...... I use EZ clear the power butten and the rest butten are realy easy to get to so i dont need to cut my self or even get out of my chair!! lol overclocking has now become even more fun

Nice work Steve. I will have to get that PA120.3 and DTeK fusion when I get back from CES. I found some settings that work well for 3.9GHz at 1,55 v on mine. I just have to wait to get another Rad before I push any higher. 71C is a little much

oh man this is so annoying... SPi32M passed at 3826 MHz at the same settings, but OCCT fails just 4 mins or so into the 1/2 hour test... I tried reducing the FSB down from 1700QDR (425) down to 1656QDR (414) in 4 MHz steps... 1684-1696 (421-424) was about 30 seconds more stable, not really conclusive enough... 1668-1680 (417-420) didn't load the OS, 1658-1664 (414-416) wouldn't even POST... this is with bios F_061... different bios versions have different FSB holes... bios 10/16 (DA18) doesn't actually have the capability of reducing the multi even though it's an option in the bios, the multi always ends up as 9x, this is the same for other earlier bios versions too including 4CTST521 which is supposed to be really good... I might try that for 9x425=3825 and see if it's any more stable than 10/16 (DA18)

clocked back down to 8x425 = 3400 MHz, ram at 425 4-4-4-12-1T... running Prime95 (Blend) overnight to see if the OCZ FlexXLC can run 1T like the Geil Ultra... I might settle for 3400 if 3825 really isn't going to happen... but knowing it's near stable is just so frustrating... I might have to trying tweaking some of the 12 voltage controls and see if I can gain some stability from the VTT/GTL/NB/SB voltages... LDT (HT Link) is best at 1.37v, every other value is less stable

time for bed now though lol

EDIT:

lol, not time for bed, couldn't settle for the same OC I had on air... kicked it up to 8x437.5 = 3500 MHz, ram @ 437.5 4-4-4-12-1T, running Prime95 Blend again, lowered the vCore to 1.550v x 104.5% which shows as 1.568v in CPU-Z

Load temps during OCCT run at 1.600v x 106%


SPi 32M @ 3826 MHz, 1.600 x 104.5%

hey nrg, would this switch work for one of those cmos switch mods? if not can you post a link for one that would? thanks!

It looks to me like That Switch is just ON or OFF for the Cmos you need a Switch that is ON or ON the middel pin is common and is switched to meet the pin on the left or the right that switch just.


If that switch works like my drawing (1) it wont work


if the switch works like my drawing (2) you will have to join the pins on the right hand side together to make a common and when the switch is held down it will clear the Cmos and when the Switch is Released it joins back as it should be.


and the switch Nrg used is like the pic (3) when the switch is joined one way it links the common and a pin and will clear the Cmos and when its linked the other way pic (3a) its links common and the other pin it will just be as it should be.






I hope that makes some sence to you...

and this would work type "K"


Edit: you never asked me but that my answer

You need a SPDT with two connectors to switch to. Doesn't matter power wise, a CMOS header doesn't stress the switch out any...

See this site for explanations http://www.kpsec.freeuk.com/components/switch.htm

Here's one the will work just fine. Toggle Switches are your friend as they usually can be mounted through a hole, and screwed in easily. Provided of course the mounting medium is thin enough, a optical bay cover, or the back of a case will work fine - plus they're cool.

Just depends on your preference.

http://www.goldmine-elec-products.com/prod...sp?number=G1485