The making of STRYKER II

[CC_Pirate]

[Flushing the system]

I got all the connections clamped and ready to rock.
Because I am using a STORM in thi system, I decided to flush it before filling it. This because the STORM has "mini jets" inside it that can easily get clogged. So I decided to flush the system with spring water in order to make sure its clean.

Here are the things I used for the flushing:

I put the case upside down and flipped the reservoir over. Then I connected a new pice of tubing to the intake of the pump. Then I let the hose coming from the outlet of the pump hang down in the bucket. Then I poured water into the hose connected to the pump intake, and once there was a decent amount of water in the tubes I turned the pump on. Then I ran one and a half gallon of springwater through the system.

I got the PSU running thanks to the no-mobo powerup-mod:

No the system is clean and ready to be filled. Coming up next.

 

[CC_Pirate]

[Filling the system]

Now I'm testrunning the system for 24 hours to make sure there are no leaks.

I started out mixing one gallon of distilled water with a bottle of Cool-Cases-USA UV dye. I filtered the UV dye to make sure no small particles got into the loop and clogged the STORM.

Pics speak for themselves:

Overall, the process wasn't as tricky as I had expected. And the D5 pump is impressive, so efficient and so silent.
The water didn't turn out as "blue" as I wanted to, so I'll be adding a little more UV dye later when I get a hold of some.

Hopefully, there is not a drop of water in the case tommorrow morning. Then I'll throw in the hardware, and the action really starts.

 

[Volucris]

Looks excellent. Keep up the good work and batten down the hatches capitan!

 

[MantisIT]

That's quite a soon to be rig ya got there, top of the line stuff!

Anyhoo, was wondering if that WC System came as a kit or did you order everything seperately? I wanted to get something just like the one you have (I like the way it looks / performs) Thanks!

 

[qbackin]

Its the swiftech Apex kit.
Very kick ass kit.

 

[Pkirk618]

ditto...I like what you've done. I wish I was WC savy to build like that...

...in the mean time, I watch and learn.

NJ!

 

[drizzt81]

why would you 'flush' a case with spring water? For all i know the price of Distilled and Spring water is the same (at least here in PA) and the distilled water should have less 'junk' in it that the spring water, considering that there are no minerals.

 

[CC_Pirate]

thanks for all compliments fellas.

yeah drizzt81, I bought the springwater at the gas station at night. It was way more readily available than distilled, thats why I used it. It doesnt really matter tho, in terms of cleanness/bacteria. I mean all I did was flush

 

[jros]

you got some nice stuff in that case!

 

[drizzt81]

yeah drizzt81, I bought the springwater at the gas station at night. It was way more readily available than distilled, thats why I used it. It doesnt really matter tho, in terms of cleanness/bacteria. I mean all I did was flush

it's was not meant as an insult/ attack, i was just curious... appreciate the explanation.

 

[Manuelj43]

Good pace of work you got goin there, really. Gives time for improvements. I hate having to rush when building a PC, customers aswell as time can be a problem. I would have gotten all matching knobs for the front, imo.

Also, I would have suggested the Asus Extreme N7800GTX TOP, still a good decision on the MSI.

Props for the work. I'm adding this thread to my favorites list, I'll be watching for new pics.

 

[CC_Pirate]

Good pace of work you got goin there, really. Gives time for improvements. I hate having to rush when building a PC, customers aswell as time can be a problem. I would have gotten all matching knobs for the front, imo.

Also, I would have suggested the Asus Extreme N7800GTX TOP, still a good decision on the MSI.

Props for the work. I'm adding this thread to my favorites list, I'll be watching for new pics.

Thanks for the compliments. I was considering the Asus TOP, however, I realized that all 7800GTXs are very similar; they deviate very little from the nVidia reference design. All you really pay for with the TOP card is that huge cooler, which I would have taken off anyway. I think the MSI and the Asus would have gotten pretty much the same OC results.

 

[CC_Pirate]

[Installing the hardware]

I've made quite a bit of progress on this rig during the last few days, but I haven't had time to update this worklog. Anyway, here we go.

As the watercooling setup proved leakproof, it was time to slap in the hardware. Not to much to say about that, pretty straight forward, although I did spend quite a bit of time on the cable routing and management, trying to make it look decent.
Remember, its not all done yet, but here are some pics:

I got a little kink between right after the outlet of the GPU block, the tubing here should have been a little longer . Fortunately it was easily fixed using two cable ties:

 

[CC_Pirate]

[Preparing the lights]

As you might have seen earlier in this thread, I have some serious lighting planned for this case.
I plan to have the following lightning in the following locations:
- 1x12" BLUE + 1x12" UV cathodes at the top of the motherboard section of the case
- 1x12" BLUE + 1x12" UV cathodes at the right side of the motherboard section of the case
- 1x12" BLUE + 1x12" UV cathodes at the bottom of the motherboard section of the case
- 1x12" BLUE cathode between the radiator and the fan filters at the bottom of the bottom section of the case
- 2x4" BLUE + 2x4" UV cathodes above the reservoir, underneath the CD-ROM

A problem I ran into is that in order to fit the cathodes in many of these places, I need the cathode only, that is, the thin cathode only and not the protecting tube around it. The tubes are too thick.

So I had to find a way to fit them where the space was to small, and the answer was obvious: Extracd the cathodes from the tubes.
This is easier said than done. At first I tried (stupidly enough) to simply crack the caps off the ends of the tube using a pair of wrenches. Lets just say it didnt go too well:

I found out that the best way to extract them was to melt the top of the tubes using a heatgun and pull the cathodes out of there. Pics:

These damn cathodes are VERY FRAGILE. I would be lying if I said I did not brake quite a few of them during the process of extracting, sleeving, and installing.

Then I had to splice some wires to make them long enough.

And finally I hat to sleeve them to make them look neat. I sleeved some fans too while I was at it:

When the cathodes were prepared I had to get the inverters ready. I labeled all of them, and mounted them all in a 5.25" bay with the included velcro:

The fans are gonna be controlled by a rheobus fan controller I had lying around. I hooked the inverters up to it like this:

Then I slapped it in there, and hooked up all lights and inverters:

Lets hope it all just works (probably not )

 

[Volucris]

That fan controller looks like a 4 year old in China made it. Horrible, the heatsinks are what really throw me off.

 

[Manuelj43]

I think you should have gotten dip switches for the front instead, but the knobs'll do.

 

[CC_Pirate]

Yes, its exactly four years old. And of course its made in China, what do you think, they make 'em in Hollywod these days? That controller is perfect for the job of controlling the lights, and honestly, I don't care what the heatsinks on the PCB of my fancontroller looks like.
BTW, the design of that one is exactly the same as the one I bought at FrozenCPU three weeks ago. It just supports 3W less epr channel.

 

[Volucris]

Yes, its exactly four years old. And of course its made in China, what do you think, they make 'em in Hollywod these days? That controller is perfect for the job of controlling the lights, and honestly, I don't care what the heatsinks on the PCB of my fancontroller looks like.
BTW, the design of that one is exactly the same as the one I bought at FrozenCPU three weeks ago. It just supports 3W less epr channel.

Eh, if it works, use it.

 

[Manuelj43]

Any update?

 

[Pkirk618]

I like what you've done. I'm a bit concerned about all the splicing and areas of those boxes for the lights. I think, and I'm no electrician, but it seems dangerous. Then again, you wouldnt be hard is you werent

 

[drizzt81]

I like what you've done. I'm a bit concerned about all the splicing and areas of those boxes for the lights. I think, and I'm no electrician, but it seems dangerous. Then again, you wouldnt be hard is you werent

I am no expert either, but if CCFLs are similar to fluorescent (sp?) lights, they should not be put on dimmer circuits. It appears to me that you are using the rheobus as a dimmer for the inverters, which isn't a great idea either.

Then again, someone with more electrical knowledge than me may clear this up, since I am not sure about it.

 

[FallenOne]

yah heard the same... plus i heard that it would not be that good if you extend the cables from the inverter to the cathodes -> thus leading to loss of brightness... but that's just a rumor i heard but the dimming-control thing is a bad idea because of the... uhm.. (trying to direct-translate from german to english) backward-current running through there...and the inverters might need a little room to "breathe".. you know.. there are holes in the inverter-case for a reason *g*

just my 0.02$

Falli

 

[EzekialRage]

just don't be too suprised if you hear a loud pop when you power them on for the first time

GL, and nice hardware

 

[drizzt81]

maybe the lack of updates here by CC_Pirate tells us something about the success (or lack thereof) of the project

 

[CC_Pirate]

heh guys Ive been a little busy for the past week.

The lights worked somewhat. The dimmer works, no loud pops, but not all lights are running on full brightness. I am unsure why, but its either poor "splicing" jobs or too long cables between the lights and the inverters. Is it really true that longer wires between inverter and lights equals less brightness?

Anyways, I think I am going to order a set of these and use them instead of the rheobus. I was just trying to cut some cost and reduce the work required but what ever.

thanks for all tips on the lights, I really need it.

 

[MONST3R]

Looking awesome so far, keep it up

 

[FallenOne]

i'd say it's because of the lengthened wires... i mean.. if it would not matter how long the cables are.. don't you think they would sell those with much longer cables from rod to inverter? (that would be much cooler cause i have a bigtower and problems with to short wires )

cya and keep it up
Falli

 

[CC_Pirate]

[Fixing the lights]

When installing the lights, I ran into some problems. First of all, when extracted, those damn cathodes are incredibly fragile. Therefore, I eventually realized that I had to leave them in the protective tubes. So now the only light that is extracted from the tube is the blue 12" underneath the radiator, and I had to change my plans regarding location of the different tubes.

Another problem that people helped me figure out in this thread is that splicing the wire between the inverter and the CCFLs to make it longer is a bad idea as it will result in a loss of brightness. Because of this, all the inverters are no longer gathered in that one 5.25" bay, but put in various locations in order to be close to the CCFLs.

Some people said earlier in this thread that using a rheobus to control the lights was not a very good idea. I thought about it and considered using some SPST switches instead. However, after doing some testing I must say that the rheobus works perfectly. It lets me dim the lights from about 5% to 100% brightness instantly. So now all the inverters are hooked up to a rheobus.

Let's take a look at how I ended up placing the lights and inverters:

Here is just a little preview of the lights in action. More UV madness to come.

 

[CC_Pirate]

[Heat issues]


I ran into a few issues regarding heat when building this rig. Here they are in detail:

Problem 1. Northbridge temp
Originally I had planned to cool the chipset passively using the Zalman heatsink. Quickly after booting up the computer I realized that this was not sufficient cooling. At stock speeds, the northbridge temperature reached about 54C when idle, not leaving me much headroom for overclocking.

Here is a pic of the original setup:

The solution, of course, was simple and involved putting a fan on the heatsink. I ordered a 5000rpm 40mm fan and mounted it to blow air across the heatsink from the side.
Pics (sorry about the bad one):

Problem solved. The NB temp efficiently dropped to about 40C idle on stock speeds, and gets to around 47C during load. I hooked the fan up to the fan controller so I can pretty much silence the fan when working in Windows etc.


Problem 2. RAM temp
I have done some experimenting with overclocking (I will cover this later), and the RAM I use (Crucial Ballistix Tracer) is fantastic performing memory. My sticks proved to perform best on 2.85 volts on my mobo. The only problem is that they run extremely hot when overclocked at these voltages and active cooling was practically essential in order to make MemTest stable. I measured the temperature of the surface of one of the RAM sticks using the Matrix Orbital, and the temps rose to about 56C during load. To solve the problem I took the 70x10mm fan that came with the Heatsink for my X2, and mounted it to blow air across the RAM sticks.
Pics:

The fan is actually held nicely in place by the ATX PSU cable (hehe). Temps are now at about 35C idle and 47C load, and the FSB can go higher. The fan is connected to the fancontrolelr as well (same channel as the NB fan), so I can silence it when using non-intensive applications.


Problem 3. Exhaust
I thought that with watercooling, case air temps would not get too high. Apparently I was wrong, as the northbridge, RAM sticks, and videocard memory produces significant heat. The 120mm fan that came with the case did a decent job of getting rid of the heat, but more exhaust power was needed. I replaced that fan with a 120x38mm Panaflo fan that pushes a lot of air.
Its not that easy to see the difference but here are some pics. Before:

After

I have no accurate way of measuring the effect of this, but the NB and RAM temps are lower and the exhaust air is WARM during load. I am considering getting one of those 2x80mm PCI-slot exhaust fans, but as of right now I dont think its necessary. Maybe with SLI it will be.

 

[Carbon_Le]

Its looking really nice...good work.

 

[CC_Pirate]

[Noise dampening]

Just a quick sidenote here.
I realized that a large majority of the noise generating elements in this rig are located in the bottom section of the case. These include both harddrives, 3x120mm fans, PSU, and the pump. Therefore I decided to ass some noise dampening material down there. Here are some pics

 

[CC_Pirate]

[Finishing up]

Allright, the rig is done now. Everything is installed properly and cables are managed nicely. Ive been using it for a while now, and (of course) I absolutely love it. Here are some pics for you to feed on. Later on I will get back to benchmarks and overclocking.

Oh by the way, I got a 2nd monitor, Dell 1704FVP 17". This is a killer dual monitor setup. More on that later.

Time for some niteshots:

Remember that I painted the PCI bracket slots, thumbscrews, northbridge heatsink, videocard memsinks, MOSFET heatsinks, and 120mm grill with that blue UV clearcoat paint? I think the stuff worked pretty damn well:

These next two shots were taken with a two-second exposure time to accent the UV effect:

The keyboard has some nice lights in it too:

 

[CC_Pirate]

[More on the Matrix Orbital (and some videos)]

The Matrix Orbital MX6 LCD is pretty sick. It can output just about anything. I hooked up two temp probes to it, one that measures the temp of the surface of the Ballistix sticks, and one that measures the temp of the water.

Here is a pic of this simple screen I made that outputs all temperatures; CPU, northbridge, GPU, RAM, and water:

Some more pics:

Here is a little video I recorded of a few of the many different screens the MX6 can display (size: 10MB):
Matrix Orbital MX6 video

While were at it, here is a video of the water flowing in the reservoir:
Waterflow

And finally a video of the Tracers in action during some BF2 gameplay:
Tracer-action

 

[Carbon_Le]

Damn really nice work...man that UV looks so crazy.

 

[TehRoflcopter]

Woah. Those temps are a tad high for that CPU man... Was that picture taken during full load of Prime 95 or something?

 

[CC_Pirate]

Damn really nice work...man that UV looks so crazy.

Thanks!

 

[CC_Pirate]

Woah. Those temps are a tad high for that CPU man... Was that picture taken during full load of Prime 95 or something?

I assume you are referring to the pic of the Matrix Orbital. That pic was taken during gaming, and the fans are running at low to medium speed. The CPU temp is 43C, which, in my opinion is not high at all. Are you sure you read it right?

 

[CC_Pirate]

[Overclocking and benchmarking]

I didn't go all out on the water cooling to not overclock. So here's what I've made so far:
(I am sorry but the pics here have to be that big in order for you to be able to make any sense of them)

1. CPU
Stock speed: 2000Mhz, 10x multiplier, 200Mhz FSB/HTT
I closely followed (cf)Eclipse's excellent A64 OC guide to get the most out of my components.
After testing RAM, CPU, and mobo seperately, this is what I figured out:
- My mobo topped out at 320HTT
- My CPU topped out at about 2780Mhz, using a 10x multiplier and 1.5volts.
- My RAM was a little difficult to be sure about. 2.85v seemed to work out best, and with active cooling it ran memtest #5 stable at about 260Mhz.

So technically I should be able to do about 2800Mhz, using for example HTT 305Mhz, 5:4 Memory divider, and 9x CPU multiplier (memory speed 305*0,83=253Mhz).
However, the system would not boot past BIOS with a HTT anything above 245Mhz using a 5:4 memory divider. I did tons of testing, and after getting a lot of help from the fantastic (cf)Eclipse in this thread, I concluded that it is a board limitation.

Therefore, I am currently running at 2450Mhz now, using HTT 245, 1:1 memory, and a 10x CPU multiplier. I could run at 2500Mhz, but I dropped 55Mhz to be on the safe side.
I am absolutely satisfied, 450Mhz is a formidable overclock, giving me above 4800+ speeds. Maybe in the future I will buy a new mobo if I need it and get those last 300Mhz.

2. 7800GTX
Stock speeds: Core 430Mhz, Memory 1200Mhz.
After some testing, I was able to run 3DMark using 570Mhz on the core of the MSI 7800GTX. However I had to drop it down to 520Mhz to get rid of all artifacts. Memory is artifact-free at 1380Mhz. These are very nice overclocks, and I can't really expect more without doing a voltmod.


_BENCHMARKS:

Lets see what the overclocking gained me in terms of performance.

- 3DMark05.
Without any overclocking I got about 7850 3Dmarks with this system. Here is what I with the stable OC speeds explained above:

A very nice gain if you ask me.

- SiSoft Sandra:
I did not really do any tests with Sandra before the system was overclocked so I have no scores to refer to, but Sandra's built in comparisons provide decent info.

Arithmetic CPU bench:

As you can see here, my scores are pretty much identical to those of a AMD X2 4800+. From what I've seen in benchmarks, the extra cache on the Toledo core doesn't really make a very big difference performance-wise, especially in terms of gaming. The way I see it, I saved 400$ by getting the 3800+ (not considering the watercooling, which I would have gotten anyways).

Cache and memory bench:

I'm getting decent scores, but obviously the 4800+ and the P4EE with their extra cache tear me a new one on this one.

Memory bandwith bench:

Sadra doesn't provide comparisons with todays top of the line systems here, but I'm kicking ass again

CPU Multimedia bench:

Slightly higher than the 4800+

Finally, some file system benchmarks on the Raptor:

Pretty much any RAID setup kicks the Raptors ass here. That said, I have noticed a great decrease in f. ex. game load times with the Raptor compared to the 7200rpm disks I am used to.

 

[revlimit]

Base on the pictures, the fan for the PSU is on top so isn't there a air restriction problem there? Would it make a difference if you cut some holes on the the horizontal plate? I have the same case and I have a OCZ modstream PSU. I'm think I might make some type of opening there.

 

[CC_Pirate]

[A few updates and changes, intro]

Not that anyonw cares, but I've done some changes to my system in order to reduce noise and increase performance. I succeeded. Here's how:


_WHAT NEEDED CHANGE
1. The tiny fan on my chipset was too loud. I wanted to get rid of it. The northbridge was also runnung VERY hot under stress if the fans werent running at full speed.
SOLUTION: Swiftech MCW30 Waterblock

2. The RAM was running too hot, and the fan I had blowing over the sticks was not powerful enough.
SOLUTION: Put ramsinks on the heatspreaders and use a bigger fan.

3.The hot air that blows into the lower part of the case thru the radiator did not get sufficiently exhausted out of the case, resulting in the rad fans blowing the same hot air over the radiator again.
SOLUTION: Make front 120mm exhaust suck air out of the cas instead of into it. Also seal the radiator to the bottom of the case.

4. The volt regulator and the RAM on the 7800GTX was running too hot and limited my overclocks.
SOLUTION: Put bigger ramsinks on the underside of the board, and have a 70x10mm fan blow over the card from the rear.

5. My Ballistix RAM couldnt handle speeds over 255MHz as advertised, and held my CPU OC back.
SOLUTION: Get a new pair of Ballistix.

6. The coolant I used in my loop stained the tubing and I had a feeling it clogged the blocks a little. After dissasembling the loop, I realized I was right.
SOLUTION: Get two bottles of PC ICE UV Blue nonconductive coolant.

7. The way I had my loop set up, more tubing than necessary was used. The loop went Reservoir -> Pump -> Radiatior -> CPU -> VGA-> Reservoir
SOLUTION: New loop: Reservoir -> Radiatior -> Pump -> CPU -> VGA-> Chipset -> Reservoir


Read on to see how it turned out..