• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

[WIP]SG10 Build Log

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

SeeThruHead

Member
Joined
Feb 23, 2011
Location
Toronto, Canada
[WIP]DIMASTECH NANO Build Log

Time to update the first post with recent developements.
Switched to Dimastech Nano test Bench. Most of the superficial stuff is done, lots more fine tuning with regards to cabling and whatnot to come.
Here is a link to my dropbox album with all the pictures I've taken of the build.
Instagram
DropBox

DSC_0091 (Medium).JPG

close.JPG

Corner.JPG

DSC_0130 (Medium).JPG

Full.JPG

Socket.JPG



So I recently tore apart every computer that I own. Stuffed my main rig into a Silverstone SG10. But now I've got the water cooling itch again. Air cooling, even the noctua nh-d14 i have in this sg10, just isn't satisfying.
Terrible picture of my original build: 2013-07-07 20.17.56.jpg
Random Picture of my new case:
silverstone-sg10-open-side_slideshow_main.jpg sg10-side.jpg

The only thing I'm going to be reusing is my bitspower fittings and my mcp35x.
Preliminary Parts List:
XSPC Raystorm Black ED with Red Faceplate. raystormblackred.jpg
2x XSPC GTX 670 Blocks
Swiftech MCP35x2 Pump setup
Bitspower Watertank, whatever the largest tank I can fit vertically.
Mo-ra3 420 PRO With both sides with fan grill(140x9)
4x 200mm Silent fans: I would like to run 4 fans at 600rpm to cool this entire rig while gaming. No mining or benching will be done. Actually the hope is to have no fans on at idle. And have fans turn themselves on and ramp up in speed as the water temp rises.
Aquaero 5 LT Controller
Chipset Blocks and ram blocks will also be used (for looks)
Ek Ram Block
Bitspower Black compression fittings.
Aquaero 5 LT will be mounted insude the fan grill of the mo-ra3, if it will fit. Hopefully it will, otherwise I might mount it inside the case, on top of the PSU, though it wouldn't get any airflow there.

PC Parts:
Maximus IV gene
2600k
Corsair Dominator GT 4x4gb
2 GTX 670 Reference
2xAdata 256GB SSD
Seasonic x750


Some notes. Will be sleeving with MDPC black and color-x. Heatshrinkless style. My first sleeving job so I hope I can figure it out.
I'll be covering up the PSU on the inside with either some aluminum or painted acrylic. This will also cover the front I/O cables and the res will be mounted to it.
The Three fan intakes on the side will be covered with a panel of brushed aluminum to match the front of the case.
I want to get rid of the Silverstone branding on the front of the case but I'm not sure how I'm going to do that just yet.
The top half of the side panel is going to get a smoked acrylic window surround on the inside with some white LED's or ccfl, will have to see how each looks before deciding.

Looks are the deciding factor for everything in this build. I've created a terrible MS Paint picture of my proposed tube routing, decided entirely because I think its the best looking tube routing.
With RES drawn in
Case with res.png
With RES Cylinder transparent
Case with ram.png

I feel like this should work well. Though depending on my GPU test I may have to add something on the mosfet/ram line to increase resistance to increase flow to the GPU's. I'll see when I get there.

It's going to be a very slow build. MDPC sleeving and some tools are already ordered and on there way. But I'm also in the process of making and mATX freenas Box which I'll probably be posting up for your perusal at some point.
 
Last edited:
hate to say it but if i've interpreted your schematic correctly your gonna have some nasty short ciruiting of gluid at the gpus. my suggestion is to have the flow through the pumps then cpu then the gpus then out of the case to what im assuming is a rad. then back in and up to the mosfet cooler then back to res
 
I'm confused as all heck... You have the fluid coming out of the reservoir to the pump AND to the GPU1? How is that going to work?

I'd suggest something similar to rustyfender, but a more direct approach with less tubing. Res>Pumps>Rad>GPUs>MOSFETs>CPU(or CPU>MOSFETs)>Res

Loop order doesn't really matter in the long run other than res before pump, so I'm sure you'll see more benefit from shorter tubing runs than from loop order.
 
I supposed I should have clarified. The tubing coming out of the block labelled mosfet is going into the ram block. The tubing going into GPU1 is coming out of that same ram block, not the Reservoir.
Loop is as follows:
..................................................GPU1
Res -> Pump -> CPU -> RAD ->GPU2 (All three in parallel) -> Res
.................................................Mosfet/RAM

I've edited my OP with a paint picture with the RES cylinder transparent.
Tubing length is of no real importance of me. Whether I have a 8 or 10 degree delta doesn't really matter to me. I chose those tubing runs because I feel they provide the most symmetry in tubing runs, and look the best as a result. My only concern is the gpu's not getting enough flow because the mosfet/ram line might have less restriction than a gpu block. I will however first be testing the temps in this current setup before deciding whether or not to either change the tube routing or add a valve to increase pressure on the mosfet/ram tube line.
 
Last edited:
Ah, ok, I see what you're doing now. But I still don't think it will work. The problem with parallel loops is the water will take the path of least resistance, so with your planned setup, chances are it'll go from the rad to GPU2, back out of GPU2 and into the res. GPU1, the RAM and the MOSFETs will be starved for any flow.

With parallel GPUs, you can really only have one inlet and one outlet on the blocks to the rest of the system.

What I suggest is having the GPUs in parallel, but having the RAM and MOSFETs in series with the rest of the loop.

Technically you don't really need to cool the RAM or MOSFETs, but if you're doing it for aesthetics I completely understand.
 
I wouldn't do parallel anything. If you want to split the flow use two loops.
 
I wouldn't do parallel anything. If you want to split the flow use two loops.

This. Parallel is too dodgy due to unequal path restrictions. You won't know or be able to easily control how much fluid goes down either path. It won't be equal due to the fluid preferring to follow the less restrictive path.

There's also no reason to water cool your RAM. It adds complication, restriction, and cost with no associated benefit.

What are the lines out the back going to and coming from?

EDIT: Oh, the radiator. KISS, man. You've overcomplicating this.
 
Ah, ok, I see what you're doing now. But I still don't think it will work. The problem with parallel loops is the water will take the path of least resistance, so with your planned setup, chances are it'll go from the rad to GPU2, back out of GPU2 and into the res. GPU1, the RAM and the MOSFETs will be starved for any flow.

With parallel GPUs, you can really only have one inlet and one outlet on the blocks to the rest of the system.

What I suggest is having the GPUs in parallel, but having the RAM and MOSFETs in series with the rest of the loop.

Technically you don't really need to cool the RAM or MOSFETs, but if you're doing it for aesthetics I completely understand.

Well you're partially right. The flow will take the path of least resistance. Since the resistance of each GPU block is the same they will receive the same amount of flow regardless of anything else in the loop. The question is, which is lower resistance between the GPU block and the Mosfet/Ram line.
Its essentially the same as this:
komodo-TRIPLE-SLI-PARALLEL-SETUPx800.jpg
Except that instead of a third GPU block I have the ram/mosfet line.
Now in a triple gpu setup each block will receive the same amount of flow because each block has the same resistance.
I have two scenarios to think about.
Scenario 1: The Ram/mosfet line has more resistance than 1 gpu block.
In this case the gpu blocks will both get equal flow, and more flow than the ram/mosfet line. This scenario is ideal as the gpu's need more cooling than the mosfet/ram line.
Scenario 2: 1GPU block has more resistance than the ram/mosfet line. This would result in the ram/mosfet getting a larger portion of the flow through the loop, and cooling potential of the GPU would be lower. Now this is actually ok, as long as GPU temperatures are within reasonable limits (under 65 degrees C)
If there is not enough flow to keep the GPU's under 65 (which I doubt, since they should be fairly easy to cool) Then I will have to balance the pressure between the GPU Block and the ram/mosfet line. This can be accomplished in a number of ways but the way I'm thinking is to use a flow limiting valve on the Ram Block intake. This will add more resistance to the mosfet/ram line and increase flow to the GPU's.

If you've even seen the Silverstone SG10 in person you'd understand that two loops is pretty hard and would end up incredibly ugly. Which is going against my entire design philosophy for this particular build.
 
Last edited:
Ah, much clearer now. Well I'm curious how this build will pan out and the results you get :)

I don't believe I've seen anything like this before so I'm definitely interested. Post pics and results!
 
Ah, much clearer now. Well I'm curious how this build will pan out and the results you get :)

I don't believe I've seen anything like this before so I'm definitely interested. Post pics and results!
I'll be posting pics (though I have no good camera) throughout the build. And little updates as I receive stuff for the build. Just ordered a Z87 Gryphon armor kit. (For the USB/PCIE dust covers) so I should be posting some pictures of how that looks when I go to pick it up.)

The build will be progressing slowly mainly due to funds, my Freenas build which is in the process of collecting parts (I have 3x 4tb HDD and an m1015 nothing else) is taking priority when it comes to parts purchasing. But that build isn't really flashy or fun to show off.
 
I do think you're over complicating things but that's just me. lol

This should be interesting. Sub'd for pics.
 
I do think you're over complicating things but that's just me.

It's not just you. :chair:

I hope it works out for him, he seems to have a workable plan. It just seems the plan is more complicated than it needs to be to do the job.

I don't like complicated. It's one of many reasons that I don't date. :D
 
i've got a bit of an update for you guys.
Got my GPU blocks in and decided to hook them up using some WC gear I had lying around from my last build. I also stuck a kuhler 620 on the CPU in place of the noctua. The noctua I found just way to large for the case. Couldnt change ram, couldnt plug fans into headers. Just too large. I also got a package in from lutro. Fast shipping there.
This is what I got from Lutro, Knipex stripper, sleeving tool, connectors + pins and some wire.
2013-07-24 16.57.37.jpg

2013-07-24 16.57.56.jpg

2013-07-24 16.58.01.jpg

This is what My pc looked like when I woke up today. Crammed! Could barely get in there to push the tab to let me take out my GPU's.

2013-07-24 17.05.12.jpg

2013-07-24 18.32.24.jpg

2013-07-24 19.05.22.jpg

2013-07-24 19.48.02.jpg

2013-07-24 21.28.18.jpg

2013-07-24 21.28.24.jpg

2013-07-24 21.42.25.jpg

Basically I'll let the pictures speak for themselves.

Now you may think that I was going to add fans to that rad. Well, looks like I'm not. Haven't seen either card go above 41C with no fans on the rad. Cpu is back at stock for now with this khuler 620.

I really wish I had a better camera so you guys could see what's going on inside there. Can't wait to expand this to my full loop once I can afford the rest of the parts.
 
Thanks Con. As you can probably tell it's a real pain in the *** to work inside such a small space. I'm thinking of getting some smaller tubing or maybe bending copper pipe for this build. Copper pipe would allow me to get the bends without the bulky 90 degree fittings (which wouldn't even fit anyway on the lower ports of the gpu blocks. Anyone have recommendations on copper pipe and fittings?

As for GPU temps, looks like they rise really really slowly, up to 55C I've seen but it doesn't seem to get much higher than that. I don't think the rad is getting much airflow, just incidental airflow from the two 92mm noctuas I have mounted on the side panel.
 
I've done some searching. Lots of great builds with copper pipe around the net. Anyway, my initial thought is to use some 12mm copper pipe and bits power crystal link fittings. Paint the pipes a nice glossy red. Well see how that works out.
 
Back