Parallel In The Black

Looks fantastic!

Cheers :)

Another image of fittings.

fittings_02.jpg


They come in varying types and designs, also notice how the thread length can vary - the 45 degree fitting (far right of bottom section) was incompatible with the waterblock I used for the GPUs, the thread was too long.
 
This looks like an absoloute monster of a system :)
All the plans and hardware design make me think it should be in a powerplant! Have you thought about installing a coal generator or some kind of brimstone power facility? :D

Seriously though, I do envy the copper pipe look and your work ethic to be building such a brilliant system
 
Amazing job, are you not seeing any temperature difference between gpu 1 and 2? I would be worried that there could be a path of least resistance and therefore one really efficient part of the loop at the expense of the others.
 
This looks like an absoloute monster of a system :)
All the plans and hardware design make me think it should be in a powerplant! Have you thought about installing a coal generator or some kind of brimstone power facility? :D

Seriously though, I do envy the copper pipe look and your work ethic to be building such a brilliant system

Thanks for the kind words and let me get back to you on the brimstone power plant :).

i love this, nice to see you didn't just bend pipe and put it in.

Keep it up

As the window is always within sight I know I'd always look at any section that wasn't right and think why didn't I put it right. So it's measure 3 times, cut 6 times and replace once!

Amazing job, are you not seeing any temperature difference between gpu 1 and 2? I would be worried that there could be a path of least resistance and therefore one really efficient part of the loop at the expense of the others.

Cheers. The system is very well balanced, there's only 1 or 2 degrees C different between the GPUs and I think that's due to the right hand card dealing with the comms to the PLX chip on the motherboard. When graphic cards in SLI are mounted horizontally the top card is usually the hottest, and while hot air rising does play some part, in my case with the cards mounted vertically and no other source of ambient heat around, I've attributed the slight difference in temperature to the PLX comms.

However, there was one worrying moment after removing one of the cards. Once connected and booted up I noticed the GPU temp wasn't at the usual 28C idle temperature, but at 49C. A quick (and I mean quick!) run of Heaven made the GPU run up to 90C, my custom fan profile ramped the fan up to 100% but of course that would do nothing due to the waterblock and not the stock heatsink being on the GPU. Why was it doing this? The waterblock was just fine before I drained the system to remove the other graphics card. Then I realised why - AIRLOCK.

So I have discovered a pitfall of parallel configuration when compared to serial. In serial if you get an airlock (which is unlikely) you'll know about when testing the system for leaks (before powering on) because of no returning flow. With a parallel system you only know it's happening when the system is running and your temperatures aren't what they should be.

Thankfully airlocks are pretty easy to solve. Ramp the pump up to 100% and give the case a wiggle and a wobble :).
 
Glad it is working for you. I would be very tempted to have a temp.sensor on each return pipe to check the temperature in the whole system. it would mean you can check for any inneficient parts of the system. If you have a temperature differential at stock clocks it might magnify later if you dont pick it up now.
 
Glad it is working for you. I would be very tempted to have a temp.sensor on each return pipe to check the temperature in the whole system. it would mean you can check for any inneficient parts of the system. If you have a temperature differential at stock clocks it might magnify later if you dont pick it up now.

Thanks. I've settled on two water temperature sensors; water in and water out (see my water cooling schematic in my first post). This gives me a lot of information such as - how efficient my radiator set up is (delta 1: water out - water in), and how much heat the water is absorbing (delta 2: water in - water out). That's all I need to know as far as water temperature goes. CPU and GPU core temperatures are far more important and since I can tap in to these with the Aquaero (virtual/software sensors) I can monitor them in Aquasuite (how I first noticed the odd GPU temperature) and set the system to respond appropriately, or me thinking wtf in this case!

If I have uncharacteristically high core temps at stock clocks due to an airlock, (which like I mentioned happened earlier) it only happens once - right after filling the system up. Once I've got the air out of the system, it's gone and it's not returning. It takes a good gulp of air to stall a waterblock, but as I've found out, it's more likely to happen in a parallel config than a serial loop (highly unlikely!). I'll show some screenshots of Aquasuite later in the build to show how I've got it configured and how it performs.

Loving the shiny and chrome look, Immortan Joe would be pleased.

Yeah, I admit I had to google Immortan Joe :).

First time i have ever seen anything like this.

Not usually a fan of metal piping but this looks great.

Cheers, glad you like it, more stuff to come :).
 
Update

this is next level stuff keep it up cant wait to see it :)

Having seen the case mods in this forum I consider that a compliment!

Another day, another update :)

First, a nice easy modification to the Silverstone FT02 - upgrade the USB ports on the top of the case to USB3.0. I bought the case so long ago before it came with USB3.0 at the top.

FT02_case_work_10.jpg


Next, is some retrospect, although you could consider it cheating :). The next few parts are things I did last year but are relevant to this build to illustrate the project as a whole.

I modified the Corsair AX1200 PSU and the air intake of the FT02 by changed the original fan of the AX1200 (because it was crap and making a high pitched whine) to a Noctua, much better, and dremelled out the restrictive fan guarding of the FT02. Made sure the filter still fitted though.

FT02_case_work_11.jpg


Again, another part done last year but this is more relevant - the cooling part of the build - the combination of the Watercool MO-RA 4x180 radiator and Aqua Computer Aqualis reservoir. I think that's enough radiator and fans to handle overclocked i7 4770K @4.7GHz, 2x 980Ti's and a motherboard chipset :).

FT02_case_work_14.jpg


In the image below, 1. Aqua Computer Aqualis and USB D5 pump, 2. the bracket mounted to the radiator, 3. the Aqualis base and pump fitted, and finally 4. the borosilicate glass reservoir fitted (the tubing was not the final routing).

FT02_case_work_15.jpg


This was connected to the case by use of some Koolance Quick Disconnects (QDCs).

FT02_case_work_12.jpg


FT02_case_work_13.jpg


Next is some maintenance to the above radiator and reservoir, then moving on to some more mods, thanks for reading!
 
Update

Update time again :)

In order to do the modifications to the 980Ti's that I have in mind but still keep it reversible for when I sell them in the future, I needed a sacrificial lamb or two, so bought a :

980ti_waterblock_mod_18.jpg


Just the cooler from an old-school Titan. As ever the inquisitive one, I stripped it down to see the components.

980ti_waterblock_mod_19.jpg


Have you ever wondered how Nvidia get the 'GEFORCE GTX' moniker to light up so uniformly? Yeah, so did I, so thought I'd find out as I wanted to know if it would be useful for my intended mod.

In reference to the image below:
1. The victim
2. Start to peel back the tape
3 & 4. The tape fully peeled back carefully so as not to damage the silver strip on both edges.

980ti_waterblock_mod_20.jpg


There's no LEDs in this thing - I reckon it's an electroluminescent panel. You can't see it in the images below as my camera isn't good enough but there is a very fine pattern in the slightly flexible acrylic.

1. The EL panel lights up with 5 to 6Vdc, any less and it just turns off. That makes me think it's PWM controlled as you can vary the brightness of this strip (and make it pulse) via the Nvidia software.
2. On the other side of the EL panel you can see the copper tracks, not very complicated but still very effective.
3. That signature glow.

980ti_waterblock_mod_21.jpg


This may be of use to my mod, some experimenting to be done still :).

The disassembly of the Titan cooler gave me some further inspiration, but the Titan cooler's mounting is incompatible with the 980Ti. So hunting began again and I sourced some more little lambs :).

titan_x_02.jpg


A pair of reference Titan X's - just the coolers I must add! In brand-spanking-new condition too. There won't be much left of them once I'm done with them; only the fan, the screws and the Titan name plate will be left over. Perhaps I could mix up the letters anagram fashion and make another name badge such as ATTAIN TINT, which would give a clue as to my intentions :).

Until next time...
 
Update

Hey guys, sorry for the lack of updates, life got in the way as it does but I have been busy modding behind the scenes. This post is going to be a bit wordy but with plenty of interesting pictures.

I really wanted to crack on and get the mods to the 980Ti’s that I intended. There will be destruction! I still wanted to use the computer so thought I’d only remove one 980Ti, mod it, then swap it. As the loop is parallel a simple bypass in place of the removed card would just have meant the water would go straight in and right back out again, so I had to “dead end” the pipework where the removed 980Ti once was fitted. I thought this would have done it, but oh no…

pipe_install_10.jpg


So I went with a simple return pipe and crushed the middle of it so that no water could flow through it. This worked :)

pipe_install_11.jpg


Here is a summary of my intended mods for the two 980Ti's, some of which I have already mentioned:

980ti_modifications_01.jpg


I shall cover each of the above mods in more detail when I get to them in the flow of the project.

First up, modifying the vapour chamber heatsinks. I first started on one from a Titan (non X) cooler that I got earlier, this type is bare metal, silver in appearance.

In reference to the image below:

1. The stock cooler from a Titan.
2. Marked up the required cuts using the mounting bracket from the water block.
3. Rather than use a Dremel I tought I'd save some time and use my circular saw. As I had no idea as to what was in the "vapour chamber" of the heatsink, I wore all the PPE - goggles, ear defenders, gloves and double filter mask - See no evil, hear no evil, feel no evil, smell no evil! :)
4. Unfortunately it's not the best quality cut finish.

980ti_modifications_02.jpg


Then I thought I'd use the circular saw to cut the thicker area of the heatsink that contacts the GPU core.... this didn't go so well!

980ti_modifications_03.jpg


The saw blade must have caught something; it ripped the heatsink from the clamp, almost taking my finger with it, thankfully wore my rubber coated gloves. Hey folk, always use PPE! Just as well it was the test heatsink, it's scrap metal now.

980ti_modifications_04.jpg


Scapped the idea of the circular saw and went back to the trusted Dremel.

980ti_modifications_05.jpg


Both Titan X sacrificial heatsinks cut. The masking tape was in place to see the marking for cutting but also to stop the fins from vibrating while cutting.

980ti_modifications_06.jpg


Have you ever wondered what's in the vapour chamber of the heatsinks? So did I, and I had to flatten them anyway for mounting, so time to open one up.

980ti_modifications_07.jpg


980ti_modifications_08.jpg


Emm...interesting..


That's all for now, more soon!
 
Hahaha awesome destruction! So wtf is that sand-like powder?

It does look like sand, but from what I've read it's sintered copper powder and acts as a wick to move the heat. The small rods in the middle are provided for structural integrity and to assist in this movement of heat. There was no liquid in the vapour chamber that I saw, so unless it evaporated or was a minuscule amount it's likely to have been a gas, possibly nitrogen.
 
Back
Top Bottom