Overclock.net › Forums › Case Mods & Cases › Builds & Case Mods › Build Logs › [Build Log] - IRONBEAST - A fully watercooled, X-99, Quad-GPU, 900D build
New Posts  All Forums:Forum Nav:

[Build Log] - IRONBEAST - A fully watercooled, X-99, Quad-GPU, 900D build - Page 125

post #1241 of 1259
Thread Starter 
The wrap, hex-mesh and dust filter material are here, as are most of the required magnets.


More of the wrap than will be needed I suspect, but better to have too much than not enough. There’s a second mesh panel here also, which should be more than sufficient.


These are the countersunk magnets, a 1.2kg linear pull in both cases. 20x10mm ferrous for the side panels, slotting into the spacer frame. 10x2mm neodymium for the front and rear bezels, used because the holes that they slot into need to be shallower owing to the chamfer on the frame. All of the former are have arrived, still waiting on some more of the latter.


This is a comparison between the brushed metal effect of the wrap and the genuine article from the front of my 900D. On a close inspection and under this light you can clearly see the difference, it just isn’t as beautiful as the real deal, but zoom out and the overall effect is impressively close. As much as I would have loved to have found a way to get this done in genuine metal and glass, this and the Perspex should make for a passible imitation.

Quote:
Originally Posted by stocksux View Post

Subbed

Welcome to the club!
post #1242 of 1259
Thread Starter 
Quick update:

All the fabrication drawings for the case panels are now done. There are a couple of other unrelated ones for another project left to finish and then the order can be sent off for manufacturing. This evening I also tried out the brushed metal effect wrap on the nearside dust filter plate, largely as an experiment to try and work out the best procedure to use for the panelling.


That’s the finish the vinyl wrap gives.


This one was the first attempt, where the panel is completely wrapped so that all the visible edges are covered. In short - total rubbish. Even snipped into tabs the curves in particular just don’t work satisfactorily.


This is the second attempt, far better. The front is covered and the edges folded over as before, but this time the centre section is simply cut away with a scalpel. It means the narrow interior face is left exposed, but the overall result is so much neater. The spray painted black of the Trespa is close enough in colour to the brushed metal that you really have to go looking in order to notice it, so that seems acceptable to me.

I think therefore that this is the technique I will use with the case panels; ignoring the cut edges and trimming the wrap to follow the exact shape of the face. The bezels will probably get a bit complicated, they may have to be done in multiple pieces, but this method should work very nicely for everything else.




Edit: Distracted by all the work getting the part drawings done, it entirely escaped my notice that we passed two years since the start of this build log. A massive thank you to everyone who has followed my utterly preposterous project, especially those who have been here right from the start, it really means a lot. I promise you will see it finally finished, and I promise never to let future projects take anywhere near this long…
Edited by OCDesign - 3/5/17 at 11:20am
post #1243 of 1259
Quote:
Originally Posted by OCDesign View Post

... I promise you will see it finally finished, and I promise never to let future projects take anywhere near this long…
Hah, be careful making promises you can't keep! biggrin.gif

I've been here since the very beginning, and it has been a true pleasure. The road to the finished product has in this case (no phun intended) been as delightful as the finished product.
post #1244 of 1259
Thread Starter 
Quote:
Originally Posted by bkvamme View Post

Hah, be careful making promises you can't keep! biggrin.gif

I've been here since the very beginning, and it has been a true pleasure. The road to the finished product has in this case (no phun intended) been as delightful as the finished product.

Thanks very much. Nice to know I’ve managed to keep the build log entertaining, despite it dragging on for all this time. And you’re right - it will no doubt prove extremely foolish of me to have said that. You’d have thought I would have learned my lesson by now, everything always takes far longer than it should, but then things would have to be going pretty badly for me to have to break a promise that vague…
post #1245 of 1259
Thread Starter 
Apologies that there hasn’t been an update from me in a bit. The case panel drawings - amongst other things - will go for manufacture on Monday. (The plans for the other unrelated stuff took a lot longer to draft than I thought. Fabrication drawings, being stark black on white, can be quite literally a headache if you stare at the screen too long).

At the moment however, there is another problem to deal with - the cooling loop isn’t working properly. GPU-side flow rate is fine, but the CPU-side throughput is way down on what it was. It also appears to be getting progressively worse. There is enough – just – for the hardware to keep working ok, but it’s definitely not good. Now this might be an easy fix, or it might be horrible. It could be something as simple as an air pocket somewhere, or that it’s just time for the rig to be flushed & cleaned again. But given that the water looks perfectly clear and that the GPU-side is running as before, I am a little worried it could be a more serious problem with the centre pump.

In any case, when there’s some free time I’m going to drain the loop and refill – since its probably an appropriate moment to do so anyway - and we’ll see if that helps.
post #1246 of 1259
You definitely don't have to apologize for not updating us more frequently. It's difficult balancing life, job and hobbies. Therefore instead of you apologizing for not updating more frequently, let us thank you for updating at all. So, thank you. biggrin.gif
post #1247 of 1259
Thread Starter 
Quote:
Originally Posted by Lee0 View Post

You definitely don't have to apologize for not updating us more frequently. It's difficult balancing life, job and hobbies. Therefore instead of you apologizing for not updating more frequently, let us thank you for updating at all. So, thank you. biggrin.gif

It's much appreciated, truly. Since the start of the build log I have done my best to keep up a reasonably regular pace of progress updates for you (albeit snail’s pace progress) and to reply to things reasonably promptly, so that there was (and is) at least something new to read every few days. Once that gap starts to get into double figures I can’t help feeling like I’m letting you all down a bit.

Still there is at least a passable excuse this time, namely that Ironbeast has been offline for the past few days: - ‘offline’ in this case being a polite euphemism for 'in pieces all over the floor'. I mentioned last time that there has been an issue with very weak CPU-side flow recently, and that I was going to drain the system down to try to fix it. Well the evening before, just too add to my problems, the fan backlighting also decided to fail. Needless to say I was not particularly happy about that, so do please forgive me if this account takes on a slightly exasperated tone in places.

When the LED strips were powered up there would be a split-second flash along the whole line and then nothing, suggesting a fault in the data connection. Having bypassed the first set of strips (there’s a plug below the bottom front fan linking the two) the rest worked perfectly, clearly indicating the problem was somewhere in the nearside lighting chain. As such the entire contents of the subfloor – all the awkward power connections, the PSU itself, both lower radiators and all their associated piping – had to be removed, leaving Ironbeast looking more than a little eviscerated. Then the nearside radiator fans had to be unscrewed, the LED backlighting taken out, and then the whole lot carefully checked to find out which bit wasn’t working. One lousy solder joint, that’s all it was. Evil thing. And so: set up soldering iron; spend about ten minutes trying to remember where I’d put the solder; detach, clean up, and remake the relevant connection; reassemble everything; fit it all back into the rig. I’m sure it was a perfect stress-free evening…for somebody.

Anyway, turning back to the flow problem. The drained coolant water still looked remarkably clean so that didn’t seem to be the issue. As such I worked my way through the system, progressively taking out and checking individual sections of pipework to try and find anything obviously wrong (thank you D-plugs). Long story short, one of the check valves appeared to be jammed. Thus it was taken out, swapped over for the other one (since the GPU-side valve isn’t vital), and the pipe was reinstalled. Cue an hour of keyhole surgery trying to get the PSU cables back in, refill the loop and…

…same problem.

So the system is now operational again and the CPU-side flow is marginally better than before – manageable, if still not as it was. (Hence why this log entry is rather more than the single gloomy Android-driven sentence from my tablet which I feared you might be getting). Since this little pitstop hasn’t done much good, and I can’t find the issue from a visual inspection, I’m going to try some experimental work with the system running. There is some spare tubing round here somewhere, so I’m going to progressively bypass individual sections of the loop – hopefully without getting too much water everywhere – and see how the coolant flow behaves. However I am becoming increasingly concerned (and rather depressed) that it may indeed be the No.2 pump. And that, owing to the wiring routes and the bottom GPU, would not be an easy fix.
Edited by OCDesign - 3/27/17 at 1:05pm
post #1248 of 1259
Thread Starter 
A very quick update, just to keep you all in the loop.

In short, nothing I’ve tried over the last week or so to fix the CPU-side flow problem has worked; with all evidence suggesting that it is indeed an issue with the centre pump. As such the rig is going to have to be drained down (again) and partially disassembled – to the minimum degree necessary, but that’s still quite a lot – in order to disconnect and take out the second D5. I’ll open it up, see if there’s anything obviously wrong, and then swap it over for No.3 pump (which doesn’t have to work as hard). Hopefully that will correct things and get Ironbeast running properly again before all the case panel pieces arrive sometime towards end of next week.
post #1249 of 1259
Quote:
Originally Posted by OCDesign View Post

A very quick update, just to keep you all in the loop.

In short, nothing I’ve tried over the last week or so to fix the CPU-side flow problem has worked; with all evidence suggesting that it is indeed an issue with the centre pump. As such the rig is going to have to be drained down (again) and partially disassembled – to the minimum degree necessary, but that’s still quite a lot – in order to disconnect and take out the second D5. I’ll open it up, see if there’s anything obviously wrong, and then swap it over for No.3 pump (which doesn’t have to work as hard). Hopefully that will correct things and get Ironbeast running properly again before all the case panel pieces arrive sometime towards end of next week.

Hope you manage to get it all working again buddy. You've got this. smile.gif
Ncase M1
(18 items)
 
   
CPUMotherboardGraphicsRAM
4790k@4.7GHz, 1.296v (4.5GHz uncore) Maximus VII Impact MSI 980 Ti Gaming 6G Corsair Dominator Platinum 16gb 2133 CL9 
Hard DriveHard DriveCoolingCooling
2x 840 EVO 500GB Toshiba XG3 1TB (M.2 NVMe SSD) PE-240 Rad EK Supremacy Evo & VGA 
CoolingCoolingOSMonitor
DDC 3.2 PWM pump 3x Gentle Typhoons 1850rpm windows 10 Acer Predator XB270HU 
KeyboardPowerCaseMouse
Corsair Strafe RGB Corsair SF450 Ncase M1 Logitech G502 
Mouse PadAudio
steelseries Bose companion 20 
CPUMotherboardGraphicsRAM
i7-4710MQ W230SS gtx 860m HyperX Impact 16GB 1600MHz CL9 
Hard DriveOSMouseMouse Pad
840 EVO 500GB windows 8.1 cyborg R.A.T-5 steelseries 
  hide details  
Reply
Ncase M1
(18 items)
 
   
CPUMotherboardGraphicsRAM
4790k@4.7GHz, 1.296v (4.5GHz uncore) Maximus VII Impact MSI 980 Ti Gaming 6G Corsair Dominator Platinum 16gb 2133 CL9 
Hard DriveHard DriveCoolingCooling
2x 840 EVO 500GB Toshiba XG3 1TB (M.2 NVMe SSD) PE-240 Rad EK Supremacy Evo & VGA 
CoolingCoolingOSMonitor
DDC 3.2 PWM pump 3x Gentle Typhoons 1850rpm windows 10 Acer Predator XB270HU 
KeyboardPowerCaseMouse
Corsair Strafe RGB Corsair SF450 Ncase M1 Logitech G502 
Mouse PadAudio
steelseries Bose companion 20 
CPUMotherboardGraphicsRAM
i7-4710MQ W230SS gtx 860m HyperX Impact 16GB 1600MHz CL9 
Hard DriveOSMouseMouse Pad
840 EVO 500GB windows 8.1 cyborg R.A.T-5 steelseries 
  hide details  
Reply
post #1250 of 1259
Thread Starter 
System restored. Sort of.

After a somewhat extended period of down time (notably longer than had been intended - for a very long list of reasons, largely unrelated to the machine’s problems) the rig has been reassembled and Ironbeast is back up and running. The system was drained, unwired, and then taken apart (again) to a rather distressing degree. The pumps were inspected, changed over, and the whole lot reconstructed. Water back in (with no leaks this time I am delighted to say), loop purged of any lingering air bubbles, everything wired back up, and this room looking a lot tidier than it has in weeks.

So, problem fixed? No…same issue still.

That kills the theory that it’s a pump problem, and leaves me a bit stuck for ideas as to what’s left that it could be. However, on the positive side, in the course of the reassembly I took out the second of the two check valves – in case it was that which was playing up. It wasn’t, but removing it has improved the flow rate a little and thus rendered the previously critical issue manageable again. (Ironbeast’s circulation seems to be at least stable and consistent now, if still diminished, so the machine is usable - without me panicking at weird noises and checking the fluid levels every 5 minutes).

Given that it’s now running well enough to do its job and still remain cool (if not as well as it did before) I’m going to press on, do the case panels, and finally see this as close to ‘finished’ as it is ever going to get: for my own self-esteem (such as it is) if nothing else. Most of the pieces for that are here, ready and waiting, and unless the problem is in fact embarrassingly obvious then it is not going to be something that I can get to with anything short of total disassembly. I’m really not in a position right now to reduce this thing right back to its components and rebuild, so on we go. One last effort.

You lot must think I’m completely mad…even more than you already did.
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Build Logs
Overclock.net › Forums › Case Mods & Cases › Builds & Case Mods › Build Logs › [Build Log] - IRONBEAST - A fully watercooled, X-99, Quad-GPU, 900D build