Taking the Water Cooling Plunge

Page 6 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

dma0991

Platinum Member
Mar 17, 2011
2,723
1
0
Do you think it would be wise to move it?
I'd say keep it that way rather than spending more time purging the loop, refilling and air bleeding. The only reason I'd ever want a flow meter is to check if I have zero flow due to a dead pump or a drop in flow rate when a single pump dies in a dual pump configuration.
 

Aikouka

Lifer
Nov 27, 2001
30,383
912
126
I'd say keep it that way rather than spending more time purging the loop, refilling and air bleeding. The only reason I'd ever want a flow meter is to check if I have zero flow due to a dead pump or a drop in flow rate when a single pump dies in a dual pump configuration.

That makes sense. I can always leave it alone for now until I have a good reason to drain my system.

Speaking of a good reason to drain my system ( ;) )... I was curious whether I could be causing the turbulence in my system that's causing my noise problems. I found this thread where the guy is pretty much having the same issue as me. My reservoir isn't 100% full (about 80%?), and the reason why I haven't filled it more is because when I shut the system down, the excess from the line drops into the reservoir, which pretty much fills it to the brim. I'm going to be changing out my fill port hardware, so it would be nice to avoid making a mess when I take the current one off. :p

So, I could either fill up the loop some more or move the return line to the bottom of the reservoir. Other than all the draining required, the negative about the latter is that I've only got two ports on the bottom, and I'm already using the second one for my temperature sensor.
 

Aikouka

Lifer
Nov 27, 2001
30,383
912
126
So, it's pretty safe to say that I'm not done yet, but I'll give you guys a little sneak peek at what I've got in store for this. It's definitely safe to say that this is a bit different than anything else that I've done!

http://i.imgur.com/HLaRNNE.jpg
 

Aikouka

Lifer
Nov 27, 2001
30,383
912
126
Well, apparently, since you guys showed so much interest in what I was working on, I'll show you how I screwed it up:

http://i.imgur.com/JDoqSi1.jpg

I was attempting to build a "tray" to put the pump on so it wouldn't have to sit next to the radiator. However, at one point during the design, I decided to extend the one end from 4" to 6" so I could set the entire pump on it. Unfortunately, that gets in the way of a small support arm for the motherboard tray that connects to the divider between the side panel and the lower bay. So, I would have had to cut a piece of it out anyway. As you can see, I bungled the hell out of the other bend. I had a feeling it wasn't going to work given my vice was simply too small and I lacked some good angle iron to throw in there. However, having to cut back to fit in that support brace will help me out since the size of both ends will be equivalent (4"), and I can bend that. Although, I wish I could get a nice, flat bend. :( If you've ever done anything like this and have any advice, I wouldn't mind hearing it!

Anyway, I'm going to cut out a new shelf and try again. I'm really putting this poor rotary tool through its paces. I did consider buying a jigsaw since it might be easier, but given it's just sheet metal, I should be fine. Now, if I start cutting 2x4s, I'll get a bigger saw. :p

I also got a new item in today:

http://i.imgur.com/1SEQnlF.jpg
http://i.imgur.com/iCRwjtD.jpg

Although, I am rather disappointed in ASUS. They advertized the built-in water cooling on this board, but they cheaped out with an aluminum heatsink. I wasn't too worried about using it as I know it isn't necessary, but... why go through the hassle and the cost of building in a feature and just half-assing it like that? Who would subject their loop to the extra restriction and the reduced performance from additives to resist galvanic corrosion?
 

dma0991

Platinum Member
Mar 17, 2011
2,723
1
0
Is that an aluminum plate? You can use a vice but it usually makes for sub par bends or scuffed surface. You could try hitting it with a rubber mallet while clamped to get it much closer to a 90 degree bend. If you could fabricate a simple sheet metal brake, it'll be a lot easier and get better results.

That aluminum could be an issue but I'm more concerned about how effective is black paint at transferring heat and the possibility of it flaking over time and develop black paint specks in the loop. No point in complaining about a feature that you knew is flawed. Should've thought twice before buying it. :p
 

Aikouka

Lifer
Nov 27, 2001
30,383
912
126
One thing that I've heard about recently is that when your radiator is mounted vertically, you should try and use the port on the bottom as the inlet. Supposedly, this helps the radiator bleed air easier. Is this something that I should worry about? My updated plan was to use the upper port of the 240mm radiator as the input as that would be the closest to the pump (when it's on the shelf). However, that means the bottom would be the outlet, but that easily leads to the 480mm radiator's bottom port, which would be the inlet.

Is that an aluminum plate? You can use a vice but it usually makes for sub par bends or scuffed surface. You could try hitting it with a rubber mallet while clamped to get it much closer to a 90 degree bend. If you could fabricate a simple sheet metal brake, it'll be a lot easier and get better results.

Hmm I thought about trying to make a brake, but I figured that I'd screw that up too. :p But yeah... I do get angled bends, which really don't help much as it makes my overall length a little too short and makes the shelf a bit higher than intended. I've also read that heating up the aluminum in an oven (recommended given the relatively precise temperature) or with a heat gun / torch (not a recommended) can make it far easier to bend. This is .08" aluminum, which is kind of thick... maybe even too thick. The overall pump setup is about 3.5 pounds, so it definitely needs to hold some weight.

Although, I did think of a possible easier solution: why create a bend when I can buy one? I could get angled aluminum and attach it to a flat surface. Boom... instant bend! I just have to figure out the best way to connect it and consider that it won't be as strong. Although, we're only dealing with a couple pounds, so that's okay.

I've also considered switching to acrylic to make the flat portion of the shelf, but from what I've read, I would pretty much need a jigsaw, because a rotary tool will heat up the plastic too much.

That aluminum could be an issue but I'm more concerned about how effective is black paint at transferring heat and the possibility of it flaking over time and develop black paint specks in the loop. No point in complaining about a feature that you knew is flawed. Should've thought twice before buying it. :p

Would you judge me if I said that I partly wanted it because it's purdy? :$ I considered using the water cooling if it was copper or brass just for the heck of it, but as people here have told me and even ASUS said, it's not really necessary. That is one of the things that kind of bugged me about it. An ASUS rep on their ROG forums flat out said that you don't really need it so it doesn't matter.

I've also been a few BSODs recently, and if I had to guess the culprit, I would say the motherboard is causing the problems. That also kind of pushed me toward switching.

EDIT:

I'm also working on another piece that ought to be much easier! I know I brought up doing this earlier...

http://i.imgur.com/0SsOz1n.jpg

I'm going to drill the holes, clean them up and paint them black. I bought some slightly longer screws and spacers (a little extra support) that will connect to the reservoir mounts. I bolded that 's', because that's the reason why I'm doing this. I originally drilled holes to connect a single mount, but I didn't like the fact that the reservoir could easily shift a lot around that axis. There just isn't a good place to drill for another mount, so my best bet is my original idea of using little brackets.
 
Last edited:

dma0991

Platinum Member
Mar 17, 2011
2,723
1
0
One thing that I've heard about recently is that when your radiator is mounted vertically, you should try and use the port on the bottom as the inlet. Supposedly, this helps the radiator bleed air easier. Is this something that I should worry about? My updated plan was to use the upper port of the 240mm radiator as the input as that would be the closest to the pump (when it's on the shelf). However, that means the bottom would be the outlet, but that easily leads to the 480mm radiator's bottom port, which would be the inlet.
Well, trapped air bubbles with a bottom up flow is far easier when air always wants to rise. It definitely makes bleeding easier but I don't think it matters too much if you can get the shortest/best path.


I've also considered switching to acrylic to make the flat portion of the shelf, but from what I've read, I would pretty much need a jigsaw, because a rotary tool will heat up the plastic too much.
SDC10090.jpg
SDC10108.jpg


This is badass. Make this. :p
 

Aikouka

Lifer
Nov 27, 2001
30,383
912
126

Haha I first saw this, and wondered, "Why is he showing me a toilet?" :p

I cut out the angled aluminum, and it should work well. I was able to get some 3/4" x 1/2", which should work. I was hoping for something longer while keeping the 1/2" (for the part that connects to the case), but no such luck! The one nice thing about this setup is it allows me to change the actual shelf part pretty easily.

If only I would have tried to build this part before I actually built the PC. I keep wanting to avoid tearing down my PC to only sit there and rework something. :p
 

Aikouka

Lifer
Nov 27, 2001
30,383
912
126
One thing that I forgot to ask about. I was thinking about my fan orientation at the bottom, and I was wondering if I have it setup in the most optimal way.

As a refresher, I have three radiators: two at the bottom (480, 240) and one at the top (360). At the bottom, the 480 is on the window-side of the case pulling air in, and the 240 is on the opposite side pulling the air out. Should that 240 be pulling air in too? If it did that, I'm guessing that the warm air would just travel upward where it combines with the 2x120 front fans that are pulling air in. I also turned the rear 120mm fan around so it pulls in air for the 360 at the top to push out.
 

smackababy

Lifer
Oct 30, 2008
27,024
79
86
One thing that I forgot to ask about. I was thinking about my fan orientation at the bottom, and I was wondering if I have it setup in the most optimal way.

As a refresher, I have three radiators: two at the bottom (480, 240) and one at the top (360). At the bottom, the 480 is on the window-side of the case pulling air in, and the 240 is on the opposite side pulling the air out. Should that 240 be pulling air in too? If it did that, I'm guessing that the warm air would just travel upward where it combines with the 2x120 front fans that are pulling air in. I also turned the rear 120mm fan around so it pulls in air for the 360 at the top to push out.

I would probably use all radiator fans to pull air inwards, and expel the air with any remaining fans. My current setup as two radiators pulling air in and a single 140 pushing air out the back.

You generally want the coldest air possible coming into your radiator, but realistically, I don't think you'd see a difference pulling in air on the bottom 360 and pushing it out on the bottom 240. I'd keep the top sucking in air and pushing air out the back of the case though.
 

Aikouka

Lifer
Nov 27, 2001
30,383
912
126
I would probably use all radiator fans to pull air inwards, and expel the air with any remaining fans. My current setup as two radiators pulling air in and a single 140 pushing air out the back.

You generally want the coldest air possible coming into your radiator, but realistically, I don't think you'd see a difference pulling in air on the bottom 360 and pushing it out on the bottom 240. I'd keep the top sucking in air and pushing air out the back of the case though.

The 360mm radiator at the top is currently pushing air out. I swapped the rear fan around so it's pulling air in rather than exhausting it, which should be feeding the 360mm radiator with cooler air.
 

smackababy

Lifer
Oct 30, 2008
27,024
79
86
The 360mm radiator at the top is currently pushing air out. I swapped the rear fan around so it's pulling air in rather than exhausting it, which should be feeding the 360mm radiator with cooler air.

The air from the single 140 in the rear isn't going to be enough for the 360 at the top to always be using cool air. I'd switch those around. That will also give you positive pressure and help keep dust from getting in any openings.
 

Aikouka

Lifer
Nov 27, 2001
30,383
912
126
The air from the single 140 in the rear isn't going to be enough for the 360 at the top to always be using cool air. I'd switch those around. That will also give you positive pressure and help keep dust from getting in any openings.

There are two other 120mm fans at the front if those help any. Although, the drive bays are in the way, so they probably don't do that much, and the drives probably heat them up a little bit.

I can swap the fans around though. Fortunately, that radiator isn't connected to the case through the fans, so I can swap it easily!
 

Aikouka

Lifer
Nov 27, 2001
30,383
912
126
Spent some time ensuring my "shelf" was the right size, and I tore most of my PC out. I've currently got the shelf "ghetto rigged" in the PC with scotch tape holding the brackets up. This gave me a good chance to see exactly what I might need to go from the pump to the radiator, and it looks like all I'll need is one of my bulkier 90-degree fittings and some tubing... nice! As for the usual bummer of the build, I noticed that I must have bumped into my 480 rad a time or two because some fins were bent. :(

Anyway, here's a photo of my wonderful tape job: http://i.imgur.com/TeLkh1p.jpg

Once I get all the holes drilled for the brackets, I'll drill the mounting holes for the shelf, and then the holes for the pump's mount.
 

Aikouka

Lifer
Nov 27, 2001
30,383
912
126
I thought that I'd give a bit of an update. Everything is up and running at this point. Although, I'll admit that it wasn't really as easy as I expected. The big issue with my little shelf is that it goes right over top of where the ports are on my 240mm radiator, which meant it was that much more fun to deal with those fittings/adapters. :p I think I'm going to cringe whenever I see my 480mm radiator again, as I know I bumped into the fins a few times with my hand! Those things are like fragile little flowers. :|

I had some slight issues when I tried to get it working at first. During the first fill, the 480mm's inlet was leaking for some reason. I got that cleaned up, and decided to swap the fittings used on that tubing run. Well, I thought I had it on... until I saw that the 240mm's outlet was leaking! :p I realized that was because I simply had a lot of trouble tightening the 90-degree swivel adapter that I had to use there, and my makeshift shelf wasn't helping either. The only way I could easily fix that was to swap back to the Monsoon fitting, which worked fine. After that, things were mostly fine (more on that in a second). The best part about this swap is the seriously large reduction in tubing. I had one run (480mm to GPU) in my previous setup that had to be close to two feet long! :eek: In my current setup, I think the longest run (GPU to 360mm) is maybe 6-8 inches.

Unfortunately, I do still have one issue that I can't seem to figure out. I use two Swiftech MCP655-PWM pumps, and as you'd expect, you're supposed to plug them into a PWM header. However, if I plug them into my Maximus VI Formula's CPU-FAN header, the pumps turn off. Yes, they don't just slow down or something, they shut off completely, and it doesn't matter if the computer is on or not. Fortunately, if the pumps aren't connected to a PWM header, they just run at 30% power, which is good enough for me. :p

I took a bunch of photos of the build that I'll post later today.

So, as smack said, you're never done, and I'm already thinking about what I'd like to do at some point in the future (i.e. not soon). My biggest problem is pretty much where to put the pump. My shelf solution is okay, but as I mentioned above, it has some drawbacks. Some possible solutions to this would be removing the 240mm radiator and dropping the pump straight down to the case's floor. If I want to alleviate the reduction in performance, I could upgrade the top radiator to a 480mm. In that case, I would just lose 120mm of radiator space. Another possible solution would be to go back to a bay reservoir, and I would pretty much have to use the dual-pump Koolance model. The one thing that would keep me from going back to a bay reservoir is that they tend to cause rather long tubing runs. The only way that I could probably alleviate that would be with a split-radiator, but those have worse performance.
 

Aikouka

Lifer
Nov 27, 2001
30,383
912
126
The pump problem could be how your MB is handling them. I'd check some of your bios settings.

I thought about that, but even after tinkering with some BIOS settings, nothing seemed to help. It was fun switching between the fan settings and the temperature monitor page to ensure my CPU wasn't going to burn up in some inferno. :p It only ever got to around 52C before I would shut the PC off, so no problems there.

Anyway, I'm not so sure if it's a BIOS thing because when I jump-started the power supply with only the pumps powered, they still wouldn't turn on with the PWM header plugged in. There are two things that I ought to try: (1) plugging the PWM cable into another motherboard (like my old ASRock Z87) to see what happens (2) plugging the PWM cable into a different PWM header on the motherboard. The manual states that all headers but CPU-OPT support PWM.
 

smackababy

Lifer
Oct 30, 2008
27,024
79
86
Hmm. Perhaps the pump itself is awaiting some kind of confirmation from the MB and that is why they are not running.
 

Aikouka

Lifer
Nov 27, 2001
30,383
912
126
Alright, I promised some photos, and here we go! To note, I won't be inlining them for those that may be on a slow connection or whatever:

Full, Frontal Shot - You can see most of the runs in this photo.

Reservoir + Fill Port - To note, the white 90-degree adapter will be replaced eventually. I normally using a swiveling 90-degree adapter for the fill ports, but I needed that one for the CPU. :(

Reservoir to Pump - In this photo, you can see the return line coming from the CPU block to the reservoir and the reservoir's line going straight to the pump.

Dual Pump - Here are the two Swiftech pumps sitting on my somewhat shoddy custom shelf.

240mm Radiator - The pump goes directly into the 240mm radiator below, which on its outlet, branches out in a t-line. The t-line has a quick disconnect for easy draining and also a 90-degree swiveling adapter for the transition to the next radiator. That swiveling adapter was where I had my second leak, and I pretty much was unable to tighten it until I replaced the fitting.

480mm Radiator - I really like how the 240mm to 480mm line just seemed to line up rather nicely. The fitting on the inlet is where I had my first leak, and I'm still not sure what wasn't properly tightened. I replaced the fitting and put it all back together.

480mm to GPU - This is a pretty short run that actually worked out rather well. It uses a slim 90-degree adapter that transitions to a 30-degree adapter on the Multi-GPU bridge.

GPU Outlet - Here's the outlet on the GPU that heads up to the 360mm radiator. This is probably the longest run in the system!

360mm Radiator - You can see the run coming in from the GPU to the radiator's inlet and the outlet going to the CPU via 30-degree adapter.

CPU Block - Here's the CPU block, which is currently using two swiveling 90-degree adapters. The one on the top is the inlet, which comes from the 360mm radiator, and the one on the bottom goes back to the reservoir. I don't really like having the swiveling adapter on the top, because I don't really like the awkward arcing bend. So, the plan is to replace it with an extension + slim 90-degree adapter. I originally had that there until I pulled that adapter off to go on the reservoir's return line.

Behind the Motherboard - You can see all the wonderful cable routing and the dangling PWM cables that aren't attached. I'm not too keen on where the pumps are plugged in (those molex connectors on the right), but I'm not too sure where else to put them right now. Thinking about it, I might remove the fancier, individually-sleeved cables and just use the thinner black ones that came with the PSU.

Custom Reservoir Bracket - You can see the back plate for the custom reservoir brackets here. There are two of those plates per mount, and they're separated by the grommets. I'd really like to buy spare grommets and modify them so I can use spacers to help support the brackets, but at this point, you cannot buy spare grommets for the 900D. Here's an example of what I was originally going for with the spacers. The finished setup also uses rubber washers between the reservoir mount and the inner bracket, and the brackets have been sanded down a bit.
 

Aikouka

Lifer
Nov 27, 2001
30,383
912
126
Oh, goodness... Emma Stone is giving me the thumbs up? Be still, beating heart! :wub:

Haha yeah... it's definitely come a long way. There are things that I would prefer to do like my own sleeving. I actually like that black look with only a wire or two in the color that you want, but that would have taken forever on the 24-wire ATX power cable. ...or, I could just buy a black, pre-sleeved one and simply change a few wires to the color I want. :hmm:

Although, every time I look at my photos, I see the pump, and I don't really like seeing it. It probably doesn't help that while the sponge thing is effective at reducing vibrations, it's rather unsightly! So, I may end up doing the whole drop it down thing at some point, which will include the removal of the 240 and the replacement of the 360 with a 480. That shelf also blocks one of the paths from the PSU area up the back of the case, so getting rid of it will clean up the wires a little bit. Good thing I spent all that time and money building that shelf! :biggrin:
 

Aikouka

Lifer
Nov 27, 2001
30,383
912
126
So, I found that if I plug the PWM plug into the Chassis Fan 2 header, it works... sort of fine. Oddly enough, it seems that no matter what I adjust, the pumps are always running at what appears to be max speed. Not like it does much though as my CPU temps are pretty flippin' terrible for being on water. Either I must have been a little too reserved with my PK-3 thermal paste or something is a bit awry. I'm pretty much doing nothing, and my cores are varying between 34-40C. I got better than this with the stock cooler. :|

EDIT:

Honestly, this whole thing is just frustrating the hell out of me. It seems that no matter what I do, something is always broken or not working properly. I've got a persistent, random BSOD problem, and even with my fifteen years of working with PCs, I can't seem to even find out what's wrong. I've got two PWM pumps that don't work in the CPU header, operate at max speed in any other header, or at 30% when not plugged in. I've got a 90W CPU that runs hotter than my 200W GPUs when under less load, and I have no idea if I didn't use enough thermal paste, or I didn't tighten it enough, or if I'm just being screwed by Haswell's ridiculous heat output. I'd like to delid the CPU, but until I can figure out what's causing my BSODs (it could be something weird in the CPU and delid = no more warranty), that's off the table.

In other words, if I just left the GPUs on water and put the CPU on air, I would be so much happier. :p
 
Last edited:

dma0991

Platinum Member
Mar 17, 2011
2,723
1
0
Its the TIM under the IHS that is causing the temp difference between each core. As for the high idle temps, still highly dependent on ambient temperatures. Water cannot go lower than ambient unless its chilled. Mine idles about 27-35C on average at stock settings.

Note the large variance in temperatures of my Core i5 3570K cores. The variance is not this big on my Core i7 2600. If I had to give an estimate, SB varies about 1-5C while IB is about 5-10C, because of its smaller footprint for heat transfer and made worse by the crappy TIM.
 

Aikouka

Lifer
Nov 27, 2001
30,383
912
126
Its the TIM under the IHS that is causing the temp difference between each core. As for the high idle temps, still highly dependent on ambient temperatures. Water cannot go lower than ambient unless its chilled. Mine idles about 27-35C on average at stock settings.

I think it's the wild difference between GPU and CPU temperatures that gets me. When my GPUs are barely doing anything, they're usually within about 2-3C of the water temperature. When I play a game and observe their usage, both GPUs being used between 50-80% usually results in around 38-44C. I stressed my CPU with the IBT, and I saw temperatures at about 70-71C. When you've got AIO water cooling users posting mid-to-high 70s, that's a real buzzkill. Am I pretty much stuck with this poor performance level until I delid?

I also recall that my temperatures were better before this rebuild, which makes me wonder if my skimpier TIM application may have something to do with it. Prolimatech's instructions for PK-3 state that you're supposed to spread it yourself unlike most TIMs where you let the heatsink's compression handle that. Every time I removed my block in the past, it seemed like there was a little too much, but perhaps I was just over-thinking it.

Also, I think I'm just going to unplug the PWM cable. I had no noise at all with the pumps at 30%, and I don't just mean pump noise, but also any sort of noise from air in the loop. With them at full speed? I get that dastardly air-in-the-loop crunching noise. I've read that turbulence can cause air that may be in the reservoir to go into the loop, which is probably what's happening. I have the Primochill ZenCoils in my reservoir, but they don't seem to do much.
 

dma0991

Platinum Member
Mar 17, 2011
2,723
1
0
Am I pretty much stuck with this poor performance level until I delid?
The use of TIM instead of solder is a proven cause of unnecessarily high temps in IB and HW. Delidding is the easiest method (complex = phase change, etc) of getting lower temps but its entirely up to you if you're willing to void warranty and/or end up with an expensive keychain. I didn't delid mine because I needed it to be reliable and trouble free.