Although I have been building my own PCs for the better part of twenty years now, the liquid cooling application is not one I have ever implemented. When these first started popping up in the general market I did not see much need as someone that at the time did not overclock (in point of fact, I no longer bother with OC-ing anymore). There was also the fear of (insert expletive) it up and destroying my rig… or at a bare minimum creating a nasty mess and a few moments of panic. Of course last but most certainly not least: the cost is extremely inhibitive. Sadly, I did not even realize the full tally until I started buying up my pieces. Honestly, if you have never done this and want to do so... think long and hard. Even if I figure for inflation and other factors... the cooling blocks, tubing, radiator, pump, PC gnomes to monitor temps, etc. added up to more than I have previously spent on most PCs (I know, FWP right). Again, I knew it was costly… I just did not realize exactly HOW costly.

Anyway, even with a fair amount of experience under my belt, this attempt has me a bit nervous. I really have no interest in flooding my chassis and frying my components. I also know that it will not literally “fry” them… but water and electronics have never really been the greatest of bed fellows so I still have my reservations here.

What I am hoping some of you might have is some, "wish I would have known this back then" kind of wisdom for this rookie before I start the project this afternoon or tonight. Tips for priming would be useful as I seem to find varying philosophies on that from one thread/forum to the next (and I thought that would be a sort of one size fits all kind of procedure). Another question I have is in regards to mapping. Logistically it appears I will have to run from RAD ► CPU ► RAM ► GPUs. However, it would seem to me that in practice hitting the RAM last would be ideal. However, as I mentioned logistically that does not seem practical... it would leave me with a lot of tubes crisscrossing up and down through the chassis also making it look poorly planned (IMO). My CPU sits upper left/center, and is poised between the DIMM slots (4 to the left, and 4 to the right), and the dual SLI standard for this MB is slots 1 and then 3 and only using slot 2 in a three-way (Mmmm…. Three-way). Rather than try to paint a pic, I suppose I will just give you one here.

Three things with regards to the .png. I am clearly not a graphic designer. Second, I do not believe the arrows accomplish portraying the potential eye-sore I see in Option 1, nor do I feel it accurately displays the additional real estate I think the tubing would occupy. Finally, I will not apologize for the cheesy Font… it stands out… that’s all I really needed it to do. With regards to the second concern, try to image how polluted this may look with 5/8” tubing in place of the arrows. It may also be worth noting that instead of rigid tube or a bridge, I am opting for a terminal connection between the two graphics cards. I have found fairly consistent opinion that these provide greater flow rates than the alternatives. Ultimately, I am primarily concerned with optimal cooling here… so does anyone have any strict opinions on the flow routing here? Other suggestions tied to the flow (i.e. throw in another radiator)?

Seriously, any help will be appreciated. I really want to do this right. Feel free to ask for any pertinent details you feel are necessary to impact your perspective.

P.S. Should I feed the gnomes... or do they receive their sustenance through the consumption of all the single socks they jack from my clothes dryer?

9 years ago*

Comment has been collapsed.

I know nothing about stuff like this, so have a bump

9 years ago
Permalink

Comment has been collapsed.

Thanks, I am grateful for the bump... anything to get more eyes and possibly some insight again is greatly appreciated!

9 years ago
Permalink

Comment has been collapsed.

Just get this and call it a day - Click Here

9 years ago
Permalink

Comment has been collapsed.

That is actually what is presently cooling my CPU. It is sufficient since I do not overclock any more these days. However, moving to the open v. closed loop is more of an endeavor to satisfy my curiosities. Not to mention the fact, that it is my second closed loop cooler, and as someone that really enjoys the challenges of a build... I am more interested in the new challenge at this point that anything. I suppose I'll end up tossing my H100i on Craigslist for $50 or $60 (more if I don't toss back on those noisy stock fans).

9 years ago
Permalink

Comment has been collapsed.

If you're not overclocking your CPU, I would just stick with some non-conductive thermal paste and the Hyper 212 heatsink. Though I suppose if you have the lust to build exotic things then it can't be helped. To me, 'exotic' is a micro-ATX case, haha.

9 years ago
Permalink

Comment has been collapsed.

Having built quite a few watercooled rigs in my time, here's my two cents.

  • First, your biggest asset/biggest hurdle will be your case. Investigate and invest wisely. Corsair currently makes the best for watercooling imo.
  • Second, make damn sure the tubing and the connectors are compatible on every. single. component. Try to get your parts from one dealer/manufacturer if you can. Oh, and for the video cards, don't use a connector block. Use tubing that can be clamped off. If one of them fries and you need to dismantle the system, that block will be a major pain in your ass. This means don't mix pre-installed and after-market watercooling components.
  • Third, mapping the water flow is definitely important, but don't neglect air flow as well. You'll a push/pull system to maximize flow through the radiator. Know your fan directions.
  • Fourth, reconsider cooling your RAM. You probably don't need to overclock it, and if you don't the sticks shouldn't run too hot. A properly ventilated case may be sufficient. Get an air cooler at most.
  • Fifth, are you getting a reservoir? If not, get one immediately. And make sure there's a system in place so you can easily refill the circuit as needed. You will need to refill after a few weeks/months, as the initial bleeding process won't get rid of everything in the first go. If you're not refilling by opening the reservoir, such as if you have input tubes like I do, remember that you'll need to open the system somehow so that when you pour it doesn't take 10 hours to trickle through the air bubble in between you and the reservoir. That means two tubes, one to fill, one for pressure release.
  • Fifth, stock up on extra water additive. Preferably pre-mix the stuff and keep it in storage until needed. Add a little more than the recommended mix. You don't want algae buildup in there.
  • Six, if you're doing everything correctly, it'll never leak. Ever. Still, if you want to test, get saran-wrap, some tape, and cover your components while you test/dismantle the loop.
  • Seven, stock up on extra parts and get tube clamps. Have the tools you'll need to dismantle the system, and pray you'll never have to.
  • Eight, mapping has no appreciable impact. The temperatures in the loop will be relatively stable throughout. Go for a balance of the most straightforward and shortest path. Also don't worry too much about elbows. Avoid if you can, don't despair if you can't. A longer total line kills flow more than a couple elbows will. That also means don't add valves. Just use clamps when you need to cut off flow.
  • Nine, TEST YOUR HARDWARE WITH STOCK COMPONENTS BEFORE YOU ADD WATERCOOLING. Last thing you want is to find that your video card is defective or that the mobo doesn't support RAID on those particular plugs or whatever when you already spent however many hours putting it together.

I may come up with more stuff later. For now, I think this'll do.

9 years ago
Permalink

Comment has been collapsed.

Awesome SimonSays... that is hands down the most comprehensive set of notes I have been able to get... and these are exactly the kind of things I am looking for right now!

I am especially interested in the input tubes. Sadly this is the first I have heard of those... and I really have been running searches like a madman on the subject. I probably should have just opted to find a reliable book or the like. I picked up a two-pump reservoir combo, but it will reside in the 5.25” bay… and refilling there would likely prove a greater pain than I would prefer to manage. I also picked up two 1000 ml bottles of the pre-mix (just over a quart a piece). I opted for two because FrozenCPU was out of stock, and Micro Center only had the two bottles… best to be prepared in case I ran short and then nobody had any on hand!

Anyway, some of the things you mentioned above I have already done or put in place. When I went into this build I really wanted to ultimately water cool (even though I still had not 100% committed to the investment at that point). Therefore I started eyeballing the cooling components and cross referencing cases for compatibility. My final chassis choice was the Corsair Carbide 780T. It provides excellent ventilation, and I can house either a 360mm or 280mm radiator topside or front. I have opted for the top mount, and my air cooling fans will remain in front.

This brings me to your ninth point (forgive me if I kind of hop around a bit). The rig has been up and running for a few weeks now, but admittedly that is only by coincidence as my GPU blocks were back ordered. So thanks for the tip. I am certain it seems like a no brainer to you, but honestly... knowing myself... I think that is one of those lessons I would have needed to learn the hard way. The air cooling is sufficient, but I have not run any full load testing to see how the temps fare under stress. Thus far the hottest temp I have recorded is 73° C on my second GPU. It's not high; but I think there are two things worth noting here. The first is as I mentioned in my reply to StarshipGhost's advice above, my primary focus here is on the experience in the build. Admittedly, my current components are not likely to require the water loop unless I do some heavy OC-ing. The Haswell-E is rated at low average temps, and the GTX 980’s are running on 30% less juice and are inherently (and significantly) cooler than their competitors and predecessors. Just the same, that 73° was hit while my son was tackling Garden Warfare… not exactly a title I would call a card killer. The primary card was still under 50° C, but perhaps the second card will benefit for the liquid cooling under higher strains.

Now, back to points two and three, everything (with the exception of the RAM spreaders) is EK and my tubing and fittings are all in sync. I also intend to maintain my front 140’s and my rear 140 in a push (front) pull (rear). If I opt to pull my HDD cage I also have the option to insert two additional 120’s on the bottom, but I am not considering that option at the moment. Hopefully they will not be necessary.

In regards to the RAM, I will not be overclocking. In fact, with the DDR4 performance (IMO) is likely to be sufficient for a while with high demand games. Again, the decision here really rested on my desire for the experience more than anything… but I did consider a return to some direct cooling fans… I just figured, “I’m all in” at this point, hence the RAM blocks.

Another insight you have given to me of great value is the tubing suggestion between the cards. Again, this is one of those lessons I would otherwise likely have learned the hard way. My guess would be that the terminal would still perform well as all three blocks are EK and compatible… still, your point makes me consider the value in being able to easily cut off that flow. I will be weighing that one heavily now… but it gives me an excuse to head back to Micro Center which I generally enjoy like a fat kid loves cake. I will also make a point to pick up some extra parts. I had considered that… but with costs rising quickly… I was justifying my failure to pick those up at the start… you’re definitely right though, and I am certain the regret later would be worth avoiding.

Once again, thanks for the tips. Looks like I will be holding off a little bit longer while I weigh out how I want to proceed with the GPU link. I had actually hoped to be constructing this last night, but a failing power steering pump diverted my attention… thank god that part’s placement on my current vehicle was readily accessible so it only took my evening hours… I’ve spent an entire day on one of these for someone because they freakin’ buried the part… and pretty much all access to the engine for that matter. Sorry… totally off topic there! Again, you have my gratitude and I will be checking back in throughout the day to see if you have any other thoughts and advice.

9 years ago
Permalink

Comment has been collapsed.

"I am especially interested in the input tubes. Sadly this is the first I have heard of those... and I really have been running searches like a madman on the subject. I probably should have just opted to find a reliable book or the like. I picked up a two-pump reservoir combo, but it will reside in the 5.25” bay… and refilling there would likely prove a greater pain than I would prefer to manage."

My experience has always been with stand-alone reservoirs and pumps. My current setup has two large cylindrical reservoirs, similar to EK's Multioption X2 if they aren't already that specific model. Because of the multiple flow-path options I opted to connect an extra tube from each routed to the back panel, because direct access to those reservoirs was unfeasible. Then I learned first-hand how air-pressure works. I have to take small-nose pliers to open a cap on each reservoir in order to fill them in a timely manner.

If you're wondering why I have two reservoirs, it's because I have two completely separate loops running on my rig (one for video cards and motherboard chipset/voltage regulators, another for CPU and RAM). Yes it's monumental, and yes it's monumentally stupid. The Corsair 800D I pack these in is a massive case, and even it could only barely handle everything I put in it. Sure on a rainy day with the window open this thing will never rise above body temperature even under load, but it weighs a ton and is a pain in the ass to move and dismantle. A lot of what I said earlier is based on my experiences with this setup.

"My final chassis choice was the Corsair Carbide 780T. It provides excellent ventilation, and I can house either a 360mm or 280mm radiator topside or front. I have opted for the top mount, and my air cooling fans will remain in front. "

This will probably work well for you. If you hadn't already got it I'd have advised a larger case such as the 900D, just to have more space to work with. Then again I've never used drive bay reservoir/pumps. Smaller cases are also easier to ventilate and drain. I'm a little concerned that there are only two drive bays available. One for watercooling, another for what I presume will be an optical drive. No fan controller. I myself wouldn't be able to live without my Zalman ZM MFC3. Then again, fan controllers are typically just more wires to tangle with and inevitably fuck up at least twice.

Top-mounting the reservoir is usually the best option in most cases. If you want to go really bonkers, do what I did and cut holes for another intake fan in the side-panel plexiglass. It's delicate work but having cool air flowing directly onto the motherboard is reassuring.

"The air cooling is sufficient, but I have not run any full load testing to see how the temps fare under stress. Thus far the hottest temp I have recorded is 73° C on my second GPU. It's not high; but I think there are two things worth noting here. The first is as I mentioned in my reply to StarshipGhost's advice above, my primary focus here is on the experience in the build. Admittedly, my current components are not likely to require the water loop unless I do some heavy OC-ing. The Haswell-E is rated at low average temps, and the GTX 980’s are running on 30% less juice and are inherently (and significantly) cooler than their competitors and predecessors. Just the same, that 73° was hit while my son was tackling Garden Warfare… not exactly a title I would call a card killer. The primary card was still under 50° C, but perhaps the second card will benefit for the liquid cooling under higher strains."

Haswell-E. 980's in SLI. Not exactly a potato is it. Having water cooled for so long (since '04) I can't recall ever having a component at 73 degrees. That sounds functional but not healthy. Another benefit of watercooling that many people miss is that lower temperatures help components last longer without failure. Usually. Theoretically.

Also, look into upgrading to a G-sync capable monitor at some point in the future. Haven't had the chance to experience it for myself but all reports say it's The Future™.

"Now, back to points two and three, everything (with the exception of the RAM spreaders) is EK and my tubing and fittings are all in sync."
That's what I would have recommended. EK are awesome.

"In regards to the RAM, I will not be overclocking. In fact, with the DDR4 performance (IMO) is likely to be sufficient for a while with high demand games. Again, the decision here really rested on my desire for the experience more than anything… but I did consider a return to some direct cooling fans… I just figured, “I’m all in” at this point, hence the RAM blocks."

RAM these days is usually never the bottleneck, even at standard speeds. And then only for games that are not graphically challenging. With a Haswell-E and two 980s, the 10fps in Skyrim you'd gain from fast RAM won't be noticeable since you're already pushing over 100 anyway.

DDR4 is a bit of a mixed bag at the moment, much like how DDR3 was when that first came out. Sure it's fast, but the currently available latency timings are trash in comparison so there's little noticeable performance gain over DDR3. That will be fixed down the line though. DDR4 does run at lower voltages though, i.e. cooler. So that's great. You'll also likely not have to replace your motherboard when you build your next rig a few years down the line and DDR4 becomes mainstream.

"Another insight you have given to me of great value is the tubing suggestion between the cards. Again, this is one of those lessons I would otherwise likely have learned the hard way."

I learned it the hard way so you wouldn't have to. One of my video cards started blue-screening when SLI was enabled. It was a two-man job to pry the whole contraption out, with each of us prodding the pci-e locks with screwdrivers. And since EVGA's replacement card wasn't the same model (and had an extended PCB that my EK block wasn't compatible with) I had to cobble together extra tubing to go to the one remaining card.

But I think I may have exaggerated the scale of the problems associated with triple-blocks. When the card failed, it's a massive inconvenience. But only an inconvenience. I had an over-sized case, stuffed with an insanely overcomplicated system, having suffered hardware failure that may have been my idiotic tampering, and inconvenience from EVGA due to the PCB dimensions. None of these conditions apply to you. As a means of cutting tube clutter and improving aesthetics it was marvelous when everything was working, so that experience won't stop me from re-using it in my next rig. I wouldn't recommend a triple-block for beginners, but only because I recommend starting with loops being as compartmentalizable as possible. You already have the block, so just use it. At least you know what to do if something does happen.

"still, your point makes me consider the value in being able to easily cut off that flow."

It sure beats having to drain the entire loop whenever something needs changing. But again, get tube clamps, not valves. And while a shorter loop is better, always give some generous slack.

"I will also make a point to pick up some extra parts. I had considered that… but with costs rising quickly… I was justifying my failure to pick those up at the start… you’re definitely right though, and I am certain the regret later would be worth avoiding."

Spare parts are like guns and condoms. It's better to have and not need them than to need and not have them.

Oh, and a few more things.

  • If you have trouble slipping the tubing onto the radiator ports, trying putting the tubing under very hot running water to soften and expand it. This won't save your fingers from getting sore but it will save time.
  • Never, ever, ever used colored tubing. Likely not an issue in your case, but for anyone else reading, when algae starts growing in your loop you'll want to know about it.
  • On the subject of algae, don't flush chemical cleaners (alcohol, vinegar, whatev) through the line. Those will corrode your blocks. You'll need to drain, dismantle, and manually clean your system if it's in the tubes.
  • Put rubber between the pump/reservoir and whatever it's resting on/against. It's going to vibrate and if it's not properly set you'll have to suffer annoying noises. There's no general solution. giant pads, tiny feet, you'll need to experiment.
9 years ago
Permalink

Comment has been collapsed.

As has already been pointed out, overclocking your RAM isn't really worth it. The performance gains once you go above 1600 are minimal, and really if you have a well-ventilated case (and perhaps RAM with heat fins? I dunno) then its temps really shouldn't be a problem. RAM is weird like that. Above 8GB you won't see much use unless you're a heavy multi-tasker or into video editing or 3D rendering. GPU RAM I'm not so sure of, but the GPU generally benefits nicely from liquid cooling solutions so that should be taken care of as part of your set-up anyway.

And of course, to state the obvious (even as a non-liquid cooling guy), it's nearly always better to pay a little too much than not quite enough, especially when it comes to quality construction. Like you said, liquid and electronics don't really mix!

Just promise me you won't get into that 'submersed motherboard' thing. I still cringe just at the thought.

9 years ago
Permalink

Comment has been collapsed.

Sign in through Steam to add a comment.