Custom 3 Loop System

[Eddie Current]

This has been the final plan to my water cooled rig. The purpose of doing this three loop system is three fold. One is to do away with all fans other than on the radiators, the second is be able to OC the different components without affecting the others (in the WC loop) and the third is so I can use any case I want (small/med) without changing the cooling systems main components.

 

[GTXJackBauer]

This has been the final plan to my water cooled rig. The purpose of doing this three loop system is three fold. One is to do away with all fans other than on the radiators, the second is be able to OC the different components without affecting the others (in the WC loop) and the third is so I can use any case I want (small/med) without changing the cooling systems main components.

View attachment 171720

Bare with me ED.

1. You're not doing or adding less when you're going multi-loop. You're adding way more than it needs to be, especially cost. If you're doing it for aesthetics and have deep pockets, it's one thing but if you're doing this for .05c-1c+ savings, well than, you're nuts or haven't done enough research at the right places so you can see once the system with the right configuration, it will reach a equilibrium and won't need as much as you think it will need or show such a huge change for it.

A single 480mm rad has no problem cooling a CPU + GPU loop. You also don't need to cool the MB.

2. There's more than enough rad in a 480mm on a CPU + GPU loop for OC. Add another 240mm tops if you're going for a very high OC (5.0Ghz+) and want a silent rig and want great delta temps. You'll must likely hit diminishing returns with that much rad me thinks for only a CPU + GPU loop on 120.6 for example.

3. A small or medium case will never fit a triple loop. Only if you're configuring your route externally only to cool the mains internally. Even a super case might run into a cluster (beep).

Think this through again. Don't over think this and ask away.

Also recommend to check out Skinnelabs and Martinsliquidlabs for more valuable data.

 

[Eddie Current]

Bare with me ED.

1. You're not doing or adding less when you're going multi-loop. You're adding way more than it needs to be, especially cost. If you're doing it for aesthetics and have deep pockets, it's one thing but if you're doing this for .05c-1c+ savings, well than, you're nuts or haven't done enough research at the right places so you can see once the system with the right configuration, it will reach a equilibrium and won't need as much as you think it will need or show such a huge change for it.

Well I think .05 to 1.0c is a lot to leave on the table, but .05 to 1.0c is a very conservative number in my opinion. But everyone has a right to their opinion.

A single 480mm rad has no problem cooling a CPU + GPU loop. You also don't need to cool the MB.

2. There's more than enough rad in a 480mm on a CPU + GPU loop for OC. Add another 240mm tops if you're going for a very high OC (5.0Ghz+) and want a silent rig and want great delta temps. You'll must likely hit diminishing returns with that much rad me thinks for only a CPU + GPU loop on 120.6 for example.

Don't let the radiators in the diagram keep you from see the big picture. They could be all 240's or two 240 and a 120, these are what I have and have chosen to run. Keep in mind that if I want to "some day" run two GPU and a multi CPU MB I'm covered.

3. A small or medium case will never fit a triple loop. Only if you're configuring your route externally only to cool the mains internally. Even a super case might run into a cluster (beep).

A small to Medium size case has a hard time with a single loop, remote all the pump(s), radiator(s), reservoir(s), misc., add a few quick connects and bang run any case you want.

Think this through again. Don't over think this and ask away.

Jack its time to step away from the Kool-aid dispenser and return to the out of the box thinking!

Also recommend to check out Skinnelabs and Martinsliquidlabs for more valuable data.

All good reads Jack, they are a little outdated but good reads and I read them way before ever coming here.

 

[sporktar]

Well I think .05 to 1.0c is a lot to leave on the table, but .05 to 1.0c is a very conservative number in my opinion. But everyone has a right to their opinion.


Don't let the radiators in the diagram keep you from see the big picture. They could be all 240's or two 240 and a 120, these are what I have and have chosen to run. Keep in mind that if I want to "some day" run two GPU and a multi CPU MB I'm covered.


A small to Medium size case has a hard time with a single loop, remote all the pump(s), radiator(s), reservoir(s), misc., add a few quick connects and bang run any case you want.



Jack its time to step away from the Kool-aid dispenser and return to the out of the box thinking!


All good reads Jack, they are a little outdated but good reads and I read them way before ever coming here.

I wouldn't dismiss Jack so quickly, he definitely knows what he's talking about, and I tend to agree with him here. If you have a lot of money to throw onto this build then by all means go for it, but your main takeaway from this is going to be aesthetics, and there are several downsides. it's important you know this going into this project. Temperature effect from split loops is pretty minimal compared to a single loop with adequate rad space (~1c range) and there are other inefficiencies that come with split loops such as clutter, maintenance, rad space heat distribution, and cost

Even for fairly heavy overclocking, passive motherboard cooling is fine, watercooling the motherboard is pretty much for looks only. You can make a better argument watercooling ram, but unless you're pushing into extreme overclock territory, you really don't need it, it's just for show. If you are still deadset on watercooling these components, then by all means do so (many people do), but they certainly don't need their own loop, they won't even come remotely close to utilizing the full potential of a 240 rad. that rad space could be better spent sharing the heatload of the gpu or cpu. Managing all those pumps and res's is going to make small cases impossible to work with, and mid-size cases will at the very least be a pain, and a huge mess. Moving the components outside of the smaller cases to free up space makes the flexibility of smaller sized rads pretty much irrelevant, and you still have to manage power going to the external enclosure, along with 6 tubes going to and from your case. Not impossible to manage, but not exactly easy ether.

I would also recommend against trying to plan an entire loop for future builds. By the time that comes around, you will probably need to replace the tubing (assuming it's soft tubing since you're going for modular design) anyways, and tubing management is near impossible if you aren't custom fitting the lengths to your current project. Add that to the potential for having to buy different waterblocks for different cpu mounting layouts, GPU's, and the cost of quick disconnects (i have 2 on my build and they cost me nearly 50 dollars in total) the costs far outweigh the benefits.

I think you should really consider your priorities with this build and make sure this is the right plan for you. The cooling benefits of splitting this loop up are negligible, and the cost is significant in comparison to a single, or even dual loop. That money could be better used to upgrade components, or buy fancier watercooling hardware. If you still think that your plan is the course of action you want to take, then do it and post pictures for us to see!

edit: i just realized that the radiators listed are 480's. Even with 2x 480's you could run your fans at 25% speed dead silent on a single loop and be perfectly fine.

 

[EarthDog]

Bare with me ED.

1. You're not doing or adding less when you're going multi-loop. You're adding way more than it needs to be, especially cost. If you're doing it for aesthetics and have deep pockets, it's one thing but if you're doing this for .05c-1c+ savings, well than, you're nuts or haven't done enough research at the right places so you can see once the system with the right configuration, it will reach a equilibrium and won't need as much as you think it will need or show such a huge change for it.

A single 480mm rad has no problem cooling a CPU + GPU loop. You also don't need to cool the MB.

2. There's more than enough rad in a 480mm on a CPU + GPU loop for OC. Add another 240mm tops if you're going for a very high OC (5.0Ghz+) and want a silent rig and want great delta temps. You'll must likely hit diminishing returns with that much rad me thinks for only a CPU + GPU loop on 120.6 for example.

3. A small or medium case will never fit a triple loop. Only if you're configuring your route externally only to cool the mains internally. Even a super case might run into a cluster (beep).

Think this through again. Don't over think this and ask away.

Also recommend to check out Skinnelabs and Martinsliquidlabs for more valuable data.

I wouldn't dismiss Jack so quickly, he definitely knows what he's talking about, and I tend to agree with him here. If you have a lot of money to throw onto this build then by all means go for it, but your main takeaway from this is going to be aesthetics, and there are several downsides. it's important you know this going into this project. Temperature effect from split loops is pretty minimal compared to a single loop with adequate rad space (~1c range) and there are other inefficiencies that come with split loops such as clutter, maintenance, rad space heat distribution, and cost

Even for fairly heavy overclocking, passive motherboard cooling is fine, watercooling the motherboard is pretty much for looks only. You can make a better argument watercooling ram, but unless you're pushing into extreme overclock territory, you really don't need it, it's just for show. If you are still deadset on watercooling these components, then by all means do so (many people do), but they certainly don't need their own loop, they won't even come remotely close to utilizing the full potential of a 240 rad. that rad space could be better spent sharing the heatload of the gpu or cpu. Managing all those pumps and res's is going to make small cases impossible to work with, and mid-size cases will at the very least be a pain, and a huge mess. Moving the components outside of the smaller cases to free up space makes the flexibility of smaller sized rads pretty much irrelevant, and you still have to manage power going to the external enclosure, along with 6 tubes going to and from your case. Not impossible to manage, but not exactly easy ether.

I would also recommend against trying to plan an entire loop for future builds. By the time that comes around, you will probably need to replace the tubing (assuming it's soft tubing since you're going for modular design) anyways, and tubing management is near impossible if you aren't custom fitting the lengths to your current project. Add that to the potential for having to buy different waterblocks for different cpu mounting layouts, GPU's, and the cost of quick disconnects (i have 2 on my build and they cost me nearly 50 dollars in total) the costs far outweigh the benefits.

I think you should really consider your priorities with this build and make sure this is the right plan for you. The cooling benefits of splitting this loop up are negligible, and the cost is significant in comparison to a single, or even dual loop. That money could be better used to upgrade components, or buy fancier watercooling hardware. If you still think that your plan is the course of action you want to take, then do it and post pictures for us to see!

QFT...

This is over complicating things A LOT. Jack is, as usual, spot on with a great follow up by Spork!

That is entirely too much radiator for a dual CPU system and dual GPUs, even overclocked. You have 10x120mm worth of radiator listed. The blocks can only put out so much heat in the first place.

EDIT: You are also missing a pump in your mobo/ram loop. So tack on another $80 for that...

This is just... this is, well, not my money. The waste here and the technical justification behind it just do not jive I am afraid. It will not accomplish what you think it will accomplish. Science says so, not an opinion.

 

[metloaf]

I ran a 2 loop system previously and I did it because I had left over parts and was able to get a few extra items from a friend for free. I had a separate loop for the CPU and my video card at the time. I will tell you from personal experience that it wasn't worth it, as far as performance and cooling gains. It wasn't noticeable at all. I only did it because I could, and it didn't cost me any extra, and the "it's cool factor." If I would have had to pay for it, then no way. It simply wasn't worth it for me. I ended up switching it to a single loop on a 360 rad without any issues and identical performance. Will it look cool, yup! Will it make any difference in performance, nope. But if you want to, then go for it!

 

[Eddie Current]

I wouldn't dismiss Jack so quickly..

I was in no way dismissive of Jack. A little tongue in cheek maybe but with respect.

Managing all those pumps and res's is going to make small cases impossible to work with, and mid-size cases will at the very least be a pain, and a huge mess.

Yes, exactly! That's why you move all the main components to a "Cooling Tower".

Moving the components outside of the smaller cases to free up space makes the flexibility of smaller sized rads pretty much irrelevant, and you still have to manage power going to the external enclosure, along with 6 tubes going to and from your case. Not impossible to manage, but not exactly easy ether.

No and no, it won't be irrelevant because the concept makes it so a person with a small case could WC his/her monster CPU and 1, 2, 3, 4 GPU's without replacing their current case. What I'm doing with three loops someone else might only want or need one - two loops, its still the same concept. It really not that complicated when you have been somewhat of a fabricator all you life and you have a garage full of tools. Seriously, the plan is to use some cooper pipe I have been storing in the attic for years.

I would also recommend against trying to plan an entire loop for future builds..

Dude, planning in advance is part of my real job and my life. That would be like stopping a moving train with a VW bug, just not going to happen.

I think you should really consider your priorities..

LOL, to late

EDIT: You are also missing a pump in your mobo/ram loop. So tack on another $80 for that..

Look again maestro, that's a pump/res combo. The one I use for my first loop remember

Will it look cool, yup!

You should have lead with this! But seriously, I have 99.99% of the parts its getting done and when I'm done and if I don't like it I'll make it one loop with two D5 pump in series or parallel.

 

[Conumdrum]

D5 in parallel. So if one pump fails the back flow kills the loop. No flow. Smoke.

 

[ProgramGuy]

He's clearly doing this for the bling factor and he has the parts... so go for it. This is an "extreme" build and for one I would like to see how it comes about. I would also like to know you thoughts during the build AND a few months after the build... ie, would you do it again.

 

[sporktar]

Dude, planning in advance is part of my real job and my life. That would be like stopping a moving train with a VW bug, just not going to happen.

I'm not saying you can't plan or keep in mind the future build, I'd just say by the time that comes around you'll likely need different blocks for new cards/cpu, and soft tubing needs to be replaced every so often (Personally i replace mine every time i build a new system, or every 2-3 years). With tubing replacements, the quick disconnects just end up restricting the loop and adding unnecessary cost.

You seem to have committed to this build, and since you've already gotten most of the parts, I'm assuming the cost wasn't much of an issue here. I'm excited to see how it pans out. This build is overkill on several different levels, meaning it would make for a great build log in the buildlog subforum. I love reading through these rad overkill builds, and i'm working on one myself. At the very least snap some pictures for us!

 

[ProgramGuy]

I love reading through these rad overkill builds, and i'm working on one myself.

I like reading these as well... Good luck Eddie... Keep us posted.

 

[Eddie Current]

Had a few set backs, had a mystery flood Thanksgiving weekend, work called and I had to go into the shop and work more days than I had planned. But on the bright side, my two Extech thermometers came in so I can do some more benches and have more accurate air temp +/- 1.8f. and I was able to pickup the material needed for the construction of the "Cooling Tower".

 

[Eddie Current]

UPDATE​

Had a little time to spend on my prototype "Water Cooling Tower / Test Bench" today so I figured I would post a few pictures.

 

[Silver_Pharaoh]

God I love that 1" dowel eh? Easy to work with.

Subb'd for more cooling goodness!

 

[sporktar]

I like the idea of not using an actual case in this build. A test bench style rig would better suit the watercooling loop for sure.

 

[Eddie Current]

I like the idea of not using an actual case in this build. A test bench style rig would better suit the watercooling loop for sure.

It should make for a very flexible platform. When I'm done, I will be able to cool at least three PC's ( or just one ) and one test bench.

God I love that 1" dowel eh? Easy to work with.

Subb'd for more cooling goodness!

Much quicker than using metal

 

[Eddie Current]

Perforated Shelf Tops​

 

[GTXJackBauer]

Very nice Eddie. Sub'd.

 

[GTXJackBauer]

I'm not too familiar with D5s but I believe the top portion of the pump is the intake so in your illustration, you're pumping water into the reservoir. If it was switched around it would be fine as I don't see any issue with the loop order as it won't matter since your reservoir is pumping fluid to your pump which is the most important thing and that it will reach an equilibrium, say after 30 mins of load.

 

[Janus67]

Eddie, I went ahead and merged the two threads together. I think they're similar enough they don't warrant separate threads to cause confusion over the same build discussion.