My first water cooling loop - No reservior

[pj-schmidt]

I just completed my first custom water cooled loop. My case has no side windows, so it only has to be functional, not pretty. Having said that, I did try to keep it reaonably clean.

Flow order: Pump ==> GPU ==> CPU ==> Radiator ==> {pump}...

As noted in the title, I elected to not use a reservior. I decided early on that I wanted to try to mount the pump inside of the 5.25 drive cage. Due to my optical drive, this was a tight fit.

With this goal in mind I elected to use a tall (60mm) radiator in X-flow configuration (Found an open box unit in perfect condition for a really good price). Compared to a "normal" U-flow radiator with both inlet and outlet on the same end, X-flow configuration reduces the distance the coolant travels by half, but it also flows at half of the speed. This half distance / half flow rate results in the coolant spending the same amount of time in a X-flow radiator as a conventional radiator giving it equivalent cooling ability. The advantage for me with the slower flowrate is that it makes it easier to for the air bubbles to escape the flow and not continue circulating in the system. I also selected a radiator with 3x ports on each end to allow me to vent air without needing to disturb the functioning connections.

Above you can see the fill port in the top of the radiator that I can use to vent air and add coolant while the system is running. The X-flow configuration also has the added benefit of providing an inlet and out closer to the expected points of connection (CPU and pump) eliminating a long horizontal tube run. The reduced flow restiction of the X-flow radiator was an additional consideration, but I doubt it has a significant effect on my system.

With the thicker radiator, I elected to go with a push-pull fan configuration. I like the idea of running more fans at lower rpm, so I'd probably have done this even with a thinner radiator regardless of cooling performance.

Additional pictures:

Thermistor plug in the outlet of the GPU for loop temperature monitoring using the motherboards T_probe header. The additonal plug on this manifold also gives me a convenient way to drain the loop near the lowest point without disonecting any hoses.

Filling and removing air from the system was silly easy. First I filled it until the radiator appeared to be about 3/4 full. Then I cycled the pump for 10 seconds before noting a bubble at the inlet of the pump. I then stopped the pump and let it sit for 30 seconds with the hope that the bubble would move up to the radiator. This worked as when I restarted the pump the bubble was gone and the pump was nearly dead silent (scary quiet). I then let it run for 10minutes while occationally tipping the unit to encouage any other bubbles to move. The radiator was then filled to max. That was nearly a week ago. Since then I have done multiple stress tests (Prime95+Unigine Heaven) which showed very good cooling ability. I have not seen any change in coolant level in this time and there are no leaks. It is certainly not a show piece, but I'm pretty happy with it.

Questions, thoughts, suggestions or critiques?

Phillip

Main Cooling Components:
Alphacool NexXxos UT60 240mm X-flow radiator
P.LOTOR Low Noise Cooling Pump
EK Supremacy MX AMD CPU block
XSPC Blade - GTX 1080 / 1080 Ti GPU block
EK CryoFuel Clear

 

[Zarathustra[H]]

Looks like it works.

Personally I prefer having a reservoir, or at least a T-line to keep some buffer so the fluid level doesn't run low. Coolant will evaporate out of the system through the tubing over time, and when you have a reservoir, the risk of running too low is much smaller, so you may have to check and fill levels more often, but that sounds like something you are aware of and willing to live with.

The single 240mm radiator should be sufficient for basic cooling, but not much more. The rule of thumb is to never go below one 120mm fan slot of radiator capacity per component, so you have observed that which is good. People don't usually build custom loops because they want basic cooling though, so I am not used to seeing that.

What do your temps look like at load?

 

[pj-schmidt]

Looks like it works.

Personally I prefer having a reservoir, or at least a T-line to keep some buffer so the fluid level doesn't run low. Coolant will evaporate out of the system through the tubing over time, and when you have a reservoir, the risk of running too low is much smaller, so you may have to check and fill levels more often, but that sounds like something you are aware of and willing to live with.

The single 240mm radiator should be sufficient for basic cooling, but not much more. The rule of thumb is to never go below one 120mm fan slot of radiator capacity per component, so you have observed that which is good. People don't usually build custom loops because they want basic cooling though, so I am not used to seeing that.

What do your temps look like at load?

After 4hr of Prime95 + Unigine Heaven:

CPU with a 4.2 OC = 68c in AMD Master
GPU at stock clock = 43c (haven't tried oveclocking yet)

I may add a sight glass for checking coolant level, or change my fill tube to a clear ridigid tube a couple of inches long to provide a mini resivior. I'm really not too owrried about it given how easy it is to check right now.

 

[Zarathustra[H]]

After 4hr of Prime95 + Unigine Heaven:

CPU with a 4.2 OC = 68c in AMD Master
GPU at stock clock = 43c (haven't tried oveclocking yet)

I may add a sight glass for checking coolant level, or change my fill tube to a clear ridigid tube a couple of inches long to provide a mini resivior. I'm really not too owrried about it given how easy it is to check right now.

That's actually better than I was expecting.

I keep forgetting that unlike my old 32nm 3930k overvolted to 1.445v, these newer chips run a lot cooler. (Well, unless you get a Theeadripper)

Maybe it's time to upgrade after all. I haven't needed to from a performance perspective, but maybe the lower temps wouldd be worth it.

 

[FlawleZ]

That's actually better than I was expecting.

I keep forgetting that unlike my old 32nm 3930k overvolted to 1.445v, these newer chips run a lot cooler. (Well, unless you get a Theeadripper)

Maybe it's time to upgrade after all. I haven't needed to from a performance perspective, but maybe the lower temps wouldd be worth it.

Tell me about it. My 3970X makes my H110i work hard to keep temps reasonable.

OP, congrats on your first WC loop. Did you originally run this on air to compare your results?

 

[pj-schmidt]

Tell me about it. My 3970X makes my H110i work hard to keep temps reasonable.

OP, congrats on your first WC loop. Did you originally run this on air to compare your results?

Thanks. I've been building my own computers for forever (think AMD Socket A era). I've had doing a custom water cooling loop on my buck list for almost as long. After my latest upgrades to a 2700X and a 1070Ti, now seemed to be the right time.

I did cool it on air for a while, but I don't have a good comparison. I didn't OC the cpu on air as it seemed pointless. I haven't OC'd my gfx as I'm already running 200+ fps in the only game I currently play (R6:Seige).

 

[pj-schmidt]

Temperature correction:

My previous CPU temp was from my memory and I got it wrong. The actual CPU temp with a 4.2OC @ 1.425v is 76c at full load after 4hr.

 

[Zarathustra[H]]

lol wut, that's called a leak. not evaporating.

Nope. No material on Earth is perfectly air or water tight, especially not plastic tubing. Evaporation occurrs very slowly straight through the solid material of the tubing, even when there are no leaks. This is pretty much standard knowledge in water cooling circles.

The dryer the air (and presumably, the warmer the coolant in the loop) the more quicky it happens. If you live in a climate where it is humid year round you might notice less of it. In Boston I notice none at all during the summer, but lose maybe a quarter of an inch at the top of my reservoir every winter.

It takes several months to a year to be noticible (and some would argue you should change your coolant more often than that anyway) but it is very real.

 

[Zarathustra[H]]

OK I'll have to take your work you that.

You could also google it. I'm far from the only person in watercooling history to mention the phenomenon.

 

[Zarathustra[H]]

Been water cooling since we used NPT fittings and soldered blocks, this is the first I ever herd of it.

Water passing through a solid object sounds like a port from a si-fi move.

All materials have non-zero values of permeability. (Article on permeability of polymers here)

In most usage scenarios the permeability is low enough that it is completely irrelevant, but in a closed water loop where you go months or years with a fixed amount of coolant, while the permeability is very slow (well below the rate of evaporation, so you'll never see water drops) it does occur, and over time it ads up to noticeable level drops.

We used to fight this problem when I worked on underwater sonar arrays. We would completely pot over our transducers with a few inches thick proprietary urethane blend, and then fully enclose the urethane slug with an adhesive rubber boot, and even then we'd eventually get water ingress permeating through the material to the sensor inside.

Permeability is a bitch.

 

[noxqzs]

I also run mostly reservoir-less systems and it usually takes about half a year before fluid level is low enough that you hear bubbles pumped through the system.

 

[Peachdown]

I just completed my first custom water cooled loop. My case has no side windows, so it only has to be functional, not pretty. Having said that, I did try to keep it reaonably clean.

Flow order: Pump ==> GPU ==> CPU ==> Radiator ==> {pump}...

View attachment 80429

As noted in the title, I elected to not use a reservior. I decided early on that I wanted to try to mount the pump inside of the 5.25 drive cage. Due to my optical drive, this was a tight fit.

View attachment 80430

If you don't mind me asking, how did you mount the P.Lotor pump? I am using the same one, but it is just sitting at the bottom of my case.

 

[Nobu]

If you don't mind me asking, how did you mount the P.Lotor pump? I am using the same one, but it is just sitting at the bottom of my case.

Not sure what he has screwed to the bottom of it, but below that mount it looks like he used double-sided tape to attach it to the drive cage.

 

[cyberguyz]

While that setup looks cheap enough I went for a bayres wit a built in D5 pump top. It takes up about the same space and does have space for a little more coolant. It is 2x 5-1/4 bays high, but only half depth. I would have liked to see at least a 3/4 depth reservoir nut it seems they are none with a D5 pump head built in

.

 

[Zarathustra[H]]

I went with a HUGE res in my build and I am very happy with it.

The larger volume of water means that temps rise more slowly. It takes much longer to reach the steady state. This means it smooths out the peakea and valleys, and allows me more flexibility and time to adjust my fans manually.

 

[cyberguyz]

I went with a HUGE res in my build and I am very happy with it.

The larger volume of water means that temps rise more slowly. It takes much longer to reach the steady state. This means it smooths out the peakea and valleys, and allows me more flexibility and time to adjust my fans manually.

This is true. The more liquid you have in your system the more stable your temperatures become die to the slower rise and fall of the liquid temps.

 

[lightsout]

Is that a push/pull setup with the push fans inside the case? Or just two fans in pull?