• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

I'm out of the loop for watercooling...

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.
Your filling issue may have had nothing to do with your power supply. Sometimes it takes a while to get the water flowing through the system, especially when the loop isn't closed off yet. It's still good to use a 2nd psu that isn't powering the PC for the initial filling though, as it keeps your system off, drastically reducing the risk if you happen to leak or spill water on your components. It also lets you cycle the pump more easily, which is helpful when you're trying to get things flowing. Regardless, I'm glad you were able to get everything up and running!

Deionized water is neat, but you really only need to be using distilled water with phn nuke added to prevent any bio growth or corrosion. Deionized is total overkill.
 
As said, the water is old and happened to be lying around so might as well use it. I'm using it as a flush not a final solution. I have 1L of premix on hand, as well as 100ml of concentrate if I need to make more, but would need to get more water for that.

Doesn't look like any leaks overnight, so now the temptation is to drain and refill for serious use. I know guides seem to pick 24h as the leak test time, but how long is long enough? By the time I've had my coffee it would have had around 12 hours test running. Is there likely to be a failure between 12-24h that wouldn't happen in 0-12h? My concern is this is an unrealistic load, since under real work there would be much more heat in the system.

My last water cooling system, which put me off custom loops, had a weird leak that only happens after running it for some days. I suspect it was temperature related and didn't happen when cold. The number of times I had to wash a 6800 Ultra under a tap wasn't fun... I never managed to seal the leak after multiple attempts and sold the kit on, entering my air cooling era.
 
The fun never stops in my house. I just had a flood, but nothing to do with the watercooling :) One of my other hobbies is keeping fish...With that out of the way I returned to the room where most of my computers are to find a faint "hot metal" smell. Everything is working but still not a good sign, unless I overcooked lunch more than I thought I did.

Anyway, still no leaks so I'm about to drain and refill with the proper liquid.
 
System refilled and running on the desktop again now. I forgot I had a spare 5.25 bay in the back I could fit the 3.5" HD in there with an adapter I need to order.

CPU ([email protected], not sure V) under prime95 small fft fluctuated around 52C. On the old AIO it would be high 50's, possibly touching 60's.
980Ti stock OC with Furmark defaults was getting to around 33C before hitting the power wall. The surprise here was running SteamVR test hit 35C. The latter I knew was easy to drive into thermal throttling, which was safely not the case here. I just did the TimeSpy stress test which went smoothly.

Next, I do have one of the Unigene things installed so will do the temperature testing as described earlier in the thread, again with prime95 small fft in background to load CPU too.
 
The key here are temps when the loop is saturated. You need to test for at least 30 mins to an hour to see max temps in a custom loop. :)

So far so good!
 
I've got the first half of the data now, and just flipped the fans over for 2nd half. I had aida64 to display a time history (not running stress) and when the CPU looked flat for a while, I assume it's reached equilibrium. GPU is loaded by Unigine Heaven extreme defaults. CPU is loaded by 6 threads of 8-64k in place FFT. I chose 6 threads since 4 is enough to max out CPU power, if the Windows scheduler was sensible. It isn't, so I'm kinda over-provisioning with 6 threads. The remaining two threads are free for Heaven and other stuff.
 
Results are in:
In the following, I'll give the 1 in 5 out temp first, then the 3 in 3 out.
GPU: 37, 37
CPU: 55, 56
Case air: 23.8, 28.7
External air: 21.8, 22.2

Air temps measured by a dual probe. Case probe suspended above the CPU, not touching it. External air is in the air to the side of the case not in the flow of any of the case fans.
CPU and GPU temps are the averages reported by hwinfo64. I reset the counter when it has reached equilibrium and left it for many minutes before looking at the average. CPU is hottest single core, not average of cores.

In performance, it looks like no significant difference in CPU/GPU temps. Air temps, as I expected, are significantly elevated with the balanced scheme. I kinda wished I noted the fans speeds too now, and see if there is any difference there. Actually, since my original arrangement seems better I'll revert to that and take more numbers then.
 
I just also tried case panel off, with my 5 exhaust fan arrangement. So as some predicted, that gave lower temps than with panel on but we're talking about up to 2C.

Fan speed wise, it is more complicated. All fans are connected to the mobo, and I left it on defaults which varies speeds with temperature. So while there is some variation, we're looking at about 1% min to max so I don't think that is significant.

Right, what's the GPU OC software that's most popular right now?
 
MSI Afterburner. 4.3.0 beta 4 is the latest. I typically get it from guru3d.

Something to also consider about your temp readings is that with the rear fan on intake, it's blowing ambient air right across that sensor. Makes sense it's lower considering the other way it has to go through a radiator and other items before it hits the sensor. If you can read motherboard/vrm and gpu vrm temps, you will have the data you are looking for. With just one intake, the system as a whole isn't really getting much fresh air, particularly to anything just below socket level on down...whereas with the air coming in the front, even though it's a few c warmer after the loop normalizes, it's still better airFLOW through the case.

Are both 980ti's in the loop? You are running more wattage than I was too.. which would make a difference I would imagine. :)
 
Last edited:
I only have a single 980Ti. Was looking at a 2nd if I saw one at the right price, but that hasn't happened yet.

Good point on the air. I didn't note the numbers, but on aida64 I also had a "mobo" and SSD temp showing on the chart too. Both increased on the balanced intake/exhaust arrangement, and lowered again on the exhaust heavy configuration. The SSD is mounted lower down on the mobo and not in the air path of the rear fan. No idea where the mobo sensor is.

This is turning into a tradeoff between hotter but higher airflow, or lower but cooler airflow. I'm sticking to the latter, as I've seen no evidence to the otherwise. I'd say again, the case has a LOT of vents, so I don't think a balanced fan config matters. The single intake fan is not the only air path to outside.
 
Running out of time once again for another night. I just messed around with unplugging the fans, and inserting the noctua resistor cable. I need to do either that, or actually both setting sensible fan speeds in bios for normal running, and see how that affects actual temperatures vs. noise.

I did find it interesting though... if I unplug the front fans totally, the SSD temperature went up, but the mobo temp went down. Plug fans back in, SSD temps drop, mobo increases. Where is that sensor??? It's not a big effect at up to 2C, but still a curiosity.
 
So I used a work toy on the setup... CPU was loaded with 4 threads of World Community Grid work, I think OpenZika at the time. GPU was loaded with Heaven. GPU temp did hit 41C today, but it is a lot warmer today than it was yesterday. Note the colours are adaptive between each image, so apart from brighter being hotter, they're not necessarily the same.

cpu.jpg
The CPU block is cool compared to the mobo around it.

gpuhot.jpg gpuidle.jpg
This is the back of the GPU, left under load, right idle. Note the hot region which is the back of the voltage regulators.

pump.jpg
Pump also generates a bit of heat

rad.jpg
Top rad seen from the inside

front.jpg
Front fans are extracting heat.

top.jpg
Top fans also extracting heat, through the case top vent.
 
Hard drive caddy arrive so at some point I need to put that back in.

How much level drop is expected? The level has certainly gone down, and I attributed it to perhaps the odd air bubble from somewhere finding its way out eventually. The loop is closed so there shouldn't be evaporation, and there's no evidence of leakage that I can see.

I'm also wondering what to do about the fans. I never changed the settings, so the noise is still a bit higher than I'd like, and two of the fans seem to be going into resonance or something and making a bit of a horrible whine at certain speeds. I think ideally I just want a fixed speed for normal running. It would give "enough" cooling under full 24/7 stress loads. I could separately turn it up for benching when I have time to revisit that. I did play with the mobo bios a bit, but that was horrible and useless. I think I need a separate controller, one that is NOT software controlled.

Also there is a faint high pitch noise. I've not localised it to anything yet but that needs fixing also. I've been debating moving the PC from the desk next to me, to the floor. Increased dust potential, but adds distance for the noise.
 
wetpinky.jpg

Did some tinkering with the system today and re-installed the LED strips I put in ages ago. I should add, this is not the colour I'm leaving it! Just happened to boot in that state. Dust on the window has never looked clearer.

While I was fitting that I had a look at the watercooling. The reservoir was never perfectly vertical and I wanted to straighten it. That's when I saw some white-ish stuff at the back. Could it be dust on the outside? Nope, it was inside. It was stuck to the inner surface. Looking at the rest of the loop, I saw similar residue on the clear CPU block, partially obstructing some of the fins. I can't see inside the GPU block as I didn't go for a clear one. Temperatures are still fine, so I don't need to fix it right now, but that's a job to plan for the future. If I had to guess, I don't think it is biological, but more likely chemical. Since I didn't have significant light in the case, I never noticed it before.

To recap, I did give it a quick flush when installing with deionised water. After that the system was filled with EK's premix. So... did I not flush it enough?
 
When setting up it was filled, circulated, drained of de-ionised water. After that, I filled it with premix which remains in there now.

When the time comes, I still have some of that premix left over, as well as concentrate to make more of the same.
 
Well, if its buildup, you'll want to get to the bottom of it, otherwise it will just move it around to other clean surfaces. If its just residue from the premix, I think you should be fine in hopes it just dissolves.
 
You must log in or register to reply here.

Similar threads

U
360mm good quality AIO vs custom loop for ~250-300W CPU
2
Replies
38
Views
2K
User5566
U
Mario1
  • Question
Best s775 CPU to overclock?
Replies
9
Views
295
freeagent
freeagent
T
Air Cooling for ROG Strix Z590e
Replies
5
Views
215
tomdean
T
M
pc building noob
Replies
1
Views
565
Zerileous
Zerileous
E
Upgrade options and AIO solutions
Replies
1
Views
162
EarthDog
EarthDog
Back