Custom Water Cooled PC breaks down

Associate
Joined
18 Sep 2020
Posts
17
This is our first water-cooled PC and it seems like we did something wrong. I posted this here and was referenced to this forum.

CPU: AMD Ryzen 9 3950X (16x 3,5 GHz)
Board: Gigabyte Aorus Master X570
GPU: Gigabyte Aorus GeForce RTX 2080 Ti Xtreme Waterforce WB 11G
RAM: 32GB G.Skill Trident Z RGB DDR4-3600 DIMM CL17
SSD: 2 x 1TB Crucial P1 NVMe SSD
Power: 1000 Watt Corsair HXi Series HX1000i
Case: Lian Li PC-O11DW Dynamic Midi-Tower

Fans: 5 x 120mm
Radiators: EK Water Blocks EK-CoolStream PE 360 + EK Water Blocks EK-CoolStream PE 240
Distro (Pump + Reservoir): Lian Li O11D Distro-Plate G1

(Full specs incl. water-cooling here)

After a while the first pump of the Distro Plate died, then after an upgrade to a pump twice as strong the second one. Then the mainboard died (because of a bad power supply which got replaced) and got replaced and now the third pump which was connected to an external power brick, died as well. We think we seriously did something wrong, but we can't seem to find the reason.

Our first loop was from Distro Plate split to GPU and CPU where each went into a separate Radiator and were merged into the Distro plate after this, but as we thought it might be the reason of the pump not working anymore, we restructured into two semi-separate loops one going directly into cpu and rad 1 and after that going into gpu and rad 2.

Thanks a lot for your help in advance!
 
Couple of things
You shouldn't run the pump outside the pc like that
Even for testing running a pump dry is a bad idea
The liquid is what lubricates it
Thanks for the advice!

Some of your tubes bends are kinked
Ie flattened out thats probably causing flow restrictions
Its not easy to see on my phone which are the in
Yeah, one or two tubes aren't bend perfectly, but could this really be the issue?


Are you bleeding the air out of the loop?
What do you mean by that? In the beginning, we fill the loop until all air bubbles disappear and leave it running for a couple of hours. I don't think there are air pockets in our loop, at least I couldn't see one.
 
As you put liquid in the air is displaced by it
That air has to come out
Its not always visible as the air bubble could be in the pump or cpu block etc
The distro plate should have some sort of bleed screw
To let air out
Yes it does and we had this open during the filling process.

1 its the flattened bends severely restricting flow
Is highly possible.
2 the system is air locked
We did let the air out during filling and the first hour or so, so this shouldn't be it.
3 you put a tube on an in port that should go on an out port
Or vice versa
We redid the loop and the layout of it 2-3 times now, this shouldn't be it. Our first combination was very suboptimal as two streams did come together exactly where the pump was, but this is fixed since Mk2 and there were a lot of issues after that.


Why was the third pump ran off an external power brick?
Because until Mk3 we apparently had a faulty power supply (found out after measuring) and this probably destroyed the mainboard. The second mainboard didn't always tell the pump to kick in (unknown reason) and so we tried to solve this problem by not using the mainboard to power or control the pump but to always run it via this power brick.

A distro block should allow for clean, straight lines, I'm not sure what you were thinking when working out the runs, the kinks will not be help, some look extremely restrictive, esp gpu to rad :o
Like I mentioned this is our first hard tube water-cooled build and we didn't know that the flow is impacted by such bends so much. Yes this is a scuffed line (gpu-> rad)!

Thanks for all your help so far!

@Defy Belief Our loop is scuffed like this, as it was partially rebuild 2 times now. But the PC is so unstable, that we think about just rebuilding the whole loop. Clean lines weren't our priority, because we thought it would be beneficial if we always went heat > cooler > heat > cooler (as it is now). Any tips are welcome!
 
Possibly you should have used a 90 degree fitting on the
Lower gpu port and on the bottom radiator
To avoid some of those flattened bends
Yeah we probably do that.

Thanks for helping us out! I think we will come back with plans on how the loop will look like.
 
Personally, and just a suggestion. Given the amount spent on this loop, and assuming you have excess tubing about, why not connect a simple loop outside the case using the distro plate purely for the pump (so 1 inlet and 1 outlet), and see if all is working.
The problem with that is that the PC works one week or two and then breaks down - so testing would need to sustain for something like 4 weeks to really conclude that it's the tubing. We will redo the tubing anyways now, but thanks for the advice!
 
How high were the cpu and gpu temps
When it worked?
In idle GPU 40°C, CPU around 40° too with maxing around 65°C GPU and CPU about 70°C if I recall correctly. So nothing concerning - at least what I thought.

@GilesGuthrie Thanks a lot for your post, this and the other posts will help us improve the layout of our loop for this.

What do you mean? Are temps absolutely crazy? Is the pump shutting off? Have you got your BIOS setup to shut the system down in the event of pump failure?
No, yes, no. With unstable I meant, that it worked for a couple of weeks and another part broke down every time the last 3-4 months this PC exists.

I’d say it’s likely down to the piping looking at your photos. No offence but that is some really janky pipework there almost like you’ve tried to bend the whole pipe. Only really needs 90/45’s or even a given angle and the rest straight. Can easily be bent with a bending kit.
None taken, like I said, it is the result of multiple change of plans after the fact and no experience in hard tubing at all.

For most of the bends we used an insert, just for the first two (one is the kinked down at the gpu) were done only by hand. We always wanted to redo them, but didn't find the time yet.
 
Back
Top Bottom