[SOLVED] Leak testing

[FSCGeek]

Hello! I have just made my first watercooled loop and now I am leak testing it for 24h. I don't know if I should keep the plug from the top of the reservoir off to let the air out or if I should plug it in.
Here is a picture of the loop:

what is image hosting

Should I let the plug off or not?
Thank you!

 

[rubix_1011]

Once you have purged the air from radiators, etc, you would want to cap it for the long haul, but I have left it open for a while to allow air to continue to bubble out. If you are still in the process of tipping and rotating the case, definitely make sure it is sealed.

If you are planning on doing a long leak test, I would seal it. This will allow the system to run in a state where the flow pressure is indicative of what you will encounter during daily use.

How are your GPUs setup, serial or parallel? I haven't seen that distro block before, but just wondering how it is feeding coolant to both cards.

 

[FSCGeek]

Once you have purged the air from radiators, etc, you would want to cap it for the long haul, but I have left it open for a while to allow air to continue to bubble out. If you are still in the process of tipping and rotating the case, definitely make sure it is sealed.

If you are planning on doing a long leak test, I would seal it. This will allow the system to run in a state where the flow pressure is indicative of what you will encounter during daily use.

How are your GPUs setup, serial or parallel? I haven't seen that distro block before, but just wondering how it is feeding coolant to both cards.

The GPUs are in serial, it was the only option that would allow me to use the NVLink bridge, as you have to change the terminal to Alphacool's SLI connector, otherwise the bridge won't fit.

Also I have tried to purge the loop by shaking the case but there still are a lot of bubbles in the reservoir. What other things I should try to purge it?

 

[rubix_1011]

The small reservoir bubbles will work their way out, don't worry much about that. Radiators like to trap air in the end tanks and you want to get that out, if you can. This should make its way to the reservoir to be displaced by water.

 

[Azzyasi]

How about a different method of testing:
Air pressure testing:
Hook a valve from a bicycle tube into a cap on the resvoir or other place where you can hook into (a T fitting or something else). Remove all water. Pump to 1bar or (2bar careful not to blow fitings apart an no bulging on tubing, so keep the presure low) and see if you still got that after 1h or so. (better if you have a pressure gauge hooked all the time to see if there is any significant drop in pressure

Also this air pressure test can be done with water in the loop as well but air pressure will ramp quickly being less volume of air, and has less of a spring effect so it might not push the water through a small leak and also a minor change in volume of water will not translate rapidly into loss of pressure of air above. When air escapes by it's ow, even a small leak is easily identified by the hiss and feel and sopy water brushed around the suspect fittings (will bubble if present) . That's why is recommended to have only air in the loop while doing this.

Also for such testing the pump should be off.

There is also vacuum testing but unless having rigid metal tubing, vacuum pump, vacuum gage.. it's harder to do.

 

[FSCGeek]

The small reservoir bubbles will work their way out, don't worry much about that. Radiators like to trap air in the end tanks and you want to get that out, if you can. This should make its way to the reservoir to be displaced by water.

Well, I guess that I'll leave it overnight in order to complete the leak test and then I'll drain the distilled water out and fill it with the coolant.

What worries me a bit is the fact that it's a pretty big bubble in the top corner of the cpu block and it doesn't seem to disappear.

 

[rubix_1011]

How about a different method of testing:
Air pressure testing:
Hook a valve from a bicycle tube into a cap on the resvoir or other place where you can hook into (a T fitting or something else). Remove all water. Pump to 1bar or (2bar careful not to blow fitings apart an no bulging on tubing, so keep the presure low) and see if you still got that after 1h or so. (better if you have a pressure gauge hooked all the time to see if there is any significant drop in pressure

Also this air pressure test can be done with water in the loop as well but air pressure will ramp quickly being less volume of air, and has less of a spring effect so it might not push the water through a small leak and also a minor change in volume of water will not translate rapidly into loss of pressure of air above. When air escapes by it's ow, even a small leak is easily identified by the hiss and feel and sopy water brushed around the suspect fittings (will bubble if present) . That's why is recommended to have only air in the loop while doing this.

Also for such testing the pump should be off.

There is also vacuum testing but unless having rigid metal tubing, vacuum pump, vacuum gage.. it's harder to do.

No, please don't suggest doing that. We aren't testing for a tire puncture.

The pressures introduced by the pump moving coolant in your loop are not subjected to the same kinds of pressures as you would be testing by those methods. In fact, there is a very good chance you would discharge tubing from a fitting and force a leak, even if the loop is empty - you could dislodge an o-ring or separate a fitting.

 

[rubix_1011]

Well, I guess that I'll leave it overnight in order to complete the leak test and then I'll drain the distilled water out and fill it with the coolant.

What worries me a bit is the fact that it's a pretty big bubble in the top corner of the cpu block and it doesn't seem to disappear.

Cap the reservoir and ensure all your ports are snug. Tip and tilt the case to get the bubble to work its way towards the CPU outlet port and see if it will dislodge that way. Follow the path of the flow and continue to move the case to get the air to the reservoir.

 

[FSCGeek]

Cap the reservoir and ensure all your ports are snug. Tip and tilt the case to get the bubble to work its way towards the CPU outlet port and see if it will dislodge that way. Follow the path of the flow and continue to move the case to get the air to the reservoir.

Tried that and it worked, the bubble has gotten significantly smaller. Thanks for your help. After I finish the leak test and replace the water with fluid I'll post a pic with the pc done.

 

[rubix_1011]

Good deal. I would expect that you might have similar issues in the GPU blocks, but maybe not due to how they're mounted horizontally. You should be able to tip and tilt the case quite a bit - just don't let the pump suck in any air from the reservoir while it is running...this just injects air back into the loop that you're trying to get out.

 

[Azzyasi]

No, please don't suggest doing that. We aren't testing for a tire puncture.

The pressures introduced by the pump moving coolant in your loop are not subjected to the same kinds of pressures as you would be testing by those methods. In fact, there is a very good chance you would discharge tubing from a fitting and force a leak, even if the loop is empty - you could dislodge an o-ring or separate a fitting.

Fair enough, it's a wild idea and not the normal way. Should not be experimented without knowledge and confidence as it can mess things up if done wrong.

Spoiler

Yet at 1bar i reckon is about the max pressure the loop will have full load by the pump.. so if O-rings got dislodged by 1bar of air, is bad because at some point it will be 1bar of water.
I pulled the value out of my hat, but really 1bar is nothing even for a garden hose with usual garden hose fittings.. they hold 10bars of water, and vynil tubing i'm sure is fine at 1bar, slight stiff at 3bar and burst at 5+bar.
The main concern is not to crack the resvoir.. but then again a 2l PET bottle holds 10bar before rupture.
This is exacly how brake lines are bled. Pump new brake fluid over the old one in the resvoir at 1-2bar enough to push the old one out, low enough not to crack the plastic brake fluid reservoir.

This was just an engineers perspective

L.E: look like i'm not the first to think about air testing: View: https://www.reddit.com/r/watercooling/comments/8ex78i/air_pressure_testing_almost_time_to_fill/

This individual used 0.5 bar instead of suggested 1. Might be a starting point for experimentation.

 

[rubix_1011]

You're correct, it isn't the first time it has been mentioned and since both air and water are technically fluids, it stands to reason you could do something like this. However, for the sake of ensuring that everyone keeps both eyes and all their fingers, messing about with pressurized systems can be an issue, especially if someone doesn't understand the dangers of high-pressure systems. If 1 is OK, certainly 50 is better!

My aim wasn't to completely shoot you down, it does have merit...it's just that for the normal builder it isn't necessary and leak testing can be done without the additional setup and danger to your components or people living in your residence.

 

[FSCGeek]

Okay guys, after a long wait and luckily no leak, here she is:

Thank you for your help!

 

[rubix_1011]

How are temps? Everything working well?

 

[FSCGeek]

How are temps? Everything working well?

So far, everything working well, other that the NVLINK bridge doesn't show up in RGB Fusion, but I'll check on that tomorrow.
Haven't got the chance to check gaming, but on idle both GPUs and the CPU were in the low 30s.

 

[rubix_1011]

Those are good temps, even if it is idle. You'll probably see more of a CPU spike under load, but GPUs probably will stay relatively cool, even while gaming. My RTX 2080 rarely sees temps higher than 45C and usually sits in low 40's C.