Is this my Fault?

After putting in all the work to link all my family's gaming systems into one mega loop that is cooled by my pool, suddenly there was a leak. Water spilled all over my Ryzen Gaming systems, my NAS servers and my UPSs. I reached out to EK Waterblock for help, but the source of the issue is still a mystery. Am I screwed?

Purchases made through some store links may provide some compensation to Linus Media Group.

FOLLOW US
---------------------------------------------------

MUSIC CREDIT
---------------------------------------------------
Intro: Laszlo - Supernova

Outro: Approaching Nirvana - Sugar High

CHAPTERS
---------------------------------------------------
0:00 Intro
2:05 The Shaggy Defense
2:49 Inspecting the water blocks
5:22 Fittings + Mounting bracket?
6:30 Corrosion in the GPU block?
7:08 Inspecting the rest of the systems
9:56 How to drain your loop
10:55 Disassembly continues
12:26 Seems OK, but why?
13:10 Plz Help
13:33 Any survivors?

[RobertPanechetta]

Mechanic here! I heard you say FUEL lines. Some fuel lines use STEEL bands to reinforce them. My theory is some of the water was in contact to the fibers in the fuel line UNDER the worm clamp since it prob stretched and exposed the steel in the hose

Edit: Alot of people keep saying fuel lines with steel reinforced fibers arent a thing. They are very much a thing but not common on non german vehicles. It is also pretty uncommon to use now a days BUT last time i checked they are half the price as fiber reinforced in autoparts stores. The lines used for the 1 corroded rig look different from the other 4 units.

[MrJimodoom]

Linus son is ridiculously chilled out. Loved the interaction "how'd you get to be so sarcastic"... "bad parenting I guess"... killed me

[darschpugs]

Sorry for your lose man, my dad just passed away yesterday. we never know when we will lose someone, anyone that reads this, call your family members and talk to them, never let yourself fall out of touch with them.

[TrashPanda-zx]

Linus, I am so sorry for your family’s loss.

I have my Master’s in Chemical Engineering. You reference galvanic corrosion, but there are different “types” of corrosion and I think there could be multiple possible causes.

1. “Pitting Corrosion”- Minute imperfections in the zinc coating could have caused localized potential differences which over time became significantly worse. The zinc coating should passivate the brass, but after enough small-scale pitting corrosion, you might see the brass of the fitting involved in the corrosion. The rate and degree of corrosion strongly suggest this type.

2. “Crevice Corrosion”- This could be the corrosion taking place in the GPU block where the corrosion appears to be happening with the non-metal parts of the GPU block. If even a small amount of fluid became trapped (like under a gasket or in very sharp corners), it could induce this type of corrosion. This type of corrosion is slower than pitting, so I do not think this is the root cause, but should still be mentioned.

3. “Selective Corrosion/Leaking”- Potential differences, the driving force behind corrosion, can develop on the same piece of metal/object; you do not need to have two different metals. Essentially, the zinc-coated brass fitting(s) corroded itself. If possible, send the corroded fittings to EK or someone else to have the metal content analyzed.

If you’d like to talk further about these, let me know. I have contacts in the material science field who might be able to assist.

[washellwash]

Benefit of being Linus' son: cool stuff in the house
Drawback of being Linus' son: if the cool stuff breaks, it takes forever to fix

[AshleyRosemarry]

Another prime example that you NEVER know what is happing to the people you see online, or on the street. They might be showing a smile, but they might be literally dying inside, emotionally and/or physically speaking. Always be kind to others

And so sorry for you loss Linus. Honestly, you should consider taking a break if needed. Your mental is worth more than the company.

[brentdavidson]

The rusty screws in the GPU block may be a red herring, but that's where I would start looking. My reasoning is this: I once bought an inexpensive oxy-acetylene brazing torch that was mysteriously equalizing pressure between the oxygen and acetylene lines with all handle valves off. Turns out there was a microscopic crack in the brass handle casting that was allowing the gasses to mix before the metering valves. If there was a microscopic crack in the delrin it could have been allowing water to contact the steel screws while the threads were tight enough in the holes to create seal that didn't leak.

A second possibility is the fittings themselves. Those fittings looked like they were possibly copper or brass that had been plated with another metal to make them shiny. That was likely electroplating. If the fitting was dropped and there was a flaw in that plating, no matter how tiny, that allowed the water to contact the base metal, then that could have provided a source for the corrosion.

[davidknightx]

"I'm the kind of person who honors his promises, so I'm not promising anything." Yeah, I'm stealing that line.

[INDIANBEAT]

So sorry for your loss Linus. As someone blessed with a younger sister, I can't even imagine how hard it must be losing someone you grow up with, fight with, play with, but also love them the most.

[erdem--]

Electrical engineer here,
There must be a ground loop from the pump and server. Even 5-10 volt difference can make it. In the case of pulling thousands of watts from servers, I guarrentee you there was a 10-20 volt difference between neutral wires of the pump and server rack.

You need to isolate the pump with 1:1 isolation transformer. And add a sacrificial anode to be safe

[SplitRensonator]

Chemistry PHD here that has designed computer liquid cooling systems for about 30 years. Linus, I need to correct several things you said in this video. First of all, stainless steel is compatible with copper. If you want a good reference for materials in liquid cooling, check out the OpenCompute guidelines. You can google “Guideline for using water based transfer fluid in single phase cold plate based liquid cooled racks” . As you can see on page 10, stainless and copper are just fine. Next, you said at the beginning you ONLY use copper and brass, but nearly every wetted material you showed us was silver colored. Copper is red and brass is yellow. That means you have metals in your system that you didn't anticipate. What it looked like to me is that your copper (or brass) fittings are coated with another metal. It's either chromium or nickel most likely, both of which are compatible with liquid cooling systems (see above Opencompute spec). What likely happened is that the coatings are bad or were damaged, exposing the copper and/or brass to the coolant. But you might think...copper and brass should be fine right?? It depends. Does your coolant have a copper corrosion inhibitor in it like tolyltriazole? Is the brass >15% zinc? The brass question is really important, since high zinc brass will undergo galvanic corrosion with copper in the system (zinc-copper couple) that results in dezincification of the brass and failure of the material over time. That's maybe what happened here, i.e. bad nickel plating on your high zinc brass caused dezincification and failure of the pipe as the zinc was removed, leaving a copper sponge with no real mechanical integrity. This is a lesson for everyone out there...coated metals don't necessarily work, it depends on the quality of the chromium or nickel coat. There are other reasons this could have happened, but this seems like the most likely scenario.

[DDryTaste]

Firstly, my sincere condolences.

In regards to your system; install a level switch in your reservoir that initiates emergency shutdown and power disconnect when the level goes too low. You can get switches with a little bit of headroom for evaporation but you should top that off anyway. You can also put flex PCB leak detectors sandwiched between the systems.

[MrBeetsGaming]

Sorry to hear about your sister, I lost one to a drunk driver a little over a year ago. Almost lost my dad in January who is my only living parent and now I have to take care of him 24/7, which I don't mind but it keeps me from working.

[florihae]

Mechanical Engineer here. We design evaporation plants, so you guessed it, a lot of corrosion. I suspect you had an electric potential applied from external instead of chemically. A few Volts is enough to create massive corrosion. Check all the grounding, add more wherever you can. Including water lines.

[streetjustice]

never seen a dad and his kid be the same age before

all jokes aside, sorry for your loss brother

[HoneyTwee]

Sorry to hear about your sister Linus. ❤

[yufeipan]

hey linus I work at ASML San Jose and last year one water cooled door on one of our water cooled compute cabinet broke and flooded our data center. We have all coppor loops. We did so much research into it and ultimately confirmed it is some bacterial colonies sitting in some low flow area region of the door consumed the copper and corroded it through. Apparently it is a very rare kind of bacteria that can live near the lead free solder joints of the copper pipes in the door. They eat, dug, live and thrive in cavities in coppor. Not sure how they got in there but the infection was very localized and we were able to isolate the bacteria from the regoin. luckily after flushing our loops with biocide it did not happen again. My personal theory of what they eat are the organic solder paste residues from manufacturing. These aux water cooled cabinet doors were always supplied with chilled water and maybe part of it got to a comfortable temperatures for the bacteria after been heated by warm exhausts from equipments and they built a colony there.

[Orcsmustcry]

Dude, I think I speak on everyones behalf when I say, we are so sorry for your loss. You did not have to share that and thank you for allowing us in.

[MrBlakester]

When you brought your sister on to build a pc together she seemed like the sweetest person ever. I cried hearing about your loss brother. Been watching you for a decade maybe even more and seeing you distraught broke me a little. RIP

[Connor_Conntroller]

13:42 Hey Linus, as you asked, inorganic chemistry major here. Even though brass and copper are generally very compatible, it could be that due to minor imperfections and similar debris/metal filings to the control loop a small electrochemical potential formed. Things like slow changes in the pH of the water, biofilm forming and temperature diffrences could be enough to accelerate this process immensely. I would check the metals under a microscope to look for impurities or attacked areas, using a multimeter to check all the isolators, check the pH, add more iodine or other desinfectant agents and lastly check the heat load of the diffrent systems. If the one has been used quite heavingly this could add to the other accelerations.
Hope this could help, and i wish the issue gets resolved quickly. And my condolences for your loss of both family and tech.
Edit: changed anorganic to inorganic. Even though they can be used synonymous inorganic seems to be used more often