Seasoned marine electrical engineer here. We specialize in galvanic and electrolytic corrosion prevention. Galvanic corrosion is caused by DC, not AC. Isolation transformers are used to derive a separate ground to prevent DC eddy currents of 2V or under from destroying metals. You CANNOT connect the ground on either side of the isolation transformer, or it does absolutely nothing. I'm not sure what you bought, but most of those benchtop isolation transformers do not derive a separate ground. It seems counterintuitive, but the DC current will just travel though the ground and the isolation transformer would be completely useless. The excessive rack grounding wasn't really necessary either as normally these racks are already grounded through the chassis of the equipment in the rack (if in doubt, use a multimeter and ohm it out). If this fails, look up a marine galvanic isolator. This little $ 50 device would have completely taken care of the problem, but i supposed it would have made for a less interesting video lol. They work well, and these are installed on the shore power inlet on boats. The AC ground on a boat is also bonded to the DC- battery ground and the sacrificial anode (along with other bits of metal in the water). Add in some saltiness to the water with a few other boats in a marina with some faulty wiring, and you have a recipe for many thousands of dollars of damage to engines and anything else touching water. These galvanic idolators prevent that and are nothing but a couple of diodes in parallel (and opposite polarity) which will pass an AC fault current easily tripping a GFCI, ELCI, or main breaker while blocking the harmful DC eddy currents that are destructive. You can even make one yourself, but you will need diodes rated to the circuit ampacity which is dictated by the breaker size. Love your videos!
It's been a long time since Chemistry and Galvanics for me, but wouldn't the water need to be in contact with electric? Like two metals of different types might make a battery, yes, but only if their potential / the difference of them is high enough; so checking the compatibility should have prevented that. The other way would require electrodes, i.e. metals connected to voltage touching the water, no? Wouldn't it then be a way bigger problem that you have water, carrying hundreds of volts and amperes, ready to short/fry anything if it ever leaked? Also obviously this only works if the water contains ions, but i guess at enough of a difference in potential the main point is that any metal would shed ions?
Industrial controls engineer here…. You should add some sort of electrically operated valves in the lines before the pumps. If you shutoff the pumps due to a leak, gravity may still allow flow through the pump
This type of corrosion is why Grounding Systems for Homes are no longer allowed through the copper piping (at least in Germany). Back in the days they would just use Water or gas pipes to be used as the ground. These days old homes get a ring ground and new homes get a foundation ground. To prevent corrosion on the piping system and to prevent voltage differences we have to run a 6mm² wire to all the pipes in one big loop in the basement and connect them to the ground System. This way the current will run through the wire and bring it all to one "voltage".
Galvanic corrosion is why none of your metal pipes are made of anything but copper or metals of similar nobility. Any more/less noble metal in contact with the electrolyte (water) and electrically connected to the copper would cause corrosion. Grounding has nothing to do with it, and in fact, your wire is still an electrical bond that would _introduce_ corrosion if there _was_ a dissimilar metal. The reason you're not allowing it as a ground is almost certainly the same reason the US requires that there be an alternate grounding system -- it's far, far too easy for it to wind up insulated.
the reason the leak sensor isnt reading exactly 0V while its dry is to be able to tell the difference between a zero singal and a broken sensor. Not as important in a home setup but in industry applications where alot of sensors are at work the risk of one breaking is so high that having any signal as 0V or 0mA isnt something that is done.
Nice! Recommendations: -set up a testing schedule for yourself. Protection systems that aren't tested won't work when the demand occurs. Failures are like a watched pot--won't boil if you're paying attention! -your shutdown of the pumps should be "latched". It really really shouldn't turn them back on when the leak sensor clears. You want someone to go and physically check on it and reset. This is fun. Can't wait 'til y'all just buy Jake a PLC and go full industrial.
110% the pump should require a reset before restarting. If the leak just dried up but the tank was still empty, that would be a really bad time. They really should get an Automation Direct PLC or something. Even used ControlLogix chasses are pretty cheap if you go with an L6 processor and an ENBT module.
I'm really impressed with your meticulous approach to diagnosing and handling the problem. Using isolation transformers to prevent galvanic corrosion is brilliant.
On the powershell command: you should always force the command to shutdown, otherwise the computers can be stuck due to an app, and that's something you wouldn't want to happen in case of a disaster. What I always do when I have to script things like that is always declare everything. Usually, when I want to shutdown the computer/server immediately, I always do this: shutdown -s -f -t 0 That way, you -Shutdown, -Force -Timer 0, meaning that it'll force the shutdown command, ignoring other programs forcing them to close and do it immediately. You TECHNICALLY don't need the -f and -t together, but I always do it regardless.
Bingo, Jake used only the /s switch thus the default timer is a non-zero value and hence the message. If /t 0 was also added it would immediately take effect and /f is required to force Windows to close all running apps to shutdown.
Jake might have made the best explanation of why using a pull up resistor is necessary. I use mstp networks frequently and have struggled to understand the need for them and now I get it.
Actually it's the worst i've seen. This would have a been Perfect moment to explains ohm's law and do a quick current path analysis/schematic/animation. Just looks like someone who knew what it was about told him to do so, so he did but without understanding how it works.
@@SamuelCarrier92uh, nope it would not have. There's a right amount of time to spend on each step and that ain't it. They were already pushing it tbh so I'm glad people are getting value out of it.
I would assume it is the iodine in the water you mentioned in the last video. Nickel gets oxidized to nickel ions (Ni → Ni²⁺ + 2e⁻), and iodine is reduced to iodide ions (I₂ + 2e⁻ → 2I⁻). So, the presence of iodine directly accelerates nickel corrosion.
For those wondering, when you have a power source that outputs a specific voltage at specific amps, it means that it can output that voltage, and it can supply current from 0 amps up to the rated amps. It doesn't mean that the power supply will force X volts and Y amps through you. The amount of amps supplied is determined by the resistance of the what is consuming the power. The higher the resistance, the lower the current, the lower the resistance the higher the current, up to the capacity of the power supply. When you short the outputs of a power source, you introduce a very tiny resistance between the output terminals and you should get a lot of current flowing. Good power supplies have protections against this situation to avoid things overheating, blowing up, catching on fire, etc. They do this by measuring the output current and turning off the power when the threshold value is reached, which is usually a bit above the rated capacity. Bad power supplies usually just blow up when they are shorted or the current draw is too high.
What i like about these videos is unprofessional professionalism. He is playing, sitting on the ground, making jokes, it looks like two friends fixing the problem they have created. All the best guys
I actually called it! 9:35I apparently was wrong about the loop corrosion, but I definitely called vendors lying about the metals in the parts. So, I'll call that a partial win.
Well... now if i ever build a rack system i'm making sure any watercooled stuff has a drain shield of some sort so if it ever leaks water rolls around everything else
@@deathtrooper2048 In fairness, the cause of the problem in this situation could have happened to anyone. They literally got scammed by whoever they bought the plumbing valve from.
@zxqhyr sure it can be fun but for my daily gaming pc im def keeping it air cooled after all ive seen. +cheaper +reliability +mainenance free +higher resale value (gpu) -maybe a couple degrees hotter Id make that tradeoff anyday since none of my components are throtteling anyways. Plus my mb gets better airflow too :)
Exactly what I do, so it goes down the sides instead of through everything. Also, I use leak sensors, and leak frogs (Remember those?) just for extra protection.
plated brass fittings lol - brass is a garbage base material and there is a reason its used outdoors only, where leaks dont matter. Leads me to think the fitting was over-torqued and cracked the plating.
What do you mean? EK execs seems to be good at carefully considering how to open up foreign bank accounts 😂 without reporting them to regulators/their debt collectors. At least according to Gamer's Nexus's latest hardware news
"That might be the best thing you've ever done" "My kids are right there..." Linus is such sweetheart, I never had such type of compliments from my parents. 10/10 parenting
He really is a good parent. Love him or hate him, every time he's talked about his kids or his parenting style on the WAN show I've always been super impressed. And we can even see the results now with vids like upgrading the gaming van. I understand why he doesn't want them on camera constantly but I'd love to see more.
2:00 The screwhole in the middle on the right side does seem to be discolored. Probably not due to corrosion though, I'm sure Linus has checked everything thoroughly
It is SUPER IMPORTANT that all grounds end at the SAME PLACE. If one is much longer, or you have extenders/connectors or splits the resistance of the wires is different. That means the grounds are at different effective voltages. And thats where ground loops happen.
Actually in this particular case if you want to completely isolate two different circuits, you do NOT want both of them grounded because if they share a common ground they are be definition, not isolated.
This is wrong. Grounds can have different lengths or splits as long as they’re in the same circuit. Ground is ground and there is not different effective voltages whatever that means. Ground is always ground anywhere in the same circuit.
They didnt even strip the paint of the case on the grounding connection point. This is so unprofessional and it will fail again as it does nothing right now.
@@johhnyknoxville3948 That *Really* depends on the paint. A lot of case paints are conductive, literally to prevent this issue. (could be an issue on the ones they painted themselves, though) However, the threads probably haven't got any paint on them after the first time a screw was screwed in anyway, so the paint on the outside of the case wouldn't even be a problem in the first place seeing as the connector is touching the screw directly. What I'm saying here is; Unbunch your panties. Its more than good enough for a non-industrial use-case like this, and while its "unprofessional" it does what it needs to do. So calm down your mammaries.
@@The_Keeper paints are not conductive what are you even on babbling about and no, the inside of the screw hole isnt enough as inodized parts are indized as a whole. Also this screwhole isnt intended for grounding so the connection between this panel and the rest is unknown and probably never existed. Lower your tone when you have never worked with industrial stuff kid.
After finding a massive leak like that I would have reconsidered how my systems were setup. I would isolate the different water-cooled systems with verticle mounts which would prevent multiple systems from being damaged if a leak were to happen again (or have his team create a custom server rack that would keep the machines separated and not stacked). Maybe even do a monthly checkup to confirm if the problem has persisted.
I work for a semiconductor brush manufacturer, all tubing/fittings (compression fit) we use for our process is teflon and we flush system with DI water in order to maintain a extremely low particle count. This also helps prevent corrosion and keeps our equipment extremely clean. The only downside is the cost.
That's true. In my opinion it's not even "water-cooling", more like "water-transfers-heat-to-somewhere-else-which-will-then-be-actually-cooled-by-air", so the water is only used to transfer the heat to the radiators which do the actual cooling by air 😁
Random suggestion for Jake, using CAT is damn handy as an "everything wire", but if you use a little pigtail/patch cable for your "custom end", and leave the RJ45 on the other side you can use dirt cheap Ethernet Couplers to connect it to an unmodified cable of whatever length is needed. Repeat at the other end, and you've got an easily repairable/replaceable/testable cable.
@@DanKaschel Can be used to power stuff if the current is low enough like 100mA. PoE injectors/splitters are pretty useful for such a usecase too, you can use a step-up DC booster to convert your 5V/12V to like 24..48V for transmission, then on the other end use a step-down to transform it back to the desired voltage. Obviously this is most useful for long wire runs, but a standard CAT5e cable is great for transmitting data (RS-485/RS-422) and power together, to like another building floor, the yard, etc.
@@DRSDavidSoft yeah, low amp stuff is fine. I just mean you can't use cheap Ethernet couplers to transmit 24V power and not expect your house to eventually burn down.
22:46 I want to clarify a common misunderstanding here: The current is determined by the voltage and the resistance, so the important part is the voltage, but that's not the whole story. Usually, you wouldn't notice 5V through skin, but since your tongue is wet it conducts electricity much better. Thus, the resistance of your tongue is much smaller so you have quite a bit higher current. Now when he says that he licked a 9V battery before doesn't necessarily mean he actually licked 9V. Batteries have an internal resistance so if you lick them, they won't output the full 9V. Licking something that is plugged into the wall is much more dangerous since the internal resistance is usually much much lower here, so you will most likely get the full voltage. So is licking something with 5V that is plugged into an outlet dangerous? Probably not, but I wouldn't recommend it. Especially because there could be a hardware fault which delivers over 100V directly into your tongue.
Just a note the water sensors that work off conductivity wont work with ro water or pure water. You run into this when the local water is garbage and the only way to clean it up is with ro and then you have to condition it after to reintroduce a set amount of minerals. Floats work in any. Also RO water will leach even stainless steel and corrode it over time.... Just from working with steamers and what not for work in the past.
Linus. Rack will be grounded over your UPS. AS long as UPS is grounded using the 3rd wire in the power grid and it is touching the rack metal on metal it will ground the rack as well.
Guys! The chassis of the servers ARE grounded via the electrical connector! Those small cables made because you need good electrical connection on every part of the rack, so for example: the doors, the side panels, the top stuff... Those should be connected to ground, because if something became live (a server chassis, a faulty cable) then the power will go to the ground, and not through you :) But if the door not connected properly with cable, and may or may not have good connection, then is the door get live somehow, you will be zapped!
Fun fact, that isolation transformer makes all that grounding pointless since it'll prevent the RCD/GFCI from working anyway. Linus should put an RCD/GFCI on the server side of the transformer and ensure it is bonded with those ground lines. Though I guess it will still help prevent the corrosion issue.
@TheAkashicTraveller The server is plugged into wall power, it's the pump on the other end of the transformer. Also they plugged the transformer into the server UPS to keep water flowing when there's a power outage
@@TheAkashicTraveller GFCI's aren't bonded to ground lines. They respond to a difference between live and neutral. Still a good idea to put one on the output of the transformer. The original RCD will work though. Transformer can't put more power out than what's going in.
Guys, the wires are for equipotential bonding (to pull all floating voltages down to the same earth potential, important for electronics also steel structures and copper piping), not protected earth bonding (which is for an RCD/GFCI in case of an electrical fault).
Just a quick PSA, galvanically isolating things via a 1:1 transformer is generally safe, but there is the possibility using one will prevent your GFCI protections from working properly. I am no expert in this field, but it is worth looking into before you get one.
It probably will prevent it from working (that is kind of the idea behind galvanic isolation) but you can also not short yourself or other stuff to ground. Though if you touch phase and neutral there pretty much is nothing to save you so the pump probably should be labeled "spicy" at least.
@@LeafBoye just because he isn't an expert in the field doesn't make his advice not accurate and in this case not having GFCI is kinda important to know
One thing I would maybe keep in mind is the leak sensor might lose connection to the leak when the water is stopped after a few moments and that could turn it back on causing more water then the leak would be detected, you get the point. Might be a good idea to either manually turn the water back on or put a delay of at least a minute for false positives to turn back on automatically
Careful with the Isolation transformer and those relays. The load rating on the relay is for a resistive load, the transformer (especially of this size) is a massive inductive load. There is a possibility on huge flyback voltages that will absolutley destroy those tiny relays.
How about using two flow sensors. One monitoring the flow to the rack and one measuring the return from the rack. If you then subtract the two readings (with a little filtering), you could be able to detect a slower leak. Another possibility is to monitor the fill state of the header tank. If it starts dropping, you’re losing water.
How precise are the sensors though? It's not uncommon to have something that is a few percent off and that would definitely make estimating a leak pretty difficult (unless it's quite massive).
@@patrick_test123 I think measuring the amount of water in the tank should be good enough, and they're doing that with the floaters. You'll know when water goes missing.
@@Keiuran True to be fair. Most of the EK stuff falls on four top level management people at least according to Gamers Nexus. They never complained about the quality of products besides the CEO not owning up for a bad defect in one and instead extorting a small mom pop shop for fixing the broken products to sell
@@KeiuranAccurate. I work in a company that does both engineering analyses and legal strategy for injured people suing companies, and after many, many years of reviewing discovery I can confirm that 99% of the time the harebrained ideas come from the bean counters in leadership, not the engineers.
When I saw an uncensored QR code I had to go back. Then I saw it was sharper than the rest of the screen. My first thought was "this goes to the LTT store doesn't it." Glad to find out I was right lol.
IEC 60364 cries loud here in special in 7:14 should be everything written in IEC 60364-5-54 (Low-voltage electrical installations - Selection and erection of electrical equipment - Earthing arrangements and protective conductors)
Linus you'll need some state keeping for that leak sensor such that when it stops detecting leak, you'll have to manually turn on the loop, instead of the loop immediately powering up again - picture this: the loop leaks, leak sensor triggers shutdown of loop, water on leak sensor evaporates, pumps restart - you'll still dump all the water out eventually.
I worked with pretty powerful water-cooled lasers mounted inside racks. We always add a tray below them, either to contain or at least divert any leakage from the rest of the system (computers, data acquisition hardware, ups...)
in a pc rack you would need to add another U height for each server to do this i am not saying it is impossible but it depends on how many U's you have and how many you are using
@@111smd exactly, but in practice new racks for often have some spare room. And if your tolerance are not too tight it is possible to shift elements by 1/3 U for a low profile tray
Have a low flow rate warning. Sometimes the impellers in flow sensors get gunked up and record lower flows. In some flow sensors you can remove the impeller to clean them, though it does void the warranty and you'd hope it would remain clean throughout the loop
They did put a filter in place to prevent the turbine flowmeter from being gummed up. Still, mag flow meters can be had pretty cheap now, which I prefer because mechanical parts will fail.
Grounding - for proper connection use a combination of washers inorder: grounding bar, external toots lock washer, the crimp connector of the wire, external toot lock washer, normal washer, nut. This gives a proper connection to the buss bar.
I love these videos where Jake nerds out on custom projects or cool networking gear. He seems like a guy who genuinely enjoy both learning this stuff himself and teaching others. His passion makes these videos all the better.
This makes me glad I stick to fans and I don't overclock anything. Having to worry about all this stuff seems like a headache and possibly a money sink. Worst case scenario for me is a fan starts to make strange noises, and I go replace it.
I work with high voltage DC, 400v to a little over 800v. On our equipment most of our grounds are exposed flat braided steel. As long as you're using protected circuits with updated (to code) components like an AFCI breaker, the breaker will break before the ampacity of your ground loop is exceeded, especially on a standard house circuit since those breakers are extremely sensitive.
A lot of the sensors could just be incorporated to HAOS then MQTT over to pulseway with an ESP32 via the ESPHOME addon for HAOS. No special boards, no bs, just pure automation. Not a fan of premade solutions as that can get very pricey very fast.
@LTT nice idea with the water sensor, just beware that it might no work with demineralized water as the conductivity is lower. That is if demineralized water is used. What could also be done is add 2 Flow meters in the loop before and after the rack and compare for difference in flow with a margin of error. If there is it would indicate a leak.
@@SamuelCarrier92 would agree in a closed loop in an open loop system the pressure (which is resistance against flow) will be almost none in the return line and probably instable in the feed line. I am not entirely sure how the system is built but from what I gathered it is open loop. The flow stays constant no matter the pressure in a liquid.
Linus and Jake, Zooz makes a Valve control unit that can shut the valve when a command is sent to them. You should check it out. Also, FYI about Z-Wave, it can only send a signal at most 4 hops, so be careful about that with his Z-Wave network.
@LeafBoye I have had this happen to me, and discovered that fact when trying to solve my own problem with my Z-Wave network. You can look it up yourself if you want to, I am happy to be given newer, accurate info if I am wrong.
I was working on a project for my college which involved flow sensors, leak sensors and relays. I know how to use them but this video really helped me connect some dots and it dropped on such a good time hehe
Not sure why you went with water cooling for your server considering a lot of the videos I've watched from you mentioned that air cooled is more reliable and accessible. If you really wanted to make use of your pool, you could have gone with water sourced heat pump and cooled the entire room. Or ducted it only to your server rack. Still would have made a great video.
You have to be mindful when grounding stuff to a painted rack. You need to make sure that you use grounding points that aren't impeded by the paint. Usually, there will be a dedicated copper threaded stud, but sometimes you have to grind off some paint and use some barbed washers.
Another suggestion is to get a switched rack PDU (if you don't already have one). This would allow you to physically disconnect power to systems in case of a leak, which is obviously safer than just switching them off.
Oh you could totally use a poe esp32 with esphome for the leak sensors, I'm surprised this wasn't the first option! Much less to go wrong than a full rpi OS
Okay let's see if I understand this; The reason that the corrosion happened is because the rack was not properly grounded which means the rack was using the water in the loop as a ground and since there was a electronic differential in the rack versus ground then it caused the electronic galvanic corrosion. Correct? Or at least correct on a basic level?
@@SSYoung125 I mean that's correct to what they thought it was until they found out the faucet head to the resevoir tank was made out of cheap metal and not 100% brass like they thought and was causing the corrosion. Oh and not from not being grounded persay but just not isolated so using the water to carry a very small electric differential.
I would say wrong. I would have loved to have them measure the resistance between rack and the ground of the house for them to realized it was 0 ohm. With that amount of grounded box in a rack directly touching metal to metal and screwed in making a good ground connection. So rack was as grounde as it is after adding the grounds. unless there was something isolating the cases for example pc cases were all plastic and that the scree weren't metal the you could have a ungrounded rack ( and it wouldn't matter). The ground not making a difference in my opinion I would agree with you that this video misses an electrical drawing or visual that explains what their theory is with regards to the voltage difference on the 2 parts of the system. how would on pump being fed at 120V and the pc being at 118V how would that affect corrosion? through induction? I'm not sure I understand their theory after that video.
@@SamuelCarrier92 Water does weird things when energized even the slightest. Having an inconsistent ground doesn't help. I personally think it partly has to do with a mixture of issues like the ground issue and voltage generated from moving water in plastic pipe and mixture of metal.
@@SamuelCarrier92 A rack standing on the ground does not mean it's grounded, tying it into the panel is a good move. As for the cases and the rack being the same potential, probably, though paint can be a bitch. But then again, it doesn't seem like they removed/scratched the paint for the grounding wires either.
I used to manage an irrigation and chemical spray store. We sold fittings that were used in chemical sprayers and would either be a brass alloy (brass with other resistant metals) or nylon. Both were great and extremely resistant to corrosion. Stainless steel is great as well.
This should come in Handy when everything freezes solid this winter. Jake’s a smart cookie. Always enjoy videos with him. For a totally different reason than why I enjoy Alex Videos.
You know what makes me mad when they say they're going to do it the right way in the future, but you know they're never going to do it until something else catastrophic happens
Is there anyway you can 3D print or buy some acrylic sheeting and thin piping which can go between the computers and drain to the floor to catch any future leaks so they don’t wipe out multiple computers? Oh and on the red handled gate valve, they are really crappy. I was draining an expansion vessel on a calorifier two weeks ago and I had one of these not catch the thread and refuse to open back up, which effectively turned the water off for half of a factory that produces bearings for nuclear reactors. I managed to open the valve by fully turning closed the valve as hard as I could and then opening it as softly as possible; the noise the system made as water flushed back through was tremendous. Replace the gate valves with ball valves. Bung up the tank (Google plumbing bung) so no more coolant can go down the pipes and you won’t need to drain the tanks.
As everyone else is chipping in, why not. Isolating the pump, sure why not. The issue here is that there's a voltage difference somewhere causing the ox/redox reaction to happen somewhere. AC won't cause this, but funky stuff can happen with oxides, as you can create diodes. Another issue you have is that you are using a DC bus, and not a single star groud over your house. Yes it does connect at the pannel but you only need a couple mA on the ground run to induce a voltage difference across the groud lead. Moreover if one supple is leaking DC to the ground. You'll also have capactive and inductive coulpling, loops everywhere and whatnot.
Hey water heaters have a sacrifice anode rod for the minerals and other stuff to eat then eating the metal boiler in a water heater. Maybe use a 30 gal UC (under counter) water anode rod in one of those plastic square tanks to reduce alot of that corrosion
Do not use multimeter to check that the voltage is zero. It can give you false reading. Use special tools like Fluke T110 voltage tester which has resistor to drain any residue voltages and gives proper reading.
Yeah presumably it will be replaced by a real electrician in a few weeks. Or it won't be and they're showing the most realistic form of home improvement: a jank solution that becomes permanent lmao
this is shown to trigger Electricians as they and rest of the professionals are almost as bad as fanboys in the comment section and they will feed the youtube algorithm.
Don't forget the statement right after this... "its just a ground, it should be fine" , my boss would make a whip from ground wires and whip me out of consciousness if I ever say that.
Engineer here too(this seems to be a thing here) - this was the thinnest seperate grounding wire i've ever seen. Would have use at least 10 or 16mm² - convert this to your freedom units
I'd like to mention that you could also install anode rod's in the water tanks just like they do in water heaters to prevent corrosion in the tank lining from harsh water quality. I know the tanks you have installed are plastic, but it might help protect other things like metal fittings in the tanks.
Always trust Linus to do a Waterloop leak video once every few months. Consistency over a decade of videos, I love it.
Honestly what I was thinking when this came up 😂
He could save himself so much time by just making his rack into an aquarium rack and cool the mineral oil
@@andrew1898 I've been waiting for LTT to do another mineral oil PC video for 10 years now.
Same thought
It's a sign of how unreliable they are.
Seasoned marine electrical engineer here. We specialize in galvanic and electrolytic corrosion prevention. Galvanic corrosion is caused by DC, not AC. Isolation transformers are used to derive a separate ground to prevent DC eddy currents of 2V or under from destroying metals. You CANNOT connect the ground on either side of the isolation transformer, or it does absolutely nothing. I'm not sure what you bought, but most of those benchtop isolation transformers do not derive a separate ground. It seems counterintuitive, but the DC current will just travel though the ground and the isolation transformer would be completely useless. The excessive rack grounding wasn't really necessary either as normally these racks are already grounded through the chassis of the equipment in the rack (if in doubt, use a multimeter and ohm it out). If this fails, look up a marine galvanic isolator. This little $ 50 device would have completely taken care of the problem, but i supposed it would have made for a less interesting video lol. They work well, and these are installed on the shore power inlet on boats. The AC ground on a boat is also bonded to the DC- battery ground and the sacrificial anode (along with other bits of metal in the water). Add in some saltiness to the water with a few other boats in a marina with some faulty wiring, and you have a recipe for many thousands of dollars of damage to engines and anything else touching water. These galvanic idolators prevent that and are nothing but a couple of diodes in parallel (and opposite polarity) which will pass an AC fault current easily tripping a GFCI, ELCI, or main breaker while blocking the harmful DC eddy currents that are destructive. You can even make one yourself, but you will need diodes rated to the circuit ampacity which is dictated by the breaker size. Love your videos!
I'd have preferred to see the rack units grounded at the (internal) fittings: the heat sink inside the box might not be bonded to the chassis.
Shhh! if you tell him how to do it properly we wont see the next instalment . . . should be around mid December 🤞🤞
@@Mark-M72wouldn’t want to end this series prematurely
Nah. Just change out the flux capacitor.
It's been a long time since Chemistry and Galvanics for me, but wouldn't the water need to be in contact with electric? Like two metals of different types might make a battery, yes, but only if their potential / the difference of them is high enough; so checking the compatibility should have prevented that. The other way would require electrodes, i.e. metals connected to voltage touching the water, no? Wouldn't it then be a way bigger problem that you have water, carrying hundreds of volts and amperes, ready to short/fry anything if it ever leaked? Also obviously this only works if the water contains ions, but i guess at enough of a difference in potential the main point is that any metal would shed ions?
Thanks LTT!
I now know to use air cooling for everything.
@@cameront8539 ya glad I watch these videos. I will never water cool anything.
I don’t need “content” in my daily life.
Yup.
JaysTwoCents;
Shows you why watercooling is awesome.
LTT;
Shows you why it really isn't.
any liquid will eventually leak :3
so that should be accounted for
Industrial controls engineer here….
You should add some sort of electrically operated valves in the lines before the pumps. If you shutoff the pumps due to a leak, gravity may still allow flow through the pump
At this point they just need to put in a PLC lol
Monkey here, can a brother borrow a banana?
and put some kind of draining system for when it happens again.
especially with a massive hot water tank on other side of room ;-) - needs to upgrade to continuous hot water... geez what kind of geek is he?
or have another sensor to activate a bilge pump to pump water outside if he can't get a drain on the floor...
This type of corrosion is why Grounding Systems for Homes are no longer allowed through the copper piping (at least in Germany). Back in the days they would just use Water or gas pipes to be used as the ground. These days old homes get a ring ground and new homes get a foundation ground.
To prevent corrosion on the piping system and to prevent voltage differences we have to run a 6mm² wire to all the pipes in one big loop in the basement and connect them to the ground System. This way the current will run through the wire and bring it all to one "voltage".
Galvanic corrosion is why none of your metal pipes are made of anything but copper or metals of similar nobility. Any more/less noble metal in contact with the electrolyte (water) and electrically connected to the copper would cause corrosion. Grounding has nothing to do with it, and in fact, your wire is still an electrical bond that would _introduce_ corrosion if there _was_ a dissimilar metal.
The reason you're not allowing it as a ground is almost certainly the same reason the US requires that there be an alternate grounding system -- it's far, far too easy for it to wind up insulated.
the reason the leak sensor isnt reading exactly 0V while its dry is to be able to tell the difference between a zero singal and a broken sensor. Not as important in a home setup but in industry applications where alot of sensors are at work the risk of one breaking is so high that having any signal as 0V or 0mA isnt something that is done.
By now Jake should own a quarter of the house. He already built/rebuilt/repaired half of it.
Damn my contractor should get the deed then
Maybe if he did it for free...
Brother. Let's fight for the workers
Lol silly, it's vancouver, people under 50 can't own houses there.
That's... not how paid labor works, but okay lmao.
Nice!
Recommendations:
-set up a testing schedule for yourself. Protection systems that aren't tested won't work when the demand occurs. Failures are like a watched pot--won't boil if you're paying attention!
-your shutdown of the pumps should be "latched". It really really shouldn't turn them back on when the leak sensor clears. You want someone to go and physically check on it and reset.
This is fun. Can't wait 'til y'all just buy Jake a PLC and go full industrial.
Ladder logic is dead simple and you know the equipment is reliable.
These are great suggestions ngl! I'd say they should listen to them, but more leak videos is always a plus!
That is not a comment - it is a life lesson.
110% the pump should require a reset before restarting. If the leak just dried up but the tank was still empty, that would be a really bad time. They really should get an Automation Direct PLC or something. Even used ControlLogix chasses are pretty cheap if you go with an L6 processor and an ENBT module.
@@Katia413 A PLC seems a bit overkill but they should definitely add a restart button for when the leakage sensor triggers.
4:51 editor missed a trick not adding Jake to the family lineup
I'm really impressed with your meticulous approach to diagnosing and handling the problem. Using isolation transformers to prevent galvanic corrosion is brilliant.
On the powershell command: you should always force the command to shutdown, otherwise the computers can be stuck due to an app, and that's something you wouldn't want to happen in case of a disaster.
What I always do when I have to script things like that is always declare everything. Usually, when I want to shutdown the computer/server immediately, I always do this:
shutdown -s -f -t 0
That way, you -Shutdown, -Force -Timer 0, meaning that it'll force the shutdown command, ignoring other programs forcing them to close and do it immediately. You TECHNICALLY don't need the -f and -t together, but I always do it regardless.
I would also add -d xx:yy and -c "comment" options to leave a trace in the event viewer for the reason why the shutdown happened
that is a command prompt command, not powershell
stop-computer is the powershell native command...
Bingo, Jake used only the /s switch thus the default timer is a non-zero value and hence the message. If /t 0 was also added it would immediately take effect and /f is required to force Windows to close all running apps to shutdown.
"shutdown /p /f" also works, and does the same thing. Immediate shutdown with applications killed.
Heh yeahhhhhh not shutting down because notepad is open and had one character written into it would not be good....
Jake might have made the best explanation of why using a pull up resistor is necessary. I use mstp networks frequently and have struggled to understand the need for them and now I get it.
Actually it's the worst i've seen. This would have a been Perfect moment to explains ohm's law and do a quick current path analysis/schematic/animation. Just looks like someone who knew what it was about told him to do so, so he did but without understanding how it works.
@@SamuelCarrier92uh, nope it would not have. There's a right amount of time to spend on each step and that ain't it. They were already pushing it tbh so I'm glad people are getting value out of it.
I’m surprised that breakout board didn’t have PU/PD resistors, as that’s a pretty common requirement when hooking up sensors.
@@SamuelCarrier92bruh i would’ve turned the video offl
It's like adding weak spring to one end of string, so when it's not loaded, it's always in defined position, pulling it up :)
I would assume it is the iodine in the water you mentioned in the last video. Nickel gets oxidized to nickel ions (Ni → Ni²⁺ + 2e⁻), and iodine is reduced to iodide ions (I₂ + 2e⁻ → 2I⁻). So, the presence of iodine directly accelerates nickel corrosion.
@@GunniBusch1 they have been known to use tap water instead of distilled so it makes sense
@@rar23424ironically distilled water is more corrosive than tap water
@mzamroni they use floride at least in the US for that. They probably use the same in Canada
For those wondering, when you have a power source that outputs a specific voltage at specific amps, it means that it can output that voltage, and it can supply current from 0 amps up to the rated amps.
It doesn't mean that the power supply will force X volts and Y amps through you. The amount of amps supplied is determined by the resistance of the what is consuming the power. The higher the resistance, the lower the current, the lower the resistance the higher the current, up to the capacity of the power supply.
When you short the outputs of a power source, you introduce a very tiny resistance between the output terminals and you should get a lot of current flowing.
Good power supplies have protections against this situation to avoid things overheating, blowing up, catching on fire, etc.
They do this by measuring the output current and turning off the power when the threshold value is reached, which is usually a bit above the rated capacity.
Bad power supplies usually just blow up when they are shorted or the current draw is too high.
What i like about these videos is unprofessional professionalism.
He is playing, sitting on the ground, making jokes, it looks like two friends fixing the problem they have created.
All the best guys
I actually called it! 9:35I apparently was wrong about the loop corrosion, but I definitely called vendors lying about the metals in the parts. So, I'll call that a partial win.
Well... now if i ever build a rack system i'm making sure any watercooled stuff has a drain shield of some sort so if it ever leaks water rolls around everything else
How about just keep water away from electronics. LTT always gets water damage.
@@deathtrooper2048 In fairness, the cause of the problem in this situation could have happened to anyone. They literally got scammed by whoever they bought the plumbing valve from.
@zxqhyr sure it can be fun but for my daily gaming pc im def keeping it air cooled after all ive seen.
+cheaper
+reliability
+mainenance free
+higher resale value (gpu)
-maybe a couple degrees hotter
Id make that tradeoff anyday since none of my components are throtteling anyways. Plus my mb gets better airflow too :)
All it takes is a thin plastic shield with drains on the sides, kinda like a sunroof drain in a car.
Exactly what I do, so it goes down the sides instead of through everything. Also, I use leak sensors, and leak frogs (Remember those?) just for extra protection.
"EK carefully considered"
Best oxymoron there
Yeah, that 'minor' issue with some nickel plated parts.
plated brass fittings lol - brass is a garbage base material and there is a reason its used outdoors only, where leaks dont matter. Leads me to think the fitting was over-torqued and cracked the plating.
What do you mean? EK execs seems to be good at carefully considering how to open up foreign bank accounts 😂 without reporting them to regulators/their debt collectors. At least according to Gamer's Nexus's latest hardware news
they "carefully considered" stuff in about the same way Fanatec carefully considered stuff
"Dumbledore said calmly"
Its a good sponsorship because it ties to the project on video and what it can do was shown with the thing actually being used and useful.
“Don’t try this at home”
Linus: *tries literally everything at home”
That explosion when Jake opened the switch at 16:10 had me in stitches. Well done editor
when he plugged the cord in too lol 22:20
I love these little bad explosions editing 😂
@@BladeScraper I jumped at that one..
@@Lem_On_Lime lol me too
"That might be the best thing you've ever done"
"My kids are right there..."
Linus is such sweetheart, I never had such type of compliments from my parents.
10/10 parenting
He really is a good parent. Love him or hate him, every time he's talked about his kids or his parenting style on the WAN show I've always been super impressed. And we can even see the results now with vids like upgrading the gaming van. I understand why he doesn't want them on camera constantly but I'd love to see more.
2:00 The screwhole in the middle on the right side does seem to be discolored. Probably not due to corrosion though, I'm sure Linus has checked everything thoroughly
also one on the left side
Ye I noticed the left side before that.
It is SUPER IMPORTANT that all grounds end at the SAME PLACE. If one is much longer, or you have extenders/connectors or splits the resistance of the wires is different. That means the grounds are at different effective voltages. And thats where ground loops happen.
Actually in this particular case if you want to completely isolate two different circuits, you do NOT want both of them grounded because if they share a common ground they are be definition, not isolated.
This is wrong. Grounds can have different lengths or splits as long as they’re in the same circuit. Ground is ground and there is not different effective voltages whatever that means. Ground is always ground anywhere in the same circuit.
They didnt even strip the paint of the case on the grounding connection point. This is so unprofessional and it will fail again as it does nothing right now.
@@johhnyknoxville3948 That *Really* depends on the paint.
A lot of case paints are conductive, literally to prevent this issue. (could be an issue on the ones they painted themselves, though)
However, the threads probably haven't got any paint on them after the first time a screw was screwed in anyway, so the paint on the outside of the case wouldn't even be a problem in the first place seeing as the connector is touching the screw directly.
What I'm saying here is;
Unbunch your panties.
Its more than good enough for a non-industrial use-case like this, and while its "unprofessional" it does what it needs to do.
So calm down your mammaries.
@@The_Keeper paints are not conductive what are you even on babbling about and no, the inside of the screw hole isnt enough as inodized parts are indized as a whole. Also this screwhole isnt intended for grounding so the connection between this panel and the rest is unknown and probably never existed.
Lower your tone when you have never worked with industrial stuff kid.
After finding a massive leak like that I would have reconsidered how my systems were setup. I would isolate the different water-cooled systems with verticle mounts which would prevent multiple systems from being damaged if a leak were to happen again (or have his team create a custom server rack that would keep the machines separated and not stacked). Maybe even do a monthly checkup to confirm if the problem has persisted.
I work for a semiconductor brush manufacturer, all tubing/fittings (compression fit) we use for our process is teflon and we flush system with DI water in order to maintain a extremely low particle count. This also helps prevent corrosion and keeps our equipment extremely clean. The only downside is the cost.
Next video idea: Linus gets a distiller to make distilled water for his custom server loop.
All these water-cooling videos remind me that going with an air-cooler was such a wise decision.
That's true. In my opinion it's not even "water-cooling", more like "water-transfers-heat-to-somewhere-else-which-will-then-be-actually-cooled-by-air", so the water is only used to transfer the heat to the radiators which do the actual cooling by air 😁
Yeah, go back to an air cooled car. Weird comment for an exotic cooling video.
@@NPurvis7622 Comparing a PC to a car is laughable. You’re the one with the weird comment here bud.
@@NPurvis7622 i like my air cooled car though
And it kinda makes me regret getting an AIO liquid cooler for my most recent build. One day, I'm going to have a problem.
Random suggestion for Jake, using CAT is damn handy as an "everything wire", but if you use a little pigtail/patch cable for your "custom end", and leave the RJ45 on the other side you can use dirt cheap Ethernet Couplers to connect it to an unmodified cable of whatever length is needed. Repeat at the other end, and you've got an easily repairable/replaceable/testable cable.
Good hack for signal wire, but don't do this for anything used for powering a device.
@@DanKaschel Can be used to power stuff if the current is low enough like 100mA.
PoE injectors/splitters are pretty useful for such a usecase too, you can use a step-up DC booster to convert your 5V/12V to like 24..48V for transmission, then on the other end use a step-down to transform it back to the desired voltage. Obviously this is most useful for long wire runs, but a standard CAT5e cable is great for transmitting data (RS-485/RS-422) and power together, to like another building floor, the yard, etc.
I love this hack, will deploy it in a project of my own.
@@DRSDavidSoft yeah, low amp stuff is fine. I just mean you can't use cheap Ethernet couplers to transmit 24V power and not expect your house to eventually burn down.
@@DanKaschel Yes, exactly, it's meant as a data transmission cable.
22:46
I want to clarify a common misunderstanding here:
The current is determined by the voltage and the resistance, so the important part is the voltage, but that's not the whole story.
Usually, you wouldn't notice 5V through skin, but since your tongue is wet it conducts electricity much better. Thus, the resistance of your tongue is much smaller so you have quite a bit higher current.
Now when he says that he licked a 9V battery before doesn't necessarily mean he actually licked 9V. Batteries have an internal resistance so if you lick them, they won't output the full 9V.
Licking something that is plugged into the wall is much more dangerous since the internal resistance is usually much much lower here, so you will most likely get the full voltage.
So is licking something with 5V that is plugged into an outlet dangerous? Probably not, but I wouldn't recommend it. Especially because there could be a hardware fault which delivers over 100V directly into your tongue.
The frequency of these water loop leak videos is all I need to never want to water cool ever
Just a note the water sensors that work off conductivity wont work with ro water or pure water. You run into this when the local water is garbage and the only way to clean it up is with ro and then you have to condition it after to reintroduce a set amount of minerals. Floats work in any. Also RO water will leach even stainless steel and corrode it over time.... Just from working with steamers and what not for work in the past.
0:52 Linus is going to rectify high blood pressure? 🤔
Rectumfry.....
mmmfgh@@matt7662
@@matt7662
(totally unrelated)
just watched LOTR: FOTR on APV
I cried like a baby when Boromir died (don't laugh)
Linus. Rack will be grounded over your UPS. AS long as UPS is grounded using the 3rd wire in the power grid and it is touching the rack metal on metal it will ground the rack as well.
I was wondering, the rack seemed to have a finish on it, it's not on bare metal. Is the finish conductive?
@@DrowslayerQC Their whole rack grounding seemed funnily unnecessary tbh, must've been jake's idea.
@@DrowslayerQC The only reasonable answer is: you never know, therefore you assume the worst.
18:00 As a curious guy, that QR code was calling my name... Bravo Editor, Bravo!
was it a rickroll?
@@TheBackyardChemist It goes to the LTT Store 😂
Thank you Jake for showing me Sequent Microsystem. I saved their website and may use them for prototyping for work.
Guys! The chassis of the servers ARE grounded via the electrical connector! Those small cables made because you need good electrical connection on every part of the rack, so for example: the doors, the side panels, the top stuff... Those should be connected to ground, because if something became live (a server chassis, a faulty cable) then the power will go to the ground, and not through you :) But if the door not connected properly with cable, and may or may not have good connection, then is the door get live somehow, you will be zapped!
Hence why grounding and bonding are two separate things. I wish they would have explained it better, or at all.
Fun fact, that isolation transformer makes all that grounding pointless since it'll prevent the RCD/GFCI from working anyway. Linus should put an RCD/GFCI on the server side of the transformer and ensure it is bonded with those ground lines. Though I guess it will still help prevent the corrosion issue.
@TheAkashicTraveller The server is plugged into wall power, it's the pump on the other end of the transformer. Also they plugged the transformer into the server UPS to keep water flowing when there's a power outage
@@TheAkashicTraveller GFCI's aren't bonded to ground lines. They respond to a difference between live and neutral. Still a good idea to put one on the output of the transformer. The original RCD will work though. Transformer can't put more power out than what's going in.
Guys, the wires are for equipotential bonding (to pull all floating voltages down to the same earth potential, important for electronics also steel structures and copper piping), not protected earth bonding (which is for an RCD/GFCI in case of an electrical fault).
Just a quick PSA, galvanically isolating things via a 1:1 transformer is generally safe, but there is the possibility using one will prevent your GFCI protections from working properly. I am no expert in this field, but it is worth looking into before you get one.
If your not an expert why are you even giving electrical suggestions?
@@LeafBoyeonly experts are allowed to know things.
It probably will prevent it from working (that is kind of the idea behind galvanic isolation) but you can also not short yourself or other stuff to ground. Though if you touch phase and neutral there pretty much is nothing to save you so the pump probably should be labeled "spicy" at least.
@@LeafBoye just because he isn't an expert in the field doesn't make his advice not accurate and in this case not having GFCI is kinda important to know
@@LeafBoyebecause other not experts showed on their video that they are using big ass anti gfci device?
One thing I would maybe keep in mind is the leak sensor might lose connection to the leak when the water is stopped after a few moments and that could turn it back on causing more water then the leak would be detected, you get the point. Might be a good idea to either manually turn the water back on or put a delay of at least a minute for false positives to turn back on automatically
Honestly pretty solid sponsorship. Like a genuinely good use of the product, and didn't feel forced
Careful with the Isolation transformer and those relays. The load rating on the relay is for a resistive load, the transformer (especially of this size) is a massive inductive load. There is a possibility on huge flyback voltages that will absolutley destroy those tiny relays.
16:12 legit scared me
Same here 😅
How about using two flow sensors. One monitoring the flow to the rack and one measuring the return from the rack. If you then subtract the two readings (with a little filtering), you could be able to detect a slower leak.
Another possibility is to monitor the fill state of the header tank. If it starts dropping, you’re losing water.
How precise are the sensors though? It's not uncommon to have something that is a few percent off and that would definitely make estimating a leak pretty difficult (unless it's quite massive).
@@meneldal If the other method is to wait for a puddle at the bottom of the rack you should be able to find a sensor that can do better than that.
@@patrick_test123 I think measuring the amount of water in the tank should be good enough, and they're doing that with the floaters. You'll know when water goes missing.
This must have been made before the recent EK news. EK and "carefully consider" don't really belong in the same sentence now lololol.
+1
EK top management carefully considers how to threaten and extort their employees and dodge government officials 😂
It's not good to equate company leadership with company engineers. Engineers are often doing their job properly regardless of where they work.
@@Keiuran True to be fair. Most of the EK stuff falls on four top level management people at least according to Gamers Nexus. They never complained about the quality of products besides the CEO not owning up for a bad defect in one and instead extorting a small mom pop shop for fixing the broken products to sell
@@KeiuranAccurate. I work in a company that does both engineering analyses and legal strategy for injured people suing companies, and after many, many years of reviewing discovery I can confirm that 99% of the time the harebrained ideas come from the bean counters in leadership, not the engineers.
Whenever I feel bad about my self, I come here and watch you guys do stuff totally wrong most all the time and it makes me feel better.
One of the best "nerdy" videos in a long time! I really want to see more electronics content, and using Pulseway was a great tie in!
18:00 that qr code leading to ltt store has to be the funniest shit I've ever seen
LOL I was looking for this comment :D
When I saw an uncensored QR code I had to go back. Then I saw it was sharper than the rest of the screen. My first thought was "this goes to the LTT store doesn't it." Glad to find out I was right lol.
@@superscatman4236I thought it was strange it wasn't censored.
I was waiting to be rick rolled to be honest
@@superscatman4236 Same but my thought was either the LTT Store or Rick Roll
@10:15 Jake makes his own Valve. HL3 confirmed.
There can be no doubt on this one...👍
IEC 60364 cries loud here in special in 7:14 should be everything written in IEC 60364-5-54 (Low-voltage electrical installations - Selection and erection of electrical equipment - Earthing arrangements and protective conductors)
16:03 lmao good one😂
22:19 never gets old!
Impressive work by the way... Let's hope nothing goes wrong~ 🎉
Linus you'll need some state keeping for that leak sensor such that when it stops detecting leak, you'll have to manually turn on the loop, instead of the loop immediately powering up again - picture this: the loop leaks, leak sensor triggers shutdown of loop, water on leak sensor evaporates, pumps restart - you'll still dump all the water out eventually.
I worked with pretty powerful water-cooled lasers mounted inside racks. We always add a tray below them, either to contain or at least divert any leakage from the rest of the system (computers, data acquisition hardware, ups...)
in a pc rack you would need to add another U height for each server to do this
i am not saying it is impossible but it depends on how many U's you have and how many you are using
@@111smd exactly, but in practice new racks for often have some spare room. And if your tolerance are not too tight it is possible to shift elements by 1/3 U for a low profile tray
Have a low flow rate warning. Sometimes the impellers in flow sensors get gunked up and record lower flows. In some flow sensors you can remove the impeller to clean them, though it does void the warranty and you'd hope it would remain clean throughout the loop
They did put a filter in place to prevent the turbine flowmeter from being gummed up. Still, mag flow meters can be had pretty cheap now, which I prefer because mechanical parts will fail.
this intro from the "pee myself" to the intro music is too great 1:06
*computer peed itself
Grounding - for proper connection use a combination of washers inorder: grounding bar, external toots lock washer, the crimp connector of the wire, external toot lock washer, normal washer, nut. This gives a proper connection to the buss bar.
I love these videos where Jake nerds out on custom projects or cool networking gear. He seems like a guy who genuinely enjoy both learning this stuff himself and teaching others. His passion makes these videos all the better.
23:48 M539 Resturation.....in my Town
@@psy_harkn5100 ??
Missed opportunity for the perfect segway 12:55
I can't believe I got jumpscared by explosions twice lol. It really does feel like that when working with electronics and wiring and all lol
saaaaaame
This makes me glad I stick to fans and I don't overclock anything. Having to worry about all this stuff seems like a headache and possibly a money sink. Worst case scenario for me is a fan starts to make strange noises, and I go replace it.
I work with high voltage DC, 400v to a little over 800v. On our equipment most of our grounds are exposed flat braided steel. As long as you're using protected circuits with updated (to code) components like an AFCI breaker, the breaker will break before the ampacity of your ground loop is exceeded, especially on a standard house circuit since those breakers are extremely sensitive.
i love that he actually immediatly states that us viewers helped diagnose the problem :)
Galvanic corossion, galvanic isolation are cool and all but what about galvanic square steel?
eco friendly wood veneer is electrically isolating, might be worth considering
galvanized square steel tubes ftw
This is the best series LLT has had since Sketchy Heat Sinks. Even my wife loved SHS, and she's an Communications Teacher.
A lot of the sensors could just be incorporated to HAOS then MQTT over to pulseway with an ESP32 via the ESPHOME addon for HAOS. No special boards, no bs, just pure automation. Not a fan of premade solutions as that can get very pricey very fast.
Pretty cool setup. Thanks for showing, as you go, how to implement each bit. Really neat.
@LTT nice idea with the water sensor, just beware that it might no work with demineralized water as the conductivity is lower. That is if demineralized water is used. What could also be done is add 2 Flow meters in the loop before and after the rack and compare for difference in flow with a margin of error. If there is it would indicate a leak.
That's right! I completely forgot about that! Water in of itself is non-conductive. It's the minerals in the water that makes it conductive!
the sensor reads very low conductivity. (even the air as it read 0.1)
I'm guessing a differential pressure transducer would do the job even more easily that measuring differential flow with 2 sensors.
@@SamuelCarrier92 would agree in a closed loop in an open loop system the pressure (which is resistance against flow) will be almost none in the return line and probably instable in the feed line. I am not entirely sure how the system is built but from what I gathered it is open loop. The flow stays constant no matter the pressure in a liquid.
I think jake poured demineralised water over sensor and read 0.5V only, so it still reads something.
Couldn’t you put a solid metal “drain pan” or something between each thing on the rack?
So if one leaks it won’t rain down on the stuff below it?
yeah i was thinking of that too. good idea.
Might mess with air flow and heating.
@@whykuable Airflow? For Computers in a Water Cooled Rack?
@@whykuable Airflow in racks is usually from front to back
Yeah the tolerances between each unit of the rack probably isn’t enough, and using a whole 1u of space in between each server would not work
Linus and Jake, Zooz makes a Valve control unit that can shut the valve when a command is sent to them. You should check it out. Also, FYI about Z-Wave, it can only send a signal at most 4 hops, so be careful about that with his Z-Wave network.
God I love UA-cam comment experts
@LeafBoye I have had this happen to me, and discovered that fact when trying to solve my own problem with my Z-Wave network. You can look it up yourself if you want to, I am happy to be given newer, accurate info if I am wrong.
I thought Linus said he was going to just ban people who make useless comments?
I was working on a project for my college which involved flow sensors, leak sensors and relays. I know how to use them but this video really helped me connect some dots and it dropped on such a good time hehe
Sorry to hear about your sister! I was looking forward to her next upgrade.
this could have been a colab with electroboom :(
Jake missing the '/t 0' in the shutdown command weirdly upsets me Lol it was a great video though
Yeah me too! I guess Jake isn't much of a Windows person, mostly working with *nixes.
Not sure why you went with water cooling for your server considering a lot of the videos I've watched from you mentioned that air cooled is more reliable and accessible. If you really wanted to make use of your pool, you could have gone with water sourced heat pump and cooled the entire room. Or ducted it only to your server rack. Still would have made a great video.
You have to be mindful when grounding stuff to a painted rack. You need to make sure that you use grounding points that aren't impeded by the paint. Usually, there will be a dedicated copper threaded stud, but sometimes you have to grind off some paint and use some barbed washers.
I really like Jake's presenting style in this video. Knowledgeable, enthusiastic but not over the top. Thanks for being a cool guy Jake!
18:01 love the QR code for ltt store
"Wrap it before you tap it" SENT me 🤣🤣🤣
Should have called it "Addressing the huge LTT leak"
Another suggestion is to get a switched rack PDU (if you don't already have one). This would allow you to physically disconnect power to systems in case of a leak, which is obviously safer than just switching them off.
Oh you could totally use a poe esp32 with esphome for the leak sensors, I'm surprised this wasn't the first option! Much less to go wrong than a full rpi OS
Okay let's see if I understand this;
The reason that the corrosion happened is because the rack was not properly grounded which means the rack was using the water in the loop as a ground and since there was a electronic differential in the rack versus ground then it caused the electronic galvanic corrosion.
Correct? Or at least correct on a basic level?
@@SSYoung125 I mean that's correct to what they thought it was until they found out the faucet head to the resevoir tank was made out of cheap metal and not 100% brass like they thought and was causing the corrosion.
Oh and not from not being grounded persay but just not isolated so using the water to carry a very small electric differential.
Pretty much. I've literally touched fittings on cars for heater hoses that just crumble in my hand from a bad engine ground.
I would say wrong. I would have loved to have them measure the resistance between rack and the ground of the house for them to realized it was 0 ohm. With that amount of grounded box in a rack directly touching metal to metal and screwed in making a good ground connection. So rack was as grounde as it is after adding the grounds. unless there was something isolating the cases for example pc cases were all plastic and that the scree weren't metal the you could have a ungrounded rack ( and it wouldn't matter).
The ground not making a difference in my opinion I would agree with you that this video misses an electrical drawing or visual that explains what their theory is with regards to the voltage difference on the 2 parts of the system. how would on pump being fed at 120V and the pc being at 118V how would that affect corrosion? through induction? I'm not sure I understand their theory after that video.
@@SamuelCarrier92 Water does weird things when energized even the slightest. Having an inconsistent ground doesn't help. I personally think it partly has to do with a mixture of issues like the ground issue and voltage generated from moving water in plastic pipe and mixture of metal.
@@SamuelCarrier92 A rack standing on the ground does not mean it's grounded, tying it into the panel is a good move. As for the cases and the rack being the same potential, probably, though paint can be a bitch. But then again, it doesn't seem like they removed/scratched the paint for the grounding wires either.
3:29 Linus Sewing Tips
The color match between Linus's and Jake's hair right now is insanely suspicious. basically admitted it at 5:38. Linus is Jakes dad
I used to manage an irrigation and chemical spray store. We sold fittings that were used in chemical sprayers and would either be a brass alloy (brass with other resistant metals) or nylon. Both were great and extremely resistant to corrosion. Stainless steel is great as well.
This should come in Handy when everything freezes solid this winter. Jake’s a smart cookie. Always enjoy videos with him. For a totally different reason than why I enjoy Alex Videos.
You know what makes me mad when they say they're going to do it the right way in the future, but you know they're never going to do it until something else catastrophic happens
It's called "content".
In a pool of water stand a server rack. On its side, these words appear: My name is Linus, King of Tech; Look on my cooling, ye Nerdy, and despair!
Is there anyway you can 3D print or buy some acrylic sheeting and thin piping which can go between the computers and drain to the floor to catch any future leaks so they don’t wipe out multiple computers?
Oh and on the red handled gate valve, they are really crappy. I was draining an expansion vessel on a calorifier two weeks ago and I had one of these not catch the thread and refuse to open back up, which effectively turned the water off for half of a factory that produces bearings for nuclear reactors. I managed to open the valve by fully turning closed the valve as hard as I could and then opening it as softly as possible; the noise the system made as water flushed back through was tremendous. Replace the gate valves with ball valves. Bung up the tank (Google plumbing bung) so no more coolant can go down the pipes and you won’t need to drain the tanks.
Thats all so sick, what a time to be alive. Such good toys
As everyone else is chipping in, why not. Isolating the pump, sure why not. The issue here is that there's a voltage difference somewhere causing the ox/redox reaction to happen somewhere. AC won't cause this, but funky stuff can happen with oxides, as you can create diodes.
Another issue you have is that you are using a DC bus, and not a single star groud over your house. Yes it does connect at the pannel but you only need a couple mA on the ground run to induce a voltage difference across the groud lead. Moreover if one supple is leaking DC to the ground.
You'll also have capactive and inductive coulpling, loops everywhere and whatnot.
16:11 why did that get me 😂
Regarding the computer signing out and shutting down in less that a minute at 19:55; instead of 'shutdown /s', try 'shutdown -t 00 -s'
One zero is fine
Came here to say that. Otherwise it shuts down in 60s by default if my memory is correct
Shutdown /s /t 0
Should have the water looped racks vertical that way if one leaks it doesnt drench the racks below it.
Jake, a man of culture with the M539 apparel! If you know, you know
Hey water heaters have a sacrifice anode rod for the minerals and other stuff to eat then eating the metal boiler in a water heater. Maybe use a 30 gal UC (under counter) water anode rod in one of those plastic square tanks to reduce alot of that corrosion
8:45 Jake contaminates pump. 💀
Do not use multimeter to check that the voltage is zero. It can give you false reading. Use special tools like Fluke T110 voltage tester which has resistor to drain any residue voltages and gives proper reading.
7:00 Electricians looking at this in absolute horror!
If/when the house gets electrically safety tested; it will fail due what is shown here.
Yeah presumably it will be replaced by a real electrician in a few weeks.
Or it won't be and they're showing the most realistic form of home improvement: a jank solution that becomes permanent lmao
this is shown to trigger Electricians as they and rest of the professionals are almost as bad as fanboys in the comment section
and they will feed the youtube algorithm.
Don't forget the statement right after this... "its just a ground, it should be fine" , my boss would make a whip from ground wires and whip me out of consciousness if I ever say that.
Engineer here too(this seems to be a thing here) - this was the thinnest seperate grounding wire i've ever seen. Would have use at least 10 or 16mm² - convert this to your freedom units
I'd like to mention that you could also install anode rod's in the water tanks just like they do in water heaters to prevent corrosion in the tank lining from harsh water quality. I know the tanks you have installed are plastic, but it might help protect other things like metal fittings in the tanks.