Chiller development team member here. The compressor is not cycled in a laser application, they vary the cooling capacity by use of an electric expansion valve. It can be seen behind the compressor by the evaporator (which is not in the tank, it's attached to the back panel) Btw, SMC (a competitor to Thermo Scientific) offers rack mount chillers up to 5kW and with the 0.1 degree tolerance.
@@noah-wn8kh Fun fact, to save energy costs they sometimes run the compressor on an inverter too. Those are normally high end models for semiconductor manufacturing, or higher cooling capacity. But that's mainly for energy savings, not to vary the refrigerant flow.
This is absolutely wild. I wish they made this a 30-40 minute video with more explanation of the coolant temp effect on the CPU and GPU performance, and did some more serious CPU/GPU benchmarking with comparison graphs to show the improvement possible vs their normal water loop. Really they should just do a part 2.
I imagine longevity of a piece of equipment would be just as important as performance when looking at temps though - if we do get data on performance relative to coolant temp, it'd be important to temper it with information on CPU and GPU health.
I think this is exactly what Labs can provide. A more in depth look at what is happening with actual data. Waaaay beyond what the average user would want but well within what the curious among us would be interested in.
If you mix propylene glycol into the water you can run below 0 c. Also, I'm pretty sure somewhere in the manual it will say not to run below something like 14 c without adding antifreeze. If you were able to look into the reservoir you would see the evaporator coil is likely coated in ice at the conditions you're running.
The chiller likely won't go that low without modification to the controller. I have an aquarium chiller and it stops at 0C. I installed pot parallel with the thermistor to offset the reading to get it to run lower.
Next up - testing it on a completely delided 13900K. Also, this is an effective way to test the difference between thermal compounds. Since the CPU can thermal throttle with this on it, you can use the scores to see which is better. And can be used to test the most efficient water cooling blocks with the same logic.
@@Prophes0r You get that with or without the IHS but the IHS already introduces two more material interfaces that make you lose efficiency. Of course the issue with the die being too small for the wattage will remain but at least you remove two bottlenecks.
So glad this all came together! Turned out awesome!!! I'm happy you guys are getting some use out of my garage junk :) Edit: There's a link to this channel on the footer of my website, yes this is legit.
This fits broadly into my favourite subgenre of LTT videos - "Alex Builds Something Scuffed" but oddly this one didn't have much of Alex building something scuffed. Hopefully this is just the first of a series of increasingly scuffed projects starring this chiller.
It would be great but the camera/editing kind of sucks. They keep zooming in to show the temps or graphs but then cut away after like half a second so you have to go back and even see what they are talking about, like at 11:53
CAREFUL running just water in that cooler. If you set the coolant at 5ºc (I don't know what Freon is inside) the gas will be evaporating at -2ºc. You will destroy your evaporator with ice. P.S. the evaporator in the black box with 4 tubes in the bottom right of the machine, the plastic container is just a reservoir. I hope to help, thanks a lot for your content guys!
Sure, but even a traditional ac with a fairly "weak" refrigerant will freeze up if no house air is blowing over the evaporator, but with some heat from the computer traveling through the other side of the exchanger I would imagine you won't freeze up at that temp. This thing does have a bit much on the side of capacity though, so you may be right.
The black "box"with 4 tubes isn't a evaporator, it's a heat exchanger between the liquid in the chiller and the liquid in the attached system. That's why the tank is so small. it holds just enough to compensate for expansion inside the system.
This may very well be one of my fav LTT videos in a while. The genuine good time, funny left in moments and Linus actually liking a thing.. 10/10 love this stuff!
This is it, the mother of all chillers! Those 18C full load on that 4090 made me giggle like a child in a candy store. This thing is just stupid bonkers, we need a whole series with that boy!
Yeah... Until someone sets the building on fire or gets seriously hurt. im surprised he showed that cable in the video. does canada have an OSHA? bc im sure they wouldnt be happy lol
Год назад+46
It's not the first video with that cable and they are waiting for replacement that is safe(or so they said)
Not a lawyer, but I don't think that's really how it works. Businesses can have dangerous/sketchy things (so long as it isn't straight up illegal). If they were using this cable to say, power their server racks permanently, or power some other equipment deployed in the production environment, then CCOHS (Canadian OSHA) would probably be interested. But just for one off testing/R&D sort of work? Very few regulations exist, as they would fundamentally prohibit the design and building of "dangerous" things. For example, how would an appliance company design new microwaves if they aren't allowed to build untested prototypes involving high voltage transformers (extremely dangerous)?
This is cool because we use these chillers everywhere I work. You can use distilled water or glycol and they are pretty easy to maintain. That threaded brass rod on the back will adjust the pressure and make sure that bag filter is clean. On the other side of the compressor there should also be a solenoid on the copper pipes that lifts a magnet inside which allows flow.
That cooler we use for industrial applications haha! Linus, the Low T warning you can set in settings. Some lasers have a min temp around 5C for condensation issues. Also some of these coolers have a Modular speed pump. If not use a bypass (T connectors between the Inflow and Outflow with a valve so adjust flowrate and bypass the main circuit) on the circuit to lower the flow to the CPU so the water gets more time to take the heat away from the block. You can set your cooler really cold.. but if the water doesn't have time to collect the heat it is worthless.
I can’t wait for the video where they delid, liquid metal, and use Corsairs new direct die CPU water block on a 13900KS, for some insane cinebench and perhaps even gaming results with this chiller. Throw in some GPU over clocking, and that’s a sick video!
It would be really interesting to see the same tests done with the CPU de-lided. That would really put into context the impact of the IHS. Really cool video!
I've used similar Thermo chillers at work, it's cool to see inside one. The small coolant reservoir means it reaches operating temperature quickly when turned on.
Just a little tech tip guys. @2:55 I'm pretty sure evaporator is in the bottom right of the unit in the form of a plate type heat exchanger. The white box looks to just be a reservoir/buffer tank to limit short cycling to me. It also looks to be a fixed speed reciprocating compressor, with duty controlled by the little black mushroom shaped electronic expansion valve (by the heat exchanger). Still, looks like a nice bit of kit, with lots of potential for various applications!
As someone who works in the HVAC industry these videos are always my favorite. The mini split video in particular as I sell that same brand and when I showed my coworkers they were both amazed and horrified
I've made sketchier 3-phase cords coming from opposite ends of the building. If it works, it works. Just mind your connections and insulation. A loose wire at 240+ means fire.
This is why I love you guys. $15,000 water chiller for scientific and manufacturing lasers? Toss it on a PC loop. LMFAO. I love both your ideas btw. I think a cool project would be to watercool the entire lounge, but I def see the productivity capabilities Linus was talking about and I think long term it'll be a great tool for those reviews. Let us see more of Shackleton (famous artic explorer).
We need an hour-long documentary detailing the conception and life of the janky 208V power cord featuring LMG employee testimonials and expert opinions. Would 100% watch that.
I just finished building my first rig. I can’t tell you how helpful the channel as well as PC centric was in helping me get it all together. You guys give easy to understand info to people just getting into the PC world and it’s been a major help.
In case someone else hasn't already mentioned it; you can spool your teflon tape around a pencil first, and then use the pencil to wrap your fitting. Should be an easily searchable tip if you want to see a demonstration. It's mostly useful for hard to access threads like something that is permanently mounted in an awkward spot, but honestly it helps even with completely loose fittings like you had here.
Alex's grin as Linus walks into the frame is such a clear indication of how this video is going to go 😂 The way Alex antagonizes Linus all the time is hilarious. Thanks for the entertainment, LTT!
Alex's PC cooling videos are some of my favorites, the fact that were halfway through the video and linus is asking "what even is this video" says it all.
What if you de-lidded the 13900 and then made a custom lid with a water block built in? I bet you could get a much lower temperature with the heatsink being the water block
Honestly at this point thats the bottleneck, doesn't matter how cold the get the collant if they simply can't transfer the heat fast enough off the cpu directly. They've reached the capacity the materials themselves ability to transfer heat. They may actually with their fancy new metal 3d printer be able to come up with something that could work, but I suspect it can't print in materials that would be better mediums for heat transfer. I did many years ago see a fully submerged pc setup, but I don't know if you could build such a thing for this, and even then I don't know if you'd still need to de-lid the cpu and if even doing so would damage it or if it could survive the submersion in that state.
@@cgi2002 you realize the comment literally talks about getting better heat transfer by doing direct die cooling. what they tried in the video was with the factory heatspreader on the cpu. if they do direct die with liquid metal they should get significantly better heat transfer
@@Naturalhighz yes I know and I agreed, my thoughts was they'd need to make a custom fitting for that, but that comes with the complication that their metal 3d printer may not be able to use the materials ideal for it. You still end up with some solid metal between the water and the cpu, even if using a liquid metal medium between the cpu and the cooler itself, that solid metal replaces the lid as the hard cap (infact without seeing the materials used in the current cooler and the lid, it may already be the hard cap) . The only way around that would be running an entire liquid cooling setup full of the liquid metal and making a sealed enclosure around the cpu, but that comes with the seperate issue of ensuring your liquid metal is non-conductive of electricity, exceptionally conductive of heat, and not even slightly corrosive under high pressure, oh and cheap because your going to need several gallons of it.
I'm willing to bet the motherboard it what taps out first in a truly unlimited headroom scenario. Intel's latest CPUs can handle absolute tons of current, but the VRMs can only go so far.
I used to build these systems for a competitor to Thermo Scientific. The evaporator is a braze plate heat exchanger in the bottom right corner by the compressor. If they're claiming 0.1C accuracy, then they are probably using a hot gas bypass system that uses the compressor's hot gas discharge to balance the load and maintain pretty stable temperatures. The systems I worked on could maintain temp even with no heat load on the system. Pretty cool to see one of these get out into the wild.
Overclocking wouldnt be limited by cooling, just how good luck they had with how much can that chip handle. I cant wait to see them put this againts liquid nitrogen cooling and see if it can even compare to that, if it can, this might become the "stantard" for extreme overclocking
@@username8644 Yes i know all that, but the difference is that on both you could potentially overclock to point of it being just up to how well that chip can handle shit and this would after initial setups be much less work and trial and error to get to working, like instead of having to vaseline your components and have paper etc everywhere to deal with condensation, you can just smack it on and leave it. The difference would be nice to see, how much would it actually be
That chiller seems like a game changer. I wanted to see you push the whole set up even further. Do a youtube mini-series on things you're going to use the chiller for! It's a beast!
Awesome vid! A few notes from someone who works in a physics laboratory with these regularly: cooling a 5000W fiber laser doesn't mean it has 5000W of cooling power since unlike cpus, most of the power in a laser is emitted in light form, not in heat form. Just something to keep in mind if you try to push it to the limit. Also we have building-wide processed chilled water loops for equipment cooling with those exact yellow valves
A note from someone who uses those chillers. if the chiller specifies 5KW of capacity they can remove the equivalent of 5KW of heat from a system. It has nothing to do with a laser. Linus just happens to have gotten only the laser power figure for reference.
Well it depends if the cooler is claiming that it is for cooling a 5000 W laser or if it is claiming that is has a 5000 W cooling capacity, if it’s the latter then yes it able to handle up to 5000 W of heat energy
Yup-and I like the fact that they constantly demonstrate that nerds are 100% proper men. "DEAR GOD THIS SMELLS HORRIBLE. Come smell this!" "OH DEAR GOD, YOU ARE RIGHT, THIS SMELLS HORRIBLE. You want another wiff?" "Of course I do-DEAR GOD!"
@@Prophes0r Direct die mount of the block would eliminate the thermal impedance of the IHS, so it would help - every W/mK counts with power densities like this. You absolutely don't need to go all the way down to direct die liquid contact, although TSMC have been working on metal layer deposition technology for that exact trick. No need for actual high pressure water though - turns out you can do it with only slightly higher pressure than typical existing pumps.
@@smokyz_ For that to work they'd need to spread the actual heat-producing circuit elements on the die out further, to reduce thermal power density. The problem there is that signal propagation times and structure locality need to be as short as possible for performance reasons, particularly with things like cache being close to the cores.
You could use a heat exchanger between the cooler and a tank and pump that water. This is common in our industry. It's actually amazing to me you haven't already done it.
Maybe next time try a glycol pump from a kegarator or something like that ? You could be easily cold enough to be condensating ice around the pipes if you wanted... This would definitely work.. and you can find them pretty cheap
In the lab we use two of these to keep the -80C freezers cold. One takes the temp down to -30 and then the other takes it down to -80, two phase changers required to reach the final temp. Absolutely insane tech that is yet so simple.
I absolutely love the Alex written videos! You can tell he’s gotten a lot more professional but I love the sketchy chaos vibe he brings to the videos! 😂
Its pretty neat seeing that chiller outside of where i work. We have ran many of them for years for things. The valve on the back is a bypass valve to control pressure. Be sure to press the arrow down button and tune it. They can reach in excess of 85psi water pressure.
I could see this thing actually being useful for the lab. Absolutely normalizing the fluid temps to test the maximum-balls-to-the-walls settings on devices is interesting.
Great video. I'm so glad you guys have this chiller now! With all the stuff Linus has bought recently, I can't believe it has taken this long to get a chiller like this (and it was only because of "John"). Those "gas" valves are labeled on the side (7:07). 600 WOG. That stands for 600 PSI rated water, oil, or gas. You can use that valve for water, oil, or gas. It should be noted that it may not be lead free, so while WOG means you can use it on water, you may not want to use it on a pipe that someone will be drinking or cooking from. All this is true in the USA, so I imagine it's mostly true up North a little ways as well. Also, I love the "cause that to go erect" with the high pitch engineering laugh shortly after. "You know what's fantastic Linus? We cannot kill ourselves with this, mostly" is a close second.
I have to make my insane gratidute for everything LMG does known. I mean I couldn‘t care less about some fancy cooler. But somehow the incredible video team and Linus manage to produce videos that are sooo fun to watch and have such good production quality. I‘m so thankful everytime we get a little look behind the scenes it baffles me time and again too see the amount of effort poured into making these videos! And it‘s free to watch 🤯 - btw I got two water bottles and they are the best I ever had.
There is all in one 4 way units that is air to water. I have one of those to my house. It can give 16kW maximum. It uses 32mm (1 and 1/4) tubing. The guy I bought it from told me to not cheep out on pipes. When I powered it on I understand why. Dam that water flows fast.
you should put a small plate frame heat exchanger on the loop for the computer and run glycol through it. It'll allow you to chill down to a lower temp without running the risk of having freeze up
My mother-in-law still has a pc with the first cooler you showed in this video. I had to change it to a bigger cooler and fan, because the original started making noise. Now it makes more noise, but more consistent :D
Episode idea: try different materials for your water block to increase thermal conductivity. Copper is at 4.0 W/cm K but silver is at 4.3 W/cm K. Make one expensive silver water block and see if you can draw more heat off a processor.
Great idea, silver is expensive but cheap enough to justify that experiment. How much do you think you would realistically need? I've never seen anyone machine silver.
@@lemagreengreen silver is about 100x the cost of copper so clearly expensive, but within the realm of possibility for the LTT organization. Plus it's not like it would go to waste - just remelt it and sell it. You wouldn't want to machine the silver, make a mold and pour it. Maybe grind a few thousandths of an inch off the mating surfaces for a better surface finish and thus better thermal transfer. Get this idea to Linus because I think it might be a cool project.
I suspect the better approach is no-material. Copper conductivity might be low, but grinding down to almost nothing is even lower. Silver is so soft that I would guess you would need more than 10% thicker.
This is a big improvement over the way we cooled the laser at my first job, which was to hook up the water inlet to a utility sink in the next room, and the have the outlet just drain out a hose into that same sink. Every morning I had to turn on the faucet enough to cool the laser, while checking for a little bit to make sure the sink wasn't filling faster than it could drain. At the end of the day I had to turn off the laser, and do the process in reverse. Not exactly a gold star for conservation.
I use something extremely similar to this to cool my ICP-MS and they're quite fun to work with. I've often had the same thought about getting one and using it to cool an entire home gaming pc rack. You should look at other industrial equipment, there are almost certainly loads of nerds willing to help you out wherever technical equipment like that is made. Try getting a nitrogen generator or a zero air generator and using it to displace all of the air in a case for a subambient loop. No water vapor, no condensation.
That beast just blowing the air out of the loop is so satisfying. I don't know how much time I've spent bleeding watercooling loops of air and this monster just does it in a second.
There are a couple units like this at my job 2 that are larger at 7000 and 10000w. They getting ready to replace the 7k one (standard life cycle replacement t still works perfectly) and I’m terrified to ask how much they want for it because if it’s not insane I might do this 😂. My computer room is next to my workshop so I could run the hosing through the wall
i build a device like this for my university graduation project... the goal was to have a lot of cooling but stable temp at the same time... back then invertor tech was not a thing so we used heater to stabilize the temperature... nowadays you only need an invertor compressor and some smart controller. an improvement can be a variable speed pump or maybe a heat exchanger with a control valve.
The chillers used in the tool for the lens, stage, and many parts that need stable temperature control for sub micron registration, uses chillers with much tighter temperature control by an order of magnitude.The temperature controller uses fuzzy logic to maintain the temperature within 0.01 degrees C. Among other rebuild tasks, I rebuild those as when they run 24/7 the pumps, valves, controls, and compressors wear out.
If you want to go complete over the top, try to get your hands on a Huber Unistat. They go down to -120 °C (depending on the model) and have an accuracy of 0.01 °C
That's a giant pain in the ass for PC computing bc you'd have to figure out a dew point detection system that regulates the output of the cooler, otherwise you risk shorting the entire PC due to condensation
@@mashakos1 There are already proved and tested ways to insulate a computer on a testbench and protect it for a few hours at a time. Of course you can't use water or water with additives like ethylene glycol. But you can use ethanol or some silicone oil
You guys should do more scuffed/experimenty videos like this. They’re my favourite on this channel. Doesn’t need “The god of computer coolers” either. Like those custom case builds. SO cool.
I’m an hvac technician. I’d like to make a suggestion. Build a chamber with gaskets out of Plexiglass. Add service access ports and pressurize it it pure nitrogen to test for leaks. Put a pure helium and maybe nitrogen mixture in it. The helium as with hard drives creates less drag. Then you can should be able to maintain a moisture less environment. Then you can add a cpu cooling evaporator and a evap with variable speed fans. You will be able to cool every thing in a pure environment. That and didn’t change your txv out to one that can handle sub zero temps? Real cryogenic freezers use a different refrigerant for process cooling. Or you can do a geothermal geothermal heat pump to provide 50°F water. Either that or get a mini chiller and go with hydronic cooling Also it depends on your sensor location, typically most people move their txv sensing bulb to areas where the most ice builds .
The peak draw is huge, but unless you are sinking 5000w of heat into it, the long term draw is much, much smaller, and mostly from the big coolant pump.
Eyyyyyy! You guys finally got just an actual chiller! Good! Now you don't have to have a mini split that gets hamstrung by trying to run it through a tiny tube. ... Can't wait to see how you mangle it. Edit: that GPU is running at similar clocks to my CPU.
I hope they don't. For all the fun you can have taking something apart and messing with it, LTT is ultimately a business and their LTT Labs wants to get work done with working equipment, not waste time by potentially breaking a piece of equipment for the LOLs.
It’s also worth noting that because the coolant was sub ambient, the radiators in the computer would have actually been adding heat to the coolant, since all they do is help hot areas (in this case, the air) dissipate their heat into cold areas (in this case the coolant). Not that it makes much difference, since the temperature delta would have been relatively small, and the cooler can definitely handle the extra load.
The sad reality is that there are currently no hot-swappable server racks available in the market. However, it is possible to design a server rack with hot-swappable components such as CPU, GPU, PS, RAM, and storage, each with its own built-in cooling block. The cooling lines could be designed to automatically connect and disconnect without leaking, making it easy to replace components without disrupting the entire system. A control system on the main board could be used to connect up to 25 storage drives, 2 RAM drives, 3 CPU drives, and 1 GPU drive, allowing for flexible configurations. Since each drive could be connected via an optics connection, the data transfer would be direct and fast. Creating RAM on a separate board could allow for more RAM to be installed in a drive. The server racks could be designed to be either rack-mountable or blade-type, with each segment being about 6 inches tall, 1.75 inches wide, and 2 feet long. This would allow for all the tubing and electrical connections to be placed in the rear. A full rack could have up to 8 sections with 10 units in each, providing a lot of flexibility for different configurations. While designing a hot-swappable server rack with these features would be a complex and expensive undertaking, the benefits in terms of flexibility, ease of maintenance, and scalability could be significant.
I’ve set up tons of these TF-5Ks out in the field and the cooling overshoot from 5 degrees down to 3 is from the PIDs needing tuning. Since these units are so versatile, they have PID tuning parameters in the menus so you can tuning the compressor duty cycle to whatever heat load you have. The menus make no sense btw, I always had to have the manual handy to decipher them. 😂
Linus, I run a 13900K on a Corsair H100i Elite. I hit 82C at full load running Prime for an hour. I don't know what the F guys are doing to the chips that you supposedly find them IMPOSSIBLE to cool....
Awesome video. I actually run a Thermo5000 as the closed loop source for an SMC HRS in place of facility water and I've wanted to run one of the Thermo2500s which wouldn't cut it on the mill in a PC loop so badly...unfortunately company equipment and all that. Thank you for letting me realize that dream through you. Probably not so great that you put it in my head that's its actually feasible and a good(?) idea....
![Gunna - HIM ALL ALONG [Official Video]](http://i.ytimg.com/vi/w7Wf0h1Q8a4/mqdefault.jpg)
As a Thermo Fisher Scientific Employee, its awesome to see our products being used by you guys :)
Hot damn! Do i see a collab incoming?
Using this for testing waterblocks and the like seemed like a great idea!
Now make one with RGB /s
Seriously, if you don't manage to get some nice advertising out of this I'll have to question your managements' competence :D.
@@RetiredRhetoricalWarhorse Thermo Fisher is a big name, just not in these parts :)
Send them some -80°C Ultrafreezer from Thermo and let's see how CPU's behave in there.
Linus is probably gonna run an entire industrial complex just to cool a computer one day
Isn’t that quantum computing?
Vid should be out next week IIRC.
liquid cooling with a liquid nitrogen loop
You can *BET* on it
Liquid oxygen system used for MRI magnet purge?
Chiller development team member here. The compressor is not cycled in a laser application, they vary the cooling capacity by use of an electric expansion valve. It can be seen behind the compressor by the evaporator (which is not in the tank, it's attached to the back panel)
Btw, SMC (a competitor to Thermo Scientific) offers rack mount chillers up to 5kW and with the 0.1 degree tolerance.
Those things are beasts, being quite impressive. Last time I saw one was in use to chill an ICP, maybe ICP-MS. I can't remember the type of analyzer.
There’s a ton of chillers that have more capacity than this, especially if you run engineered fluid
@@jamesladeau419 Definitely, and ones that could go sub-zero. I would love to see Linus do a submerged build with galden.
finally someone understands IT"S NOT INVERTER it has an EEv for varying the amount
@@noah-wn8kh Fun fact, to save energy costs they sometimes run the compressor on an inverter too. Those are normally high end models for semiconductor manufacturing, or higher cooling capacity. But that's mainly for energy savings, not to vary the refrigerant flow.
This is absolutely wild. I wish they made this a 30-40 minute video with more explanation of the coolant temp effect on the CPU and GPU performance, and did some more serious CPU/GPU benchmarking with comparison graphs to show the improvement possible vs their normal water loop. Really they should just do a part 2.
...and you've just subscribed at floatplane.
I imagine longevity of a piece of equipment would be just as important as performance when looking at temps though - if we do get data on performance relative to coolant temp, it'd be important to temper it with information on CPU and GPU health.
I think this is exactly what Labs can provide. A more in depth look at what is happening with actual data. Waaaay beyond what the average user would want but well within what the curious among us would be interested in.
@@elbeetlebeasto do you mean de-lidding it and bolting on a custom heat transfer plate? fuckin wild.
If you mix propylene glycol into the water you can run below 0 c. Also, I'm pretty sure somewhere in the manual it will say not to run below something like 14 c without adding antifreeze. If you were able to look into the reservoir you would see the evaporator coil is likely coated in ice at the conditions you're running.
They should definitely add inhibited glycol inside the cooling loop (>25%). Ethylene glycol should be fine too and has better thermals
@@FavoritesAG Ethylene Glycol is more toxic though in comparison to Propylene Glycol.
The chiller likely won't go that low without modification to the controller. I have an aquarium chiller and it stops at 0C. I installed pot parallel with the thermistor to offset the reading to get it to run lower.
@@Aaron48219 Only when ingested, or are you a thirsty boy who likes to drink coolants?
@@Aaron48219 mOrE ToXiC. Grow a pair
I love how the first thing Alex wants to do with it is basically the whole room water cooling project again, lmao.
that was my first thought. WRWC done right, and not pissing into a bathtub
i might acualy work with that cooler
That was a fun series, really hope they choose to revisit it now that they have this thing.
Yeah. Linus looks like Edward Witten next to Alex.
Its kind of Backwards though this would be a Room Full of PC instead of a Whole Room full of Cooling
Next up - testing it on a completely delided 13900K.
Also, this is an effective way to test the difference between thermal compounds. Since the CPU can thermal throttle with this on it, you can use the scores to see which is better.
And can be used to test the most efficient water cooling blocks with the same logic.
Would love for them to do a direct die cooling solution
Yes. I would love to see this
YEEEEEEEEEEEEEEESSSS I WAS THINKING THIS THE WHOLE TIME
@@Prophes0r You get that with or without the IHS but the IHS already introduces two more material interfaces that make you lose efficiency. Of course the issue with the die being too small for the wattage will remain but at least you remove two bottlenecks.
Yeah great idea
So glad this all came together! Turned out awesome!!! I'm happy you guys are getting some use out of my garage junk :)
Edit: There's a link to this channel on the footer of my website, yes this is legit.
Hero
Thanks for making this video happen!
Not sure if this is the real thing, your channel was created today?
Thank you for your help john! Awesome content from them!
Thank you for your help john! Awesome content from them!
This fits broadly into my favourite subgenre of LTT videos - "Alex Builds Something Scuffed" but oddly this one didn't have much of Alex building something scuffed. Hopefully this is just the first of a series of increasingly scuffed projects starring this chiller.
Also my favorite genre
It would be great but the camera/editing kind of sucks. They keep zooming in to show the temps or graphs but then cut away after like half a second so you have to go back and even see what they are talking about, like at 11:53
@@rubiconnn Thats a soild 10 seconds, more then enough time to read it, or atleast pause the video
Did you imagine it would be a bed? LOL
CAREFUL running just water in that cooler. If you set the coolant at 5ºc (I don't know what Freon is inside) the gas will be evaporating at -2ºc. You will destroy your evaporator with ice.
P.S. the evaporator in the black box with 4 tubes in the bottom right of the machine, the plastic container is just a reservoir.
I hope to help, thanks a lot for your content guys!
Sure, but even a traditional ac with a fairly "weak" refrigerant will freeze up if no house air is blowing over the evaporator, but with some heat from the computer traveling through the other side of the exchanger I would imagine you won't freeze up at that temp.
This thing does have a bit much on the side of capacity though, so you may be right.
take a shower
@LinusTechTips pls read this don't damage your cooler
I'm not sûre about what you are saying, but it's definitely not freon, that shit's illegal
The black "box"with 4 tubes isn't a evaporator, it's a heat exchanger between the liquid in the chiller and the liquid in the attached system. That's why the tank is so small. it holds just enough to compensate for expansion inside the system.
This may very well be one of my fav LTT videos in a while. The genuine good time, funny left in moments and Linus actually liking a thing.. 10/10 love this stuff!
same same same!! shockingly 10/10 perfect result
I love watching enthusiasts being, well, enthusiasts. We’re all nerds drag racing computers at the end of the day
@@upforellie skkkkkrrttttttt *insert engine noises*
The fact that the entire intro is Alex giving us a shit-eating grin that he _cannot contain_ is how you know this video is gonna be good.
This is it, the mother of all chillers! Those 18C full load on that 4090 made me giggle like a child in a candy store. This thing is just stupid bonkers, we need a whole series with that boy!
'Will it Chill?'
@@Outland9000 Good series idea. xD
Yeeessssss
R u a furry
@@Tiger11246 Wrong comment?
For engineers linus must be a dream employer, the amount of "i don't care how dangerous it is, if it works i'm fine with it" is bonkers.
Yeah... Until someone sets the building on fire or gets seriously hurt. im surprised he showed that cable in the video. does canada have an OSHA? bc im sure they wouldnt be happy lol
It's not the first video with that cable and they are waiting for replacement that is safe(or so they said)
@@tbag6600 party pooper
Not a lawyer, but I don't think that's really how it works. Businesses can have dangerous/sketchy things (so long as it isn't straight up illegal). If they were using this cable to say, power their server racks permanently, or power some other equipment deployed in the production environment, then CCOHS (Canadian OSHA) would probably be interested. But just for one off testing/R&D sort of work? Very few regulations exist, as they would fundamentally prohibit the design and building of "dangerous" things. For example, how would an appliance company design new microwaves if they aren't allowed to build untested prototypes involving high voltage transformers (extremely dangerous)?
@@_trbr Yea one needs to do scatchy things wen developing or testing new shit.
Regulations come in wen u wanne ship that shit to the public.
This is cool because we use these chillers everywhere I work. You can use distilled water or glycol and they are pretty easy to maintain. That threaded brass rod on the back will adjust the pressure and make sure that bag filter is clean. On the other side of the compressor there should also be a solenoid on the copper pipes that lifts a magnet inside which allows flow.
Glad to hear they are easy to use!
I work for Thermo Fisher
Distilled water is not the best, can freeze and destroy everything, I have seen coolers thrown away like that
That cooler we use for industrial applications haha!
Linus, the Low T warning you can set in settings. Some lasers have a min temp around 5C for condensation issues.
Also some of these coolers have a Modular speed pump.
If not use a bypass (T connectors between the Inflow and Outflow with a valve so adjust flowrate and bypass the main circuit) on the circuit to lower the flow to the CPU so the water gets more time to take the heat away from the block. You can set your cooler really cold.. but if the water doesn't have time to collect the heat it is worthless.
You can see Alex just love his job as a engineer doing fun things to linus.
Horny engineer
Is Alex only engineer on the team?
@@mmert138 No, according to the linus media group site they have at least five engineers
@@mmert138 apart from labs he might be
@@mmert138 He's the only mad scientist. Don't know if that counts, but it should. :)
How Alex has managed to survive this long is truly impressive.
Canadians are built different. Just look at ElectroBOOM.
Its called being a narcissistic psycho who is a kiss ass towards your boss without actually showing it.
I really hope he never goes away because he's my favorite LTT employee. If Alex is in the video you know it's going to be entertaining!
I can’t wait for the video where they delid, liquid metal, and use Corsairs new direct die CPU water block on a 13900KS, for some insane cinebench and perhaps even gaming results with this chiller. Throw in some GPU over clocking, and that’s a sick video!
They did it
They did, but with EK x Der8auer Direct Die block, and they used a regular 13900k instead of a KS
It would be really interesting to see the same tests done with the CPU de-lided. That would really put into context the impact of the IHS. Really cool video!
Yeeees! De-liddddd
Someone call der8auer right now!!!
Linus is addicted to insane cooling tech😂
Him and Alex both 💀
But then they usually bodge it just for entertainment purposes instead of giving us actual valuable data
intel is making ovens and linus provide the juices !
annoyed they didn't do direct die cooling
Alex is and ropes Linus in who is also enjoying it.
I've used similar Thermo chillers at work, it's cool to see inside one. The small coolant reservoir means it reaches operating temperature quickly when turned on.
You can see the excitement on their faces and I'm all for it
Thats why we are here 😁💪
I love when Alex is allowed to be as sketchy as he wants to be
If Linus doesn't stop him from using that power cord.. I'd say he's got pretty free rein to do what he wants. :)
You can see it in the smile on his face! Lol Never get to see a happier Alex.
@@Yuriel1981 truly a mad scientist
Just a little tech tip guys. @2:55 I'm pretty sure evaporator is in the bottom right of the unit in the form of a plate type heat exchanger. The white box looks to just be a reservoir/buffer tank to limit short cycling to me.
It also looks to be a fixed speed reciprocating compressor, with duty controlled by the little black mushroom shaped electronic expansion valve (by the heat exchanger).
Still, looks like a nice bit of kit, with lots of potential for various applications!
As someone who works in the HVAC industry these videos are always my favorite. The mini split video in particular as I sell that same brand and when I showed my coworkers they were both amazed and horrified
Ahh yes, Alex’s sketchy cord. My favorite reoccurring guest in this series returns.
I've made sketchier 3-phase cords coming from opposite ends of the building. If it works, it works. Just mind your connections and insulation. A loose wire at 240+ means fire.
Just combine your sketchy cord with the comfy chair and you have a deadly combo going!
@@nobodynoone2500 a loose wire at 240V+ means a blown fuse.
Just the same as 120V.
I was just gonna say that cord is part of the team now
This is why I love you guys. $15,000 water chiller for scientific and manufacturing lasers? Toss it on a PC loop. LMFAO.
I love both your ideas btw. I think a cool project would be to watercool the entire lounge, but I def see the productivity capabilities Linus was talking about and I think long term it'll be a great tool for those reviews. Let us see more of Shackleton (famous artic explorer).
That thing costs 15k? wow I think Im making chillers now. The parts of that machine probably cost like $1500-$2000.
We need an hour-long documentary detailing the conception and life of the janky 208V power cord featuring LMG employee testimonials and expert opinions. Would 100% watch that.
It's a 208V power cord. Not sure where you're getting 240 watts from.
@@TehButterflyEffect thanks for catching that, not really sure why it popped into my head either.
it's a suicide cord lol. Not cool
@Blake Belladonna I was wondering too. the previous owner ran this somehow, so there has to be a proper cable.
You can get a C13 connector from Digi-Key (part number 486-1107-ND) and wire that onto the end of the 208V cord.
You guys need to do this with direct die cooling, would love to see if you can get it cooled that way!
That is exacly what thought, take out the IHS and just pressure mount the tubing to the dye. which would be hard to engineer but possible...
AND LIQUID METAL
Excellent ideas! 👍
I just finished building my first rig. I can’t tell you how helpful the channel as well as PC centric was in helping me get it all together. You guys give easy to understand info to people just getting into the PC world and it’s been a major help.
In case someone else hasn't already mentioned it; you can spool your teflon tape around a pencil first, and then use the pencil to wrap your fitting. Should be an easily searchable tip if you want to see a demonstration. It's mostly useful for hard to access threads like something that is permanently mounted in an awkward spot, but honestly it helps even with completely loose fittings like you had here.
Alex's grin as Linus walks into the frame is such a clear indication of how this video is going to go 😂
The way Alex antagonizes Linus all the time is hilarious.
Thanks for the entertainment, LTT!
alex's constant snickering in this video reminds me of muttley from wacky races
Alex's PC cooling videos are some of my favorites, the fact that were halfway through the video and linus is asking "what even is this video" says it all.
What if you de-lidded the 13900 and then made a custom lid with a water block built in? I bet you could get a much lower temperature with the heatsink being the water block
Honestly at this point thats the bottleneck, doesn't matter how cold the get the collant if they simply can't transfer the heat fast enough off the cpu directly. They've reached the capacity the materials themselves ability to transfer heat.
They may actually with their fancy new metal 3d printer be able to come up with something that could work, but I suspect it can't print in materials that would be better mediums for heat transfer. I did many years ago see a fully submerged pc setup, but I don't know if you could build such a thing for this, and even then I don't know if you'd still need to de-lid the cpu and if even doing so would damage it or if it could survive the submersion in that state.
Der Bauer makes a direct die frame and delidding tool...
@@cgi2002 you realize the comment literally talks about getting better heat transfer by doing direct die cooling. what they tried in the video was with the factory heatspreader on the cpu. if they do direct die with liquid metal they should get significantly better heat transfer
@@Naturalhighz yes I know and I agreed, my thoughts was they'd need to make a custom fitting for that, but that comes with the complication that their metal 3d printer may not be able to use the materials ideal for it. You still end up with some solid metal between the water and the cpu, even if using a liquid metal medium between the cpu and the cooler itself, that solid metal replaces the lid as the hard cap (infact without seeing the materials used in the current cooler and the lid, it may already be the hard cap) . The only way around that would be running an entire liquid cooling setup full of the liquid metal and making a sealed enclosure around the cpu, but that comes with the seperate issue of ensuring your liquid metal is non-conductive of electricity, exceptionally conductive of heat, and not even slightly corrosive under high pressure, oh and cheap because your going to need several gallons of it.
@cgi2002 It could also be cnc'd though, as metal 3D printing is only necessary as a convenience
I am kinda shocked they didn't delid the CPU for this one. I really want to see how far you could push this thing.
I'm willing to bet the motherboard it what taps out first in a truly unlimited headroom scenario. Intel's latest CPUs can handle absolute tons of current, but the VRMs can only go so far.
yep, direct die cooling needed here.
@@DigitalJedi Yes! Fry the VRMs! That would be a hell of a video.
They went as far as cooling it with an industrial chiller, why not delid, I’d really really be interested to see thermals in that scenario.
@@DigitalJedi your probably right. New title “How far can we push this motherboard”.
I used to build these systems for a competitor to Thermo Scientific. The evaporator is a braze plate heat exchanger in the bottom right corner by the compressor. If they're claiming 0.1C accuracy, then they are probably using a hot gas bypass system that uses the compressor's hot gas discharge to balance the load and maintain pretty stable temperatures. The systems I worked on could maintain temp even with no heat load on the system.
Pretty cool to see one of these get out into the wild.
I'd really like to see an update with direct die cooling. That must be insane😁
Was a bit disappointed they didn't even bring it up
Overclocking wouldnt be limited by cooling, just how good luck they had with how much can that chip handle. I cant wait to see them put this againts liquid nitrogen cooling and see if it can even compare to that, if it can, this might become the "stantard" for extreme overclocking
@@username8644 Yes i know all that, but the difference is that on both you could potentially overclock to point of it being just up to how well that chip can handle shit and this would after initial setups be much less work and trial and error to get to working, like instead of having to vaseline your components and have paper etc everywhere to deal with condensation, you can just smack it on and leave it. The difference would be nice to see, how much would it actually be
@@onebacon_ Yup, at the beginning of the vid, I hoped so, too...
0:20: There is nothing like Alex's smile after a "most people think it's stupid" to know it's going to be a GREAT video.
That chiller seems like a game changer. I wanted to see you push the whole set up even further. Do a youtube mini-series on things you're going to use the chiller for! It's a beast!
Awesome vid! A few notes from someone who works in a physics laboratory with these regularly: cooling a 5000W fiber laser doesn't mean it has 5000W of cooling power since unlike cpus, most of the power in a laser is emitted in light form, not in heat form. Just something to keep in mind if you try to push it to the limit.
Also we have building-wide processed chilled water loops for equipment cooling with those exact yellow valves
A note from someone who uses those chillers. if the chiller specifies 5KW of capacity they can remove the equivalent of 5KW of heat from a system. It has nothing to do with a laser. Linus just happens to have gotten only the laser power figure for reference.
Unless laser efficiency has improved since I last looked but aren't most laser less than 1% or effecient?
Well it depends if the cooler is claiming that it is for cooling a 5000 W laser or if it is claiming that is has a 5000 W cooling capacity, if it’s the latter then yes it able to handle up to 5000 W of heat energy
Judging by the size of that compressor its extremely likely that the cooler has in fact 5000W of cooling power and must have about 20% of head room.
I would actually love to see Alex cool the lounge PCs with that chiller. That thought is so exciting.
Whole room water cooling v2!!!
@@dylanandersen9318 whole room water cooling 2, electric boogaloo
@@astro143_ it even rhymes!
Whenever you see this duo in a video, you know the project is going to be good!
Yup-and I like the fact that they constantly demonstrate that nerds are 100% proper men. "DEAR GOD THIS SMELLS HORRIBLE. Come smell this!" "OH DEAR GOD, YOU ARE RIGHT, THIS SMELLS HORRIBLE. You want another wiff?" "Of course I do-DEAR GOD!"
I'd love to see you try this with direct-die on the 13900K.
@@Prophes0r Direct die mount of the block would eliminate the thermal impedance of the IHS, so it would help - every W/mK counts with power densities like this. You absolutely don't need to go all the way down to direct die liquid contact, although TSMC have been working on metal layer deposition technology for that exact trick. No need for actual high pressure water though - turns out you can do it with only slightly higher pressure than typical existing pumps.
@@Prophes0r Why can't they just make silicons bigger to make the surface area bigger?
@@smokyz_ For that to work they'd need to spread the actual heat-producing circuit elements on the die out further, to reduce thermal power density. The problem there is that signal propagation times and structure locality need to be as short as possible for performance reasons, particularly with things like cache being close to the cores.
you can always tell how much a wild video idea costs by how big Alex's smile when he appears
You could use a heat exchanger between the cooler and a tank and pump that water. This is common in our industry. It's actually amazing to me you haven't already done it.
This is NUTS! We need a whole series with this cooler!
Maybe next time try a glycol pump from a kegarator or something like that ? You could be easily cold enough to be condensating ice around the pipes if you wanted... This would definitely work.. and you can find them pretty cheap
Everyone reply to this so we get their attention so they look into this experiment
The only downside would be frost, I guess? But you could absolutely pick up a glycol cooler being sold off by a bar or beer company.
@@gsuberland I have always wondered for years if you can prevent frost by removing all humidity
@@remingtonrojas Technically, but you'd need a perfectly sealed chamber.
@@remingtonrojas you can in theory but that's pretty hard to do
In the lab we use two of these to keep the -80C freezers cold. One takes the temp down to -30 and then the other takes it down to -80, two phase changers required to reach the final temp. Absolutely insane tech that is yet so simple.
Every Linus and Alex video in the workshop is a must-watch for me.
I absolutely love the Alex written videos! You can tell he’s gotten a lot more professional but I love the sketchy chaos vibe he brings to the videos! 😂
Amazing!! You should definitely find a cooler that allow the liquid to go to negative temperatures and use car coolant. It would be really cool to see
condensation would fry it
Its pretty neat seeing that chiller outside of where i work. We have ran many of them for years for things. The valve on the back is a bypass valve to control pressure. Be sure to press the arrow down button and tune it. They can reach in excess of 85psi water pressure.
Like many have said already direct die cooling with this would be a pretty cool follow up video. Maybe with testing different TIMs too
0:19 The way camera started pointing towards Alex after Linus's introduction in a cooling video - is absolutely epic
I could see this thing actually being useful for the lab. Absolutely normalizing the fluid temps to test the maximum-balls-to-the-walls settings on devices is interesting.
Great video. I'm so glad you guys have this chiller now! With all the stuff Linus has bought recently, I can't believe it has taken this long to get a chiller like this (and it was only because of "John").
Those "gas" valves are labeled on the side (7:07). 600 WOG. That stands for 600 PSI rated water, oil, or gas. You can use that valve for water, oil, or gas. It should be noted that it may not be lead free, so while WOG means you can use it on water, you may not want to use it on a pipe that someone will be drinking or cooking from. All this is true in the USA, so I imagine it's mostly true up North a little ways as well.
Also, I love the "cause that to go erect" with the high pitch engineering laugh shortly after. "You know what's fantastic Linus? We cannot kill ourselves with this, mostly" is a close second.
How rapidly Linus unzipped the temperature display sticker without emotions broked my heart 💔
I have to make my insane gratidute for everything LMG does known. I mean I couldn‘t care less about some fancy cooler. But somehow the incredible video team and Linus manage to produce videos that are sooo fun to watch and have such good production quality. I‘m so thankful everytime we get a little look behind the scenes it baffles me time and again too see the amount of effort poured into making these videos! And it‘s free to watch 🤯 - btw I got two water bottles and they are the best I ever had.
Jesus. Calm down dude. This was mildly entertaining and moderately interesting at best
@@hjvh78ho "standard"
@@hjvh78ho Isnt he the one who gave the chiller?
@@cesargonzalez2011 nope
@@ExTr3Me_Cobra pretty standard
I'm just waiting for the day where he cools a computer with the HVAC system of an entire office
There is all in one 4 way units that is air to water. I have one of those to my house. It can give 16kW maximum.
It uses 32mm (1 and 1/4) tubing. The guy I bought it from told me to not cheep out on pipes. When I powered it on I understand why. Dam that water flows fast.
@@matsv201 Pipe in a few quick disconnect loops near your desk... you know... for science and/or/in engineering.
already done
It's not going to matter when you're heat transfer bottle neck is still going to be the water cooling blocks
Tech Ingredients did this and it was extremely effective. More so than a water chiller they compared it to.
you should put a small plate frame heat exchanger on the loop for the computer and run glycol through it. It'll allow you to chill down to a lower temp without running the risk of having freeze up
I can't wait to see this in a direct die cooling loop!
My mother-in-law still has a pc with the first cooler you showed in this video. I had to change it to a bigger cooler and fan, because the original started making noise. Now it makes more noise, but more consistent :D
6:51
If the valve says 'WOG600,' it means Water, Oil, and Gas at 600 psi, even if it's color-coated for gas. So it's okay :)
Episode idea: try different materials for your water block to increase thermal conductivity. Copper is at 4.0 W/cm K but silver is at 4.3 W/cm K. Make one expensive silver water block and see if you can draw more heat off a processor.
Great idea, silver is expensive but cheap enough to justify that experiment. How much do you think you would realistically need? I've never seen anyone machine silver.
@@lemagreengreen silver is about 100x the cost of copper so clearly expensive, but within the realm of possibility for the LTT organization. Plus it's not like it would go to waste - just remelt it and sell it.
You wouldn't want to machine the silver, make a mold and pour it. Maybe grind a few thousandths of an inch off the mating surfaces for a better surface finish and thus better thermal transfer. Get this idea to Linus because I think it might be a cool project.
@@Hazdazos I bet they could find a subscriber out there somewhere that is also a jeweler that would be willing to take on such a task.
I suspect the better approach is no-material. Copper conductivity might be low, but grinding down to almost nothing is even lower. Silver is so soft that I would guess you would need more than 10% thicker.
The only way to go up from here is to de-lid. Would be really interesting to see the results though.
Positive Displacement pumps are pretty wild. We had a few on the ship in the Navy and the pressures they could output were incredible
How loud they were?
@@darekmistrz4364 there are electric ones used for paint or grease dispensing. No louder than your AC return vent.
This is a big improvement over the way we cooled the laser at my first job, which was to hook up the water inlet to a utility sink in the next room, and the have the outlet just drain out a hose into that same sink. Every morning I had to turn on the faucet enough to cool the laser, while checking for a little bit to make sure the sink wasn't filling faster than it could drain. At the end of the day I had to turn off the laser, and do the process in reverse.
Not exactly a gold star for conservation.
Yikes. a big tank and a pump would suffice. Total loss systems are messed up.
Many large ice making machines work this way...water in to cool the condenser and then out to the drain.
Full room water cooling need to be brought back with this thing I love it.
I use something extremely similar to this to cool my ICP-MS and they're quite fun to work with. I've often had the same thought about getting one and using it to cool an entire home gaming pc rack. You should look at other industrial equipment, there are almost certainly loads of nerds willing to help you out wherever technical equipment like that is made. Try getting a nitrogen generator or a zero air generator and using it to displace all of the air in a case for a subambient loop. No water vapor, no condensation.
This will be the 14900K stock cooler
That beast just blowing the air out of the loop is so satisfying. I don't know how much time I've spent bleeding watercooling loops of air and this monster just does it in a second.
Their cooler videos are some of my absolute favorites. As Linus once generally said, I love watching graphs and boxes fill up.
Those coolers are available used on many industrial clearance sites. I've used them often in Pharma processing... good deals can be had
when the video opens up in the workshop with a grinning Alex, you know it'll be a good one :D
There are a couple units like this at my job 2 that are larger at 7000 and 10000w. They getting ready to replace the 7k one (standard life cycle replacement t still works perfectly) and I’m terrified to ask how much they want for it because if it’s not insane I might do this 😂. My computer room is next to my workshop so I could run the hosing through the wall
Just go ahead and ask it, some of these things get sold very cheap.
Just do it, or better run as cooling system for your PC and personal ac trough a radiator
Those are insanely expensive as it is, your electricity bill will be worse
Definitly bay one (biger ofcors) 😁😁😁
Using a 2kW heater for too long puts a big dent in my energy bill
Chiller go brrrrrrrrr!
Brr brr brr brr brr
Bro is a time traveller
Before the video was released, huh?
'Cool' joke!😂
inspired by an old thumbnail?
i build a device like this for my university graduation project... the goal was to have a lot of cooling but stable temp at the same time... back then invertor tech was not a thing so we used heater to stabilize the temperature... nowadays you only need an invertor compressor and some smart controller.
an improvement can be a variable speed pump or maybe a heat exchanger with a control valve.
Next, can you guys run that cooler with direct-Die cooling on some CPUs? I'd like to see just how much the IHS is impacting cooling performance.
I love that you can see just how giddy this makes Linus, just seems so amped to play with this idea
The chillers used in the tool for the lens, stage, and many parts that need stable temperature control for sub micron registration, uses chillers with much tighter temperature control by an order of magnitude.The temperature controller uses fuzzy logic to maintain the temperature within 0.01 degrees C. Among other rebuild tasks, I rebuild those as when they run 24/7 the pumps, valves, controls, and compressors wear out.
Every time time I think you guys thought of the last cooling video idea you outdo yourselves. I'm impressed.
If you want to go complete over the top, try to get your hands on a Huber Unistat. They go down to -120 °C (depending on the model) and have an accuracy of 0.01 °C
That's a giant pain in the ass for PC computing bc you'd have to figure out a dew point detection system that regulates the output of the cooler, otherwise you risk shorting the entire PC due to condensation
@@mashakos1 It's the same as with liquid nitrogen.
Let me introduce Alex and his pot of vaseline, he will take care of it.
@@DarthJenow lol
@@mashakos1 Idk at this point use an oil as liquid to cool. and put the whole pc in a tub.
@@mashakos1 There are already proved and tested ways to insulate a computer on a testbench and protect it for a few hours at a time.
Of course you can't use water or water with additives like ethylene glycol. But you can use ethanol or some silicone oil
Its AMAZINGLY clean to have been used for years. Must have been in a clean room
I would love to see the whole lounge cooling project
You guys should do more scuffed/experimenty videos like this. They’re my favourite on this channel. Doesn’t need “The god of computer coolers” either. Like those custom case builds. SO cool.
I’m an hvac technician. I’d like to make a suggestion. Build a chamber with gaskets out of Plexiglass. Add service access ports and pressurize it it pure nitrogen to test for leaks. Put a pure helium and maybe nitrogen mixture in it. The helium as with hard drives creates less drag. Then you can should be able to maintain a moisture less environment. Then you can add a cpu cooling evaporator and a evap with variable speed fans. You will be able to cool every thing in a pure environment. That and didn’t change your txv out to one that can handle sub zero temps? Real cryogenic freezers use a different refrigerant for process cooling.
Or you can do a geothermal geothermal heat pump to provide 50°F water.
Either that or get a mini chiller and go with hydronic cooling
Also it depends on your sensor location, typically most people move their txv sensing bulb to areas where the most ice builds .
I have about 5 of those chillers at my job that were scrapped. Never been used. I always though about using one but the power draw is insane lol.
The peak draw is huge, but unless you are sinking 5000w of heat into it, the long term draw is much, much smaller, and mostly from the big coolant pump.
Props to Alex' "janky" power cord to have its use til to this day
Linus: *a sound business improvement with the cooler*
Alex: "Anyway, what I wanna do is cool the whole lounge"
I wish you showed a diagram for how this new type of pump works. It clearly fascinated Linus when Alex showed him it on his phone
It's now new, its just not common pc cooling tech.
Positive displacement pumps are very common. Oil pumps in cars, direct injection fuel pumps, etc.
Eyyyyyy! You guys finally got just an actual chiller! Good! Now you don't have to have a mini split that gets hamstrung by trying to run it through a tiny tube. ... Can't wait to see how you mangle it.
Edit: that GPU is running at similar clocks to my CPU.
I hope they don't. For all the fun you can have taking something apart and messing with it, LTT is ultimately a business and their LTT Labs wants to get work done with working equipment, not waste time by potentially breaking a piece of equipment for the LOLs.
@@Corristo89 I mean... they kind of do?
Man, this video could have been 40 minutes long of Linus / Alex shenanigans. These type of videos are always my favourite.
It’s also worth noting that because the coolant was sub ambient, the radiators in the computer would have actually been adding heat to the coolant, since all they do is help hot areas (in this case, the air) dissipate their heat into cold areas (in this case the coolant). Not that it makes much difference, since the temperature delta would have been relatively small, and the cooler can definitely handle the extra load.
You should do a whole server room water cooling video, I want to see if it could dissipate the heat from an entire server rack
The sad reality is that there are currently no hot-swappable server racks available in the market. However, it is possible to design a server rack with hot-swappable components such as CPU, GPU, PS, RAM, and storage, each with its own built-in cooling block. The cooling lines could be designed to automatically connect and disconnect without leaking, making it easy to replace components without disrupting the entire system.
A control system on the main board could be used to connect up to 25 storage drives, 2 RAM drives, 3 CPU drives, and 1 GPU drive, allowing for flexible configurations. Since each drive could be connected via an optics connection, the data transfer would be direct and fast.
Creating RAM on a separate board could allow for more RAM to be installed in a drive. The server racks could be designed to be either rack-mountable or blade-type, with each segment being about 6 inches tall, 1.75 inches wide, and 2 feet long. This would allow for all the tubing and electrical connections to be placed in the rear. A full rack could have up to 8 sections with 10 units in each, providing a lot of flexibility for different configurations.
While designing a hot-swappable server rack with these features would be a complex and expensive undertaking, the benefits in terms of flexibility, ease of maintenance, and scalability could be significant.
I’ve set up tons of these TF-5Ks out in the field and the cooling overshoot from 5 degrees down to 3 is from the PIDs needing tuning.
Since these units are so versatile, they have PID tuning parameters in the menus so you can tuning the compressor duty cycle to whatever heat load you have.
The menus make no sense btw, I always had to have the manual handy to decipher them. 😂
Linus, I run a 13900K on a Corsair H100i Elite. I hit 82C at full load running Prime for an hour. I don't know what the F guys are doing to the chips that you supposedly find them IMPOSSIBLE to cool....
What’s the core clock
@@bumkinboi5956 Stock frequencies.
These LTT video’s are my absolute favorites, the professional jankiness is great
This is amazing. Please make many many more videos with this thing.
5:40 PLS proof this!! Would be so cool if u build ur own GOD COOLER!! PLS do this!!!
Awesome video. I actually run a Thermo5000 as the closed loop source for an SMC HRS in place of facility water and I've wanted to run one of the Thermo2500s which wouldn't cut it on the mill in a PC loop so badly...unfortunately company equipment and all that. Thank you for letting me realize that dream through you. Probably not so great that you put it in my head that's its actually feasible and a good(?) idea....
The way Alex was smiling when he first came into the shot let me know full well this was going to be a great video