CPU Cooling with BOILING LIQUID 🔥🔥

Linus Tech Tips ·Linus Tech Tips ·2019-05-06 · 2,518 words · ~12 min read
Floatplane YouTube

Transcript

JSON SRT VTT 208
0:00 so the thing with water cooling is there's a few different reasons people do it some do it for performance you
0:04 know getting them over clocks some people do it for looks although that
0:08 clearly wasn't the case here and some do it for silence
0:13 now here's the thing though no matter how
0:16 many radiators you throw at your system like you could put like six radiators on
0:20 this thing you could take all the fans off and it would easily dissipate the
0:24 heat from a CPU and a graphics card you are still going to be stuck with at
0:29 least one moving part and that's your pump now sometimes pumps take the form
0:34 of like a big standalone piece like this like with a custom loop while other
0:39 times you're gonna have it kind of built in or integrated into something else so
0:43 this aio from nzxt has the pump built
0:46 into the CPU block while others i believe ek does this for example build
0:51 it into the radiator but you're always going to have that
0:54 moving part which is a point of failure and a source of noise
0:59 unless you don't
1:02 this right here is a prototype
1:06 water cooler with no pump
1:10 how on earth does that work how well does it work
1:15 i guess there's only one way to find out isn't there
1:19 by sitting through this sponsor spot from pia private internet access keeps
1:24 your ip hidden from anyone who wants to know and allows you to access services
1:29 online as though you are from a different country check it out at the
1:32 link in the video description
1:42 so i first got turned on to the idea of pumpless water cooling when i was
1:47 reading a post on extremesystems.org
1:50 that's a blast from the past uh way back in 2008.
1:55 so a user by the name of ocz tony who
1:58 worked at you guessed it ocz at the time was doing some experimentation with
2:03 their upcoming CPU water block which i'm not sure if it ever got released or not
2:07 but it's hardly the point and he posted
2:10 that he had figured out a way to use convection just the natural convection
2:16 of the heat of the CPU warming the water
2:20 causing it to rise up one tube and then get cooled by the radiator causing it to
2:25 fall down the other tube in order to achieve enough circulation to get a
2:30 functional like custom grade water
2:33 cooling loop all without any risk of failing components and without any noise
2:38 whatsoever tony's approach had a couple of fatal flaws though so for one thing
2:43 it wasn't possible to get a ton of circulation so modern cpus like the 18
2:49 core one in here we're going to have no chance of moving enough water to carry
2:54 all that heat away and number two is that it relied on a very
2:59 specific layout of the entire system so that the
3:04 convection action could get started in the first place so the radiator would
3:08 absolutely need to be above the CPU
3:12 socket and then once you started to add multiple heat sources like you know a
3:16 graphics card or something well then life gets really confusing
3:20 fortunately there are other ways to move fluid around inside a closed system take
3:26 this for example this looks like an ordinary pc tower cooler but inside
3:31 these heat pipes are tiny capillaries
3:34 that allow the coolant to flow up and
3:38 down the heat pipe in order to allow it to transport heat much more quickly than
3:42 just solid tubes of metal would be able to do
3:45 that's part of what makes these things so efficient so in a nutshell then
3:50 capillary action is the phenomenon that you would observe if you took a cup of
3:53 water and then something thin like a straw for example and stuck it in you
3:59 would actually see that the water level inside the straw would rise a little bit
4:03 even though the force of gravity is working against it it's partly caused by
4:07 surface tension and it is the exact principle that our cooler here works on
4:13 now i only gave you guys a really brief look at this before but now we're going
4:18 to take a bit of a closer gander so first we've got a bog standard radiator
4:24 i think it's made uh i think it's made of aluminum and then we've got our CPU
4:29 block here which looks pretty normal you've got your flat copper base and
4:33 there's only really a couple things about it that strike me as
4:36 not ordinary so one is this little
4:40 kind of looks like a fill port of some sort that's been terminated by pinching
4:45 it shut kind of like you would do with a heat pipe and two
4:49 the fact that the fittings the block itself everything about this thing is
4:54 like soldered shut rather than using o-rings and screws in
4:59 order to seal it they they are serious about keeping whatever's in here
5:05 inside but will it work
5:22 now this is 10 out of 10 engineering
5:26 sample stuff uh ragin tech sent it over
5:29 but this is like this is not to be evaluated for its
5:34 performance or the packaging or anything like that this is
5:39 just a concept stage device
5:43 which means that mounting this is
5:46 probably not going to be that simple there's no instructions so
5:50 that kind of looks like it goes there
5:54 that kind of looks like it goes there and there
5:58 and there this really does just look like cobbled together from
6:03 scrap metal they found in the warehouse or something
6:07 are these thumb screws or just spacers these are these are thumb screws
6:11 uh well
6:15 you only live once especially if you're this poor CPU this
6:20 is my favorite thing about oh no did it go in the power supply
6:25 a screw in your power supply is a bad freaking time
6:30 one of the nuts appears to have gone missing well
6:34 hopefully we have an alternative nut oh heck yeah the nuts from my test bench
6:39 work that was easy i didn't have to go very far at all
6:43 we're also going to need a couple of cooling fans
6:46 did they say something about it needing to be above trying to remember
6:50 yeah now there's some for sure so because this process is heat assisted
6:55 we're not actually expecting it to start flowing right away like how would it do
7:00 that what would power it so we're actually waiting
7:03 for the heat output from the CPU
7:06 to move the water around we're converting heat energy into mechanical
7:10 energy but that means if there's no load on the
7:14 CPU then it's going to take a while for it to get hot enough down here to begin
7:18 that process i mean you can see here our CPU is clearly not putting out a ton of
7:22 heat and that is a hefty copper block on it
7:27 so it's going to take a while for it to appreciably heat up now that CPU is
7:32 pretty toasty at this point you can actually see we're getting some
7:35 more bubble arch we're getting some real bubble action here there's a big there's a big bubble
7:41 look at it go see that
7:44 it's actually starting to flow now now what's interesting is you actually see
7:48 bubbles flowing out of both sides
7:52 now why would that be
7:56 so you can see we're getting a little bit of kind of jerky coolant flow but
8:02 CPU temperatures are actually well within spec it seems to kind of
8:07 correspond actually to little bits of coolant kind of dripping
8:12 down into the block so i actually may have recalled
8:16 incorrectly this one may rely on some gravity assistance here so
8:21 let's go ahead and
8:25 provided so now this is kind of like having your radiator installed in the top of your
8:29 case oh and CPU temps are way down they were
8:32 up near 90 now they're in this upper 60s and they're more consistent
8:37 okay so theoretically we're all set up properly ish now you can see that our
8:43 CPU temps have settled in at around the
8:46 low 60s and that's not great obviously this is an 8700k by the way that's not
8:51 great for idle temperatures it's not even doing anything right now but you
8:55 got to remember that this doesn't work unless we add enough heat and it's
8:59 clearly working well enough that even though that's a high idle temperature
9:03 we're well within spec so in order to get our coolant flowing
9:07 more rapidly and to get the cooler working better we actually have to hit
9:12 it with a load so why don't we go ahead and do that
9:17 sure small fft 12 thread oh that's probably going to
9:21 crush this poor thing but let's give it a shot whoa okay so our temps
9:26 shoot up to 100 degrees basically instantly but what's interesting is you
9:32 can see there's a lot more action down here at the cooler we're also going to
9:36 want to check out what kind of clock speed we managed to maintain throughout this so we're still at 3.7 gigahertz
9:42 which means we're getting enough coolant flow that the thing's not just like
9:46 gonna immediately die we hope
9:51 now let's have a look at our coolant flow here so what's weird is we've actually got
9:57 liquid returns
10:01 on both of these
10:05 which would suggest that it's turning to vapor to move up to
10:10 this radiator so maybe the configuration
10:14 can be tweaked once more just trying to remember what their
10:18 instructions were i could have i could have looked this up before we started
10:21 but what fun would that be sorry i'm leaving now
10:25 it's possible that the design expects the motherboard to be oriented the way
10:30 it would in a typical tower case so why don't we go ahead
10:35 and do that just hold the board in place there we go
10:39 all right now let's see how she goes
10:42 oh interesting our temps are down in the high 90s range
10:49 now also our speed which had fallen down to
10:53 about 3.5 gigahertz is up to 3.63 we're
10:56 clearly turboing higher what's weird is that it appears as though it's operating
11:01 more like a heat pipe cooler right now with the thick tube
11:06 doing most of both the liquid and gas
11:10 movement why use only capillary action
11:14 when you can use convection as well we could tell that all the action was in
11:20 our top tube here but the top one was the thick one
11:25 so if we were to change it around and use our skinny one as the top instead of
11:30 having those two forces work against each other
11:34 we could hopefully have them work together
11:40 we can't use temperatures to differentiate our orientations here
11:45 because they're all going to hit 100 degrees in thermal throttle
11:49 but what we can see is how far it throttles and this actually throttles
11:54 much worse than our previous configuration
11:57 now we're back up to now now we're up to 3.8 gigahertz
12:02 what is your logic so this is clearly the hot side
12:10 this side is pretty hot too um
12:13 okay so i gave in i consulted the materials they sent over back when they
12:17 provided this unit and it appears as though this was the correct
12:21 configuration with the small tube on the bottom the big one on top and the
12:25 radiator up above so let's go ahead
12:28 and try it one last time
12:32 so this is the best result we've gotten so far we've only got one core that's at
12:36 100 degrees and we're turboing up to about 3.8 mid to 3.8 high gigahertz that
12:43 is not a bad result now it's obviously
12:47 not a perfect result which is probably a big part of the reason that this even
12:51 though they've shown samples of it as long as like
12:55 a year and a half ago hasn't turned into a commercial product
12:59 yet because there's a lot of fine tuning to make this work for one basically
13:04 endless tuning and tweaking could be done on the diameter of the tubes in
13:08 order to allow for better coolant flow for another there's all kinds of
13:13 different fluids with low boiling points which is part of what actually allows
13:17 this rapid movement that could be used to charge a system like this we don't
13:21 know what they're using inside i'm sure once they figured out a formula that
13:25 works they would hold that as a very closely guarded trade secret for another
13:29 thing remember how we showed that this whole
13:33 system was sealed up well maybe they're
13:36 maintaining it at something other than atmospheric pressure that could also
13:41 affect the characteristics of the entire system so
13:47 nothing like this is available from you know
13:51 any household brand that you'd recognize at this point in time but
13:57 i would say that this works well enough
14:01 that there is still potential for this technology and i'm personally very
14:06 excited for the truly silent water
14:10 cooler i mean ignoring that this particular one
14:14 has fans on it and that it'd be very difficult to diy something like this but
14:19 details details details
14:22 oh yeah and the bubbling does actually make noise
14:25 you can probably hear it
14:29 get access to amazing games books comics and more at humble bundle and
14:34 save money at the same time they believe that when you pay for things you should
14:38 have the ability to select how much you want to give to each party the creator a
14:44 charity or even humble bundle itself and you can either shop on specific titles
14:48 through their store page or through their bundle sets pick the bundle that
14:52 you want or subscribe to their monthly bundle to save even more so go check
14:56 them out at our partner link LMG dot gg
15:00 humble bundle we're gonna have that down below so thanks for watching guys if you
15:04 disliked this video you can hit that button but if you liked it hit like get
15:07 subscribed or maybe consider checking out where to buy the stuff we featured
15:11 at the link in the video description also down there is our merch store which
15:14 has cool shirts like this one actually this is a brand new design let me know
15:17 if you guys like it in the comments and our community forum which you should
15:21 totally join