WEBVTT

00:00:00.000 --> 00:00:06.319
most thermal compounds have a thermal conductivity of about 10 to 20 watts per

00:00:04.319 --> 00:00:10.559
meter kelvin with really good ones reaching 30 or so the king of all

00:00:09.040 --> 00:00:17.840
thermal compounds liquid metal manages a whopping 73 watts

00:00:14.160 --> 00:00:20.960
per meter kelvin but while browsing

00:00:17.840 --> 00:00:24.240
newark an industrial supplier we found

00:00:20.960 --> 00:00:28.080
this the panasonic pyrolytic graphene

00:00:24.240 --> 00:00:30.760
sheet or pgs with an incredible thermal

00:00:28.080 --> 00:00:36.160
conductivity of 1950 watts per meter kelvin

00:00:34.399 --> 00:00:43.360
but wait that would make this thermal pad 20

00:00:39.520 --> 00:00:45.920
times as conductive as liquid metal

00:00:43.360 --> 00:00:50.719
is that even possible can i do a 360 during this sponsor segue

00:00:48.960 --> 00:00:54.559
is that even possible oh yeah

00:00:52.399 --> 00:00:58.079
ting wants to help you save money by getting you to pay for only the mobile

00:00:56.160 --> 00:01:01.039
data you use stick around until the end of the video to hear about their

00:00:59.199 --> 00:01:03.600
giveaway or click the link in the video description

00:01:10.720 --> 00:01:17.840
before we get started i need to show you this because it's just absolutely

00:01:14.720 --> 00:01:19.759
amazing okay just getting a little r r

00:01:17.840 --> 00:01:25.119
and thinking about how my circuit board over here can handle all this heat in

00:01:22.400 --> 00:01:29.840
fairness though pyrolytic graphite really is a super cool substance much

00:01:28.080 --> 00:01:34.880
cooler than that video it's man-made and cannot be found in

00:01:32.000 --> 00:01:40.560
nature and while the actual process of fabricating pyrolytic graphite is a bit

00:01:37.600 --> 00:01:44.720
over our heads it sounds pretty sweet basically you take a hydrocarbon like

00:01:42.720 --> 00:01:50.240
methane and put it under a vacuum of about 1 torr or about 1 760 of

00:01:48.560 --> 00:01:56.399
atmospheric pressure then you heat up the methane to 2000

00:01:53.200 --> 00:01:59.119
degrees celsius and slowly about a

00:01:56.399 --> 00:02:04.960
thousandth of an inch per hour a layer of graphite grows this creates perfect

00:02:02.240 --> 00:02:08.720
hexagonal carbon sheets that lie on top of one another what are they going to

00:02:06.640 --> 00:02:12.640
think of next right the shape of the carbon crystals also

00:02:10.720 --> 00:02:17.280
makes pyrolytic graphite the most diamagnetic material by weight meaning

00:02:15.120 --> 00:02:21.120
that it's repelled by a magnetic field and can be levitated

00:02:19.120 --> 00:02:25.360
unfortunately we don't have a magnet big enough to try it and we'd probably want

00:02:23.120 --> 00:02:29.280
a slightly thicker sheet of graphite but look here's a cool video of someone else

00:02:27.760 --> 00:02:33.280
doing it go check it out at the link below this also makes it excellent for

00:02:31.280 --> 00:02:37.840
electromagnetic shielding for use in things like radio towers and satellites

00:02:36.319 --> 00:02:40.959
of course we're not here for the magnetic

00:02:39.760 --> 00:02:46.080
properties you haven't even opened this have you no you have no idea if it's going to work not at all we want to look

00:02:44.239 --> 00:02:49.280
at its thermal properties the most thermally conductive substance in the

00:02:47.599 --> 00:02:54.000
world is a diamond with a thermal conductivity of around 2200 watts per

00:02:51.680 --> 00:02:59.280
meter kelvin so we're getting very close here to the limits of thermal

00:02:56.720 --> 00:03:04.239
conductivity with that insane value of 1950 watts per meter kelvin

00:03:01.840 --> 00:03:08.800
apparently the thermal conductivity of this stuff is so great that you can use

00:03:06.480 --> 00:03:15.440
a sheet of it to slice through an ice cube using just the heat of your hand

00:03:11.840 --> 00:03:18.640
it's this thin with an adhesive so it's

00:03:15.440 --> 00:03:21.280
a 10 micrometer of graphite and i think

00:03:18.640 --> 00:03:25.200
six micrometers of adhesive so how do i get this adhesive off like i can't even

00:03:23.200 --> 00:03:29.760
get a fingernail in there can you see it going in what if we add more heat

00:03:27.760 --> 00:03:33.519
holy crap it's actually like going faster

00:03:31.280 --> 00:03:36.400
i can tell like easily i don't know if i have the patience for

00:03:34.799 --> 00:03:40.879
the whole thing but it definitely sliced into it check this out and you can

00:03:37.840 --> 00:03:42.319
actually tell it's sliced in more on the

00:03:40.879 --> 00:03:46.799
side that doesn't have the adhesive on it we really need to figure out how to remove that what if we scraped off a

00:03:45.120 --> 00:03:51.440
corner of the graphite yeah there you can see the plastic sheet there okay i

00:03:49.280 --> 00:03:56.319
think i've got it peeled there we go oh my goodness it's so thin i think we're

00:03:53.360 --> 00:04:01.120
good i think i think i got it apart so this insane thermal conductivity means

00:03:58.560 --> 00:04:06.080
that pyrolytic graphite is used in you know blow stakes applications like the

00:04:03.439 --> 00:04:11.920
nose cones of missiles rocket motors nuclear reactors and even heart valves

00:04:09.040 --> 00:04:17.040
of all things but of course this is Linus tech tips we want to know

00:04:14.560 --> 00:04:22.479
if it's good for cooling computers curiously one of the main selling points

00:04:19.040 --> 00:04:24.320
of pgs is its use in electronics but it

00:04:22.479 --> 00:04:29.199
seems like the marketing focuses on thicker less thermally conductive versions

00:04:27.199 --> 00:04:33.919
rather than the thinner much more thermally conductive kind that i'm

00:04:30.639 --> 00:04:36.639
holding this one is just 10 micrometers

00:04:33.919 --> 00:04:41.520
thick also curiously it seems that for the thicker thermal pads that panasonic

00:04:38.880 --> 00:04:45.840
makes they've basically glued together a bunch of thin thermal pads with a

00:04:43.520 --> 00:04:50.400
thermally conductive glue now compared to other thermal pads these offer

00:04:48.240 --> 00:04:54.000
excellent performance but neither of those are going to cut it if we want to

00:04:51.680 --> 00:05:00.560
get the most performance ah there it is out of our CPU here before we test this

00:04:57.120 --> 00:05:03.120
out though we need a baseline

00:05:00.560 --> 00:05:08.800
now we already pre-ran using Noctua's nth2 but there's already a graphite

00:05:06.880 --> 00:05:12.560
thermal pad on the market for computers this is the innovation cooling graphite

00:05:10.720 --> 00:05:17.680
thermal pad and we looked at it a couple of years ago it doesn't keep up with a

00:05:15.039 --> 00:05:22.479
high quality thermal paste but it's good enough it's reusable it creates no

00:05:20.240 --> 00:05:25.680
mess it lasts basically forever and it can withstand temperatures that your

00:05:24.160 --> 00:05:30.160
processor will never see we actually really like this especially for test benches

00:05:28.080 --> 00:05:33.039
one problem with the ic graphite thermal pad though is that it is electrically

00:05:31.840 --> 00:05:38.240
conductive apparently that is not the case with pgs

00:05:36.320 --> 00:05:41.759
my script just says from here things get loose and probably weird

00:05:40.080 --> 00:05:46.160
that sounds about right that is not a lot of resistance two ohms

00:05:44.240 --> 00:05:49.280
yeah something like that okay let's try this okay that's about the

00:05:48.320 --> 00:05:54.880
same with that out of the way frankly i have no idea what to expect now because

00:05:52.880 --> 00:05:58.560
already the spec sheet doesn't seem to be entirely

00:05:56.320 --> 00:06:02.000
forthcoming with real information about the product

00:05:59.759 --> 00:06:05.840
and there's got to be a reason that people aren't using this stuff i

00:06:03.840 --> 00:06:11.199
have concerns about the thickness yeah that's my concern as well like even just

00:06:08.639 --> 00:06:15.280
micro imperfections in the bottom of the heatsink or CPU seem like this isn't

00:06:13.840 --> 00:06:20.479
going to bridge the gap yeah and there's stuff like i think Noctua's thermal compounds not as

00:06:18.880 --> 00:06:24.639
thermally conductive as a lot of other ones but it works better because it's just

00:06:22.960 --> 00:06:27.919
better at spreading out and filling in those gaps it's also like think about

00:06:26.479 --> 00:06:32.639
how sick it would be to be able to just install one of these on the bottom of a cooler and just never worry about it

00:06:30.800 --> 00:06:36.960
again yeah wait we don't know what the thermal

00:06:34.400 --> 00:06:42.639
conductivity of that adhesive is though um it's quite high oh okay there's our

00:06:39.919 --> 00:06:49.120
innovation cooling pad now we can throw our nhd14 on here

00:06:45.360 --> 00:06:50.639
d15 is this a d15 yep well there you go

00:06:49.120 --> 00:06:54.400
it has your name on the box and you don't even know what it is

00:06:52.240 --> 00:06:58.240
i'm more of a single tower cooler guy what the dualies are the best yeah i

00:06:56.080 --> 00:07:04.240
know but they're so unwieldy that's like part of the appeal though

00:07:00.639 --> 00:07:06.560
our test bench is a core i9 10 900 k

00:07:04.240 --> 00:07:11.199
and we're running the blender classroom test which is going to take us yeah

00:07:08.720 --> 00:07:15.280
about six or seven minutes giving us a good idea of where it's going to top out

00:07:13.120 --> 00:07:19.840
yeah for the nth2 our average temperature was 86.4 degrees and the max

00:07:18.080 --> 00:07:23.599
was 90. so that's we're going to be i found this by just sorting most

00:07:21.520 --> 00:07:27.520
thermally conductive on the website

00:07:25.680 --> 00:07:32.800
and i was like wow that's really really terribly conductive

00:07:29.759 --> 00:07:34.240
expected that's a little worse

00:07:32.800 --> 00:07:38.199
so i'm doing some rough in head math and

00:07:36.880 --> 00:07:43.599
i'm going to say 93 and a half 93 not bad yes uh that's

00:07:41.840 --> 00:07:49.039
worse but it didn't thermal throttle and that's all we really need because that means we know the thermal output through

00:07:46.400 --> 00:07:53.039
the duration of the test was the same so we can shut it down now

00:07:52.160 --> 00:07:59.360
and take a crack at this i brought over a dowel pin so that we can try and smooth

00:07:57.039 --> 00:08:03.120
it out sure it's actually probably not a terrible idea i mean this whole thing is

00:08:00.639 --> 00:08:06.639
kind of not really a great idea but like should we adhere to the heatsink or to

00:08:04.879 --> 00:08:09.759
the CPU i kind of liked your idea of doing it to the heatsink yeah cause that

00:08:08.160 --> 00:08:15.759
would be so awesome to be able to just poop plunk a heatsink yeah okay once it

00:08:12.960 --> 00:08:21.440
sticks on there it's pretty sticky so that's good for us to know about uh

00:08:18.319 --> 00:08:25.120
right about now oh lordy okay

00:08:21.440 --> 00:08:27.180
so here's the plan

00:08:25.120 --> 00:08:30.960
hmm

00:08:30.960 --> 00:08:35.599
like i was just here's the plan and then nothing

00:08:34.080 --> 00:08:40.309
there is no plan that's what i'm trying to say alex

00:08:52.080 --> 00:08:58.399
good job Linus it looks like there's some like junk under there though

00:08:56.240 --> 00:09:02.880
doesn't it you know what forget it no no no no i'm out i'm out try again

00:09:00.880 --> 00:09:06.640
oh that's way better i'm glad we didn't even bother to test the other

00:09:04.000 --> 00:09:10.160
application if we discovered a fantastic and affordable new thermal interface

00:09:08.560 --> 00:09:14.800
material you can bet you'll find it soon on ltteststore.com marked up of course

00:09:12.720 --> 00:09:20.880
well the stuff is only like seven dollars and fifty cents for a

00:09:16.640 --> 00:09:20.880
sheet of it like that size wait

00:09:20.959 --> 00:09:24.959
we spent 40 dollars on this little pack

00:09:26.800 --> 00:09:32.000
if our idol temps are anything to go by

00:09:29.839 --> 00:09:35.920
this may actually work of course idol temps tell us

00:09:34.320 --> 00:09:39.360
really not much with the story we're also like just on the cusp of thermal

00:09:37.839 --> 00:09:43.360
throttling which is kind of unfortunate but

00:09:41.839 --> 00:09:48.240
well that's fine because if it doesn't perform as well as the ic

00:09:45.440 --> 00:09:52.480
pad then it doesn't it's irrelevant so i don't think it's gonna work

00:09:51.279 --> 00:09:57.680
not as well as this pitch for ltteststore.com anyway all right

00:09:55.760 --> 00:10:05.240
i want it to work because i want to be like oh we found a thing that's so cool

00:09:59.600 --> 00:10:05.240
but i my my hope level is quite low

00:10:06.160 --> 00:10:13.360
and we're done bud oh no

00:10:09.600 --> 00:10:16.240
and it's terrible oh oh we dropped like

00:10:13.360 --> 00:10:22.640
400 megahertz off our CPU what if we put another layer

00:10:19.120 --> 00:10:23.600
on the CPU to try to

00:10:22.640 --> 00:10:28.079
you know curiously the contact patch actually

00:10:25.920 --> 00:10:32.399
isn't even that bad it does only contact this area here but

00:10:30.320 --> 00:10:35.600
that's basically the die so that's about that matters it's about half of the die

00:10:34.000 --> 00:10:38.240
the die on the 10 900k is pretty big under there you know what i want to try

00:10:37.040 --> 00:10:41.360
it i want to put another layer on the CPU okay screw it how did you do this

00:10:40.320 --> 00:10:47.920
last time good luck everybody oh oh hi hey oh yeah

00:10:44.800 --> 00:10:50.480
got a fan going over there

00:10:47.920 --> 00:10:54.160
uh that's that's not great i kind of took out one of the

00:10:52.079 --> 00:10:57.760
things here if it doesn't do worse i guess i can accept that as sort of a

00:10:56.000 --> 00:11:02.640
small victory f12 to

00:11:02.640 --> 00:11:07.680
what pay respect okay okay

00:11:06.399 --> 00:11:13.440
so it's not great for this application

00:11:10.959 --> 00:11:17.040
but that's fine because nobody advertised it as being great for this

00:11:15.279 --> 00:11:22.959
application one of the other things it is apparently good for is heat transport

00:11:20.399 --> 00:11:27.600
so you might think okay well hey Linus okay it's not that great for

00:11:25.040 --> 00:11:31.760
moving heat from the CPU up to the heat sink but what if it could spread it out

00:11:30.000 --> 00:11:35.680
a lot well the problem is that it's not nearly as

00:11:33.680 --> 00:11:39.600
good as a heat pipe those can get up to a hundred thousand watts per meter

00:11:37.680 --> 00:11:44.560
kelvin because they aren't relying on just conduction instead they're using

00:11:42.240 --> 00:11:48.480
convection and phase change cooling together to move the heat around like

00:11:46.800 --> 00:11:53.040
real quick styles some random other use cases for this

00:11:50.320 --> 00:11:57.040
stuff include apparently phones heated steering wheels which i thought

00:11:54.800 --> 00:12:01.279
was pretty cool that that would be a really neat way to kind of move the heat

00:11:59.519 --> 00:12:06.079
around from the element to spread it out more evenly heated seats again to spread

00:12:04.320 --> 00:12:11.839
heat out more evenly servers very vague but apparently they're good

00:12:09.120 --> 00:12:14.560
for servers and radio antennas which we mentioned

00:12:12.800 --> 00:12:18.959
before as for using it to cool your computer

00:12:16.240 --> 00:12:25.600
well not unless you want to run at you know 98 degrees because you just

00:12:21.680 --> 00:12:27.600
missed the 90s that much

00:12:25.600 --> 00:12:32.320
our sponsor ting does mobile phone service differently there's no contracts

00:12:30.000 --> 00:12:36.079
no overage fees or any other carrier tricks you just pay a fair price for the

00:12:34.160 --> 00:12:39.360
talk text and data that you use each month it's especially great if you're

00:12:37.760 --> 00:12:42.880
stuck at home using wi-fi instead of mobile data because hey then you don't

00:12:41.440 --> 00:12:46.480
pay for that mobile data you're not using 10 gives you complete control over

00:12:45.200 --> 00:12:50.160
your cell phone account so you can set alerts and caps for every device on your

00:12:48.320 --> 00:12:54.320
account to keep your usage in check and they've got nationwide lte coverage in

00:12:51.920 --> 00:12:58.480
the u.s using t-mobile sprint and verizon almost any phone will work with

00:12:56.720 --> 00:13:02.880
ting from an ancient motorola razer sitting in your basement to the latest

00:12:59.839 --> 00:13:04.480
iphone 11 and oh that's right i've

00:13:02.880 --> 00:13:10.240
almost forgot there's a giveaway at ting.com you've got a chance to win one

00:13:06.720 --> 00:13:13.240
of 20 ting swag bags or their ultimate

00:13:10.240 --> 00:13:16.320
prize a galaxy note 10 plus enter at

00:13:13.240 --> 00:13:17.360
mobile.ting.com Linus giveaway okay i

00:13:16.320 --> 00:13:23.279
know it's a little misleading they're not giving away a Linus just the phone and the gift bags

00:13:21.120 --> 00:13:27.040
at the link in the video description if you guys enjoyed this video and crazy

00:13:25.200 --> 00:13:31.839
cooling projects why don't i why don't we throw to the five gigahertz laptop

00:13:30.160 --> 00:13:37.200
sure yeah check that all the time yeah check out the crazy uh the crazy

00:13:34.639 --> 00:13:40.480
cooling the laptop with the blowy matron server fan got it to five gigahertz that

00:13:39.120 --> 00:13:45.360
alex and i did ages ago you know we could also throw people to

00:13:43.360 --> 00:13:49.440
another thing that didn't work like when we harvested the thermal interface

00:13:46.959 --> 00:13:52.800
material from between until no let's throw them to like the macbook

00:13:50.639 --> 00:13:55.760
video how about all three