Thermal Paste Application Methods - Which one is best? - The Workshop

Linus Tech Tips ·Linus Tech Tips ·2016-05-06 · 1,913 words · ~9 min read
Floatplane YouTube

Transcript

JSON SRT VTT 139
0:00 Welcome back to the workshop. Today we're going to be checking out different methods of applying thermal paste. The
0:05 line, the P, the the X, the whatever.
0:09 We'll check them all out. This question was submitted by Nth Thrust on Twitter.
0:12 If you have a question for us to explore on the workshop, submit it to either
0:16 Twitter at Luke_lafr or just in the comments down
0:20 below this video.
0:30 Apacer's AS720 is a dual interface SSD
0:33 with USB 3.1 typeC on one end and SATA
0:36 on the other with read speeds of up to 540 megabytes per second and write
0:40 speeds of 450 megabytes per second. Check the link in the video description
0:43 to learn more. So, I watched a bunch of different YouTube videos of people
0:47 building their rigs and looked at a bunch of different build logs to try to
0:50 accumulate all the different ways that someone may or may not apply thermal
0:54 paste. So we can try all of them. The first and most common was the P method
0:57 or the dot method. Then we had the line. Then there was uh plastering it out or
1:03 spreading it around. Then there was adding way too much in whatever way you
1:08 managed to do that or way too little. Then there was the medium line where you
1:11 just do kind of a short line which I like to call the extended P. Or there's
1:16 the spiral. Or there's putting a bunch on and then putting a little plastic bag
1:20 over your finger and pushing it around.
1:24 So, we'll see if there's any difference or how big that difference is. All
1:28 right, guys. So, last time we had a workshop episode. I made a straw poll
1:32 about uh asking you guys what power supply you thought would perform the
1:36 best. This time, we're going to have two straw polls. Which thermal paste
1:41 application solution do you think is the best? And which thermal paste
1:45 application solution do you think will be the worst? Vote before you see any of
1:49 the results. Don't cheat. That's super lame. Just vote on what one you think is
1:53 going to win and then watch the video. First method we're going to try is the
1:56 dot or the P. That should probably be fine. A
2:01 pretty good rule of thumb is just look at the size of the top of the capacitors
2:05 near it. If you're around that size, you're probably good. Okay, so the dot
2:10 method has been applied. I grabbed a thermometer here to check the ambient
2:15 temperatures that are in the room and going into the case. So, I'm going to load up ID to 64, run a system stability
2:21 test on just the CPU. We will then wait
2:24 10 minutes, take a temperature reading on the ambient temperature, then take a
2:28 temperature reading on the CPU. So, the temperature that ID 64 is giving me is
2:32 about 47°, and the temperature that the ambient is in the room or the warehouse
2:36 right now is 16.8°. Next up, we're going to try the
2:40 line. The line is the next most common one. And the theory with the line is
2:44 that you want to go over top of all the things that are under the IHS or the
2:47 little metal plate there. So, as you can see from the dot, we got a nice kind of
2:51 circular pattern and it covered everything on the CPU without going way
2:54 too far out. Okay. So, yet again, we're just cleaning everything off with
2:58 isopropyl alcohol. And then we will try the line method. So, I'm going to try to
3:02 have the line be pretty thin, especially because this is really thick thermal
3:06 paste that I'm using. Here we go. So,
3:09 let's get the cooler on. So the line method also produced a temperature of
3:14 47° the exact same as the dot method and
3:18 the ambient temperatures in the room are exactly 17° which is2° higher than the
3:25 previous one. So basically everything's the same. You can see with the line
3:29 method it actually kind of sort of looks like a circle like we got from the
3:34 previous method. The dot or the P. That's just cuz of like even pressure
3:38 from the cooler. a lot of them are going to somewhat look the same. I don't know,
3:43 maybe that's the conclusion, but we'll figure that out later on. So, our next
3:46 method is going to be the X. I have never done this one personally, so to
3:51 anyone that does the X. Sorry if I do it
3:56 wrong. I don't know. That looks like a lot of thermal paste to me. Not really a
4:00 fan of the X. So the X method resulted in a temperature of
4:06 47 with an ambient of
4:10 17.2. So the temperature read out by the
4:13 computer has not gone up by one and the ambient has not gone up by one.
4:19 So yeah, time to try some other stuff. I
4:23 think I'm going to skip the medium line one. As we saw, the dot and the line and
4:28 the X were all basically identical. So, we'll try
4:31 this spiral and then we'll do some other
4:35 crazy ones after that. Oh, wow. It's not really
4:39 working. Um, no one called me an artist. So, for
4:45 the weird, awkward, horrible application method of a spiral in terms of thermal
4:50 paste, we landed with 47° C with an
4:54 average room temperature of 17.2.
5:00 So, exactly the same as the X method and
5:04 functionally the same as every other one.
5:11 So, time to move on, I guess. Next up, we're going to try
5:17 too little. So, adding not enough thermal paste at all finally gave us a
5:21 different result of 54° read out by IDA 64. The thermometer
5:27 is still reading 17.2, too. So, the ambient is the same, meaning yes, your
5:32 temperature goes a bit higher if you don't add enough thermal paste at all.
5:36 Okay, let's move on and try too much thermal paste. Warning, if you're doing
5:40 this all on your own, you want to make sure that you're not adding way too much
5:44 thermal paste because if it squishes out and goes onto your board or goes
5:48 anywhere else, it might conduct and short things out. Uh, too little thermal
5:53 paste gave us a number that was higher. So hopefully way too much gives us a
5:58 number that's higher too because I kind of think it will. And if it doesn't, I
6:02 would feel a little silly. I was wrong. It's 47° read out by I to 64 and 17.3°
6:09 ambient, which is effectively the same as basically all the other tests that
6:13 we've had. I expected this would be more because we have been told less is more
6:17 in terms of thermal paste application since forever. And even though we didn't
6:24 get that result, I'm actually going to stand by that. If you had a conductive
6:27 thermal paste, you wouldn't want it to be splooshing out so much. On the CPU,
6:31 the thermal paste coverage actually isn't too bad, which is probably why the
6:36 result ended up being pretty good. But as you can see from the block, there is
6:41 a ton of extra thermal paste gooping mainly over the sides. So, we had too
6:47 much, but I guess it wasn't like absolutely excessively too much to the
6:52 point where it's actually spilling onto the board. Okay, so now I'm going to try
6:55 to spread it around with like a card or whatever. I don't know how well this is
6:59 going to work, especially with Icy Diamond. Wow, Icy Diamond doesn't really
7:04 like to spread. I'm just going to try to squish it out onto my
7:09 finger. Okay, so I was going to do two spread
7:13 out tests. one where I uh spread it out with a card and one where I spread it
7:17 out with a my finger through a plastic bag. But I guess I just am going to have
7:23 a bunch of thermal paste on me and we're going to combine both of those tests
7:27 into just spread it out with your finger. Okay, so this is the last test.
7:33 I kind of think the results going to be 47°.
7:38 It got 47° out of IDA 64 and 17.3°
7:44 ambient. So, the conclusion, I guess,
7:48 is just don't use way too much,
7:52 especially if you have conductive, and don't use way too little, and you're
7:58 probably fine. I think the easiest one to do would probably be the line or the
8:04 P method. Uh, other than that, it's really not a
8:09 big deal. I don't really know what else to say. I guess that's actually a pretty
8:12 good conclusion for a workshop episode. That's what makes these things fun.
8:16 Crunchyroll is the site created by anime fans for anime fans. They offer the most
8:21 current episodes of new shows straight out of Japan, like Myriad Colors,
8:26 Phantom World, or Mobile Suit Gundam Ironblooded Orphans. They have a large
8:30 collection of the most popular anime series like Fairy Tale and Gate with all
8:34 the new content on their site being professionally subtitled all the time,
8:38 which is great. So, head over to crunchyroll.com/Linus and you can sign
8:41 up for a 30-day free trial of Crunchyroll Premium. If you enjoy the
8:45 many benefits of premium, like 1080p streaming, ad free service, getting new
8:49 episodes of shows straight from Japan within an hour of their premiere, and
8:52 being able to stream anywhere, anytime on a variety of devices like your phone,
8:56 tablet, game console, or computer, or whatever, you can continue your premium
9:01 membership to Crunchyroll for only $6.95 per month. So again, head over to
9:05 crunchyroll.com/Linus and check them out. Thank you for submitting this
9:10 question. If you guys have other questions, let me know at Twitter, Luke
9:15 AFR, in the comments down below down there. That'd be cool, too. Thanks for
9:19 watching this episode of the workshop. I wonder how many people are going to be
9:22 pissed cuz they've been trumping around forever that you have to do it this
9:26 specific way and doesn't really seem to be that way at all. Or how many people
9:31 are going to tell me I did the test wrong? That'll be interesting, too. We'll see that later on. Anyways, if you
9:36 enjoyed the video, like it. If you just hated it cuz I screwed everything up,
9:39 dislike it. Don't forget to subscribe. That's cool, too. These are Amazon.
9:44 These are Amazon affiliate code to shop for stuff like thermal paste and
9:49 probably not spudgers cuz it's fine. You
9:52 can just do whatever method you want. And you know, you can buy a shirt, which
9:56 isn't this one. You can buy Linus Tech Tips shirts
10:00 and uh you can become a contributor on the forum. That's another thing you can
10:04 do. Check out this video where other people were mad because I said that your
10:08 cable management doesn't really matter that much. Cool.