So for Noctua at Comp Computex, we have

a lot of stuff to talk about. All

technical as we've done the last couple

years. We actually have a whole separate

video on this thing, but I'll show it

briefly. This is a carbon nano tube

thermal pad. Basically, super cool.

We're not going to talk about it in this

specific video, but we do have a full

in-depth interview talking about that.

Uh Naka is working with another company,

Carbise, to sell pads. They also have

this lowprofile AM5 cooler where you can

see they're really maximizing the

keepout zone on the motherboard. So,

right up against kind of the the limits

of the keepout zone for AM5 near the

heat sinks and then just kind of using

that memory area. So, you need lower

profile memory uh to accommodate it. But

that's a 70mm tall cooler. It's not

ready to launch yet. They don't have a

price target yet, but I just wanted to

show it for the stuff we're actually

going to be focusing on today. So, the

thermosiphen is back. We've talked about

this at least once, maybe twice now for

uh booth visits. And Jacob off to the

side over here is going to be joining

again for now the I'm not sure how many

years running, but we'll be go going

over all the details for this, the

changes they've done in the last year on

the technical side. And then also on the

opposite side of the booth, the liquid

cooler, which is now actually coming to

market. A lot of cool technical detail

on that, too. So, this one is going to

be mostly focused on uh I I know this is

scary, but education and talking about

science, so should be pretty fun. We

brought you this video with our brand

new Wireframe V2 mouse mat on

store.ac.net.

Also accompanied by our new Micro Slop

t-shirt that's on its way to our

warehouse right now. These feature a

parody micro slop logo with a blue

screen of death frowny face, warning

marks from event viewer, and our

rendition of Tux the penguin hidden away

and of course micro slop. So everyone

you pass either thinks you work there

for now or they know your thoughts on

AI. The wireframe V2 mouse mat on the GN

store was made by Andrew on the team in

Blender. Fully 3D modeled and then

represented in a high quality mouse mat

that you see here. The Mac can easily

accommodate a keyboard and mouse, has

fine detail with the city built of

wireframe components for the heat sink,

RAM, because let's be honest, it's the

only place any of us can get any now,

and cooling tubes, and we use a matching

blue stitching for anti- fray with a

blue rubber underside for some unique GN

flare. We modeled these to ridiculous

levels of depth in that there are things

in the model that you can't even see in

the product because we went that deep

with it. like for example the springs

underneath the switch underneath the

keycap that's represented in the matte

surface. Head to store.gamersacess.net

to support our deep dive independent

research content like this directly.

Okay, so I'm joined by Jacob again.

Jacob, we've done this at least a couple

years in a row now.

>> And uh last year you had a a pretty

early version of the thermosiphen

actually. This if I remember correctly,

this was like basic it looked like that,

right? It was just metal and a sheet in

the front, a glass sheet. Um, so on the

in the past year, what's the biggest

thing that's changed for the

thermosiphen?

>> Well, I would say the most important

thing to mention is that we've made a

major leap towards reaching the

performance targets we set out. Um, if

you recall last year, we mentioned that

our performance target is to reach the

same roughly the same performance level

as today's best all-in-one water

coolers. And we're very happy and

excited to show you a prototype here

this year uh that's actually like

reaching that target. Last year we were

uh demoing on a 9,800 X3D. I stressed

that it was not a performance demo. It

was a functional demo to show you that

we are okay to run like typical

workloads, but this is like around 100

watts territory. Um this year we are uh

running a 9950X3D

at 230 watt pitting um our uh latest

prototype against our AIO

>> and looks like a OCCT here running the

test.

>> Absolutely stress test. This is all

fluctuating here a little bit. It's not

a lab environment but you can see that

the moment the thermosiphen prototype is

at 81° C. Uh the all-in-one is at 81.5.

Um the the point to demonstrate is not

that this is slightly better than the

other but just they are roughly on the

same level

>> trying to get at least performance

parody for thermals I guess. Um what uh

we've kind of gone over some of the

basics of like what you guys are trying

to do for the last year but one of the

things I remembered was you had a wall

of like six different cold plates or

something. Did you end up deciding on

what direction to go with those?

>> Yeah, I think we are are pretty set on a

specific direction now. Um last year uh

a lot of the stuff we done and we've

experimented with uh was about dealing

with the high heat flux densities of

modern CPUs and the issue these hotspots

create for two-phase coolers. Um if you

remember last year we uh discussed the

thermodynamics background of the boiling

curve and uh that point of critical heat

flux where uh vaporization becomes so

intense that uh bubbles not only rise

and uh form full streams and jets

columns of of vapor but actually uh

vapor cushions form against the surface

and thermally insulate the surface

against the working fluid. uh which then

uh causes excessive temperature exactly

at that those points where you need

cooling the most.

>> You still have a version of that chart

here today. Is that over here? Yeah,

>> we've uh shrked it down a little bit so

that we can fit in all the other stuff.

Um but we still have uh that boiling

curve here. Um that starts with the

phase of nucleate boiling isolated

bubbles then from the inflection point

on goes into jets and columns. And

here's the point of critical heat flux

from where um the uh amount of heat that

we can transfer per excess temperature.

So temperature above the separ

separation temperature of the working

fluid draps drops drastically until the

light and frost point where the film

boiling phase starts and the hot surface

is fully insulated with a vapor film. So

that's still uh the background we're uh

we're battling with. Um but we've taken

a slightly different approach in the

end. A lot of the cold plates we've

shown last year uh were focused on um

like mitigating that uh hot spot

formation on a um structural level. Um

paradoxically making the the inside

surface of the evaporator um a little

bit cooler where the hot spot is so that

we don't run into these uh kind of

phenomena. And what we ended up doing

now is that we um actually have vertical

guiding ridges that have the main

purpose of um making the fluid

circulation inside the evaporator more

controlled, more predictable. And with

having better circulation inside, we are

becoming less dependent on that specific

hotspot u optimization because we have

more fluid going over the surface

removing heat.

>> Is is there I'm sure there's a purpose

for it. the height difference here

between so walk me through that.

>> Definitely. So when we look at it at a

micro level, um the first thing that's

important to point out is that when

vapor rises, it also drags liquid fluid

up with the vapor. Um and what we want

to have is a um as smooth as possible

recirculation of that liquid down in the

evaporator and up again. And um the

different height ridges allow for

pathways of for the liquid to circulate

back down uh without being disrupted by

turbulence from the from the upstream uh

vapor and liquid with the circulation

being more predictable and more

controlled. we have uh less issues of

vapor basically shooting out that so

violently towards the top uh that it

drags up um drops of of liquid fluid

into the into the vapor line

>> basically traffic jam is what I came to

when you were describing it but you're

just talking about I guess is this uh is

this liquid here bubbles of liquid

>> droplets basically clogging the uh the

vapor line restricting its uh its

efficient diameter and thereby

increasing uh the the fluid resistance

for the vapor to rise. And any kind of

this undesired uh uh flow impedance in

parts of the system um can cause uh the

the performance to go down. And this is

why it's so important to make um both

the evaporator circulation in the

evaporator but also the condenser work

more predictably. Um because whenever we

have situations like for example in the

condenser um uh of in this case

premature condensing where in one or

more of the of the micro channels of the

condenser the fluid uh condenses before

it actually should uh that increases uh

flow resistance through the condenser

and then may end up with fluid liquid

fluid pooling in the condenser and less

fluid being available in the evaporator

which can lead to dry out conditions in

the in the worst case

>> with an AIO. Typically they'll leave a

little bit of air in the liquid cooler

for you know a number of reasons

shipping being one of them. But uh with

a thermosiphen

do you follow the same principles? Do

you leave any air in it at all?

>> No we absolutely don't want any air

inside our system uh the system uh um we

create a vacuum in the system before

filling in the refrigerant. Uh and

that's crucial because any uh

non-condensible gases would um uh

accumulate uh in the condenser and then

actually reduce the uh the performance

of the condenser insulated towards uh

towards the fins where the where the

heat where the fans can then remove the

air. Um over the years um permeation

will introduce minutes amount amounts of

air. But this is precisely why we are so

uh careful and deliberate in selecting

the tubing materials uh to make sure

that we reach a 10-year lifespan without

any significant impact of performance

due to to permeation. When you look at a

water cooling radiator, you just care

about the delta t between inlet and

outlet. How that happens, where that

happens, you don't care. With a

condenser, if you have premature

condensation in one of the micro

channels and then increased fluid

resistance in that channel, that's a

major problem. Or conversely, if you

cool down the working fluid too much,

you end up with a phenomenon called

subcooling where the returning liquid is

too cold and takes long longer to vap

vaporize. And uh basically that means

that the system is working in a

singlephase uh stage until the the

liquid reaches uh the desired boiling

point and the entire system is not

optimized for single phase. Right.

>> So you really need to uh adjust fill

rate, pressure and temperature gradients

within all parts of the system. Uh make

sure that the fluid balance is is ideal

in all parts of the system to ensure

consistent performance. So what's what's

going on here? I see non-treated surface

surface with micro layer. This is giving

me some hints. And then micro layer of

porous cin copper also sounds like some

some concepts taken from other parts of

cooling.

What what is it special you guys are

showing here?

>> Micro layer of scinted copper is very

critical to ensuring the performance

levels we're seeing now and to increase

hotspot resistance as well. Um so it's

pretty similar technically uh as what

you use in a cinder copper heat pipe. Um

so small copper particles that form a

porous uh mesh and uh the the the

benefit of this structure is quite

similar to what you have in a heat pipe

that you have very strong capillary

action but at the same time the porous

structure allows uh for enough pathways

for vapor to escape. So we're not

running into these conflicts here and

also the v vapor pressure is generally

high enough to break through um layers

of liquid fluid. compared to a heat

pipe, the technical situation is

different in the sense that we're in a

pool boiling environment, meaning that

in a heat pipe, um you have liquid flow

only through the wick and the open area

in the middle is vapor. In our case,

we're in a pool where vapor has to find

its uh its path outwards of the of the

weted surface. Um due to the high

surface area of the of the copper weak

structure um we have better heat

transfer, better nucleation and this is

not a puzzle piece to avoid the vapor

blankets forming.

>> Um I guess a vapor blanket really just

creates an does it create an insulating

layer?

>> Yeah, totally. For for any uh typical

working fluid uh the gaseous phase of

the fluid is uh um higher thermal

resistance than the liquid phase. And

that means that if you have a vapor

blanket forming somewhere, it will

insulate from a from a thermal

perspective. And that's why it's so

critical to avoid this,

>> right? You you want to use the phase

change for its benefit, I guess, and

then kind of get it out of the way as

quick as possible.

>> Yeah, absolutely. Um, if we look at it

at a bit more micro level, um, it's

actually the so-called evaporating

miniscus uh that does the heavy lift

heavy lifting. That's the the extremely

thin um minuscule um vapor to liquid

interfaces that form within all these

crevices of the of the mesh structure.

And zooming in uh deeper one more once

more. Um the way the miniscus works is

that we have an absorbed film region

where uh the the film liquid film is

extremely thin. So adhesion forces are

very high and thermal interface

resistance is very high. So not much

ever evaporation going on at all. And

then we have the thicker bulk region

where the liquid film is so thick um

that the thermal resistance is also

quite high. But the intermediate

transition uh region is where the magic

really happens. Um because the the

thermal resistance is low and when the

fluid evaporates uh it will cool off the

thin liquid layer and then that

increases surface tension and you're

getting these Maragoldi stresses or

Marangoni flows that pull additional

liquid uh downwards uh to vaporize

again. And uh yeah, the vaporizing

liquid is drawing in additional

capillary actions. And uh yeah, that's

the beauty of the of the um yeah uh

vaporization enhancing surface. This

cold plate here is actually with the

surface uh enhancement applied with the

porous copper layer. You may wonder why

it looks grayish. That's due to the

additives that we use to control um pore

size. Um tuning the pore size, the

particle size, and the the the layer

thickness of of the entire thing is

crucial to really getting the best out

of it.

>> Right. Let's uh let's hop over to the

AIO and wrap there. Yeah. Okay. So, on

the AIO, we actually we just got these

into the office. You'd sent them. So,

we'll be testing those soon and you

know, posting benchmarks uh on the

channel. So, we'll we'll have all those

numbers pretty soon, but you guys do

have some interesting stuff to talk

about here, too. One of my favorite

tests. I still haven't been able to use

a vibrometer for any kind of testing

yet, but I've seen them in a lot of

labs. And this is, correct me if I'm

wrong, I I think when we've seen these,

it's been for like video cards and power

supplies where they'll shoot lasers at

it to basically look for really, I'll

say, microscopic movement in the

surface, right, for noise.

>> Definitely. It's a contactless uh

surface vibration measurement device. Uh

and the reason why we care so much about

uh surface level vibration is because uh

the the vibration will typically

translate into airborne noise and uh

obviously we want to cut down on that as

much as we possibly can.

>> Just looking at the frequency, you know,

spectrum and spread. I'm seeing it looks

like maybe some improvement in some of

this like 2K or 20 2500, but the chart

up there it really looks like most the

improvement is going to be in that we'll

say like 4,000 plus hertz range. Yeah, I

would say we definitely have some

improvements in the lower frequency

areas as well. You can see uh those

spikes around like what is this maybe

330 uh 500 uh 6 uh 670 they are slightly

reduced but the bulk of the improvement

is definitely uh in this like 1500 to

2,000 uh hertz zone and then the 3 kHz

to 10 kHz zone. So we're really uh we're

really significantly reducing the high

frequency noise content. Um that's also

something that's clearly audible in in

the video. we can we can uh share that

maybe you can you can cut that in. It

will also be available on YouTube from

our side. Um so the important message

that we uh that we want to get across is

that uh sound pressure level reduction

in terms of like dBA is is one important

thing. We're seeing typically one to

four dBA low noise levels with uh the um

the noise damping cover installed. Um,

but the sound quality improvement is

another very important aspect uh that

may often be easier to to perceive for

customers than the actual SPL drop. Last

year we were still looking at the 3D

printed dummy. Now we have the final uh

foam here uh two layers of uh open pore

soft acoustic form uh and a dense

barrier in between to uh target the

lower frequencies. Um the whole um pump

noise absorber is mounted on these uh

floating silicon mounts uh that ensure

that uh no vibrations or as little as

possible uh as few vibrations as

possible are transmitted uh to the outer

shell of the cover. And uh as we've

discussed last year uh the entire

assembly works both as an uh acoustic

soundproofing device but also as a tube

mass damper that reduces vibrations.

>> Right. Um what are we looking at for

these like uh sort of distribution maps?

What is that showing?

>> So you can see um our uh damping cover

uh at the bottom here and and two covers

with screens above. And what we can see

from the measurements is that the covers

with screens have a much higher um uh

vibration amplitude and a much bigger

gradient in terms of high to low uh

versus our cover having a lower

amplitude in general. and uh a much

smoother gradient. So we can spread the

entire vibr v vibratory energy over a uh

smaller surface area and this uh will

transmit uh into airborne noise uh much

less than these high uh differences in

terms of uh vibration amplitude. So the

way psycho acoustics work and the way

these types of analysis work is that we

use mathematic standardized mathematical

models that have been developed um uh in

in the scientific literature and then

score um recordings based on that uh on

these on these models and uh yeah so in

this sense it is uh very controlled very

reproducible and not something

subjective in the sense of you and me

listening to the video and you say it

sounds better I say it sounds worse. Uh

this is like uh as controlled as it gets

uh for something that uh is intended to

reflect subjective perception

>> objective basis for noise. Yeah. I mean

for example like uh typically higher

frequency people find more annoying. So

there's some objective truth to that.

But

>> totally we we discussed about the

frequency spectrum and how we uh can uh

very efficiently reduce high frequency

content. Um and uh that is for example

reflected in the sharpness core.

Sharpness uh is a metric that describes

how much high frequency uh content you

have in the acoustic spectrum. And you

can see that we're doing much better

with the cover. Um loudness is another

key metric um where uh that describes

describes perceived intensity of the

sound but it's not directly translatable

into dba db or even dba that can be

converted into like physical pascal

pressure differences because it factors

in um things like um sensitivity to

critical bands um temporal composition

of the of the sound signal. Um yeah, and

those are all the the the um the

complexities of acoustics that get lost

in a simple DBA score.

>> Um you mentioned annoyance before. Um

it's uh um not uh maybe not as commonly

known as I don't know loudness,

sharpness, tonality, roughness. Uh this

is something very a lot of people know.

Annoyance is a composite score that

basically uh combines all the other key

scores. So from from our perspective

it's maybe the most important one

because it shows uh that the overall um

net result is is very positive with our

cover.

>> What's the uh unit of measurement here?

Are you

>> uh this is just for psycho acoustic uh

annoyance. It's a dimensionless abstract

unit. So it it cannot be converted into

anything because it's based on on

subjective perception.

>> Cool. Uh, one thing I just realized is

that we should um mention the

differences. I think for regarding uh

last year's data regarding the the SPL

measurements because some viewers may

remember that last year we had an

average improvement of I think it was

5.7 dBA. Now we're uh talking about 1.4.

Why is that? Did the cover get worse?

>> Was this the because you were testing

free floating rubber band suspension

versus

>> Yeah, exactly. Um so last year's data

was in a lab pump suspended on rubber

bands. Um so we have no way for the

vibration or acoustic energy to go

except uh into the cover. Um and this

data now here has been collected on a

real user setup uh AM5 in installed

inside the Antiflux Pro Noctu Edition

case. And in such a scenario obviously

we have limited control over how

vibration and uh yeah resonance uh

spread through the CPU through the

motherboard and into the case. That's

the first reason why we're um seeing

slightly lower scores. Now, the second

reason is that we changed the me

measurement methodology. Last year's

data was from a standard um uh fixed

distance between uh the microphone and

uh the noise uh radiating surface um

methodology and now we've uh switched uh

to keeping the distance between uh the

original source of noise and the

microphone constant. So that actually

means that the microphone is closer to

the noise radiating surface uh with the

cover installed just to make sure that

we we cannot be accused of like uh uh

using a test condition that would favor

favor our own product. Um so they do

have other stuff too. We already filmed

the uh thermal grizzly wire view

collaboration that's on the channel if

anyone's curious. But that'll wrap us

for the thermosiphon and for the cooler

the AIO. There's also the pad video

coming out separately. Jacob, thank you

again.

>> My pleasure. Thanks for coming.

Absolutely.