Apple’s ‘Silent’ Airflow Method - Genius or Pointless?

Check prices on Amazon below
Fractal Define 7: geni.us/20rZ3x
Noctua NF-A12x25: geni.us/DjceBwv
Video gear
Camera: geni.us/cN16f
Primary Lens: geni.us/Mfv0kQO
/ optimumtechyt
/ alisayed3
/ optimum_tech
As an Amazon Associate I earn from qualifying purchases.

[@optimumtech]

Had seen a few Mac Pro reviews mention this and was super curious whether we could re-produce the same thing on a mid-tower build. Let me know your thoughts! - Ali

[@Dave30RC]

i would be also interested in seeing the noise and temp performance when the case is divided by spacers or plastic pieces to isolate the airflow for every component

[@hermannbjorgvin]

You were measuring the wrong thing for this test I think. Loudness is just a measurement of power, but the point with the fan controller was to spread it over a larger spectrum so we don't notice it as much. It's all about the perception of noise. 3 fans all occupying the same frequency spectrum is just more noticable, not louder.

[@Dave30RC]

@@hermannbjorgvin what would be the right tools for that, a sound spectrometer?

[@optimumtech]

@@hermannbjorgvin It was measured with an SPL meter.

[@znoozi]

every PC fan spins at +/- 10% of the rated RPM. you need a tachometer to measure the real RPM to make sure it's correct for each fan.

[@WinterCharmVT]

Hey. So I'm a bit of an expert when it comes to macs, aero, and sound (actually have been working on an airflow driven SFF case that I plan to ship you a prototype of in a few months!) 1. Apple's strategy was to constantly change the fan speeds at various intervals, which eliminates any resonance effects you would have by offsetting them at a fixed interval... 2. Flatten the interval between your fan RPMs. +/- 50 or 100 rpm. It should be all that's needed to offset the different frequencies. Too big of an interval might negate the effect due to the added noise of the loudest fan... (might be a subtle effect, and the noise is confounding your measurements) 3. You almost had it!!! You were comparing 800/1000/1200 to 1000 rpm fans. You should have tried this same approach with the 100rpm stepping and 50 rpm steppings, where they are compared to 3 fans running at the *average* frequency of the 3 stepped fans. 4. The 2 spike region vs 1 spike region could be altering your results. You showed around 1400rpm, there is a second frequency spike. Because behavior changes above and below this region, consider measuring in "both" regions. 5. It's very possible this technique eliminates the "second" spike seen at 1400 rpm, but not the deeper hum of the first spike that is common below 1400 rpm, so you may not be in the right region to measure the effect in the first place. Consider trying: FIXED: 3x1000 rpm vs 1100/1000/900 rpm (100 rpm interval) 3x 1000 rpm vs 1050/1000/950 rpm (50rpm interval) 3x 1600 rpm vs 1550/1600/1650 rpm (100 rpm interval - 2 spike region) 3x1600 rpm vs 1500/1600/1700 rpm (50 rpm interval - 2 spike region) VARIABLE: 3x1000-2000 rpm (fan curve) vs 900-2000 / 1000-2000 / 1100-2000 rpm (100 rpm interval *narrows* as fan ramps up) 3x1000-2000 rpm (fan curve) vs 1000-1900 / 1000-2000 / 1000-2100 rpm (100 rpm interval *widens* as fan ramps up) 3x1000-2000 rpm (fan curve) vs 950-2000 / 1000-2000 / 1050-2000 rpm (50 rpm interval *narrows* as fan ramps up) 3x1000-2000 rpm (fan curve) vs 1000-1950 / 1000-2000 / 1000-2050 rpm (100rpm interval *widens* as fan ramps up) For the variable runs, base these fan curves on CPU temperature or something else that you can cause to change relatively quickly by running benchmarks or something, and then record the sound from the fans ramping up and look at the frequency graphs in real time. If you have software that can set the fan speeds differently based on CPU temperature (or within the Bios), you could attempt to set slightly different fan curves and run Prime95, so you get a ramp-up of the fans at slightly different intervals to try and replicate "constantly changing intervals". That may help you beat some of the resonant effects. :) Hopefully that will surface some new insights :)

[@ThrashingBasskill]

For his home setup, there are quite a looot of variables he would have to come by to make this test scenario viable I think. How do you controll the rpm of the fan so you know they really run at 1200 rpm? How do you controll the voltage of the fan or of the Mainboard header? How do you controll the db measurements? Why not using the "sone scale" instead of db, since sone measures what you are actually going for, the perceived acoustics and not the air pressure? Gamers Nexus spend years and 10-thousands to develop a process with correct data at testing cpu coolers, I think they know why. But would love to investigate in this :)

[@jasimsyriac1445]

11 year's ago I had just got a pc and playing grand theft auto vc

[@akira8817]

Idk what the fuck ur saying but you sound genius! Thumbs up

[@SubZeroGamez]

Hmm I'm courious abot this prototype of yours since you seem to have a good understanding of airflow. You should definitely send one out to him and let him check it out.

[@haroldng5897]

legend

[@colin8601]

My fans blow together in a solemn e flat minor chord. It is the most humbling experience

[@erwickdsouza]

ROFL

[@DenzCasuela]

this deserves a lot more likes

[@lemonade2473]

Lol

[@Collin0]

yay e minor

[@laxmannate07]

Mine play the brown note

[@AnythingLoud]

Noctua uses this idea in almost all of their fans. The blades can alternate between two or three different shapes in just one fan to reduce the number of blades contributing to one frequency. You can see in certain Noctua fans certain blades have a number of notches while alternating blades will have a different number of notches.

[@uss_04]

AnythingLoud Everyone talks about noctua’s total dB, but for me its all about the tone. A cheap desk fan vs a more premium fan could both be low and quiet, but the tone of a bad one can be intolerable.

[@OTechnology]

Apple's done this with their macbook fans too with alternating blade pitch and distances designed to make the noise more spread out. I would say this has some real effect but not a massive one.

[@ThrashingBasskill]

I believe the notches are kinda.. Marketing. They just have the best fan design, the best motors and the best material. Using clay to fill out the notches would probably change anything.

[@v000000000000v]

​@@ThrashingBasskill asmedigitalcollection.asme.org/IMECE/proceedings-abstract/IMECE2016/50510/V001T03A060/261359 www.patentbuddy.com/Patent/6733240 www.researchgate.net/publication/325085075_Reduction_of_Ceiling_Fan_Noise_by_Serrated_Trailing_Edge there are many papers with well documented experiments with serrated fan blades, sure they help with marketing but they are definitely functional

[@timjrvine]

Same with car tyres

[@yoshireincarnate1649]

hey, really interesting topic this. however i do feel you may have (accidentally i hope) strawmanned apples technology. my understanding was that they also modulate the fan speeds of all 3 fans at different rates to minimise resonance. which, in theory would be the piece of the puzzle which brings it all together. having 3 fans sit constantly at 3 diff speeds will still create harmonics so wouldn't really do much. and also, i'm fairly certain they would have found which fan speeds relate to which 1st harmonics and then chosen the speeds based on them. not just arbitrarily choosing 800/1000/1200. great idea for a vid but like i said, i think getting someone that has a background in audio engineering/theory would be of great help to a vid like this, because it can get seriously complicated quickly. which might actually answer your question as to why nobody other than apple has gone down this route. nice one, love your channel.

[@optimumtech]

Great insight, thanks for your comment!

[@harshitmunot28]

yeah sound is more nuanced than expected.the fans and the front panel design also may lead to different results. And also what about the difference in the thermal performance.

[@MarceloTezza]

Strawmanned, great word to describe it.

[@BTcaice]

Yoshi's correct. There is a bit more special sauce with Apple's implementation. You would probably have to use a separate microcontroller to achieve something like the variable speeds being applied. Having said that, I think noctua and some others do get us most of the way there with induced turbulence and silent motor+bearings.

[@insane5125]

@@optimumtech Marques Brownlee had a great explanation of this in his latest review of the Mac Pro.

[@meijboomm]

6:33 - that's a really fast fan

[@optimumtech]

Oops, should be 1100RPM. Good catch!

[@KuntalGhosh]

There is servers fans that can easily do 14k rpm so 11k is normal.

[@simoSLJ89]

You even said that "one-one thousand" I felt like Der8auer was speaking! :D

[@_Gecko]

Came down to the comments just for that

[@fidgetwidget9116]

Kuntal Ghosh so it’s “Over 9000” ?

[@kasakka]

I recently swapped my Fractal Define S front fans to Arctic P14. I kept hearing a kind of motor noise when two of the same fan were in the same fan header, running at low 800 RPM speeds. Replacing one of them with a Noctua A14 helped either get rid of this or mask it by running at slightly different speed. I believe this could be some sort of resonance in the case where the fan causes the case to resonate slightly at an annoying frequency. I would recommend a good fan control software like Argus Monitor to everyone for fine tuning each fan to where you get the best compromise for cooling and noise.

[@soundmanks]

You’re definitely on the right track. If a fan develops a specific frequency at 1000 RPM, and you double the number of those “sound sources”, you raise the noise level by 3dB. If you add two more of the same fans, at the same RPM, that same sound is 6dB - technically double the volume. By setting all fans at different RPM, you get more of a white noise affect, where the frequencies span across a wide spectrum, but don’t add sound energy to adjacent fans. However, a difference of 100 RPM may generate an overlap. In theory, if you have 2 identical noise sources, and can get them 180° out of phase with one another, they essentially cancel one another out. It’s the sound when 2 fans, not running at identical speeds, generates that slow pulsation sound. Twin engine prop planes do the same thing. In a absolute perfect environment, it would be possible to make 2 fans nearly silent.

[@kings_pride]

7:09 I think the main point was to loose the sine wave / humming sound when running at the same speed. At this timestamp its clearly audible.

[@carlossquire2003]

6:31 damn, 11000 RPM that’s one pretty fast fan!

[@lucaspbastos]

Was wanting to try this concept on my own build after watching the Mac Pro reviews, great video!

[@RP123lolololol]

Nice review bro, and thank you for all your videos.

[@johnretiredoagr9551]

Great video, appreciate the testing and analysis... I was hoping it would be significantly quieter... might be in the big monster cases with custom loops and 20+ fans... which is not a suggestion to test that

[@mdkenley1]

I am subscribed to a large # of channels and this is the only one I have the notifications turned on for. Always look forward to your content, Thanks!

[@coolalcoolal4165]

The apple fans constantly ramp up and down individually at different intervals which makes it near silent. All 3 of your fans were all set to specific rpms... Which still caused harmonizing. Big difference.

[@uss_04]

Looks like you’re actually making use of that Define 7 case. Not R7, as I need to keep on reminding myself. Although I do like that optional filter on the Optical opening

[@michaelmistaken2863]

Apple have been tinkering with the "detuned fan noise" idea for a while now. The blurb for recent generations of MacBook Pros say they supposedly have blowers made of varying-size/weight blades, spreading the fan's tones over a slighter broader range of frequencies is perceived as a less intense sound. Ie, it's a psychoacoustic effect, probably not machine measurable via decible metre. I assume something similar is intended for the Mac Pro front fans. And, thanks for the video. Good science!

[@KNGLeonidas99]

up

[@maxhammick948]

Spreading the energy over a broader range of frequencies to blunt the narrowband peak will show up on the high resolution Fourier transform used in this video - it'll only get hidden if you start working with energy sums like OTO levels or a total level

[@michaelmistaken2863]

Good point. Be interesting to see the graphs of offsets reflecting something closer to Apple's apparently RPM differentials (eg: 501/537/595 as I type). I have a cheap Behringer measurement mic somewhere. I might have a tinker some time. :)

[@TheRealDioBrando]

Noctua already do this too, but I think the different speed resonances might be more of a small quality of life thing rather than a huge difference, e.g the Mac Pro's fans give off a more natural tone rather than a huge overall difference.

[@enioapolinario8538]

Great video as always, curiosity is the mother of science Optimum Tech!

[@scylozy]

It seems to me that the main idea is to reduce the SPL at the fundamental frequencies related to the fans. However, the frequency plot you had shows that there isn't so much dominance in single frequencies as opposed to more of a white noise behavior. This means that the dBA level is derived mainly from turbulence or simply a wider frequency signal to begin with. Of course, doubling the sound sources would result in a +6dB increase in SPL if the sources are exactly in phase and frequency. If you are looking at a bigger spectrum you will more likely get combfiltering and end up with overall increase of +3dB in SPL, (+3dB mostly in higher frequencies, remains at ~+6dB in lower frequency due to large wave lengths). All in all, it seems to me that since the increase in fan speeds is often higher than 3dB SPL, then maybe the effect results in less overall benefit compared to just a single fan being louder. Maybe there would be some different results if the fans were just standing freely, but I doubt it. Source: I am a recently graduated acoustical engineer. I might be wrong or have not read the plot properly, but this is what it looks like to me.

[@asdf51501]

Solid video, loved the experiment. Good content.

[@primalapparel2707]

Certainly genius Quiet, sleek and minimal is something more ITX cases need to learn from. Also, have you had a chance to look at the Nouvolo Steck Case? Seems like a more improved, less premium Ghost S1 which is available. it is a DIY though but easy build.

[@DenzCasuela]

if a fan can make a corresponding frequency at certain a rpm, then three fans can make a chord

[@WinterCharmVT]

Next Video: Playing "Through The Fire and the Flames" using PC case fans >:) You'd have to write a script to change fan RPMs but with some trial and error, it *could* work :D

[@bananya6020]

@@WinterCharmVT ♪ through the fire and the flames we carryyyyy ooooonnnnnn!!!!

[@didiodulaw7417]

reminds me of the floppotron

[@ugh.idontwanna]

Proper fan controllers come with a MIDI port.

[@sc0or]

A brilliant mention. To reduce a noise there must be a phase shift. This is possible with identical sound sources precisely controlled (which us not a case), or with the same sound wave reflected and interferenced (which is also not our case). Thus Apple who all the time plays with human psychology, just shifted a tone. So customers can “confirm” they “hear” fans more silent.

[@doclogic]

The 164hz and 328hz spikes on the same line make sense since they are octaves of each other (2 x 164 = 328). I think you would need to go closer is RPM range so that you would overlap and cancel frequencies out. This would reduce the SPL levels and, if you got the right frequencies, cancel out the annoying ones between say 2k-5k (upper mid-range) that are annoying to most people. I would like to see a full range spectrum on this to at least 16khz instead of just up to say 1800hz.

[@itsdeonlol]

Great video! Very appreciated sir!

[@arnone1862]

I think you missed one key factor and that's the design of the fans themselves. the Mac Pro fans have spinning shrouds that eliminate turbulence at the blade tips. some PC case fans incorporate wings at the blade tips to achieve a similar effect.

[@CheungyNSX]

This was the design premise for the Silverstone AP123 fans, as they designed it with 3 different fan blade profiles to reduce perceived noise

[@joshlu7992]

I wonder if they are trying to keep the overall frequency being generated by the fans in a certain range. When it comes to speakers, the room can have a significant effect on acoustics, particularly in bass frequencies. For example, my room has a massive suckout between 100-120 hz that causes audio in that range to be significantly quieter. This is caused by 100-120hz frequency waves reflecting off walls and canceling each other out. I have to adjust for that by applying EQ to boost the 100-120 hz range. I believe the technical term for this is Speaker Boundary Interference Response (SBIR). I'm curious if a case could be designed to have similar reflections in order to cancel out fan noise. At lower temperatures, fans could be run at low RPM and maintain a low noise level. At higher temperatures, fan speeds would be designed to ramp up to a particularly frequency range that the case is designed to destructively interfere with. The reason to run fans at different speeds here is that the noise generated by each fan would reflect differently based on positioning, and might have a different frequency range where destructive interference is strongest. Therefore to keep noise low one would want to keep the fans at that optimal frequency range when the system is being stressed.

[@OhTyTy]

Omg, I have not watch yet but thank you for doing this video, when I heard about this with Apples new Mac Pro I was so curious about this to and wondered if this is a thing, why dont software and or fan controllers do this

[@NoHomework16]

Because it takes effort.

[@user-fc6dk4sz8e]

Well that’s why some pc builds are hella more cheaper and performs better... you don’t need over engineering to get good noise levels and temps.

[@brucemclee]

Could the unusual grill pattern of the Mac Pro also contribute to it's noise performance on top of the staggered fan speeds?

[@mike__durrett]

If I recall correctly, the Mac Pro has a passively cooled CPU. I know it wasn't part of your testing, but I think you can get away with lower fan speeds when it comes to noise, as there isn't as much to amplify with the fans. Good video! Hope you do a follow-up!

[@thomasuvano4349]

The Machine is very nice where do I find the components of the machine I seen the link for the case but not the components Thanks

[@nikoolix]

Mac pro fans also have the "ring" around the fan blades being a part of the rotating piece. That supposedly reduces turbulence or something,

[@Luke-qs2cg]

I think its the turbulence sound that they are missing as well. with all at 1 rpm, there is a much more noticeable turbulence to the sound

[@justinvalentin8983]

Question about the Define 7: Would a 5.25-inch drive fit at the top if you use only 120 mm fans?

[@Thadicus-Rex]

It sounded like there was a bit less droning noise when they were three different speeds which could sound better. It almost seems like you could harmonize them which would sound even better

[@noneminor]

Stepped fan noise seems like more natural pattern and wave like feeling, same rpm is producing more like a static sound like, highway driving tire noise. Stepped could be more more beneficial with some slightly pitched fan and cfm types. (exp; 1 noctua, 1 nzxt, 1 corsair fans.)

[@halistinejenkins5289]

i keep all my intake fans under 900rpm (140/120mm) and my exhaust (single 120/140mm) ranging from 800-1200rpm. works for my old ears (no head phones).

[@lex62lex]

I would really like to see the stepped method tested again, but with much smaller step sizes and with different random steps between the fans, like a 4 and a 7 rpm difference. Maybe that is much too small to make a difference but from what I know, bigger distances and regular distances make it easier for the produced sound to resonate, like in the energy levels and corresponding frequencies of a harmonic oscillation. Maybe a step of ~23 and 31 makes a better difference, but that would need testing.

[@abdielalfonso4556]

Now I love this video idea, how you went out to test it instead of just believe the over-engineering.

[@BenjaminSow]

One thing I noticed very obviously is with the constant fan RPM (1200), you can clearly hear the crests and trough of the fan noises which might explain the higher peaks. While the variable RPM is only marginally softer, it is a more constant hum since the crests and troughs of each sound wave produced by each fans aren't amplified on one another. Hence, the basic hypothesis is true, but to really utilize this for a meaningful outcome one would probably need a more dynamic approach.

[@Hahnck750i]

there is another reason for a lower noise level on the Mac Pro. There is no fan filters. when air needs to move through any sort of restrictions it will generate noise and at the same time lower down the air flow compare to the bare fan alone. For intake it will usually be a quieter when you do not put on fan filter or the front panel. For case exhaust, i've yet to see any case design that shape the holes for lesser restriction and noise. So far the best result i found are with a totally grill free fan exhaust and for safety add on the rounded metal fan grill on it. because blowing air through a flat metal with sharp edge always makes more noise...

[@keibohow69]

If you know someone with the apple 2019. Run it without. the shell. Also with your rid put rubber gromets in between screew head case and radiator.

[@the_realJAMES]

It would be cool to set the three fan speeds so they are enharmonic as they spin.

[@haaake]

Love that you covered this! The design and engineering really deserve way more recognition. The biggest consideration with the Max Pro fans is that they are providing airflow to a system that is properly sealed. Air only goes where it is designed to go meaning the VERY LARGE fans can run much slower than fans that are cooling poorly designed airflow paths where a lot of the air just ends up spilling out somewhere it isn’t useful. The fan design is also quite unique in that the blades and shroud are monolithic, so the blades aren’t separated from the outer shroud leaving gaps and areas for turbulence. And the actual fan shrouds are well sealed and make sure all the air that is brought in by the fans actually goes through them. The fans rarely ever need to go beyond 500rpm but can go well above 2500rpm if manually set. The blower fan that cools the RAM, SSDs, PCIe chipset and helps cool the power supply has asymmetric blades similar to what had been done in MBPs and all other blower fans Apple makes for years now. It’s also quite large, with properly sealed and completely controlled airflow paths. Not to mentioned ZERO cables to get in the way of any of this. And it also is able to pass through TB3 from its GPU modules and handle 12 TB3 ports. It’s a god damn masterpiece really. All that said, it’s so much more than just a different fixed fan speed for each fan. The most important thing PC parts companies should learn is that case design and board design should be considered together. The standard generic PC board layout doesn’t actually make very much sense. And the same can be said for power supplies and many other components. Imagine how good you could make a SFFPC if the parts were actually designed from the ground up for it.

[@haaake]

I would love to see you try to do some testing where airflow is properly controlled through a case. Make some basic shrouds and pieces to control paths of airflow, and keeps CPU and GPU airflow from mixing. NR200P with air coolers are a great example. They do much better with CPU airflow intaking from the back so you get direct cool air through the heatsink. If that airflow path was controlled so the exhaust doesn’t interact with GPU airflow at all it becomes even more efficient. The problem is that PC motherboard design is… well it’s terrible. They should entirely rethink their standard layouts, move more things to the back of the boards (RAM, nvme, chipset components), and actually consider airflow when designing them. Case companies should work directly to design custom boards for specific case concepts.

[@vipervidsgamingplus5723]

My new case is fairly silent and the case I was using sounded like a jet engine taking off, I have not switched fans

[@taung9535]

Just spitballin here. What if there were small speakers in the package case that played the inverted sound frequency of the fan sounds like how they do for noise canceling headphones. Would that theoretically work?

[@pensuke89]

I think there's a misconception on the mismatched fan speed.. first of all, it requires a very specific setup and not all pc cases are suitable for it. Why? Because in the Mac, each fans are designed for separate cooling channel. The top fan is only for the CPU, the middle fan is for other components such as PCIe cards in the middle and the bottom fan is for the GPU. Since we know that different workload will generate different heat load on CPU and GPU respectively, the fan does not need to run at the same speed. Only running at the speed it required to satisfy the cooling of the specific component is enough. If you're running CB20, it only taxes your CPU, so the top fan should run faster to push more airflow. But the bottom fan doesn't need to spin up. This will reduce the overall noise of your system. Of course, Apple Engineers also optimize the frequency produced by the fan itself. But thats entirely another optimization.

[@rizrizriz]

This reminds me of inline-6 engines in cars. They reciprocate at different intervals to minimize vibrations for maximum efficiency.

[@Maisonier]

what books do you recommend about this?

[@Jesse_wenting]

What about the audible difference at equal thermals?

[@utkarshanand9706]

What's probably happening over here is resonance. I've forgotten almost everything (and I'm not going to verify), but if the fans are producing sound that are at resonant frequencies then the amplitude comes out to be higher than expected. I guess that if the frequencies of two fan's sound waves are the same and they have the same amplitude, the amplitude of the resulting sound wave doubles (which is the maximum increment). You have to take care that the sounds aren't in resonance even with the stepped approach. The difference will be noticed between two stepped configurations, one of which are in resonance and the other isn't. Once you've figured out the amplitude, you will have to calculate the loudness (in dB) according to the formula 20*log(amplitude). In the case that amplitude is doubled, if you plug it into the formula, it becomes 20*log(2*amplitude of one of the sound waves) = 20*(log(amplitude of one of the sound waves) + log 2) = 20*log(amplitude of one of the sound waves) + 6.02. So the loudness (in decibel) should increase by 6.02 dB. I cannot guarantee either the theory or calculation, as it has been long since the last time I did it, but this should give you a start.

[@bengerber4542]

You are mostly correct. However, at these low frequencies, resonance will only be within a couple Hz or each other for it to superimpose the waveform enough to really be a problem. The other half of this side is manufacturing. Noctua states their fans have a +/- 10% manufacturing tolerance in terms of RPM. This means that the frequency will likely be off between fans by more than 2-5 Hz or so depending on frequency of the resonance. That means that they wouldn't be resonating at the same frequency unless you got REALLY unlucky. It's also a problem with stepped tests like this one. You never know how fast the fan is ACTUALLY spinning because at worst, a fan sending an 1000 RPM speed could be spinning at ~900 RPM or ~1100 RPM.

[@samwong9494]

Interesting. Not sure if it's possible to perform the test, but if you are able to map the entire frequency response to RPM, you should be able to tune the slower fans to generate a 180-degree phase-shifted frequency pattern w.r.t. the loudest fan. In theory, this should cause destructive inteference and reduce noise overall. Perhaps this was what NZXT was trying to implement with their "smart-case".

[@Larry-qz2fw]

So while I was watching I have my computer running a trans-coding load and the fan is spinning loudly. So I dropped one of the fan speed by 10% and immediately notice the fan noise feel softer. There may be some truth to this theory. However I'll have to wait and see how this effect the thermals for my system

[@beepisfartboy]

There are two changing variables so that isn't really a valid indication of the effectiveness of stepping rpm. The volume may well have just dropped because you slowed a fan down.

[@fuongbregas]

Instead of slowing one fan down, how about increasing speed of one fan

[@josearaujo8616]

Sound perception is psicological mostly... but in reality nothing changed.

[@baitboy3191]

Knowing apple, there was probably a whole team dedicated to those three fans. Everything from the bearings themselves were built specifically to maximize airflow and reduce noise. The fans in the apple case are THICC, acting more like duct fans having the air go out in a column rather than standard case fans, where the air goes out in a cone. It's amazing how dead silent those fans are even at high rpm.

[@OlavAlexanderMjelde]

Its funny I stumbled across this, because it was today that I noticed what Apple here is saying they are doing. I had some intermittent noise from my PC and I thought it was a fan going bad, my CPU is the large Noctua tower and while I am doing Folding at Home, it some times sounds as if there is a broken bearing. I thought at first it was my PSU, but then when stopping the different fans one by one, I noticed when I stopped the middle fan on the Noctua at high RPM, the noise went away. So then I stopped the one in the end (of 3 fans) and that also made the noise go away. So the protuding fan of the middle section of the 3 fans was then creating noise in cojunction with the fan at the end. I can of course control this by simply lowering the RPM in the treshold on the Asus hardware controller application, but I think I will use one of the noctua resistors for the middle fan as I have limted fan outputs from the Mini-ITX and I dont want to use the AIO header as that one reverts to 100% at each boot.

[@nurenzayyan4825]

6:37 the method works so well that it can make a 11000 rpm fan sound quieter than a 1200rpm fan

[@SirHackaL0t.]

How does that case get airflow in the front? It seems a rather solid door.

[@meischda86]

The door has a enought space to the fans when it's closed, so the air comes in from the sidevents, works pretty good.

[@michaelmistaken2863]

16 core MacPro (580x, 384Gbyte RDIMM / 1Tbyte Apple SSD / 2x970evo). Idle speeds: Front 617rpm (Top/CPU) / 545 rpm / 486 rpm. Blower (Ram/SSD/PSU): 502 rpm. Full CPU load (Cinebench r20), fans are stable at: 619 RPM / 510 RPM / 547 RPM. Blower: 498rpm. (84C CPU package, 3.3/3.45/3.7GHz min/avg/max) Full load (Cinebench r20 + Heaven), fans are normally stable at: 627 RPM / 511 RPM / 556 RPM. Blower: 500rpm. but occasional short periods of higher when 580x temp spikes to ~95ish: ~900 RPM / 550 RPM / 550 RPM. Blower: 500rpm. Front fans max at 2500rpm, PSU blower 1200rpm. Never heard the PSU blower spin up except when I set its speed manually. Only seen any of the front fans get over 650rpms when the GPU is burning.

[@Cody8P]

So they just letting their system burn up again?

[@michaelmistaken2863]

@@Cody8P Intel Power Gadget tells me that after soaking with CinebenchR20, the CPU package temp stabilises about 82C-84C on my 16 core with 3.3/3.45/3.7GHz min/avg/max frequency. That's not a wonderful temperature, but way better than the MBP's 95-98 degrees, and still near-silent unlike the laptops. With the fans all maxed the CPU package temp drops to ~55 C, but the frequency range still shows as 3.3/3.45/3.7Ghz, so it doesn't seem like there's CPU performance being left on the table by running warmer&quieter, unlike the 15" MacBookPro thermal/throttle issues back in 2018.

[@Cody8P]

@@michaelmistaken2863 good for the people who don't want to think about computer stuff i guess. Tech people probably not

[@michaelmistaken2863]

​@@Cody8P For those, there's TG Pro/other fan curve apps to do some science with. ;) (Was quite surprised the freq operating range didn't change with a 30 degree temperature drop, makes me personally happy with the default CPU fan curve).

[@niter43]

@@Cody8P 80 degrees is normal/safe range for CPU.

[@humanbeing9079]

The vast majority of the noise created by a fan is from the turbulent mixing of the air accelerated by the fan and stationary air in the surrounding. A "stepped" approach distributes the speed differential to all the fans, so each transition will be less violent/turbulent hence generating less noise *overall* . i.e. Going from stationary air->500rpm(fan 1)->800rpm(fan 2)->1000rpm(fan 3) VS going from stationary air up to 800rpm(all fans) directly. Also distributing noise to different frequencies would make it sound less noticeable to the user, while not necessarily lowering the SPL(sound pressure level).

[@ElectroMac74]

what brand microphone in use to check noise?

[@FreeRider_7]

is it right to compare 2 sound with different mic level gain??? i don't think so

[@PaulSebastianM]

I am leaning towards the idea that Apple is using different and probably varying RPM ratios to control air turbulence. So not the noise coming from the fans themselves but maybe the noise coming from the air becoming turbulent inside the case depending on what the interior of the case looks like, or what add-on cards you have.

[@joshpayne4015]

To add to the convo, in addition to speed-variance techniques, Apple also designed new fans for their approach, which I'm sure have intentional characteristics which only improve their results. This would be something you likely won't be able to replicate with your testing using off-the-shelf fans.

[@jondonnelly4831]

Noctua could build a case that synergies with one of their massive coolers. Vertical integration.

[@bunnymaid]

Compare the uncontrolled airflow of the PC case to that of the Mac (or some HP workstations) All that space above the motherboard, the empty slots is just air going past doing no work, meaning you need faster fans for the areas that actually are moving heat out. You can try this yourself with judicious application of say, cardboard to restrict airflow to where it will do the most good. Remember to keep the exhaust from one (eg, PSU) from feeding another (eg. GPU) too. Enjoy Thermodynamics for beginners!

[@RJ-pb1qx]

You have the noctua NF-F12 in the video but the link is for the NF-A12x25 in the link description? I think these are totally different performing fans if I'm correct here. So I guess my question is, which fans were you actually using?

[@brianwolfgaming1452]

I went with slower speed fans, none of them go above 1000rpm. It does though seem like 1000rpm is the sweet spot for enough airflow with not as much noise as far as case fans go..

[@DesuVR]

One piece of info missing from this video that may or may not have any impact is that the Mac Pro's fans are never at a constant RPM, they're always slowly speeding up and slowing down so no one fan is ever at the same speed. At least, that's what I've heard from a Mac Pro review video (probably MKBHD's).

[@madvinegar]

What about other factors like the fan size and structure? Bigger fans could produce big volumes of air at lower RPMs. Also, the structure of apple's fans (the fins etc.) seems to be different from the classic structure of the fans in PCs.

[@ImaMac-PC]

It's not just about different fan speeds for different frequencies for lower sound. Think of physics of airflow in the Mac Pro. Those 3 fans and 1 blower control temps in 3 different thermal chambers that each have their own flow restriction. Think of how the air, and sound, travel through those baffles. In the PC case you have, the fans are just moving air into one chamber and is not representative to achieve measurable results.

[@haris525]

What you need to do is have the fan frequency with 180 degree phase shift. In phase sound waves are +3db, phase shifted by 180 should be -3db

[@tofuprius3384]

There's a difference between absolute spl and perceived volume. If all fans are producing the same fundamental pitch there will be a series of resonant peaks that will to contribute to a perceived increase in volume, even if the total db level isn't significantly different. Think of the difference between white noise, where all audible frequencies are produced randomly, versus a single piercing tone where only one pitch is represented. At the same db level, the second tone is going to be much more noticeable and perceived as being louder.

[@Ourtown_English_Schools]

The effect is actually more psychological than physical as it produces a “whiter” noise. A similar effect is achieved with asymmetrical blades like in Apple’s laptops 💻

[@greyareaRK1]

The Sony Vaio already did this better years ago in a laptop. Fan frequencies adjusted to be inversely proportional to each other to cancel each other out.

[@joaquinrp7784]

What software do you use to control the fan speeds?? Could you link a download? please? I am dying to make my CPU fans quieter but they won't go below 1,900RPM in the BIOS. Same for my chasis fans. They are stuck at 800-900RPM. I just need a software to control them. Or anyone here in the comments help.

[@garethevans9789]

I believe the issue here is down to resonance and positive feedback. A problem with the fans operating at the same speed will cause the small imbalances of each individual fans to align together amplifying the vibration (Google the Millennium Bridge). That's not actually the worst part, the fans don't all operate at the same speed, the will have small individual differences from manufacture, meaning that they'll all spinning at slightly different speeds. Causing annoying harmonics as they cycle in and out of synchronisation with each other. There's also resonance, things like the case and the panels will have curtain frequencies that will make them resonate like a gong, again making more noise. You should be able to suspend a panel and knock it to make it 'ring'. If I were Apple, I wouldn't have all the fans using the same offset as that will also increase harmonics, plus as you mentioned the fastest fan will counter any gains. So 1200, 1183, 1150 for example (offset of 50 & 33), you don't want the fastest fan spinning much faster. I would be interested to know if the offsets were made to resemble musical chords to make the sound more pleasant.🤔

[@agustintrinchinetti8450]

What about those coloured fans on the video cover?

[@9SMTM6]

Honestly I wonder why Noone looks more into the sound transformation approach. Only usage that comes to my mind is Noiseblocker with i.e. Their eLoop Fans. Additionally to their interesting blade design they have rubber inlets in the edges that are supposed to transform high pitched vibrations to lower pitched ones, that are far less annoying, and from what I've seen that works pretty damn well, their fans are far less annoying at the same sound pressure

[@iraklimgeladze5223]

Why front of PC case is locked from air getting from front?

[@CerealReviewer]

Where your measured peak frequencies come from: 1 RPM = 1/60 Hz (just converting minutes to seconds) the fans you are using have 7 blades, so 1 RPM from the fan produces 7 pressure waves 1 fan RPM = 7/60 Hz pressure oscillation 800 RPM --> 93 Hz 1000 RPM --> 117 Hz 1200 RPM --> 140 Hz 1400 RPM --> 163 Hz These are all within 1 Hz of your measured value. Interestingly for the 1400 RPM fan you are also picking up the second harmonic at 326 Hz so there must be something else in the system resonating with that frequency. The best way to reduce the noise level is always going to be one or a combination of a few things: sound insulation used to simply damp the sounds waves before they reach your ears reduction in resonances, for example if you used the same 1400RPM fan in a different case the higher harmonic would likely not show up, but harmonics at other frequencies could be amplified. This is a complex fluids problem and something that would mostly need to be implemented during case manufacturing although other components can certainly resonant as well. Making a measurement like yours and then trying to move around components or try a different case to see the change in higher harmonics would be interesting. noise cancelling. When we think of noise cancellation now a days we think of active noise cancelling where the sound output is measured and then fed back into another sound producing component that can generate the inverse sound wave, but in reality you only need this type of feedback loop in order to cancel sound that is heavily varying. For fans in a system like this, you could implement rudimentary noise cancellation by controlling their relative phase. You want to make it so that the minimum of the pressure wave produced by one fan aligns in time with the maximum of another near by fan (with more fans it gets more complicated, but can still be done). This would minimize the amplitude of the produced sound waves and therefore decrease the volume. I don't know if this is what apple is doing, but presumably it could be done by varying the relative frequencies very slightly to keep the fans out of phase. This is probably not what apple is doing as it would be hard to implement this level of fan control. Would be pretty cool if someone did it though, my guess is that the noise reduction would be substantial. Last option is to just reduce the amplitude of the pressure waves each component makes. Essentially run your fans slower to make less noise. If you are not phase controlling as I outline above you are on average not going to create any interference between different sources. Changing the frequencies of individual fans alone will not greatly effect the total noise output beyond the sum of the individual fan contributions. You can however choose to run fans in different frequency ranges to "spread out" the noise. This could make the produced tone less perceivable, but isn't actually reducing the total amount of noise.

[@shillty]

Noise-cancelling uses an inverse wave to cancel out the original sound, maybe you could somehow use the pwm controller to actively change the wave output from the fans to cancel each other out,, A fan that was exactly half a turn ahead of the other maybe do it. Or just use normal noise cancelling.

[@benjaminolry5849]

Sorry for digging up such an old video but I think the reasoning of apple is not that the sound pressure measured in decibel will be lower, but that the human perception of the noise is less intrusive as it resembles a more spread out/ white noise-like noise. You would need to make your measurements in sone (en.wikipedia.org/wiki/Sone) in order to evaluate the perceived loudness of the fans. Apple uses a similar approach in the blade structure of their notebook fans in that they have small imperfections in the length of the blades so that the noise of the fans is more pleasing to the human hearing. But even considered all mentioned before I think that the differences are very small and that it is more of a marketing thing in order to show their attention to detail.

[@mahadevovnl]

As others have said it's not going to help if you have all fans run at a steady speed. They'll just create the same amount of noise constantly with the fastest fan creating the most. Look at it this way, if you have 1 speaker blasting at 80 decibels you can measure it; 80 decibels. Then you add 5 more speakers at 60db to the mix and measure it; still 80 decibels. The trick is having a fluctuating fan speed for all fans involved. And to make their curves never overlap. The sound might not even be measurably quieter, but it will be easier on the ears because it's no longer a constant buzzing noise, but more like a gentle vibrato.

[@JamesSam2416]

Apple probably tune their fan speeds in a more complex way than simply putting them at different speed. They might potentially running at the same rpm but in different phases, so they can cancel out each other. Like others suggested, they could have the fans running at different rpm and have the harmonics work with each other(also phase related). They probably build a quite complex acoustic model and simulate it for iterations to achieve this. It might be quite like what they did with the speakers on their MacBook Pros, fine tuning the acoustic characteristics of their chassis to achieve a amazing audio output performance.

[@markusrayan]

Also note that in a Mac Pro the middle one of the three fans spins in the opposite direction (fan blade direction reversed) to avoid turbulence between the fan edges. Clever stuff.

[@haaake]

This isn’t true.

[@MrWillypanda88]

a point here. you put the microphone near the middle fan, at exactly 10cm, maybe the mic ended up only catching the middle fan or more likely the middle fan noise is the main deciding factor. since most people didn't exatly put their ear 10 cm from their PC, maybe try doing it a bit farther?

[@HUNtastic1]

I'd assume they specifically tune the fan rpms to cancel each other's sound out. Similar to the nose canceling technology. I'm just not sure how they achieve it, when the fans are on different rpms. 🤔

[@forestR1]

Sound waves don't just compound. The can also cancel each other out while bouncing around the case (exactly how a muffler works). BUT if there's a very small difference in frequency, it will create a phaser effect. you can hear this at 6:59. that 1-2Hz phase (wurr-wurr-wurr) is more annoying than any flat loud fan because you cant tune it out. - so maybe there is something to be said for apple keeping their speeds far apart. interesting video

[@notthedroidsyourelookingfo4026]

I teach fluid dynamics at a university. And I have no idea :D

[@Farengast]

I have a degree in physics and was pretty certain this topic was WAY more complicated than could be broken down into a video like this. Glad to see my seniors in the field agree. While I suppose Apple could have done some very sophisticated airflow modeling, I suspect they instead just spent a lot more time running three fans in different ways and just found a profile that sounded and measured nicely.

[@nadrile]

Isaac L this and the amazing amount of other factors such as Size of the fans The shape of blades Distance from from panel Shape of the front panel Each fan blows in a separate compartment How the case will have a pressure difference between front and back and how it affects the airflow Haven’t seen too many videos of how the turbulence near the edge of air stream of near a physical objects affect the general flow, but all pressure changes create sound. Weird fact due to the shape of owls wings, they make less noise compared to let’s say an eagles do. But in all honesty I could only start guessing what Apple has really done. Unlikely the full on analysis of noise and airflow with software and wind chambers...

[@josearaujo8616]

@@Farengast no its not.. very simple, barely an inconvinience. Sound pressure, DB's don't add above 10 db's

[@RealGingerTea]

TLDR: From an acoustics standpoint, all fans have a +-15% in RPM (of the exact same fan) so they do this naturally. All you need to do is arrange them in the order of their variance and match the fans curves. If you want to run them at different RPMs, avoid perfect fractions/multiplications. The juice: You want to avoid the fans moving at modal intervals, in simpler terms, multiplications of each other, as this would also increase the peaks as sound less like background noise and more like a tuned pitch. For example, running a fan at 100 RPM or 100 Hz and another at 200RPM/Hz would amplify them both in much the same way, because they are an octave apart, and this would further enhance the perception of pitch. To see all of the fractions at which this occurs, please read up on the harmonic series. This is a basic concept for musicians, as this s how sound is produced. If you wish to avoid a humming sound (ie pitched) you want to create noise, so avoiding perfect fractions is the only way to achieve this. Air flow is a different story to noise though, but as always, it's a balance and trade off system. Your graphs and test equipment aren't good enough to accurately measure this. You would need a clean A/D (analogue to digital) converter (+_0.1 db @ 20hz to 20KHz) and the same for the mic preamp and the microphones frequency response, you would then need to calibrate all of these together to get a accurate reading, using sine wave sweeps and white noise. Otherwise, all we're seeing is the quality of your audio equipment, (which - I'm sorry - is not impressive) If you want to get more accurate test equipment, try an AKG C414, a DACS mic pre, and high quality A/D converter like Motu 16A or an RME fireface, and run that into pro tools (pro tools first would work fine and is free). You would also need to acoustically treat the room and a high quality loudspeaker with the flattest frequency response possible, and to account for the loudspeakers frequency response in the measurements, but this is where it gets expensive, and you're better off just using an anechoic chamber.

[@therealcarbine15]

causes low frequency baffling due to sound wave interference between the fans. i find it annoying.

[@tresnugget]

What would be awesome is if you could somehow have all of the fans running at the same frequency but be out of phase so that the sound cancels itself out to some degree. Don't have a clue if that's possible or how you could attempt it but I imagine if it was someone wouldve done it by now.

[@tresnugget]

I decided to look into it and found a professor that put a mic and speaker in his case to record his CPU fan noise and play it back out of phase. Reportedly it removed the "whine" but not the "whoosh"

[@piotrkosewski8940]

I don't own a Mac Pro and sadly don't have access to one. But given that cooling is built around separate airflow channels (much like other OEM custom workstations), it seems extremely unlikely that fans run at constant speed, let alone fixed rpm delta. Fan noise has multiple frequency peaks, so they don't have to run at the same speed to amplify. I'd expect Apple to calibrate this in factory or adjust live using a sensor in the case. Also, you don't have to struggle with complicated splitting, resistors and all that. Keep the same RPM. Use different fans.

[@nicholaslau3194]

Fan speed aside, an assumption is made that faster fan speed will lead to better cooling performance. However, it is also entirely possible that a 100-200RPM increase in fan speed makes a negligible difference , especially as a case fan

[@MrMelodyCold]

could be interesting to phase align the sound of fans

[@Asura1984]

An Inwin A1 Review would be nice

[@bluetear3372]

6:32 One of the fan is running at 11,000 rpm. Looks like the guy is making his PC fly

[@Phynix72]

Few things more to note from Apple - Component alignment Ventilation Vibration suppression Fan Design & CFM

[@MaxxPlay99]

yea, the fans have a very big circle in the middle. So less wings, less air-movement, less noise.

[@harshitmunot28]

the sound is being measured from the front so i don't think that the internal alignment of components will effect the sound. ventilation,fan design & cfm might be the most impactful things contributing towards the quietness.

[@MaxxPlay99]

harshit munot The Blades make the sound out of turbulences, right? Less Blades, less turbulences less sound. But could be possible that I‘m wrong, too, of course.

[@beauregardslim1914]

I wonder if this is just a misunderstanding of Apple's fan control in general. It is entirely possible that Apple sets the new Pro's fans to all spin up with just a slight difference in speed, but it seems more likely that they set individual fans to blow harder if that section of the box is hot. I had the original Mac Pro 1,1. It had 3 airflow compartments, 4 fans (2 in the bottom section), and a bunch of sensors. All fan speeds differed depending on temperatures in the different compartments. If this "Silent" method is a real thing, it is almost certainly a "perceived" noise that you may not be able to test with a meter.

[@currentlykent4593]

As usual, high quality video sir. Stylistically, do you consider MKBHD influential?

[@eskimolost2012]

Siblings are going college and like apple. Does 3rd gen ryzen works and what kind of gpu i should use on a portable build

[@markusstaden]

With something like this you want to avoid common factors, in this case 100Rpm. But what is way more interesting in Apples cooling system is the GPU Cooling Approach. Having all the air moving in one direction in the case makes way more sense, then our "Blow the hot gpu air vertikal to all the other air against a flat piece of pcb" system. I hope something like this finds its way in the PC Market. Also fan speed control based on CPU temp AND GPU temp would be very nice to have.