Hacker News Reader

Comment by elric | Hacker News Reader

As someone who's been obsessed with silent computing for nearly two decades, just looking at the picture of that fan makes my ears hurt. Tiny fans are invariably whiney. The surface area of the blades is barely larger than that of the motor. Running it at 3V instead of 5V will make it move barely any air.

The Pi 4 can easily be passively cooled, you just need a slightly bigger heat sink. It pains me to see that they went the quick and dirty way with that crappy little fan. That case has a huge surface area. It's what, ~9x7cm? Passive solutions would have been pretty easy, and probably not much more expensive than the $5 fan. And if you really want a fan, at least get a 50mm fan, so you can spin it slowly and still actually move air.

View on HN

When I went passive on my desktop 4 years ago after having to endure an extremely loud fan, I was constantly surprised at the beginning that one is able to do computing without noticing any sound at all. Now it's become a part of my normal expectations. I run a big workload and it doesn't make any sound. Beautiful.

Ha! I actually like the noises that the computer makes. They are sort of reassuring. A computer that is always silent regardless of whether it idles or churns a lot of compute, is very creepy. Give me clunky mechanical keyboards and noisy fans!

I miss the sound of the hard drive as an indicator of I/O activity.

My little desk-computer has a pretty dirty audio path. I can turn up the audio to hear the chirps of the power regulators over the speakers; when the CPU is leaning on the PSU. Even just keystrokes in my editor cause a flurry of chirps as the clock-rate is increased and the systems races to idle within a few milliseconds.

When a process stalls I would already know whether it's IO or computation. Depending on the chirping I may reach for either ping or top to investigate further. Sometimes I leave these chirps on for hours. It's nonintrusive feedback to me about background tasks and when they clear.

I actually worry that my next machine will not have this flaw. And now I'm thinking of a syscall-chirp generator to get an even better feel for the pulse of the machine :-)

>And now I'm thinking of a syscall-chirp generator to get an even better feel for the pulse of the machine :-)

Nullsoft (the winamp people) made something like this back in the day. I encourage you you to write your own :)

Reminds me of the overheating cpu xkcd and the emacs programmer.

Or the sound of a modem to let you know the internet is working? Nothing sounds better than two pulsing, clicking sine waves finding each other after long last, making love, and giving birth to white noise.

No thanks, I'll take all my computers and electronics as quiet as possible.

Heh, I really, really liked that sound as well.

But I can't explain why really. Could be the sci-fi nerd in me similar to the way we tend to like the feel of mechanical toggle switches.

I've two machines tucked away near my desk & a sudden rise in the ambient hard drive rumbling has occasionally been an indicator that I should look into what's happening, usually it's just one of the larger torrents but it has also drawn my attention to a few bots going over my git server.

It draws attention to an alternate sense (hearing) when it comes to systems monitoring & information delivery, Phones are the common culprit of obnoxious beeping, whistles and tones but this sound isn't an intentional interpreted interrupting noise, it's a direct line from the machine.

Oh it’s still around. I just bought a 14TB external for my desk and god it’s loud.

It’s a bad loud too. Old 160MB Quantum SCSI drives have that satisfying crispy whine and that sound like some tiny hamster is changing gears around inside of it when it does a seek.

I’m surprised someone hasn’t made a SATA MITM that does this.

I once put an LED in series on a long rs232 link so I could see them flash when I/O was happening. But they have the spare voltage to work with.

Sometimes the motherboard makes a similar, but much softer, hum.

Yup, I could tell by ear if my computer had crashed and I should hard restart or if there was still hope for the program if I waited it out.

Yeah for me sound is usually the first sign that something isn't right. A truly silent laptop seems pretty inconvenient. if the screen is black and power is on, there are tons of possible causes, but hearing a lack of coil whine immediately clues in that it's properly hosed and there is no point in being patient.

> My laptop “failed” again for the 4th time today. The first 2 motherboard replacements seemed odd, but I was given a completely new laptop from Apple on the 3rd failure. Just like before, the screen was pure black after clicking the power button and there was a slight fan sound. The only other indication that anything was alive was that the machine would still make an audible chime when plugging in power and the capslock key light could be toggled off and on. So after losing about 2 weeks of my time, >$10,000 in Apple warranty repairs (2 logic boards, new cables, and a complete replacement of a >$7000 computer), troubleshooting input from several Apple Geniuses, level 1 and 2 tech support from Apple Corporate, diagnostic tests at the Apple Store, and diagnostic tests twice at Apple’s repair facility in Texas; what was the root issue?

> The screen brightness was turned all the way down (not merely dim, but off).

https://9to5mac.com/2019/06/11/macbook-pro-problem/

My build emits some extremely slight coil whining, but for the most part, it's not noticeable, only if you put your ear within close distance to the computer and pay close attention.

Also upon boot, the bios makes its speaker emit a click. From that as well as the LEDs I can tell the computer is on. No need to listen for fan noise.

It'd be interesting to convert CPU/SSD/network/etc. activity into audio using some algorithm. It's pretty easy to do poorly, just by looking at stuff like utilization percentage, but ideally you'd base it on what operations the device is actually performing.

Nullsoft made an initial effort in the early 2000s: https://web.archive.org/web/20101031004813/http://www.nullso...

Are you using a mobile processor to keep the thermals down? Or underclocking a desktop-class processor and using a giant heat sink?

Underclocking and undervolting CPUs has been a mainstay of silent computing. AMD's Duron platform was excellent for this. Not suitable for gaming, but fine for daily computing. Big heat sinks with heat pipes work pretty well. Big, slow fans when there's no way around them. Plastic rivets instead of screws and rubber washers manage to keep vibration noise down (fans, spinning disks).

Fans are cheaper than the thermostatic equivalent of aluminum or copper.

And less over-design required for “rough-service” environments.

While inefficient. You could always put some 1n400x diodes in series with your fan and drop 5V down in 0.6V increments. Even cheaper 1n4148s could do as the current is small enough.

These days, I really prefer fans because they help cool everything. Bad caps are my nemesis...

>"As someone who's been obsessed with silent computing for nearly two decades, just looking at the picture of that fan makes my ears hurt."

I bought 2 Pi4s with different cases. Both came with the same type of fan that looks like one in the article and set of heatsinks. As a person who actually tried it I can assure you that those are absolutely silent. No discernable sound comes out.

Tiny fans aren't necessarily whiny, but the quiet fans generally don't push a ton of air. My RPi4 has a Noctua 40mm x 20mm fan that is inaudible from more than a couple feet away.

Sure, it would be fairly easy to design a die cast housing to cool the RPi silently, and the part cost would be about the same as a fan, but the tooling costs are going to be fairly high for a product that really won't sell a ton of units.

Look up 'raspberry pi 4 heat sink case' and you'll find that a couple companies have already made these.

There are massive aluminum cases for Pi 4 that turn the whole thing into a giant heat sink. They work, but they aren't exactly cheap; and they still heat up a lot.

https://wickedaluminum.com/

I beg to differ:

https://flirc.tv/more/raspberry-pi-4-case

What I don’t understand is why the external surface is flat.

Even worse, they polished them to a mirror-like finish!

You want to maximize surface area to radiate heat. Ideally they’d be rippled like a regular heat sink.

Edit: also standoffs! Don’t undervalue how much heat you can expel from the underside.

As this is sufficient to cool down the pi even at max load, I assume they considered the esthetic next

Something like this? https://www.coolipi.com/

https://desalvoinc.com/products/maker-block-case-for-raspber...

https://desalvoinc.com/products/housing-for-raspberry-pi-mod...

They offer one with a flatish top and one with hexes. I bought the flat because I think it looks better. I there is much difference in cooling because in either case you will have a lot more cooling surface than the processor without a heatsink.

But if it's already good enough to cool the pi at max load why go and make it ugly too?

The theory goes that the cooler something runs, the longer its lifespan.

“Cool enough” is highly dependent on ambient conditions. A cool basement floor vs on top of an audio receiver in a warm climate will have different results.

Pimoroni wins again if you ask me. I have too many Pimoroni cases.

https://shop.pimoroni.com/products/aluminium-heatsink-case-f...

I have both, and Wicked cases are quite a bit heavier and with more material inside, so presumably it's a better heat sink.

Not all tiny fans are loud. I built a very quiet Raspberry Pi IP camera[1] actively cooled by a little Noctua fan[2], and it's so quiet I can't hear it when I put my ear up next to it (as you'd expect from a fan that costs almost $1 per cubic centimeter).

[1]: https://github.com/milkey-mouse/basket/tree/master/hw/case [2]: https://www.amazon.com/dp/B00NEMGCIA

> The surface area of the blades is barely larger than that of the motor.

It looks like about 2.5cm blade vs. 1.5cm motor diameter, which is around 2.75× the area.

Enclosing the fan in plastic with no vents should help keep the noise down, though it makes little sense from a thermal perspective.

Reply on news.ycombinator.com