I know "completely silent" would mean using passive cooling and solid state components, either not viable or not worth it in most cases. But then, when we talk about silent we mean silent compared to the ambient noise level or speakers, silent enough that it doesn't distract you.

At this point, I've observed several kinds of fan noise.

1. Bearing noise, especially if the lubrication has dried/the fan is dirty/older fans, especially with ball bearings, and about any fan at max speed makes bearing noise. Using better fans usually helps with this, but I also want to experiment with thermistors to control my fan speed based on temperature - old PSUs and motherboards don't have the modern ethos of fan speed control.
2. Air turbulence, especially when a high CFM fan pushes against an obstruction, bad cable management, fans against components without enough clearance. This can be fixed by removing airflow obstructions like the poorly perforated exhaust ports on older PCs and replacing them with wire-based grilles. Also, adding clearance between the fan and the exit helps a lot too.
3. The case itself vibrating. Especially with cheaper and flimsier cases with bendy side panels. And with aluminum cases. The more mass your case has, the more energy it requires to move, so it vibrates less. Rubber screws and anti-vibration pads are a lot of help too.

Of course old PCs make other kinds of noise. The HDD can be noisy, especially older models. The optical drives, the floppy drive. The coil whine, especially from the PSU. Coil whine, as already noted, can be dealt with with electronic grade silicone - power capacitors, chokes, VRMs, using silicone to absorb the vibrations has proven to be quite effective, but you also don't want the material to turn into a conductor.

Sleeve bearings made from sintered bronze can often be regenerated. Just clean the bearing thoroughly with some product not leaving residues (brake cleaner, electronics cleaner, etc), then pop it in a bath of suitable oil, and heat it all up to whatever max temperature the oil can handle before its properties are permanently affected. The max temp can usually be found in the oil's datasheet. Then let it cool down to room temperature before taking the bearing out.

The idea is to make trapped air expand by the heat and exit the bearing. When the bearing cools down and the remaining trapped air contracts, creating an under pressure, oil will enter the bearing instead.

The same method works well with small ZZ ball bearings too, though, if possible to remove one of the dust shields without destroying it, it's always better to clean the bearing properly and lubricate it with grease as it was from the factory.

Also note that this is no magic way to save bearings that's been running dry for a long time wearing out. If they've just dried out during decades of storage, there's a good chance they'll become more or less as new though.

elszgensa wrote on 2024-03-15, 23:14:

Nice. The effects of removing airflow obstructions as much as possible are greatly underestimated.

A rule of thumb is that those punched grates reduces the airflow to about half, wire finger guards only by 25%. Quite an improvement 😀

songoffall wrote on 2024-03-26, 08:08:

I know "completely silent" would mean using passive cooling and solid state components, either not viable or not worth it in most cases.

Unless I'm working or gaming (retro or otherwise), most of the time (including now) I browse the web and watch youtube and such on a silent (and quite cheap) PC build around the ASRock Q1900-ITX mainboard. No fans, no moving parts, complete silence.

Grem Five wrote on 2024-03-16, 02:44:

[...]
That one is simple get old enough where you remember using these machines and chances are your tinnitus are much louder than any coil whine that has ever existed.

Sadly not true. I used to be able to hear up to 22kHz and was very sensitive to all kinds of noise (in particular coming out of CRT monitors). Now I'm lucky if I can hear 14kHz - but coil whine can be quite a bit lower. At work we got a returned (little brick) PSU the other day with a big 12kHz peak from a whining coil. I was very much able to hear it.

I second the opinion that volume isn't everything. Pitch & colour of the sound are at least as relevant - and are completely subjective. I'm not irritated by the sound of airflow itself, even if it can be a quite loud thrum in some cases, but a high-pitched jagged fan sound will irritate me to the point of breaking a device open and modding it pretty much on the spot, even if it's objectively less loud. Similarly, I have a soft spot for old chugging hard drives even if quite loud, where I'm irritated by more clicking hard drive seek sounds.

songoffall wrote on 2024-03-26, 08:08:

2. Air turbulence, especially when a high CFM fan pushes against an obstruction, bad cable management, fans against components without enough clearance. This can be fixed by removing airflow obstructions like the poorly perforated exhaust ports on older PCs and replacing them with wire-based grilles. Also, adding clearance between the fan and the exit helps a lot too.

This is a very important point that most people don't know about, but also that they can't do much about anyway. Quite often the high pitch white noise is actually a polyphony of many (not very loud on their own) "whistles" that are created when the fan blade approaches an edge of some sort. The sharper the edge the more annoying the sound can become but even blunt edges close to the blades can create that effect. Two possible fixes are slower fan rotation or increasing the gap between the fan and any mesh/grille next to it. But as I've said that isn't always possible. It's worth experimenting though, sometimes even a mm or two can add just enough gap to change the pitch and make the noise more acceptable, even if the audible power has not dropped by much.

Deunan wrote on 2024-03-26, 14:27:

songoffall wrote on 2024-03-26, 08:08:

2. Air turbulence, especially when a high CFM fan pushes against an obstruction, bad cable management, fans against components without enough clearance. This can be fixed by removing airflow obstructions like the poorly perforated exhaust ports on older PCs and replacing them with wire-based grilles. Also, adding clearance between the fan and the exit helps a lot too.

This is a very important point that most people don't know about, but also that they can't do much about anyway. Quite often the high pitch white noise is actually a polyphony of many (not very loud on their own) "whistles" that are created when the fan blade approaches an edge of some sort. The sharper the edge the more annoying the sound can become but even blunt edges close to the blades can create that effect. Two possible fixes are slower fan rotation or increasing the gap between the fan and any mesh/grille next to it. But as I've said that isn't always possible. It's worth experimenting though, sometimes even a mm or two can add just enough gap to change the pitch and make the noise more acceptable, even if the audible power has not dropped by much.

5mm seems to be the sweet spot, and that's what Noctua settled with (NA-IS1). Sadly, it seems those exist only for 120 and 140mm fans, and retro PCs usually have 60, 80, 92mm fans.

And one more thing, which comes more to me messing up 😀)

One wouldn't expect me to read the manual for PC fans, but I really should. Turns out my Noctua fans didn't come with long and short extension cables - only the long ones were actual extension cables, and the short ones were so called "low-noise adapters", which are essentially voltage dividers.. The model numbers are NA-RC-10, 12, 13, 14, etc. So instead of all fans blaring at 100% constantly, now I have my fans running at 75% only, using these adapters, and it made quite the difference. My Athlon build is now silent enough for a retro system - I think it can still be improved, but who cares. I have a MacBook Air for when I want it to be completely silent.

EDIT: as the included "low noise adapters" are different, here's the actual values for each model I have:
NA-RC6 ~ 80 Ω
NA-RC7 ~ 50 Ω
NA-RC9 ~ 100 Ω
NA-RC10 ~ 51 Ω
NA-RC12 ~ 150 Ω
NA-RC13 ~ 82 Ω
NA-RC15 ~ 27 Ω

You'll get two types with each fan and the case will describe one as a "low noise" adapter and the other as an "ultra low noise adapter". But I think it's fine to play by ear and use the one that both sounds best and satisfies your cooling needs the best. With my build, NA-RC10 on 92mm fans has turned out to be a very adequate solution, and with 92mm fans you can't go too low without losing a lot of your airflow. As for the 60mm CPU fan, I'm not slowing it down at all, though I plan to add a spacer between the fan and the heatsink to make it a bit more silent.

I put an experiment today.

I have a relatively modern MATX system (Core i7 2600K, 16Gb DDR3, GeForce GTX 1650, Samsung EVO 870 500Gb), with no fans on the case itself and the SFX PSU working as an exhaust. In a desktop configuration, because the side panel is perforated, I would assume this system to work well, and installing a GTX1650 Dual OC has essentially cut the case into two halves so the PSU doesn't affect the CPU at all. The CPU heatsink isn't great, and the fan used to be an 80mm Glacial, that I replaced with the stock AMD cooler from Ryzen 3000 era (I tend to recycle stuff if it makes sense). While the system itself didn't thermally throttle too much, the CPU fan was blaring at 100% all the time, the GPU temps weren't great, and just by touching the case you could see hot air was accumulating there.

So I added an 80mm Noctua intake (used an old dead PSU 80mm fan to build a funnel that I attached to the case, then screwed the fan on it) to provide some air to the GPU, and screwed an 95mm Noctua to the side panel right above the CPU.

Both fans are running far below their max RPM, and I can assure you, even a slight breeze inside the case changes a lot. Idle CPU temps went from 50 to 38 degrees Celsius - and it maxes out at 61 degrees, and I think doing better than that would require a better heatsink and thermal compound. At max load, it is still audible, but not as bad as it used to be. The Noctua fans on a modern system, even small ones, are essentially silent. The GPU temps are 35 idle, 75 at max load.

So fans running on minimum vs no fans makes a lot more difference than fans at min vs fans at max.

WD-40 and fans: it helps, somewhat, but if you expect to make an old noisy fan silent, forget about it. The PSU fan in one of my systems was barely moving, so it went from clicking to humming. Better than nothing, I guess.

WD-40 is well suited for getting seized fans to spin again and detaching any unwanted junk from the whole bearing/shaft/etc. assembly. For the actual lubrication part you'll want something like sewing machine or bike chain oil.

A001 wrote on 2024-04-01, 20:24:

WD-40 is well suited for getting seized fans to spin again and detaching any unwanted junk from the whole bearing/shaft/etc. assembly. For the actual lubrication part you'll want something like sewing machine or bike chain oil.

Thanks. I'm surprised you have bike chain oil; back in the day I used to lubricate my bike chain with blue lithium grease 😀) the kind used in aviation, because it didn't harden in the winter. Got it wholesale from auto shops.

songoffall wrote on 2024-03-15, 07:59:

...lubricating it with WD-70 or lithium grease might return the noise levels to stock, but you've got to be realistic - these fans weren't very silent to start with.

You meant WD-40, I guess. Anyways it's a common misunderstanding that WD-40 is a lubricant. While it's not. It's rather a lubricant/old grease remover as it contains kerosene. It may have a short term lubrication effect right after applying, but that's it.

I've read that WD-40 was actually a different product initially, and then the formula was changed. The company making it claims WD-40 does contain oils with lubricating properties. The question is how much is there and what viscosity it has, and does it evaporate or not. All that being said the jack of all trades is a master of none - it's neither a great lubricant nor an excellent penetration agent. But it works well enough in many cases and most people don't want or need a shelf full or flammable (and possibly toxic) chemicals just to loosen a bolt.

Bondi wrote on 2024-04-02, 16:51:

songoffall wrote on 2024-03-15, 07:59:

...lubricating it with WD-70 or lithium grease might return the noise levels to stock, but you've got to be realistic - these fans weren't very silent to start with.

You meant WD-40, I guess. Anyways it's a common misunderstanding that WD-40 is a lubricant. While it's not. It's rather a lubricant/old grease remover as it contains kerosene. It may have a short term lubrication effect right after applying, but that's it.

My bad, I did mean WD-40.

I'm pretty sure it contains oils that linger after the kerosene evaporates. Not that I would use it to lubricate the bearing itself - lithium grease is better for that - but it's good enough for the interior of the fan, the coils etc.

Deunan wrote on 2024-04-02, 23:28:

I've read that WD-40 was actually a different product initially, and then the formula was changed. The company making it claims WD-40 does contain oils with lubricating properties. The question is how much is there and what viscosity it has, and does it evaporate or not. All that being said the jack of all trades is a master of none - it's neither a great lubricant nor an excellent penetration agent. But it works well enough in many cases and most people don't want or need a shelf full or flammable (and possibly toxic) chemicals just to loosen a bolt.

Yeah, and if you're not doing automotive work professionally, very specialized chemicals might have to linger on the rack way past their shelf life. It is good enough for home use and has enough uses for me to finish the can, and that's fine by me.

songoffall wrote on 2024-04-03, 09:46:

Bondi wrote on 2024-04-02, 16:51:

songoffall wrote on 2024-03-15, 07:59:

...lubricating it with WD-70 or lithium grease might return the noise levels to stock, but you've got to be realistic - these fans weren't very silent to start with.

You meant WD-40, I guess. Anyways it's a common misunderstanding that WD-40 is a lubricant. While it's not. It's rather a lubricant/old grease remover as it contains kerosene. It may have a short term lubrication effect right after applying, but that's it.

My bad, I did mean WD-40.

I'm pretty sure it contains oils that linger after the kerosene evaporates. Not that I would use it to lubricate the bearing itself - lithium grease is better for that - but it's good enough for the interior of the fan, the coils etc.

Yes, it definately leaves some kind of oil film on the surface, but it's very very thin. My experience of lubricating a fan with it was not successful. I lubricated the fan of my Thinkpad X60 and it started to rattle after a few days as if there was no grease at all. I applied lithium grease after this and it has been working very silently for a coulpe of years since.

Bondi wrote on 2024-04-03, 10:33:

songoffall wrote on 2024-04-03, 09:46:

Bondi wrote on 2024-04-02, 16:51:

You meant WD-40, I guess. Anyways it's a common misunderstanding that WD-40 is a lubricant. While it's not. It's rather a lubricant/old grease remover as it contains kerosene. It may have a short term lubrication effect right after applying, but that's it.

My bad, I did mean WD-40.

I'm pretty sure it contains oils that linger after the kerosene evaporates. Not that I would use it to lubricate the bearing itself - lithium grease is better for that - but it's good enough for the interior of the fan, the coils etc.

Yes, it definately leaves some kind of oil film on the surface, but it's very very thin. My experience of lubricating a fan with it was not successful. I lubricated the fan of my Thinkpad X60 and it started to rattle after a few days as if there was no grease at all. I applied lithium grease after this and it has been working very silently for a coulpe of years since.

Got it. Time to break out the lithium can then 😀))