Computer noises

The lower-level parts of computers

General: Computer power consumption · Computer noises

Memory: Some understanding of memory hardware · CPU cache · Flash memory · Virtual memory · Memory mapped IO and files · RAM disk · Memory limits on 32-bit and 64-bit machines

Unsorted: GPU, GPGPU, OpenCL, CUDA notes · Computer booting

✎ This article/section is a stub — some half-sorted notes, not necessarily checked, not necessarily correct. Feel free to ignore, or tell me about it.

Hard drives

Complexity to reduce:

In a roomy tower: relatively simple, given space (maybe ten bucks and some airflow consideration).

In a laptop, this is basically decided for you

Platter hard drives hum (rotation), and churn while working (head movement).

Note that:

Some are designed to be quieter, and there's usually little price difference

...but it's still mechanical, so don't expect silence.

This is partly related to the physical size of the drive. Laptop hard drives are smaller and quieter.

also related to speed. Slower tends to be quieter.

physically decoupling the sound can works well

but consider heat

Decoupling the sound

Most hard drives are screwed into a metal bay mount that is itself welded/screwed into the metal case as a whole.

This is a decent heatsink, which is great.

It's also a decent sounding box, having hard-coupled a vibration to a large surface. This is the opposite of great.

Having multiple hard drives (of the same RPM speed) tends to sound a little different, give you beating vibration - a variation in the hum often with a periodicity on the order of a second or so.

There are easy and cheapish ways of avoiding much of this noise.

Most try to make some part of the mount be rubber instead of metal. Commonly:

An adapter to somehow mount the 3.5" drive(s) in a 5.25" bay, using something like rubber as an intermediate. These may cost ten bucks or so, and work fairly well.

Elastic suspension (probably the most efficient way to decouple), by specific products

Costs a little more, tend to work a little better.

Note that DIY creativity with enough rubber bands can work well

The drive is not not cooled as well, since you don't use the case as a heatsink anymore.

The easy solution to that is to put a fan right next to the drive(s), so that air has no choice but to flow past. Doesn't have to be much, but do make it continuous, and close.

Fan noise

Complexity of reducing: Often simple, but sometimes replacement is finicky. Making them spin slower works well, but comes with footnotes.

Getting lower-noise fans

Cheap cases and cheap power supplies have cheap fans. (Sometimes even not-so-generic pre-built computers do. Profit margin and all)

A halfway decent lower-noise fan costs only a few euros/dollars, so if you buy a few and replace most of your existing fans, you eliminate most fan noise.

Larger fans move the same volume air with slower rotation, and since most fan noise comes from the rotation, larger fans are are often quieter at the same job. They're barely more expensive, so if you have space for them, use them.

Computer fans were historically often 80mm (the measure is of one side; 80mm means 8cm by 8cm) Fans in power supplies are now often 90mm or 120mm.

Case fans and CPU cooler fans have also gotten larger.

If your CPU fan is annoying you, probably the easiest way to fix that is to get just a chunkier heatsink (space permitting). Having the heatsink part be 12cm to a side is useful, not only because larger implies more surface area, but also because it fits a larger standard fan.

Small fans (~30mm, 40mm) tend to rattle and die faster, and are often also not as easy to replace. (The same goes for northbridges, but motherboard designers usually opt for a slightly larger heatsink instead of a smaller one with a fan, which is nice)

PWM and undervolting

You can get most fans to spin lower, which can make noisy fans less bothersome, and low-noise ones even subtler.

Be aware that at some voltage it would will not reliably spin up - even when if it would continue to rotate if it were spinning and voltage was lowered to that.

That voltage varies, but is often somewhere around half the full-speed voltage - assume it's above.

As fans are often 12V, 7V often works. One trick is to power them between the PSU's 12V and 5V lines - most PSUs don't care if you do. But this, or a resistor on the 12V line, are pretty makeshift and not the most reliable.

External fan controllers, or built-in fan control -- which are both typically PWM -- tends to be more reliable and, since it implies you've got a chip dedicated to your fan(s), also more flexible. They may be able to do things like spin up with full duty cycle then spin to real speed.

And in server motherboards the BMC in also in charge of the sensors, so it can often be responsive and cleverer than you are at overall heat management.

Some consumer motherboards imitate this behaviour.

It's still not so common to see all your motherboard-connected fans being PWM controlled, but when that works, it's nice.

You can often leave them controlled by the BIOS, or have your OS take control.

The BIOS settings are often hard to understand. In linux fancontrol is a nice alternative. In windows it can be a little harder.

There are also electronic speed controllers - little boxes with plugs, that you can place between fan and fan motherboard fan connector. Most of them lower the voltage, a few do PWM.

Some airflow pointers

Servers and well-designed desktop workhorses will regularly have separate channels for fans - intake, hot thing, outside air again. If also designed for lower noise, they will often use larger and multiple fans.

A more general desktop often has fans removing heat from hot components and dumping it into the case. Which is why you also want fans removing hot air from the case

A single case fan may be enough for this, if it's strong enough, if there is air intake elsewhere so it actually draws in air, if outside the case there's place for the warm air to go that is not that intake, etc.

In fact, sometimes the PSU fan 'may be good enough at this, but PSUs tend to be full and not have a lot of airflow.

So most cases have space to add one 120mm, or one or two 80mm grilles.

Note: some modern hard drives run hotter, so particularly if you have more than one in close proximity, run some air directly over them is a good idea. (I find mounting them in the 5.25" bays and adding a 12cm fan there works well).

Think about where air comes in and goes out. You can often choose this by flipping fans, and sometimes that even makes sense. But don't make fans fight each other if you can avoid it.

The PSU cools itself and adds some airflow to the case.

You may want to get a PSU with a 120mm fan, although these are usually on the bottom (pushing air), which depending (on how crowded it is, its own hole/grille on the back, etc) may not actually be more efficient than a decent 80mm on the outside mostly pulling air. But will often still be quieter.

Replacing a power supply fan with a quieter one sounds scarier, but is often easy enough.

When the PSU does speed control and you trust your electrical skills, to the same wires that the previous one was on (just beware that those plugs - if they are plugs in the first place - may be wired differently).

I often connect it to the motherboard or a molex connector.

OLTT and CLTT

Coil whine

Complexity of reducing

Easy if you get lucky
...but more typically somewhat hard

Inductors deal with AC and electromagnetic fields. In computers they are mostly used around power conversion, which tends to be high speed switching.

They are physical objects, so AC will mean it exerts some small force on itself, and may cause some physical vibration. This is usually tiny, but when it happens to resonate, it can get noticeable.

Due to the adaptive nature of power conversion, this may be constant or happen only under some very specific circumstances (which are predictable in theory, but who's going to model adaptive hardware to this detail?)

It will typically sounds like a high-frequency chirp, hiss, or whine.

Note that this is a physical sound, not to be confused with, say, coupling of EM fields into the sound card (which may also happen, but then appear in your audio).

In computers, you'ld mostly look at the PSU, the CPU's voltage regulation (typically right next to the CPU socket), and the GPU.

Fan speed controllers may also do it, or LED controllers if you've got something particularly blinky, though these are often too low-current to be the top culprit.

Possible solutions:

stick some hot glue or similar on it. Cheap, non-conductive, and you'll change the physical resonant frequency.

Often limited effect, but trivial to do.

If you can control the frequency that the current draw works at, away from the physical resonance frequency, that helps.

Try tweaking motherboard voltage management stuff - enable/disable spread spectrum, give the CPU a slightly different voltage. Experiment.

Sometimes it turns up mostly in specific CPU idle states (or under high load, or under low load, may change pitch, etc.)

Try fiddling with the way your CPU sleeps (for example using RMclock). E.g. some time ago, disabling Intel chips' C4 state seemed to help.

On notebooks, this does decrease the battery life somewhat, but when the whine drives you bonkers, that may be a good tradeoff for you.

if your PSU is always noisy, it may just be low-quality. You could consider replacing it.

Design time:

In circuit board design, coil whine can be reduced by more secure physical attachment, but more often by damping materials (which tend to degrade over time, so older electronics will seem to whine more)

Interference

CD/DVD drives

Drives may vibrate if the construction is not very well balanced.

Also, drives may spin at fairly ridiculous speeds (5000-10000RPM), so it'll make noise no matter what.

You can always take out the CD/DVD, of course, but sometimes you want it to use it silently.

Various laptops let you tell the drive to spin at the lowest speed that is still enough for DVD-Video playback (or such).