amplifying computer noise: mystery sounds

[gusl]

When I turn my computer speakers really loud, I can hear background noise from the circuitry.

What's weird, though, is that it makes a characteristic sound when I maximize, minimize or restore windows. "Minimize" causes a rising sound, whereas "Maximize" sounds like an impact, followed by a dropping sound. "Restore" has its own weaker version of each of those sounds... it's almost a purring sound.

This strikes me as weird, especially because I'm using the "Windows Default" soundscheme and all of these sounds are set to "(None)".

Comments

[smandal.livejournal.com]

The video card emits a lot of digital noise. In a cheap computer, the shielding between that and the analog side of your sound card is weak or non-existent. So on gross graphical operations (like minimizing windows) there is a lot of cross-talk.

[gustavolacerda.livejournal.com]

If this is true, then my video card must be handling my window management (which sounds unlikely)... because the visual changes between maximized and restored are very minimal. I also experimented with painting the whole screen at once in Paint, but it didn't make a sound.

[smandal.livejournal.com]

The video card doesn't handle window management, but sends all the pixels to your monitor. In all likelihood, your desktop has some animation when manipulating windows; this is not one big change, but many small changes sent to your monitor. The video card makes noise when making so many small changes in a short period of time to its memory representation of the monitor image.

[gustavolacerda.livejournal.com]

So if I made a film of this, then it should make the same noise...

I tried playing MPEG films soundless, but they didn't cause any extra noise.

[smandal.livejournal.com]

MPEG is lossy compression, and won't duplicate the effect as well.

It's possible that I am wrong, and in fact the issue is between the CPU/memory bus and the sound card -- the window manager and the draw instructions sent to the video card are generated in the CPU, after all.

I get crosstalk when I scroll terminal windows with a lot of text in them; otherwise, none.

[gustavolacerda.livejournal.com]

This makes me wonder: do our brains have to deal with neural noise? Are simultaneous functions placed far enough apart to avoid interference?

If so, this would explain we are always inhibiting some kinds thinking, at any given time (Minsky says we need to avoid traffic jams: he doesn't clarify if this is a neural process or at a higher level)

[smandal.livejournal.com]

This might be why the fundamental logical element is the neuron, and not the assembly of neurons (is this right?).

It seems that most diseases like epilepsy, bipolar, etc. are due to poor chemical regulation, not interference. Like a race condition in your multi-thread program versus actual electrical interference.

[gustavolacerda.livejournal.com]

...not sure what you mean by "fundamental logical element". Most people believe that neurons are like bits (or analog versions thereof), but that symbolic (including logical) reasoning consists of many neurons acting together.

[smandal.livejournal.com]

Hmmm, so the neurons are like bits, but the neuron work together like gates and such? If so, don't they communicate by neurotransmitters, rather than actual electrical signals?

[brer-vole.livejournal.com]

My guess is that you're hearing amplified demodulated RF noise from buses that connect things in your machine. This video card being one of these things. However you can also hear network activity....

RF is emitted in relation to the edge rate of this signal and the wire length. Typically we're talking about less than 1nS edge rates (ie > 1GHZ signals) in a PC. I'd broadly estimate that brain signals have edge rates of > 100uS - and thus to radiate with a similar RF would need a wire 1e5 times longer.

Also you need to factor the receiving wire. THe PC speaker wire is relatively long and thus makes a good antenna. A "short" wire wouldn't pick up so much and so short wires and low frequencies I'd ex[ect RF cross talk to look something like (F*L)^2. Which means that for the brain expect cross talk to be 10^10 less than for the PC....