[OT] Decibels and the ear [was: Re: [plug] Hot and bothered CPU hankers for cool breeze]

[Cameron Patrick]

(I feel kind of guilty continuing this OT thread, so I'll endeavour to
reply to a few posts with this one message.  My apologies to anyone who
doesn't care about any of this and/or has a mail set-up which makes
getting hundreds of PLUG messages a day a pain.  Also to those who
expect the In-Reply-To header to have some kind of correlation with the
contents of my messages...)

On Wed, Dec 17, 2003 at 12:06:53AM +0800, Craig Ringer wrote:
| >Random pointless stats - in the last year (using "l ~d <1y" in mutt), I
| >count 909 from craig at post, 808 from devenish at guild, 715 from
| >leon at brooks, and a mere 560 from the distant fourth: cameron at patrick.
| >(Apologies if I have left out anyone else with >500 posts this year.)
| 
| OMFG. That's rather frightening.

Indeed :-)  I'm not even going to ask what proportion of people's sent
folder that represents ... except that James has already mentioned it.
Well, my PLUG posts are a touch under half of my sent folder for this
year - I'm not sure whether that's good or bad ;-)

| I'd like to think I'm answering questions at least as often as asking 
| them, personally.

And I'd like to think that I'm making silly puns and participating in
off-topic threads as much as I am answering questions ... no, wait,
that can't be right ...

... and now on with the decibels ...

On Wed, Dec 17, 2003 at 10:56:24AM +0800, James Devenish wrote:

| > > I'm not sure that the 'semantics' of aural decibels are well defined
| [...]
| > They are well-defined. Hence A-weighting for low to medium sound
| 
| After doing a bit of Googling, it seems Cameron really meant what he
| wrote for his initial comment that 10dB ~= twice as loud :-)

Woohoo!  :-D :-D

On Wed, Dec 17, 2003 at 10:21:02AM +0800, Craig Ringer wrote:
| >To be reasonably fair to yourself, you should consider wearing
| >hearing protection if you're exposed to noise levels about 75dBA for
| >more than 8 hours a day. Workplace regulations, IIRC, require it
| >above 85dBA, based on an 8-hour exposure average.
| 
| I wear earplugs at work (as I'm in here with the servers), despite the
| noise not being generally "all that loud". I suspect it's safer to
| wear them anyway, since I work in here a _lot_ and there's a lot of
| fan + HDD noise.

I'd imagine that given what others have said on this thread and
elsewhere, the fan noise is probably not as bad as the HDD noise - HDDs
tend to be annoying high-pitched whiney things, rather than fan noise
which is (hopefully) mostly white noise due to air turbulence.

On Wed, Dec 17, 2003 at 10:56:24AM +0800, James Devenish wrote:

| Heh, I used to wear earplugs when I was in a very quiet office. Maybe
| I should do the same where I am now, as I continually have nefarious
| thoughts about disabling the machine that's next to mine.

But not your own, of course, because it's engaged in much more important
activities... :-P

It is actually my experience that a constant drone or white noise (such
as those produced by computers) tends to get filtered out subconsciously
after a while, which may be another reason why Derek didn't notice his
400 servers being incredibly noisy.

... and now I prove that I'm a maths geek ...

On Wed, Dec 17, 2003 at 01:13:28PM +0800, Mike Holland wrote:
| On Wed, 17 Dec 2003, James Devenish wrote:
| 
| > wrote for his initial comment that 10dB ~= twice as loud :-) So, a
| > nominal three decade increase from 399 extra machines could be perceived
| > as eight times louder than one machine, which is the sort of thing I was
| > wondering about. Sorry everyone!
| 
| Amongst the interesting technical comments, I think the simple answer was
| lost. 400 fans are 400x louder.

True, for some values of "louder" - specifically /not/ the way that
humans perceive "loudness", though, but instead the sound pressure level
that they produced.

|   Its more sensible to humans to talk in dB, but then its wrong to speak
| of multiples.

You /can/ speak in multiples, so long as you're aware of the maths
behind the units you're using.  Logarithmic scaled such as dBs are used
precisely /because/ speaking in multiples makes more sense when you're
talking about sound, of which we can perceive such an incredible range
of intensities.

| 100dB is NOT double 50dB.

This is correct.  It's 10^5 * 50dB.  See, speaking in multiples again?

| Just as a temp of 20 deg.C is not double 10C. (more like a 3%
| difference)

You're confusing two issues here.  Forgetting entirely about human
perceptions of temperature (of which I have no knowledge at all, besides
my own personal experiences, which tell me that 20C is significantly
warmer than 10C), the reason that there's just a 3% difference in
average kinetic energy between 10C and 20C is because kinetic energy is
proportional to /absolute/ temperature (normally measured on the Kelvin scale).
10C is 283K, and 20C is 293C, which as you said is a 3.5% difference.

Note that this is /not/ an issue of logarithmic scales, and as the
temperature approaches infinity, the Celsius and Kelvin become equal.

|   It just doesnt make any sense to talk about multiples (e.g. doubling) on
| an open-ended logarithmic scale with an arbitrarty zero point.

Yes it does, because multiples are a pretty good approximation of how
the human ear perceives sound.  You're right in that it is generally not
useful to talk about multiples /of/ dBs, thoguh (e.g. I would be
incredibly dubious of anyone talking about some f(x) such that f(50dB) =
100dB and f(60dB) = 120dB - not that such functions don't exist, but
that they're unlikely to have any physical significance).

Cheers,

Cameron.