home *** CD-ROM | disk | FTP | other *** search
- Path: susx.ac.uk!peterbe
- From: peterbe@cogs.susx.ac.uk (Peter Beck)
- Newsgroups: comp.sys.amiga.audio
- Subject: Re: Paula chip and Amiga audio
- Date: 25 Feb 1996 21:51:36 GMT
- Organization: University of Sussex
- Message-ID: <4gqll8$dn2@infa.central.susx.ac.uk>
- References: <wfblanDL5rJB.IK8@netcom.com> <wfblanDLKurL.6rz@netcom.com> <4e05du$4dv@serpens.rhein.de> <judas.0ho5@tomtec.abg.sub.org> <4ekcsm$13p@news.jhu.edu> <4fk2i7$bni@nntp.texas.net> <4g0gv9$aoq@infa.central.susx.ac.uk> <4g1maa$buu@serpens.rhein.de>
- NNTP-Posting-Host: tsunx.ctn.cogs.susx.ac.uk
- X-Newsreader: TIN [version 1.2 PL2]
-
- <4gf7cj$a0i@infa.central.susx.ac.uk> <4gg9bq$r2m@serpens.rhein.de>
- Distribution:
-
- Michael van Elst (mlelstv@serpens.rhein.de) has been known to utter the following:
-
- : White noise is random. It is a special kind of noise that evenly distributes
- : over the whole spectrum.
-
- Agreed. It is also the most common form of noise, and the one which is
- usually refered too as noise.
- Because it is so random it is impossible to eliminate, unless you are
- using some sort of balanced load system.
- When you usually read about noise in technical specifics, they are
- usually refering to white noise, otherwise they could mean pink noise or
- some other weird noise definition that they have invented themselves. It
- is all down to what context the word is used in.
-
- : >Quantisation noise is still random, because it occurs due innacurracies
- : >in the sampling/playback system.
-
- : Inacurracies ? No.
-
- Depends on how you define innaccuracy.
- I was talking about any sampling system being inaccurate, because it will
- always lose data from the original analogue signal.
-
- : >an error percentage (48dB for 8 bit in fact), but the noise within that
- : >error is still random, because it cannot be predicted.
-
- : It can be be calculated. Just quantize the signal like the D/A converter.
-
- It can be calculated ideally perhaps, but real systems behave
- differently, depending on the A/D conversion method used. Cheap A/D
- systems are quite Kludgy, and are based around making approximations
- using a comparitor which finds the closed digital value by matching it
- with the analogue signal in real time. If the input signal changes
- significantly within that time, then you will get some very nasty
- results. This puts a strain on the sampling speed which you can use.
- Flash conversion is a lot faster and a lot dearer, but it still brings in
- errors due to the non-exact sizes of the resistances used in the circuit.
- Now imagine that you had a variation of 1mV between each discrete digital
- value in an AD system, and you input a voltage of 1.49mV, do you still
- hold that you can calculate what the digital value produced will be,
- considering the physical limitations of AD conversion? No?
- Well until you can, quantisation is still unpredictable, and therefore
- random noise is produced.
-