Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.1 6/24/83; site allegra.UUCP
Path: utzoo!watmath!clyde!floyd!harpo!ulysses!allegra!karn
From: karn@allegra.UUCP (Phil Karn)
Newsgroups: net.audio
Subject: Re: CD phase shift question
Message-ID: <2394@allegra.UUCP>
Date: Wed, 11-Apr-84 02:48:14 EST
Article-I.D.: allegra.2394
Posted: Wed Apr 11 02:48:14 1984
Date-Received: Thu, 12-Apr-84 04:42:36 EST
References: <920@drufl.UUCP>
Organization: Bell Communications Research, Inc
Lines: 22

Sure, filters can shift phase by as many degrees as they like.
Filters which do this as a linear function of frequency with
zero phase shift at zero frequency are called "linear phase"
and pass all signal components with equal delay related to the
slope of this phase shift characteristic.  If the filter also
passes all frequencies with equal amplitude, then it is called
a "delay line".  Whatever waveform goes in comes out exactly
the same sometime later.  As long as this is done equally in
both channels you cannot call it "distortion", as the same effect
is achieved by hitting the "play" button on your CD a few microseconds
later...

One of the characteristics of high-order filters is that they
introduce more and more delay. This is mathematically necessary
in order to more closely approximate an ideal filter, which
of course would be "non causal", i.e., respond before it saw
an input, which of course is impossible in the real world
(net.misc excepted.)  Bessel filters are just much more careful
about making this delay constant over all frequencies than
other filters, and in fact Bessel filters improve with increasing
order while most other filters get worse.
Phil