Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10 5/3/83 based; site houxa.UUCP
Path: utzoo!watmath!clyde!burl!ulysses!mhuxl!houxm!houxa!czp
From: czp@houxa.UUCP (C.PODARAS)
Newsgroups: net.audio
Subject: cd playback sampling delays
Message-ID: <415@houxa.UUCP>
Date: Tue, 15-May-84 14:40:12 EDT
Article-I.D.: houxa.415
Posted: Tue May 15 14:40:12 1984
Date-Received: Wed, 16-May-84 04:17:05 EDT
Organization: AT&T Bell Labs, Holmdel NJ
Lines: 57



a number of people have replied to an earlier article which i posted to the 
net regarding cd recommendations.  in that article, i made the following 
statement:

> if you listen to any of these on a good CD player -- one that creates 
> *simultaneous* (ie, no inter-channel half-sampling-period delay,like 
> the sony cdp 101) channel outputs thru appropriate data buffering 
> and use of dual D/A's, and has some semblance of flat-group-delay-
> characteristic reconstruction filters -- you will (assuming you like the 
> music) be *amazed* at what this medium can sound like. 


to answer some of the questions, here goes...

the standard for cd recording (ie, sony, philips, aes standards committee) 
calls for ** simultaneous ** sampling of audio channels; in the case of 
mastering for cd, this means that left and right are sampled at the same time.
the l and r channel bits are then combined (interleaved) into a single stream 
for recording into the cd medium.  for playback, this serial stream enters 
a buffer, allowing the left and right channel bits to be pulled apart, and 
hence the channels can be played back simultaneously.  following the 
standard, then, will cause l and r output signals to be coincident in time.  

the problem is that some players don't have the buffers.  (philips and marantz 
players, to mention two, are ok.)  some players (ie, sony cdp101) use a single 
d/a converter which switches very quickly from l to r channel, causing an 
11.34 us time lag between channels (half of the 22.68 us sampling period, as 
the converter switches back and forth to cover both channels for each period).
also, some other players use separate d/a's for each channel, but still no 
buffering: they switch the error-corrected output very quickly between 
the converters' inputs during a sampling period, which still causes the 
half-sampling-period lag.

this lag causes a slight high-frequency phase shift, which (pick your side 
of the fence) may be an audible degradation in stereo playback.  however, 
there's no question that problems do exist in mono playback, where (as many 
broadcasters have unfortunately found out) the high-frequency cancellation is 
** very ** noticeable.  the solution thus far has been either to use analog 
delay lines, or to switch to a "good" player.

check this out on your cd player by listening to a *stereo source* (ie, music 
disc: some test discs only have material on one track at a time) with the 
preamp playback mode in mono.  headphones recommended.  result?...bummer.  
don't mean to disillusion you early player owners; just remember there's always
a learning curve in engineering design.

ah, the old engineering "gotcha": get tricky to save $ by scrimping on parts, 
and marketing will love ya.  then murphy comes to town and lays down his law...

	(i'll buy one eventually, when you don't have to tear it apart in lab 
	to make it work)
				chuck podaras
				bell communications research
				holmdel, gnu joizee
				{..., ihnp4}!houxa!czp