NetNews Usenet Archive 1992 #31/NN_1992_31.iso/spool/rec/audio/17631

Internet Message Format | 1992-12-31 | 3.2 KB

Path: sparky!uunet!noc.near.net!news.bbn.com!NewsWatcher!user
From: [e-mail] (Kerry Shetline)
Newsgroups: rec.audio
Subject: CD/LD jitter: I certainly do get it! (Was: Kerry doesn't get it...)
Followup-To: rec.audio
Date: 1 Jan 1993 03:09:51 GMT
Organization: BBN
Lines: 52
Distribution: world
Message-ID: <shetline-311292220109@128.89.19.85>
NNTP-Posting-Host: bbn.com

PART I
------ (Something in this message is poison to my ARA modem link, and keeps
        causing me to lose my connection when I send it. I'm trying it in
        pieces to see if that helps. Aaaaargggh!)

In article <[e-mail]>, [e-mail] (Norman
P. Tracy) wrote:
  ...
> NOT. Try this. Open any CD/LD player or transport. Now point to the rows of
> SIP-RAM modules containing the mega-bytes of buffer memory.  You can not, they
> are not there.  You object, "the buffer RAM is inside the VLSICs" good we
> agree the amount of buffer is limited.  Now with a limited amount of buffer
> it has to be arranged such that the data written into it and read from it
> never under flows or over flows.  The only way to do this is a servo loop
> that includes the disk drive motor, laser head, and demodulation circuitry
> which contains the much vaunted buffer RAM.  To look simply at one aspect of
> the operation, if the buffer get close to overflow, slow down the disk speed;
> when the buffer gets close to underflow, speed up the disk.  And the data
> being read out of the buffer has to track these variations or the buffer
> will empty or fill up.

(Note: I will for simplicity refer to CD players most of the time, although
this argument applies to LD as well. I am a little less certain about the
exact implementation of LD digital audio as compared to CD, but the digital
subcarrier from a laser disc, once extracted, it is intentionally very
similar (if not identical) to a CD data stream, making the development of
combination units far easier.)

Why on earth does the output have track input variations? One does not need
to control the weather, the oceans, the lakes, and the seas to get a smooth
flow of water from the kitchen faucet. Similarly, you do not need
multi-megabyte buffers to get smooth, jitter-free output from a CD or LD.
You can take a cans of soda out of a vending machine on a completely
regular schedule, regardless of the delivery man's schedule, as long as he
shows up often enough that the machine doesn't empty out. You can take
bytes out of a buffer on a regular schedule too, as long as the buffer
doesn't empty.

CD play does not begin until the disc is up-to-speed and the
synchronization bits on the CD can be properly detected. Once up to speed,
servo motor timing is controlled, in part, by the timing of signal
transitions in the pick-up output. It is this timing which is responsible
for varying the speed of a CD from about 500 RPM at the start of play down
to about 200 towards the end of a disc. If timing goes too far astray, the
disc will simply mistrack with quite audible consequences. However, the
signal-transition-regulated data rate is quite stable by itself. (LD uses a
similar analog-based stablizing system, based, I believe, on the 3.58 MHz
color-burst signal).

(continued...)

-Kerry