Rich and I have switched to using a PIC16C622, eliminating the LM393. The new version should be available here soon. More details are available below.
The distribution now contains a theory of operation document and a schematic in HPGL format. Unfortunately the schematic capture package Rich uses doesn't generate very good HPGL, and it doesn't generate PostScript at all. For the next release I will redraw the schematic in Protel Advanced Schematic and provide it in both HPGL and PostScript.
Dan Gregson has provided a schematic in JPEG format. Thanks, Dan!
Because the field output of the LM1881 was particularly unreliable, we ended up using only the composite sync output of the sync separator, and deriving the horizontal sync, vertical sync and the field in software. The Elantec part probably generates perfectly good sync and field outputs, but it would be more trouble than it is worth to switch back.
Rich converted the code to run on a PIC16C56. The only advantage of the '56 is that Rich had a Parallax downloader for the '5x family, and that the '56 is cheaper than the '71. Microchip offers a '61 (which is a '71 without the A/D converter), which is priced fairly similarly to the '56. Rich points out that the '56 version of the code isn't very maintainable and the RAM is completely full.
Microchip's recently announced PIC16C62X series parts are an even better fit for this application, since they have two built-in comparators which replace the separate LM393. Rich and I recently switched over to using the PIC16C622, although a PIC16C621 would work as well. All future development and release of this design will use the '62X.
Since the '62X is available in as 20 MHz speed grade, the latest code is able to take nominally five samples per data bit (vs. four at 16 MHz), so in principle the reliability of the data recovery can be improved.
A much larger and more complicated design using an 8031.
A "must-read", this is a very good article regarding capturing digital data from video signals.
Mark's design is also 8031 based and uses 24 chips, but provides much more functionality. In addition to closed captioning it also decodes network time stamps, World System Teletext (WST), and North American Basic Teletext Specification (NABTS).
An earlier PostScript version of the article with additional information on WST and NABT, and source code including new support for VITC is available by FTP from Circuit Cellar Ink in VBI.ZIP.
Electronic Industries Accociation
2500 Wilson Boulevard
Arlington VA 22201
American National Standards Institute
11 West 42nd Street
New York, NY 10036
This 132-page document may be ordered for US $114 (plus shipping, handling, and tax) from Global Engineering Documents, a division of Information Handling Services.
Unfortunately, aside from Part Two (Line 21 Waveform, 3 pages), this standard doesn't give any information regarding the standard closed-caption data encoding. The rest of the standard is only concerned with extensions for fancier captioning and other data services. They make a few vague references to the NCI, PBS, and FCC standards but don't actually give specific citations.
Public Broadcasting Service
1320 Braddock Place
Alexandria, VA 22314
National Captioning Institute, Inc.
5203 Leesburg Pike
Falls Church, VA 22041
An online, searchable copy of the CFR has been made available on an experimental basis by The U.S. House of Representatives Internet Law Library. Unfortunately at this time it is limited to text searches only, so it is hard to use.
Last updated November 22, 1998
Copyright 1995, 1996, 1998 Eric Smith