Some notes on TIFAX

diff --git a/NOTES b/NOTES
index 78fdb3efb859010086bc6db4e28e9ba4f24344d0..f573325f92bf463cc6412c84eda5c961ce292e81 100644 (file)
--- a/NOTES
+++ b/NOTES
@@ -257,3 +257,36 @@ P 𜹇  Q 𜹈  R 𜹉  S 𜹊  T 𜹋  U 𜹌  V 𜹍  W 𜹎
 X 𜹏  Y 𜹐  Z ▘  [ ▝  \ ▖  ] ▗  ^ ▀  _ ▌
 
 Looks like Unicode has somehow missed out one of them.
+
+TIFAX
+=====
+
+The TIFAX XM11 was a 1975 Teletext decoder board made by Texas
+Instruments Ltd.  It used a 5 × 9 character matrix with rounding like
+an SAA5050, but implemented over rather more chips.  Unlike the
+SAA5050, the XM11 had two black pixels horizontally between
+characters, so the total character size was 7 × 10.
+
+https://www.blunham.com/Radar/Teletext/PDFs/XM11-B183.pdf
+
+The character ROM is an SN74S262, and the character shapes mostly
+match the SAA5050, so treating it as a variant might not be
+unreasonable.  There is some uncertainty over some characters as
+explained by Neil Williamson: https://p298.net/devnotes/
+
+The different character pitch could mostly be accomodated by having
+the application add extra spacing between characters, but what about
+mosaic graphics?  The XM11 generated those in a separate chip (the
+X908), and they were the full 7 pixels wide, four in the left column
+and three in the right.  Would we want an entire extra set?  At least
+the XM11 didn't support separated graphics (added in the 1976 spec).
+And what about box-drawing?
+
+The XM11 has a dot clock the same as the Teletext bit clock: 6.9735 MHz.
+Hence a nominal XPIX of 105.8.
+
+There was also a Swedish ROM, SN74S263, which was used in the ABC80, a
+Swedish Z80-based computer, so there is a purported ROM of it
+available for emulators.
+
+Also some real screenshots; e.g. https://www.youtube.com/watch?v=Jy6tQsyd4ho