David Anderson opened with a report on the archive at timexsinclair.com. For about six weeks he has had Claude work through the scanned newsletters and magazines one issue per week, using a vision model rather than archive.org’s OCR. For each issue it transcribes the articles, matches them to the existing stubs in the site’s database, tags people, crops illustrations into the pages, and types in the program listings. It tests the programs in an emulator, fixes characters lost to bad copies, captures screenshots, builds TAP or P files, uploads them to archive.org and creates computer media entries on the site. It had just finished Sinc-Link, the Toronto user group newsletter, which ran from 1982 to 1994.
Of the roughly 16,000 articles listed on the site, 6,400 are now full text, drawn from almost 1,900 issues of 174 periodicals. About 2,600 programs are listed, and 1,500 of them can be downloaded and run in the browser.
David also showed an Elecrow seven-inch ESP32 touch tablet with a small Bluetooth keyboard. Claude has ported Fuse and the TS-Pico software to it.
Jon Becker went through his R-Type port for the TS2068 in extended color mode. He picked the game because it shows up so often on lists of the best Spectrum games. His AI pipeline disassembled the Spectrum executable, pulled the original arcade graphics, matched them to the Spectrum assets and converted them to extended color with dithering. It then gave him a web page for adjusting each conversion by hand. The AY music comes from the Spectrum 128 version, and some sound effects come from the arcade through his earlier Speech-to-AY project. The game uses the TS-Pico’s 512K of RAM. It can load one level at a time or stage the whole game, about eight cartridges’ worth. The work took about a week and a half. It runs slower than the Spectrum original, mostly because of extended color mode.
While working on R-Type, John found bugs in PicoVideo. Codex located and fixed problems with extended color mode and some other modes, and that fix has been submitted back to the project.
Jon’s video clips and ZX Desk both want much more memory, so David had Claude design a four-megabyte RAM board for the 2068. It uses the dock banks the way Timex planned for the never-released bus expansion unit: up to 256 banks of 64K, each split into 8K chunks, set up through the system configuration table.
David has also redesigned the OpenSpand for the TS1000/ZX81 as an all-surface-mount board, because the original takes a lot of soldering. The new one is long and low like a Memotech pack rather than tall like a 16K RAM pack. It started out with composite video, then VGA. On Tim Horner’s suggestion it now uses the Pico 2’s built-in high-speed serial to output DVI through a micro-HDMI connector. It emulates the AY chip and keeps the joystick port, SD card and real-time clock. The only through-hole parts are a few switches, the joystick port and two LEDs.
Paul Anderson has his Larken disk system working after three days on hardware, software and, in his words, operator error. One of his two Larken cartridges turned out to be bad, and the one that came from Frank Davis works. TASword II runs. When he tried to add a 3.5-inch drive he blew a power supply, so he moved the working 5.25-inch drive into a single enclosure. He now has a large stack of 5.25-inch disks he can read. He also has his Oliger A and B boards, three 2068s, a four-slot expansion and a Kempston joystick interface he once used to play Starion to the end. David’s site has many of the Oliger manuals.
That led to GoTek floppy emulators. David has a GoTek and Willie Jones’s recreated Oliger boards but has not tried them together yet. Joe said a GoTek can emulate almost any Shugart-style drive, including hard-sectored ones, if the configuration file matches the hardware. Joe Vandezande has them in an Osborne and an Amstrad CPC running CP/M from emulated three-inch disks. Joe said a board exists for using a Commodore 1541 with the TS1000 but it is a resource hog, and he knows of nothing for the 2068.
The rest of the meeting was about printers. The printer in David’s background is the TS2080 from 1983 publicity photos, believed to come from Mannesmann Tally. Paul found a different “Timex 2080” printer on Johnny Red’s site. Tim said the Sinclair-branded QL printer was a serial model from the same family. David showed a Sinclair SP200 printer that Jack Boatwright sent him, made for the Amstrad-era Sinclair PC200. Josef described the Czechoslovak BT100, a single-pin printer that used carbon paper instead of a ribbon and took about half an hour per page. Hobbyists timed its slot wheels against CPU cycles to get finer output. He also showed a kit-built plotter driven by nylon strings, a telephone relay for pen lift and about nine transistors wired straight to a computer port.
