Skip to content

TSU Meeting SUNDAY — March 19, 2023

Watch on YouTube

David, Claudius, Carl Miles, Jeff Burrell, Jeff K, Adam Trionfo, Ryan Gray, Marcel, Ada, Gustavo, Joe, Keith, Rich

Claudius, joining from the UK, presented his "Timmy" board — a from-scratch recreation of the Portuguese Timex Computer 2048 (TC2048) main board, designed to drop into the original case and reuse the original chassis connectors. His board strips out the 12V supply (using an AD724JR PAL encoder), drops to two 64K×4 DRAMs for the upper RAM, swaps in an Atari-polarity joystick port, adds an AY-3-8910–equivalent PSG (Winbond/KCF clones, sourced new), exposes a switchable 80K-RAM mode via port 255, and uses a $2 Bluetooth audio module so tape files can be loaded directly from a phone running PlayZX. The board is two-layer, fabricated cheaply in China (~$20 bare), and has been built around five times. Future plans include a USB-serial adapter wired to the PSG I/O lines and possibly Wi-Fi via ESP12.

At the heart of the board is an SCLD replacement designed by Mark Smith (author of the "Slam" ZX Spectrum ULA replacements), implemented first in a Xilinx CPLD and later ported to an Intel CPLD to fit the PLCC-68 footprint. The chip handles the Portuguese TC2068 SCLD's digital functions plus analog video/tape, but does not match the American TS2068 SCLD — pinouts, RGB vs. YUV component-video outputs, and a missing BRIGHT pin on the US part all differ. Macrocell limits also forced one compromise: interrupt timing is Spectrum-accurate rather than the ~16-T-state-offset Timex timing. The replacement has been proven in Mark's faulty TC2068 and in Claudius's clones. Distribution is informal — Claudius can help with PCBs, but the SCLD design is Mark's IP and isn't being mass-produced; CPLDs themselves are now ~$60 each with six-month lead times.

Discussion turned to whether Mark's SCLD could be adapted to the American TS2068, and to whether a full TS2068 replacement board is worth doing — Jeff Burrell noted the 3.3V FPGA / 5V Z80 bus-mastering problem (BUSREQ/BUSGRANT) makes it tricky, and Carl Miles observed that SCLD failures on US 2068s are vanishingly rare, so the case for a new board is weak. Jeff Burrell then caught the group up on his FPGA 2068 effort: the video side is "locked down" with all standard 2068 modes plus ULAplus, a Grant-Searle-style 80-column terminal (for future CP/M), and a TMS9918 (VDP) core; a Raspberry Pi Pico interface handles SD-card loading and USB-serial download from a PC. He agreed it would make sense to spin the FPGA video upgrade out as its own standalone board ahead of the storage features. Separately he's written a Python TAP-to-audio utility to spool programs into the 2068's EAR jack via a USB audio device.

The Albuquerque contingent (Carl, Ryan, Adam) reviewed video-conditioning work on the TS1000/ZX81: the ZX8-CCB composite mod (with tearing issues that worsen as the machine warms — Claudius diagnosed a dying ULA), an upcoming build of zx.zig's NE555-based back-porch board from JLCPCB, and the vLA-81 ULA replacement as the recommended fix. Claudius pointed at the Argentinian/Brazilian TK-Pi (Victor Trucco) as an HDMI-output route for the Spectrum/Timex world. The group also touched on TC2068 vs. TS2068 cartridge-port incompatibility (different signals, not just physical), the Unipolbrit Polish/UK joint-venture 2048 with its 8255 Centronics port and modified ROM, the early ZX81 ROM math bug (squaring) fixed before the TS1000, and the Spectrum "dead cockroach" interrupt fix. Marcel from Montreal introduced himself as a new attendee asking about the Harlequin. David announced the next meeting (Monday, April 3, 9pm Eastern) featuring Robert Fischer and the Profile 2068 / RMG enterprises cartridge edition.

Highlights

  • Claudius demoed "Timmy" — a from-scratch TC2048 clone board with AY sound, Atari-polarity joystick, Bluetooth tape input, and a switchable 80K RAM mode
  • Mark Smith's CPLD-based SCLD replacement works in European TC2068/TC2048 but not the American TS2068 (different pinout, RGB vs YUV, no BRIGHT pin)
  • Jeff Burrell's FPGA 2068 video is "locked down" — all standard modes plus ULAplus, an 80-column terminal, and a TMS9918 (VDP) core; group urged splitting the video upgrade into its own board
  • Diagnosis: tearing on the ZX8-CCB composite mod that worsens as the machine warms points to a dying ZX81 ULA — vLA-81 recommended as the replacement
  • Next meeting Mon April 3 — Robert Fischer on Profile 2068 and the RMG Enterprises cartridge edition

Projects discussed: FPGA "Advanced 2068" Topics: Timex Computer (European: TC2048/3256) · Hardware projects & new boards · Video upgrades (composite, RGB, HDMI) · TS2068 / TC2068 · TS1000 / ZX81 / TS1500 · Pico / modern interfaces (UnoDos, etc.)

Summary generated from the meeting recording; the full transcript is kept in the group's private archive.