Skip to content

CPCSyntaxError v0.8.0

Choose a tag to compare

@WacKEDmaN WacKEDmaN released this 24 Oct 22:01
· 22 commits to main since this release

CPCSyntaxError v0.8.0

This release makes the CRTC, the Plus ASIC, the monitor and the disc drive more accurate. The changes come from the SHAKER 2.7 screens that were compared with photographs of real CPCs and with AmSpiriT.

CRTC

  • All chips: the raster address when C9 is past R9. The video address now uses bits 0 to 2 of C9 itself (ACCC 20.2). Before, a C9 larger than R9 was folded back into the character. This was wrong on lines where C9 runs past R9, for example the R5 additional lines and a C9 that keeps counting with R0=0 (ACCC 13.2.6). SHAKER 0/A7 and the BRETURN screens on all chips are now correct.
  • CRTC 2: R1=0 at the start of a line. The step VMA'=VMA now uses the C9 of the line that starts (ACCC 20.3.3). Before, it used the C9 of the previous line and the next row started 63 characters too far. SHAKER 2/AO/G now shows its row of black and white boxes.
  • CRTC 3 and 4: HSYNC overflow. The ASIC now uses the standard C3 overflow on C0=R2 (ACCC 15.3.1 and 15.3.2). An R3 write on the chip's own C3 count ends the HSYNC (ACCC 14.5.3). SHAKER 3/DH/B1 and B2 are now correct.
  • CRTC 3: VSYNC with R0=0. A VSYNC that is due while R0=0 starts when C0 leaves 0. SHAKER 3/AI now matches the real Plus.
  • CRTC 0: R7 written on C0=R0 counts for the next line (ACCC 16.4). SHAKER 0/B6 now prints GOOD.
  • CRTC 1: R4 and R9 on the last character. The additional management stays on only when there is an additional line (ACCC 11.2.4). SHAKER 1/AU now prints the correct value.
  • CRTC 0: interlace. The deferred IVM test starts only when R8 changes to 3 (ACCC 19.8.1). SHAKER 0/B9 now gives the frame sizes of the real CPC.

Plus

  • Interrupt acknowledge per instruction. The Plus now uses the same acknowledge timing as the CPC. SHAKER 3/BR now prints GOOD.
  • PPI port B bit 7 reads 1 on the 6128 Plus, because the 6128 Plus has no cassette deck.

Monitor and colour

  • The picture width changes gradually after a change of sync width, as on the real CTM photos of SHAKER DH.
  • The sweep drifts during a gap with no HSYNC and then comes back, as on the real photos of SHAKER D8.
  • The jump re-lock is removed. It put the first line of each band in SHAKER D7 two characters out.
  • Colour tube: chroma bandwidth. A small colour filter replaces the old blue lag, so that dithered colours look like the references.

Disc drive

  • phX loads again. Its loader switches the drive motor off and on 5 ms apart and then recalibrates. The drive needed two new turns of the disc after every motor start, so it reported Not Ready and the loader waited for ever with a black screen. Now a motor that comes back on within one turn finds the disc still at speed, as on a real drive.

Other

  • CSL scripts: crtc_select on a running CPC changes the CRTC without a restart, and a letter in a key string is typed without SHIFT.