Skip to content

Design review — control paths and I/O (sequencer logic/memory, IO, bus tester, backplane, video) — 2026-09-21

Section 6 of the 2026-09-21 control-paths and I/O review: the video card v1.1, findings 6.1-6.5. The other cards' sections ("above" and "below" in the original), the cross-card notes and the summary: hardware/reviews/control-io-2026-09-21.md.

Second pass after the mechanical one in mechanical-2026-09-21.md. Read from the KiCad netlists (reports/netlist.net, pin types from the Eagle symbols), chip values from the Eagle boards, the firmware that drives the ATmegas (embedded/bus-tester/bus-driver, embedded/sequencer-card/sequencer4) and the microcode generator (firmware/microcode/ucode-generator2). Gate pin-outs were checked against the symbol pin functions (I0/I1/O for 7400/08/32, O/I/I for 7402; 74175/74/192/244/374 pin functions as printed in the netlist).

Every finding: the trace (ref.pin -> net -> ref.pin, chip types), the failure scenario, severity (HIGH = stops the card / damages parts, MED = unreliable or latent, LOW = housekeeping), and how to confirm it on the bench. Already-known items (BACKLOG.md, video README) are not repeated except where a new detail changes them. Report only; no redesign.

Scripts used (not committed): a netlist dumper (parts + nets with pin function/type), an SV1/SV2 pin-by-pin comparator, a bus-driver-table-vs-tester-wiring check and a microcode-bit-vs-RAM-wiring check; the results of the three checks are in the "checked, no issue" lists below.


6. Video card v1.1 (hardware/cards/video/kicad/v1.1) — beyond the README / fix-6845-register-select.md items

6.1 HIGH (untested path — no 6845 fitted yet) — the 6845 E clock is generated by an RC one-shot whose capacitor has no charging path except a TTL input

Trace: -MEM-RD (B23) -> IC26 74LS86 p9, -MEM-WR (B24) -> IC26 p10, IC26 p8 = N$1 = "a memory strobe is active" (not qualified by BOARDSEL — every memory access in the machine, fetches included). N$1 -> IC1 74ALS08 p1 and -> IC27 7416 p13; IC27 p12 (open collector) = N$3 -> R1 10K -> N$4 = IC1 p2 with C1 (value "1", C025 ceramic) to GND. IC1 p3 = E -> IC17 6845 p23. There is no pull-up on N$3 or N$4.

Behaviour: at the start of every strobe the 7416 turns on and discharges C1 through R1 (τ = 10 µs if C1 = 1 nF), so E = strobe until N$4 crosses the ALS08 threshold — a one-shot, presumably so that a single-stepped access still gives the 6845 a bounded E pulse. When the strobe ends the 7416 releases, but the only current that recharges C1 is IC1's input current (74ALS08 IIL ≤ 0.1 mA, i.e. ≥ 14 µs per volt into 1 nF; 14 ms/V if C1 is 1 µF). At run speed memory strobes arrive every few µs, so N$4 sits at or below threshold and E is short, ragged or absent — the CRTC's registers cannot be written even after the RS fix. The planned "pull-ups on the 7416 outputs" must include IC27 p12; with a pull-up on N$3 the R1/C1 network becomes a real one-shot whose width is set by R1·C1 (the MC6845 wants E ≥ 450 ns; 280 ns for the 1.5/2 MHz grades), so C1's value must be confirmed. Confirm (no 6845 needed): run any loop, scope IC1 p2 (expect it stuck below ~1.5 V) and IC1 p3.

6.2 MED — Character clock is a self-clear runt; the 74LS166 may never load

IC28 74HC160 (asynchronous -CLR): CLK p2 = DOTCLOCK (SV4), A–D = GND, LD/ENT/ENP = N$73 (R10 1k to VCC). CHARCLOCK = IC19 74LS00 p8 = NAND(IC28 QA p14, IC28 QC p12) — low only while the count is 5 — and it is IC28's own -CLR p1: the low lasts one clear-propagation (~30–60 ns) and the counter divides by 5. CHARCLOCK also clocks IC22 74LS175 p9 and IC23 74LS174 p9 (rising-edge, fine) and is IC24 74LS166 SH/-LD p15, which is sampled only on DOTCLOCK rising edges (IC24 p7): the parallel load happens only if the runt is still low at the next dot clock edge, i.e. only for dot clocks faster than ~16–20 MHz. With a slower crystal (Q2 value blank) the shift register never loads and the video output is blank. A synchronous-clear counter (74HC162) holds the state for a full period. Confirm: scope IC19 p8 against IC28 p2; look for glyph bits on IC24 p13 (QH).

6.3 MED — 74HC inputs driven by 74LS outputs

IC28 (74HC160) CLK p2 comes from IC20 74LS04 p6 or p8 through SV4, and -CLR p1 from IC19 74LS00 p8. LS VOH is 2.7 V minimum (2.4 V under load) against the HC VIH of 3.15 V at 5 V: works on typical parts (LS outputs idle near 3.4 V), not by specification. 74HCT160 or pull-ups on N$57/N$59 and CHARCLOCK. Confirm: measure the high level at IC28 p2.

6.4 LOW — the JP1 read-back latch is enabled by address alone

IC2 74LS373 -OE p1 = N$16 = IC27 7416 p10 (open collector, no pull-up — known) = NOT(IC1 p8), IC1 p8 = AND(BOARDSEL·A11 (IC1 p6->p9), ADDR0 p10). No -MEM-RD term: a write to $D401/$D403 turns IC2 onto DATA0–7 against the writing card. Software must never write odd addresses in the upper half of the block; worth a line in the fix document.

6.5 LOW — MC6845 LPSTB (IC17 p3) floating

An NMOS input; tie low. MA12/MA13/RA3/RA4 unconnected outputs are fine.

Checked, no issue (video)

  • Board select: IC18 74S85 A0–A3 = SV3 jumpers with RN2 pull-ups (RN2 p1 = VCC in v1.1), B0–B3 = ADDR12–15, A=B_in p3 = VMA (IC20 p13->p12), AB_in = GND; $D000 = only SV3 5–6 fitted, as the README says.
  • VRAM IC15 right port: -CER p47 = IC19 p6 = NAND(BOARDSEL, /A11 (IC20 p11->p10)); -OER p43 = -MEM-RD; R/-WR p46 = -MEM-WR; A0R–A10R = ADDR0–10, A11R = GND. Left port -CEL/-OEL = GND, R/-WL = VCC (read only), A0L–A11L = 6845 MA0–11.
  • 6845: -RES p2 = -RESET (active low, correct); R/-W p22 = -MEM-WR (low = write, correct); CLK p21 = IC28 QC; -CS as documented. IC2's D inputs from JP1 with LE p11 from JP1 p13 (transparent when open).
  • Character generator IC25: -CE/-OE = GND; A0–A2 = RA0–2, A3–A8 = IC23 Q6..Q1 = VDATA0–5 (64-character set), A9–A11 = GND; O0–O2 unused, O3–O7 -> IC24 D–H with A/B/C/SER = GND (5-dot glyph in an 8-dot cell). VDATA6 (IC15 p22) intentionally unused; VDATA7 = inverse attribute into IC26 p2.
  • Pipeline IC22 (cursor XOR attribute, DE) and IC23 (character code) both on CHARCLOCK, CLR from N$73 (high).
  • Sync/video mix: IC27 p4/p6 (HS/VS) wired-OR on N$85 with R12 270 to VCC — the one 7416 net that does have a pull-up; video IC27 p2 through R11 33 Ω; RCA ground.
  • Dot clock: Q2 Pierce oscillator on IC20 gates A/B with R8/R9 1k, buffered by gate C; IC21 74LS90 ÷2 (QA) with R0/R9 inputs grounded; SV4 picks osc or /2.
  • IC1 74ALS08 gate assignment re-derived from the netlist: E = p1·p2; N$17 = BOARDSEL·A11; N$6 = N$17·A0; N$2 = A11·/A0 — consistent with the README's -CS analysis.
  • +5V/VCC merge and RN2 = 10k: as recorded in the README.

Last updated 26 October 2026