Design review, datapath cards (control-path pass) — 2026-09-21¶
Second pass after mechanical-2026-09-21.md (mechanical). Cards: ALU v3.2, Index Register 1.1, Memory v1.3, Mem Switch 1.1,
Mem Register 1.0. Method: every select / enable / strobe / clock / reset net was traced pin-to-pin from
kicad/<ver>/reports/netlist.net (pin names and types from the Eagle symbols) with chip types from the Eagle .brd,
and checked against the way the sequencer microcode actually drives the bus (firmware/microcode/ucode-generator2/),
because several hardware choices only work under a microcode convention. Signal polarity: - prefix = active low.
Gate delays quoted are LS-TTL typicals from the datasheets in docs/datasheets/.
Severity: HIGH = would stop the card working or damage parts; MED = unreliable / depends on a convention that is not enforced by hardware; LOW = housekeeping or documentation.
The cards' findings are in each card's own reviews/ folder:
| Card | Findings | File |
|---|---|---|
| Memory v1.3 | M1-M8 | hardware/cards/memory/reviews/v1.3-datapath-2026-09-21.md |
| Index Register 1.1 | R1-R3 | hardware/cards/register/reviews/v1.1-datapath-2026-09-21.md |
| ALU v3.2 | A1-A3 | hardware/cards/alu/reviews/v3.2-datapath-2026-09-21.md |
| Mem Switch 1.1 and Mem Register 1.0 (one section) | B1-B2 | hardware/cards/mem-switch/reviews/v1.1-datapath-2026-09-21.md |
| Mem Register 1.0 | (points to the Mem Switch file) | hardware/cards/mem-register/reviews/v1.0-datapath-2026-09-21.md |
This page keeps the parts that belong to no single card: the system-level items S1-S3 and the summary.
System-level items found while tracing (not on the five cards, recorded so they are not lost)¶
S1 — MED: no power-on reset anywhere; -RESET is a manual RS latch¶
Sequencer-logic v2.1: RESET = IC32.4 (74ALS02) = NOR(FP-RESET [switch RESET0 via R7], RUN); RUN = IC32.1 =
NOR(RESET, FP-EXECUTE); -RESET = IC36.6 (74LS04) = NOT RESET. There is no RC or supervisor. At power-up the
latch settles arbitrarily, so until the button is pressed the memory card's FORCE-ROM (IC12A PRE), all sixteen
74LS192 register counters (CLR = RESET), and the ALU carry/shift flip-flops (IC9 CLR) are undefined. Bench: meter on
C30 after a cold power-up, before touching the front panel.
S2 — LOW: reset polarity is consistent across cards¶
Bus -RESET (active low, totem-pole from IC36.6): memory IC12A PRE (active low), ALU IC9 CLR (active low), register
IC34 XNOR(GND, -RESET) -> active-high CLR of the 192s. No mismatch.
S3 — LOW: control-line idle state is "floating" bus-wide¶
See M5. Every control line on rows B/C is a 374 output with OC = -BUS-EN; no card and not the backplane pulls them.
The memory card is the one that could lose data from it (M2, M5).
Summary¶
| ID | Card | Sev | One line |
|---|---|---|---|
| M1 | Memory | HIGH (masked) | FORCE-ROM clock = ADDR15 AND VMA races the register card's address drivers; clears on the first cycle whenever the address bus floats. Hidden today by the microcode holding -VMA in every line, which also disables all -VMA gating of the chip selects. |
| M2 | Memory | MED | 28C64 -WE = -MEM-WR, unprotected; during FORCE-ROM every address selects the ROM block, so a boot-time store writes the EEPROM. Shipped monitor avoids it (BR first). |
| M3 | Memory | MED | TMP registers latch on the leading edge of -TMP-REG-LDn; depends on the microcode's source-one-line-early convention; moveRegtoTmp breaks it (unused). |
| M4/B1 | Memory + bring-up | MED (config) | Low RAM ($0000-$7FFF) cannot be jumpered out; with a bring-up card fitted (or FORCE-ROM after reset) two or three drivers on DATA0-7 at $0000. |
| R1 | Register | MED | CD4077 CMOS thresholds (VIH 3.5 V) on LS-driven -RESET, -RDSEL, -LDSEL, -HL-SWAP; ~100 ns in the 245 enable/direction path. |
| R2 | Register | MED | UP/DN = OR(select, -REG-UP): deselecting while -REG-UP is low counts the deselected register; safe only under the microcode ordering the author flagged. |
| S1 | Sequencer | MED | No power-on reset; FORCE-ROM, registers, carry undefined until the button. |
| A1 | ALU | LOW/MED | BR-COND floats when JP1 = -ALU-FUNC and -ALU-FUNC is high. |
| M5/S3 | Memory | LOW | All bus control inputs float (374 OC = -BUS-EN) before READY / on handover; no pull-ups anywhere. |
| M6, M7, M8, R3, A2, A3, B2 | various | LOW | Open LS04 inputs; JP1 must be fitted; RN5/RN6 value unknown; no same-card swap; -AC-LD-INV also flips BR-COND; ns-scale internal overlaps; 244 vs 240 documentation. |
Checked and clean: memory write-strobe ordering and 62256 -OE/-WE relation, block-jumper decode, FORCE-ROM uses raw ADDR15; register-card bus direction (the 1.2 open question is answered: correct as drawn), swap path exclusivity, load timing, address drive, 192 cascade; ALU accumulator/carry/borrow/carry-in logic, comparator cascade seeding, shift register and shift-out, 4078 output polarity, transceiver enables; bring-up cards' read/write qualification and decode.
Last updated 26 October 2026