The machines: a photographic survey
Appendices · reference
Status: Solved as a reference: all 16 frames in research/pictures/ are
indexed and 14 are read; two are not, and what they show is still unknown.
14 of the 16 frames were examined — IMG_4856–IMG_4869. IMG_4870 and
IMG_4871 could not be processed by the vision pipeline, which rejects them even
after an EXIF strip and a re-encode, so they are indexed but unread; they need a
description or a retake.
What the 14 read frames show: the CoPower-88 8088 board (12 frames,
IMG_4856–IMG_4867) and the Osborne mainboard with its Z80 interposer
adapter (2 frames, IMG_4868 and IMG_4869). Two findings are load-bearing
elsewhere: the 8088 board is the Kaypro variant — silkscreen KAYPRO-88,
software default port pair 0x7E (data) / 0x7F (status) — and the Z80 in this
machine is a plastic NMOS Z80A, date code 8408, which is what decides the
level-shifter parts for the interposer (#11).
CoPower-88 8088 board (IMG_4856–4867)
- Silkscreen: “KAYPRO-88”, “(C) SWP 1983”, “Lic. SWP Inc.”, “COPOWER-88 REV 1.8”. ⇒ The user’s board is the Kaypro variant, consistent with software default port pair 0x7E (data) / 0x7F (status).
- CPU: Fujitsu MBL8088 (NMOS 8088, ceramic DIP, 8319).
- Clock: CRYSTEK 16.000 MHz oscillator can → 8284 divides by 3 → 5.33 MHz 8088.
- Boot ROM: 2732A labelled “SWP-8088” (the 8088-side monitor — the firmware that answers the host mailbox; a dump of this ROM would fully specify the protocol. Consider reading it when hardware is accessible — candidate for the reader hardware in #14).
- RAM daughterboard: 16 x Mostek MK4564N-25 (64Kx1, 128 KB) soldered via SIP-style carrier above the main board.
- Glue visible matches 611-0003: 74LS138 x3 (RAS/CAS + port decode), 74LS244, 74LS373/375, 74LS645, 74LS257, 74LS175, 74LS153, 74LS32, MC74HC374 x2, 74LS161/393/174 etc.
- Blue DIP-package IC marked “16-2103 8140” — a chip, not a DIP switch, read off the physical part by the user. Almost certainly an 8-bit magnitude comparator (74LS688-class) doing the A1-A7 port-base compare from sheet 4. ⇒ The host port base is fixed by this device / its hardwired reference, not user-switchable. Consistent with the Kaypro software default 0x7E/0x7F. TODO: confirm the exact part number + how the reference nibble is set.
- Host ribbon: IDC at board edge (to Z80 socket adapter), plus a second 16-pin(?) white Berg-type connector (matches “BERG PINS” RDX read-strobe / expansion note on sheet 4).
- Solder side (4866): single blue bodge wire, otherwise clean.
Osborne mainboard + Z80 adapter (IMG_4868, IMG_4869)
- Mainboard marked “24187A” (handwritten/silk), assorted 74LS, socketed 6116 x2 (video RAM), WD1793? area partially visible.
- The Z80 lives on the CoPower-88 interposer adapter: a small PCB
plugged into the mainboard’s Z80 socket carrying
- a plastic Zilog Z80A, date code 8408 ⇒ NMOS (a CMOS Z84C00 would be marked differently; also 1984 predates common CMOS Z80s)
- 74HCT-series glue (74HCT04 visible) and pads labelled A4 A5 A6 A7 — address lines broken out for the ribbon to the 8088 board’s port decode (sheet 4 A1-A7 compare feeds the blue comparator IC; adapter passes them down)
- ribbon cable from adapter to the KAYPRO-88 board.
- Level-shifting conclusion for #11: this machine is NMOS Z80A (TTL-ish outputs, high input thresholds) ⇒ use 74HCT/74AHCT toward the Z80 bus or 74LVC245 with careful thresholds; per survey (#31) the safe choice remains LVC245 host-side + HCT toward host inputs. Still confirm the other two machines before finalizing.
Open items (need user)
- IMG_4870 / IMG_4871: what do they show? (They follow the CPU-area shots.) Please describe or retake — likely the missing piece for the adapter↔mainboard ribbon routing.
- Which of the 3 machines is this (OCC1 label? board rev 24187A)?
- Close-up of the blue comparator IC (“16-2103”) to read the full part number and confirm the 0x7E/0x7F decode hypothesis.
- Photos of the other two machines’ Z80 areas for the NMOS/CMOS matrix in #29.
Wanted: shot list for the next open-up session
Compiled 2026-10-13, ahead of a session that opens all three machines and the
upgrade boards together. Every row is something the tree is missing, and what
it unblocks. Shoot each as its own frame, well lit and square on, file them in
research/pictures/, and fold the findings back into this doc.
| # | Shot | Why it is wanted | Serves |
|---|---|---|---|
| 1 | UD11 monitor ROM label, each machine | the 2732’s printed label usually names the revision. This is now the only way to identify the oldest machine (its ^D test was never run) and it confirms the other two are 1.44. |
#205, #70, #222 |
| 2 | UA15 character-generator label, each machine | the 2716 in the char-gen socket. Two images circulate and which one a machine carries tracks its BIOS revision. | #70, #24 |
| 3 | Mainboard revision silkscreen, each machine | one board is marked 24187A; the other two are unknown, and #39 cannot close without them. |
#39, #29 |
| 4 | Z80 chip markings on the other two machines | NMOS vs CMOS decides 74AHCT vs 74LVC in the level shifter. One machine is a plastic NMOS Z80A, date code 8408. | #39, #11, #12 |
| 5 | The Z80 socket area, interposer fitted and removed | the interposer’s mechanical fit, pin-1 orientation and ribbon routing. IMG_4870/4871 were meant to be this and failed the vision pipeline. | #12, #11, #39 |
| 6 | The character-generator socket and its pin-1 orientation, with the wiring around it | gap 3 of docs/research-gaps.md — “char-gen socket pinout” — is solved from the schematic (docs/o1-mainboard-schematic.md §4a read UA15 off sheet 4 of 9), so this shot confirms a pin-1 orientation in a real machine rather than unblocking anything. |
#24 |
| 7 | The SCREEN-PAC board, fitted and loose | its connection into the video circuit, and its own silkscreen. | #24 |
| 8 | The double-density upgrade board | which machine carries one, and its part number and silkscreen. | #95, #70 |
| 9 | The video RAM and attribute chips | the two 6116s and the bank-3 attribute device, for the video section of the memory map. | #97, #40 |
| 10 | The P8 floppy connector and the drive interface | connector and cable detail the schematic sheets leave schematic-level. | #62, #20 |
| 11 | The power-supply board | the tree holds the PSU’s own repair manual — research/osborne1/o1astecpsac8151.pdf, the Astec AC8151 (#52), with its test set-up and the +5/+12/−12 rail table — but no supply schematic and no board-level view, which is what a photograph supplies. |
repo archive |
| 12 | A lid-off board-stack view of each machine | #39’s own “board-stack overview” item, and the frame that makes every other shot locatable. | #39, #9 |
One line of description per frame is worth the minute it costs: two of the
existing sixteen (IMG_4870, IMG_4871) could not be processed by the vision
pipeline even after EXIF strip and re-encode, and a description is what saved
them from being unreadable and unidentifiable.