Draft. Draft — assembled from the working notes by `tools/build_book.py`. The source document is the authority; this page is a copy for reading, and it is regenerated from that source.

This directory is the project’s research archive: the documentation, disk images, and photographs that the reverse-engineering and design work under docs/ and the GitHub issues depend on.

These files are reserved as safety copies of the documentation necessary to complete this project. They are committed in-repo on purpose: upstream archives move, go offline, or drop files, and this project cannot finish without these specific documents. If an upstream link dies, the in-repo copy is authoritative. Please do not “clean up” large PDFs or disk images from this tree — their presence here is the point.

Where each item was sourced is recorded below, so provenance is preserved alongside the copies.

OCR convention: most of the scans here have no text layer. When a document is OCR’d for research, the text is committed adjacent to the PDF as <filename>.ocr.txt (e.g. drive_c/Drive_C_Users_Manual.ocr.txt) so it never has to be regenerated. Regenerate with tools/ocr_pdf.sh if a better pass is needed.

osborne1/ — Osborne 1 mainboard documentation

File What it is Sourced from
2F00153-01_Osborne1TechnicalManual_1982.pdf (19 MB) Osborne 1 Technical Manual, 1982. No text layer — image-only scan. Theory of operations, banking, I/O, video, disk, keyboard. bitsavers
2F00040_service_2ndEd_1983.pdf (12 MB, 292 pp) Osborne Field Service Manual, 2nd Edition, 1983 (part no. 2F00040-00) — the whole-machine service manual: board-level theory, parts lists, troubleshooting and the schematic set. The general-purpose companion to the Technical Manual above, and the source several docs/ pages already cite by part number. §6.12 documents the ScreenPac IC-harness fly-lead solder connections (used by firmware/screenpac-video/); it also backs docs/floppy-adapter-design.md (drive A’s 150 Ω terminator, the P8 male header, DS1 strapping) and docs/o1-memory-io-map.md. bitsavers
2F00040_service_2ndEd_1983.ocr.txt OCR text layer (per the convention above), generated 2026-10-09 —
OCC1_1A2011-00_Schem_RevE.pdf (1.2 MB) Two drawing sets in one 12-page scan. Pages 1–3: DISK ELECTRONICS, DWG 1A3004-00/01 Rev C (sheets 1–3 of 3 — the drive’s own board). Pages 4–12: MAIN PC BOARD, DWG 1A2011-00 Rev E (sheets 1–9 of 9). 600 dpi bilevel; the embedded text layer is a useless early-90s OCR pass. Indexed, and the Z80 socket and P8 floppy pinouts extracted, in docs/o1-mainboard-schematic.md (#62). bitsavers (also mirrored as Osborne_1_Schematics_1A2011-00_Rev_E.pdf on RetroTechCollection/Internet Archive)
OCC1_1A2011-00_Schem_RevE.ocr.txt OCR text layer for the above (per the convention below), generated 2026-10-10 at 400 dpi. Good enough to grep for component references; not reliable for pin-level detail — read sheets visually with tools/schematic_render.sh. —
keyboard/osborne_keyboard_pinout.html + keyboard/PROVENANCE.md Verbatim capture of the pinouts.ru Osborne 1 keyboard page: P4 pinout (incl. pin 19 = +12V via J6+R21) and the full key matrix map. Used by #45/#47. pinouts.ru, captured 2026-10-08 (provenance in the folder)
floppy-adapter/ (README, LICENSE, .sch, .kicad_pcb, PROVENANCE.md) Snapshot of WayneVisser’s Osborne1FloppyAdapter KiCad project — passive Gotek adapter documenting the O1’s “Shugart-like” floppy interface (power over the cable’s unused pins). Used by #20/#49. github.com/WayneVisser/Osborne1FloppyAdapter, captured 2026-10-09 (provenance in the folder)
modem/o1techman-interface-extract.pdf (1 MB, 10 pp) Chapter 4 “Osborne 1 Interface Design” extract of the O1 Technical Manual: §4.1 IEEE-488 (P3), §4.2 RS-232 (P2), §4.3 MODEM (P1). Authoritative source for the MODEM DE-9P pinout — TTL levels, not RS-232 — incl. pin 4 MSB, pin 5 CTS, pin 8 MCB, pin 9 RI, pin 7 +12V via 22Ω. Used by #56 (and #21/#12). Also the P3 (IEEE-488) pinout: §4.1.2’s 26-pin edge table, §4.1.1’s signal directions and §4.1.5’s Centronics re-mapping, transcribed in docs/o1-ieee488-port.md §6 (#84). Image-only scan — OCR it, pdftotext returns nothing. kev.pulo.com.au/osborne1, captured 2026-10-09
o1doubledensityusersguide.pdf (2.2 MB, 15 pp) Osborne 1 Double Density User Manual. The DD option’s own manual, and the primary source for what it actually does: it is a multi-density option that recognises six formats, not just a denser one. §”formats automatically recognized” is quoted in docs/dd-upgrade-formats.md (#95). Image-only scan; OCR’d 2026-10-10 at 300 dpi as o1doubledensityusersguide.ocr.txt. kev.pulo.com.au/osborne1, supplied by the maintainer 2026-10-10
Comm-Pac_Users_Manual.pdf (3.1 MB, 43 pp) Osborne Comm-Pac User’s Manual — part no. 2F-00145-00, “Osborne Datacom Modem + AMCALL Communications Program” (AMCALL is MicroCall Services, 1982; manual by Joseph Caggiano). The modem package’s own manual: UART configuration, file transfer and its spec, Touch-Tone dialing, X-ON/X-OFF protocol, voice communication, diskette buffering, UNLOAD.COM, HELP menu. Used by #21 and #56. Has a text layer (Adobe Paper Capture, 2009), so it is greppable as-is and carries no .ocr.txt. bitsavers bits/Osborne/Osborne1/Comm-Pac/ (byte-identical)
o1astecpsac8151.pdf (3.4 MB, 11 pp) “Repair Manual for Astec Power Supply AC8151” — the Osborne 1’s own PSU. Test set-up (isolation transformer ≥500 VA, 0–28 V variac, DMMs, scope with X10/X100 probes, load board), the 115/230 V input-select wire, and a performance table for the +5/+12/−12 rails at 190/220/270 VAC. Image-only scan; OCR’d 2026-10-14 at 300 dpi as o1astecpsac8151.ocr.txt. Serves #52. kev.pulo.com.au/osborne1 (byte-identical)
o1astecpsac8151.ocr.txt OCR text layer for the above (per the convention above), generated 2026-10-14 at 300 dpi —
o1md1000-190specs.pdf (4.3 MB, 11 pp) Model MD1000-190 specifications (preliminary) — the stock Osborne 1 5-inch monitor: 127 mm diagonal, 3.80 × 2.54 in display, 55° deflection, 650 lines centre / 500 corner, 10 Hz–12 MHz bandwidth, 15.75 kHz horizontal and 50/60 Hz vertical sync, TTL sync drive, positive-white video, +12 VDC ±0.5 V. Also carries service notes (CRT 2nd-anode discharge) and a schematic. Image-only scan; OCR’d 2026-10-14 at 300 dpi as o1md1000-190specs.ocr.txt. Serves #24 and the mechanical work. kev.pulo.com.au/osborne1 (byte-identical)
o1md1000-190specs.ocr.txt OCR text layer for the above (per the convention above), generated 2026-10-14 at 300 dpi —

Basis for docs/o1-memory-io-map.md (#40) and docs/o1-mainboard-schematic.md (#62). Both scans are image-only; extraction is manual — OCR for text pages, and for schematics a native-resolution region render (tools/schematic_render.sh) read by eye. The mainboard map is independently cross-checked against the running machine in MAME; see docs/o1-mainboard-schematic.md §7 (#64).

Two further scans offered on 2026-10-14 are deliberately not here, because the archive already holds them: o1screenpacusersguide.pdf is byte-identical to screenpac/o1screenpacusersguide.pdf, and 2F00040-00_Service2ndEdition_1983.pdf is a 4.4 MB re-compression of screenpac/2F00040_service_2ndEd_1983.pdf — which is the pristine 12 MB bitsavers copy and carries the OCR sidecar, so it is the better of the two. See #245.

inspired-by/ — adjacent projects

inspired-by/README.md — other Osborne 1 reinventions (MAME cyberdeck, Raspberry Pi brain transplant, Loxley’s Gotek mod): what they did, and what we take from each. Contrast defines the project: we keep the original machine intact.

mx80/ — Epson MX-80 printer references

File What it is Sourced from
mx80/Epson_MX-80_Operations_Manual.pdf Epson MX-80 operation manual; Appendix 4 (PDF p. 97+) has the character-font tables (6×9 matrix) and ESC code reference. Internet Archive / manualsbase, captured 2026-10-09
mx80/EPSON_MX-80_Fonts_v1.0.zip + mx80/fonts/ Michael Walden’s MX-80 font pack (CC BY-NC-SA 4.0). Convenience cross-check; see the folder README for the double-strike/2×-width caveat. mw.rat.bz/MX-80, captured 2026-10-09

Also: drive_c/Drive_C_Users_Manual.ocr.txt — OCR text layer for the image-only Drive C manual scan (generated 2026-10-09 per the OCR convention above; source of the TurboPac details used by firmware/drive-c-gpib/).

References for the virtual Epson MX-80 printer personality (#19).

screenpac/ — OCC ScreenPac (80- and 104-column upgrade)

File What it is Sourced from
o1screenpacusersguide.pdf (2.3 MB, 20 pp) Osborne 1 SCREEN-PAC User Guide. The primary source for what the upgrade does, and it is more than a wider window: p. 13 (PDF p. 17) gives the three software-selectable widths (52 / 80 / 104), horizontal scrolling in 32-column blocks, and the presence flag at 2400H bank 2 that lets software detect the board. Quoted in docs/mame-emulation.md §2 and docs/o1-memory-io-map.md §2a/§3 (#73, #24). kev.pulo.com.au/osborne1, supplied by the maintainer 2026-10-10

The ScreenPac is the mechanical/electrical template for the interposer and for the video-output (HDMI) personality (#24, #9 photos).

The Field Service Manual those ScreenPac notes come from (2F00040, §6.12) is a whole-machine document, so it is filed with the rest of them in osborne1/ above — not here.

technical-articles/ — extracts from Osborne-era magazines

Only the pages that matter are kept, cut from the original and OCR’d. The full issues are not in the repo: they are 50–60 MB scans each, they are on the Internet Archive (https://archive.org/details/theportablecompanion) and on the maintainer’s own backup, and keeping whole magazines for a few pages of schematics is not worth the repo weight. Each extract names the issue and the pages it was cut from, so it can be checked against the original.

Extract What it is Cut from
TPC-Issue11-1983-pp32-osmosis-product-line-ad.pdf (0.6 MB, 1 p) OSMOSIS’ full-page advertisement, “The Next Step Forward” — the whole product line with product numbers: OSMOS 1 DD 184k/drive ($195), 1.5 ($335), 2 DS/DD 370k ($895), 3 DS/DD double-tracking 750k/1.5MB ($1050), 4 “read, write and format over 20 other manufactures CP/M formats” ($225), 5 the internal 80-column board ($250 — “includes a Hard Disk interface for future plug-in expansion”, normal/inverted video, underline, ⅔ and ½ intensity), 6 telex, 7 Winchester to 42 MB, 8 instrumentation, 10 Olivetti ET121 interface; bundle OSMOS 1+4+5 for $385. Quoted in docs/o1-upgrade-boards.md and docs/dd-upgrade-formats.md. Relevant to #24, #11/#12 and #95/#30. The Portable Companion #11, October 1983 — PDF p. 32 (printed 30)
TPC-Issue11-1983-pp54-doubling-hard-disk-memory.pdf (0.7 MB, 1 p) “Doubling hard disk memory” by Bob Gras (reprinted from the OSUG newsletter, Anchorage) — patching Corvus’s CLINK2K2.ASM so the link program works after a double-density upgrade: DELTA EQU 500H → 900H, because “more memory is required for managing the Osborne floppy drives” under DD. The same page carries the tail of John Dilday’s WordStar printer-spooler article. Quoted in docs/dd-upgrade-formats.md and docs/research-gaps.md §B1. Relevant to #95 and the hard-disk survey (gap 9). The Portable Companion #11, October 1983 — PDF p. 54 (printed 52)
TPC-Issue13-1984-pp26-31-reverse-video-and-9600-baud.pdf (3.3 MB, 6 pp) “Reverse Video and a Faster Serial Port” by Thom Hogan — changing the underline cursor to a reversed-video one, and pushing the serial port to 9600 baud, with the schematic changes. Relevant to #24 (video) and #21 (serial/modem). Worth noting the author: the same Thom Hogan whose AUTOCPM.ASM is the source of the site wordmark (#97). The Portable Companion #13, June/July 1984 — PDF pp. 26–31
TPC-Issue14-1984-pp43-47-drive-c-loader-and-archive-programs.pdf (4.6 MB, 5 pp) “Review of New Drive C: Loader & Archive Programs” by Larry Walker — a FOGHORN (First Osborne Group) reprint documenting the second-generation Drive C: software: DCL.COM (loader, replaces DRIVEC.COM), DFD.SPR (“a substitute console command processor … automatically loaded”), DCA.COM (Archive-Retrieve, replaces DCARCH.COM/DCRET.COM), QuickPac (DCQ/DCQINS/DCQ.DAT/DCQAUT→AUTOST.COM), the DC1/DCN/DCU command syntax, print-buffer sizes (16–128K or dynamic), retrieval timings, the BackPac battery, and the gotcha that a printer connected but switched off crashes the machine. Relevant to #6, #84 and #19. The Portable Companion #14, November/December 1984 — PDF pp. 43–47 (printed 41–45)
TPC-Issue15-1985-pp72-77-16bit-msdos-and-cpm-for-your-o1.pdf (4.9 MB, 6 pp) “16-Bit MS-DOS & CP/M For Your Osborne One!” by Robert J. Thomson — the Co-Power-88: an 8088 processor, MS-DOS / CP/M-86 / CP/M, and a 256K memory disk under regular CP/M, with the installation. Directly relevant to #23/#26 and the CoPower-88 work in research/copower88/. The Portable Companion #15, January/February 1985 — PDF pp. 72–77
TPC-Issue15-1985-pp32-block-cursor-with-screenpac.pdf (0.8 MB, 1 p) Letters: the Lovells’ fix making Hogan’s block cursor work on a machine with the SCREEN-PAC 80 fitted, without removing the piggy-back board — the modification moves to the upgrade board’s own U2 (74HC00) and U23 (74HC74), with a capitalised warning that the pins to cut are not the ones in Hogan’s article. Quoted in docs/o1-memory-io-map.md §2a. Relevant to #24. The Portable Companion #15, January/February 1985 — PDF p. 32
TPC-Issue15-1985-pp52-drive-c-letters.pdf (0.9 MB, 1 p) Letters on Drive C: — a user report (a 384K unit, WordStar to the dBASE II dot prompt in six seconds, 90% less drive wear) and the reviewer’s rebuttal (“Drive C: is not a good buy for the average user”). Quoted in docs/drive-c-protocol.md. Relevant to #6, #22 and #84. The Portable Companion #15, January/February 1985 — PDF p. 52
TPC-Issue15-1985-pp39-nuevo-four-upgrades-ad.pdf (0.7 MB, 1 p) Nuevo Electronics’ full-page advertisement, “Four Nuevo Ways to Upgrade”: the 80-column board (six cursor styles, reverse video, horizontal scroll, optional baud generator to 19.2K), the DD upgrade (183K), the 2 x 2 (780K, DS/DD, 128 directory entries) and the 4 x 4 (1580K, double-sided 96 tpi drives, auto-recognising SS SD/DD and DS 48/96 TPI, “read or write, even boot”). Quoted in docs/dd-upgrade-formats.md and docs/o1-upgrade-boards.md. Relevant to #95, #30, #24 and #21. The Portable Companion #15, January/February 1985 — PDF p. 39
TPC-Issue08-1983-pp38-41-understanding-the-ieee-488-port.pdf (2.5 MB, 4 pp) “Understanding the IEEE-488 port — Scientific interfacing with the Osborne 1” by Wayne Kogachi and Shahriar Kianersi, both of OCC: the machine as Talker/Listener/Controller, the 15-device bus limit, the O1’s own address 0, and the eight low-level ROM drivers (Control Out, Status In, Go To Stand By, Take Control, Output Interface Messages, Output Device Messages, Input Device Messages, Input Parallel Poll — the last cutting the bus to eight devices), with the error set and the SETUP route for a printer. Written up in docs/o1-ieee488-port.md. Relevant to #22, #84, #19. The Portable Companion #8, July 1983 — PDF pp. 38–41 (printed 36–39)
TPC-Issue08-1983-pp44-49-interfacing-external-devices-joystick.pdf (4.9 MB, 6 pp) “Interfacing External Devices — add a game joystick to the Osborne 1” by Kelly Smith (Wizard of OsZ), part one of two: the 26-pin card-edge → DB-9 adapter cable (12-conductor ribbon, D0–D4 + GND only), the Atari joystick bit map (1 north / 2 south / 4 east / 8 west / 16 fire, diagonals as sums) with the 128-character screen offsets, the CALL 0E154H ROM parallel-poll route, and JOYSTIK.ASM programming the 6821 at 2900–2903 — including EF08H as the ROM/RAM bank flag and the per-monitor-version origins EE80/ED80/EA80. Written up in docs/o1-ieee488-port.md. Relevant to #22, #84, #19. The Portable Companion #8, July 1983 — PDF pp. 44–49 (printed 42–47)
TPC-Issue08-1983-pp87-osbaud-baud-rate-generator.pdf (0.8 MB, 1 p) Brad Baldwin’s review of Advent Products’ “Osbaud” baud-rate generator ($59.95, 16 DIP-switch rates to 19200), plus “The CRT Connection” — the pre-SCREEN-PAC recipe for an 80-column terminal at 9600 baud. The review gives the four solder points (“pins 4 and 12 of the 6850 UART, pin 7 of the 74LS32, and pin 7 of the 74LS04”), which decode to TxCLK, Vcc and two grounds. Written up in docs/o1-modem-serial.md. Relevant to #21, #24, #45. The Portable Companion #8, July 1983 — PDF p. 87 (printed 85)
TPC-Issue05-1983-pp25-mega-byte-md10-hard-disk-ad.pdf (0.7 MB, 1 p) Mega-byte Inc.’s full-page advertisement for the MD-10 hard disk: 10 MB (formatted), 5¼”, $2295, “20 and 44 megabyte Winchesters also available”. The selling points are the architecture — it “does not tie up existing OSBORNE ports” and is “significantly faster than other winchester subsystems which use the IEEE port“, so the MD-10 is not an IEEE-488 device (unlike Drive C or the Trantor TSL below); portability survives (“unplug the winchester and the OSBORNE works exactly as before”); CP/M drivers cost under 2K where “other systems require as much as 6K”; a 64K file reads/writes in under four seconds; one unit takes up to eight 44 MB subsystems, a network of sixteen Osbornes, and 36 user areas; the disk becomes A: and B: with the floppies moved to E:/F:, a second unit becoming C:/D:. The footnote — “MD-10 is a registered trademark of Media Distributing” — identifies it as the product behind research/adaptec/ (the Media Distributing disk that ships MD10.COM/MD20.COM), so the archive’s “Media Distributing / Adaptec” hard disk is an MD-10. Hand-transcribed — tesseract returns only the folio from this display-type page (see below). Quoted in docs/research-gaps.md §B1 and docs/occ1-harddisk-analysis.md. Relevant to #8 and the hard-disk survey (gap 9). The Portable Companion #5, February/March 1983 — PDF p. 25 (printed 23)
TPC-Issue05-1983-pp28-trantor-tsl-hard-disk-ad.pdf (0.8 MB, 1 p) Trantor Systems’ full-page advertisement for the TSL-5 and TSL-10 hard disks: 5 MB $2995 / 10 MB $3395, “unpack your disk, plug it into the IEEE port, run a simple 5 minute installation procedure and you’re ready for use”, supplied “complete with BIOS, back-up, retrieve utilities and cable“. This is the Trantor line that docs/drive-c-protocol.md records the Drive C vendor as reselling at a discount, so the ad is the primary source for that product’s interface: an IEEE-488-attached Winchester — the opposite architecture to the Mega-byte MD-10 on p. 25. No folio is printed on the ad page. Quoted in docs/drive-c-protocol.md and docs/research-gaps.md §B1. Relevant to #6, #22 and #84. The Portable Companion #5, February/March 1983 — PDF p. 28
TPC-Issue05-1983-pp98-tra-80-100-column-video-and-winchester-ads.pdf (0.4 MB, 1 p) Two advertisements on one page. (1) The Research Associates (TRA) — “80 and 100 column video”, a video system “as an add-on to the Osborne 1” giving 80×24 and 100×24 windows into a 128×32 format on an external monitor, with soft scrolling, reverse video and blinking, composite and TTL outputs, connection “by any Osborne 1 user. Simply plug into the IEEE port and into your external monitor”, “parallel printer connection is not forfeited” (the printer moves to a parallel port on the TRA), and “a one-time, automatic disk configuration process” with vendor-supplied driver software. That is a third 80-column implementation alongside SCREEN-PAC and Nuevo (#24), and unlike either it reaches the monitor through the IEEE port while taking over the printer function. (2) Helix / Micro Computers of New Orleans, “Winchesters Now Available For The Osborne 1”: a “Z-80 adaptor card [that] leaves external ports free and increases the speed that your information is displayed”, ribbon cable disconnects for portability, 5 MB $2500 / 10 MB $3000 / 15 MB $3500. Quoted in docs/o1-upgrade-boards.md (TRA) and docs/research-gaps.md §B1 (Helix). Relevant to #24, #19 and gap 9. The Portable Companion #5, February/March 1983 — PDF p. 98 (printed 84)
TPC-Issue07-1983-pp50-51-dating-your-computer-date-in-memory.pdf (1.4 MB, 2 pp) “Dating your computer — a system’s date facility” by Walter B. Novinger: a DATE.BAS that asks for MMDDYY and stores it, a GETDATE BASIC subroutine that reads it back, and an AUTOST.COM modification (LXI B,10 → LXI B,13, command string 'MBASIC DATE',0) that makes the prompt automatic at boot. The reusable fact is the storage location: the date goes in the 12-byte page-zero block at 0050H–005BH, which the Osborne 1 User’s Guide “Memory Allocation” chart (p. R-243) marks “not used” — written in both binary (POKE 80,D, POKE 81,M, POKE 91,Y) and ASCII month name (82H–8AH) so the reader needs no machine code. It survives a warm start (RESET) and diskette swaps but not power-off. Prior art for the RTC work (#18/#7) and the boot disk (#28). Quoted in docs/rt60a-analysis.md and docs/o1-memory-io-map.md. The Portable Companion #7, June 1983 — PDF pp. 50–51 (printed 48–49)
TPC-Issue07-1983-pp56-57-programming-the-serial-port.pdf (0.9 MB, 2 pp) “Programming the serial port — modifying the serial port for scientific test devices”, signed by the Osborne Computer Corporation System Technical Support Group: how to change the 6850 ACIA’s data bits / stop bits / parity away from the factory 8-N-1 by patching the BIOS’s serial control byte — a fully worked procedure (MOVCPM 60 * for a 1.3 BIOS / MOVCPM 59 * for 1.4, SAVE 39 CPM60.COM, DDT CPM60.COM, S1FE9 and write the code at 1FE9H/1FEAH, then SYSGEN), a “Table of 6850 control codes” covering all 16 combinations at 1200 and 300 baud (e.g. 8-N-1 @1200 = 55H, 8-N-1 @300 = 56H, 7-E-1 @1200 = 49H, 8-N-2 @1200 = 51H), and a which-BIOS-do-I-have test (DDT, DA900H: C3 xx E5 = 1.3, C3 xx E1 = 1.4). Gives the exact BIOS offset the ACIA’s control register is initialised from — a second handle on the ACIA beyond the 2A00H/2A01H port pair, directly useful to the snoop/modem work (#21, #56). Quoted in docs/o1-modem-serial.md and docs/o1-memory-io-map.md. The Portable Companion #7, June 1983 — PDF pp. 56–57 (printed 54–55)

How these extracts are made

The maintainer reads the issues and hands over the ones worth keeping, giving the PDF page sequence to cut. The standing workflow:

  1. Cut the pages: pdfseparate -f A -l B ISSUE.pdf /tmp/p%03d.pdf then pdfunite /tmp/p*.pdf OUT.pdf.
  2. OCR the extract with tools/ocr_pdf.sh OUT.pdf — adjacent, per the convention above. Wait for its Wrote <path> line before committing; the file exists while it is still being written.
    • If you drive tesseract by hand, do not render into /tmp. On this machine it cannot open images under /tmp (Leptonica reports the file as not found) and, with 2>/dev/null, fails silently — a 43-page sweep over /tmp returned nothing at all and looked like “no matches”. The script is fine because it renders into a mktemp -d.
    • A display-type page can defeat tesseract entirely. Full-page advertisements set in display faces over tinted screens sometimes OCR to nothing but the folio — the #5 Mega-byte MD-10 ad (TPC-Issue05-1983-pp25-…) was the first. Do not commit an empty .ocr.txt: transcribe the page by eye into the file behind a [HAND TRANSCRIPTION, not tesseract output …] header, so the text still lives beside the scan and never has to be re-derived.
  3. Add a row below, and write the findings into the docs/ page they bear on with the extract cited.

Two things to know before cutting pages:

  • Page numbers are PDF pages, not printed pages. Every page checked so far prints its own number two lower (PDF = printed + 2) — issues 11, 13, 14, 15, and now #5 p. 25 (→ 23) and #7 pp. 50 and 56 (→ 48 and 54). So an article “on page 26” is PDF page 26 and printed page 24 — and cutting printed 26 would miss the article’s title page. The offset is not uniform, though: the #5 extract filed as pp98 carries folio 84, i.e. +14, so an issue can contain unnumbered pages. Read the folio off the page rather than assuming — pdftotext -f N -l N -layout ISSUE.pdf - | tail -3 prints it, and the extract’s own .ocr.txt usually ends with it.
  • Provenance for every article from The Portable Companion is the same collection: https://archive.org/details/theportablecompanion. Rows therefore cite the issue and the page range rather than a per-file URL.

The full issues are not kept in the repo — 50–60 MB each, and they live on the Internet Archive and on the maintainer’s own backup. Only the extracts are committed, and they are cut from files outside the repo (the maintainer’s ~/Documents tree), never from a copy inside it.

copower88/ — SWP CoPower-88 (Z80-coprocessor board being reverse-engineered)

Path What it is Sourced from
zorba/Co-Power-88guide.pdf, zorba/Co-Power-88advert.pdf, zorba/611-0003-1..4.pdf SWP CoPower-88 user guide, advertisement, and SWP technical document scans (Zorba-flavored set) zorba.z80.de — the Zorba Portable Computer archive site
zorba/*.td0, kaypro/*.td0 TeleDisk images of CoPower-88 CP/M-86 / DOS boot disks (Zorba and Kaypro versions) zorba.z80.de (exact path not recorded in repo history)
zorba/extracted/, kaypro/extracted_cpm/, */disasm/ CP/M files extracted from those images by the project pipeline (#2), and disassemblies of the CoPower RAM-disk/DOS drivers Derived in-project — not external sources
swp/Copower_User_guide.pdf, swp/Swp_ad.jpg Fuller CoPower-88 user guide (4.7 MB, not the Zorba set) and an SWP advert scan kayprojournal.com Swp_copower_88 — File:Copower User guide.pdf, File:Swp ad.jpg
swp/1600px-73356_8088_card.jpg, swp/1600px-73356_mem_card.jpg, swp/866px-73356_cpu_card.jpg Board photographs of the CoPower-88 8088 card, memory card and CPU card — held as MediaWiki thumbnails (1600 px / 866 px) of File:73356 8088 card.jpg, File:73356 mem card.jpg, File:73356 cpu card.jpg Same page: kayprojournal.com Swp_copower_88

The whole swp/ set is from that one page. Confirmed 2026-10-10 by querying the wiki’s API (list=allimages and prop=images&titles=Swp copower 88) and matching all five File: names exactly — not by reading the rendered page text. The wiki also carries File:Copower sch 1..4.pdf, a second copy of the SWP schematics we hold from zorba.z80.de.

Basis for #3/#4/#5 (CoPower-88 protocol analysis). The board’s own 8088 ROM is roms/swp-p88.rom — see roms/ below and #70.

drive_c/ — Drive C (IEEE-488 ramdisk, #6)

Path What it is Sourced from
Drive_C_Users_Manual.pdf Drive C user manual (IEEE-488 virtual disk drive), second edition June 1984 bitsavers bits/Osborne/Osborne1/Drive_C/ (upstream name Drive_C_-_Users_Manual.pdf)
Drive_C_Executive_Users_Manual.pdf Drive C user manual for Osborne Executive computers — the same product on the other host: identical hardware appendix, ECL/ECN/ECU software under CP/M+ instead of DCL/DCN/DCU under CP/M 2.2, and no DFD.SPR. First edition July 1984, second printing 9/84; cover stamped PRELIMINARY, running headers “final draft”. 111 pp, 12 MB, image-only (150 dpi grayscale re-encode — see the note below). Analysis: docs/drive-c-executive.md (#84, #6, #145) Jim Rayburn, private Facebook group, posted 2024-01-06 — posts …/2063148960719887 and …/2023247321376718; supplied to the maintainer as Executive_DriveC.pdf, 2026-10-11
Drive_C_Executive_Users_Manual.ocr.txt OCR text layer for the above — 300 dpi tesseract, per the convention below Derived in-project, 2026-10-11
DRIVE_C.IMD ImageDisk image of the Drive C utility disk (label comment Drive C for the Osborne 1 computer) same folder — not imaged in-project; corrected 2026-10-10
Drive_C_disk_label.jpg Photo of the physical disk label same folder (15 044 B, byte-identical) — not a project photo; corrected 2026-10-10
extracted/ Files extracted from DRIVE_C.IMD by the project pipeline (#2) Derived in-project
disasm/ dcl.asm + dfd.asm — z80dasm listings of the installer and the resident driver, for the protocol work (#6). Regenerate with the exact command in docs/drive-c-protocol.md §Disassembling Derived in-project

A note on the Executive manual’s size. The owner supplied a 300 dpi scan, 42.33 MB — larger than any other single file in this tree. It is committed as a 150 dpi re-encode instead (12.13 MB), the same resolution as the Drive C manual above: the document is a typewritten mono manual, and the page images carry no colour at all — every pixel is neutral grey, checked across all 111 pages — so the extra resolution bought nothing but file size. All 111 pages and the page geometry are preserved. The 300 dpi text layer was taken from the owner’s original before the re-encode and sits beside the PDF as Drive_C_Executive_Users_Manual.ocr.txt, so no text was lost. It is recorded as a re-encode rather than as the owner’s original, so the provenance stays honest: the file here is not byte-for-byte what Jim Rayburn posted.

rtc/ — RT-60A real-time clock (#7)

Path What it is Sourced from
RT-60A_Manual.pdf RT-60A RTC manual (JG Communications) bitsavers bits/Osborne/Osborne1/Osborne_1_Real_Time_Clock/ (upstream name RT-60A_Real_Time_Clock_Manual_-_Osborne_1.pdf)
RT-60A.TD0 TeleDisk image of the RT-60A utility disk — a damaged dump (27 no-id sectors); filesystem extraction fails, see tools/README.md same folder

harddisk/ — ACT/OCC1 hard disk (#8)

Path What it is Sourced from
OCC1_HARDDISK.IMD ImageDisk image of the OCC1 hard-disk system disk (label comment act-osborne / hard disk software / rev 2) bitsavers bits/Osborne/Osborne1/_floppy_images/ — not imaged in-project; corrected 2026-10-10
extracted/ Files extracted from the image, incl. hardbios.hex Derived in-project

adaptec/ — Adaptec ACB-4000 hard-disk software (2026-10)

Raw Osborne 1 SSSD floppy images (102 400 B = 40 tracks × 10 sectors × 256 B, no IMD header) each with a photo of the disk’s label. Source: Internet Archive, md-utilities-adaptec-4000-controller-ver.-1.93-osborne.

File What it is
MD Utilities Adaptec 4000 Controller Ver. 1.93 Osborne.img Adaptec ACB-4000 utilities
MD Utilities Adaptec 4000 Controller Ver. 1.93 Osborne - MD-10 MD-20.img Same, for the MD-10/MD-20 drive pair
Osborne Hard Disk Control Software Version 2.01 (1984).img Osborne’s own hard-disk control software
compatible-drives-1991-usenet.txt (23 KB) A period list of ST-506/MFM drives, not Osborne software: Robert L. Haar’s “List of replacement hard drives for 3b1” (comp.sys.3b1, 12 Feb 1991, TUHS Usenet archive), which reposts Lenny Tropiano’s “ST-506 MFM Hard Drive Manufacturer listing (LONG), model #’s and specs” (unix-pc.general/comp.sys.att/comp.periphs, 2 Sep 89), itself edited out of the full drive list on the Western Digital technical-support BBS. Per manufacturer: model, formatted capacity, cylinders, heads, reduced-write-current, write-precomp, encoding, transfer rate, access time and sectors/track — the last of which is 17 for almost every entry, where DRIVETBL.DAT’s capacity formula uses 18, and it gives PYXIS-13 320 cylinders where DRIVETBL.DAT says 321. Archived because the personality should offer any ST-506/412-class drive, not only the 126 the MD-10 tools ship (docs/harddisk-decision.md §7.2)

Together with acb4000a.rom (Maslin’s archive) this identifies the controller behind harddisk/OCC1_HARDDISK.IMD — issue #8.

adaptec/md10-system/ — the MD-10 operating system (2026-10, generated in-project)

The system the archive did not hold. docs/harddisk-decision.md §7.3 concluded that nothing in the archive initialises the MD-10’s low-RAM tables and that the utilities disk carries no system image — which left #179’s acceptance (“boots CP/M from the emulated hard disk”) with nothing to boot. It was not obtainable by reading; it is obtainable by running, and the vendor’s own installer produces it: PREP.COM option 3, “Configure operating system”, writes it to the drive. Captured off configured drives on 2026-10-12 and 2026-10-14.

File What it is
lba0-15.bin (8 KiB) PYXIS13, two partitions of 11544 blocks: the first sixteen blocks of a configured drive — LBA 0 the disk label and its partition table, LBA 1–3 the header, LBA 4 a table, LBA 5–15 the operating system
md10-os.bin (5.6 KiB) its operating system — LBA 5, 11 blocks (0x1600 bytes) — relocated 0x100 high, so it needs tools/research/md10reloc.py before it boots (#198)
lba0-15-p27.bin (8 KiB) PYXIS27, three partitions of 15399 blocks — the driver’s own static fallback table, and the only layout that matches its shared DPB. The floor pin
md10-os-p27.bin (5.6 KiB) its operating system, already at its load base (JP 0C4C8h, with LD SP,nn at the target) — no relocation step
README.md provenance, and what each file is

Derived in-project from adaptec/*.img (the PREP.COM that writes it) and the 2.01 utilities disk (the HARDCCP.SPR/HARDBDOS.SPR it writes); the PYXIS13 drive it was taken from had md5 7fa2a3836813dabdeaaa664ea88d002e, and tools/research/md10drive.py reproduces either drive byte for byte from these files (--layout md10 / --layout md20). The PYXIS27 pair was produced by tools/research/md10prep.sh 321,8,18 3 and independently reproduced by it. The recipe, the MD-10’s memory map, the boot chain and the one-DPB constraint are in docs/md10-system.md.

adaptec/acb4000/ — Adaptec ACB-4000 controller hardware (2026-10)

The controller itself — schematics, board photographs, the controller’s own firmware ROMs, and two OEM manuals. Archived because #8 selected this controller (docs/harddisk-decision.md). Source: bitsavers pdf/adaptec/ACB-4000/.

Path What it is
500504-00_ACB-4000_OEM_Manual_Mar84.pdf (9.8 MB) OEM manual — the command set (Table 6-1), the reference host-adapter register model, a sample Z80 driver and a sample format sequence
400003-00A_ACB-4000A_Users_Manual_Oct85.pdf (7.1 MB) ACB-4000A user manual. Image-only scan — pdftotext returns nothing. Carries §2.3.1 (device independence / auto-configuration) and §3.7 (power-up operation): the drive parameters are encoded onto the disk at FORMAT, in the ID field of cylinder 6, and read back at power-up — see docs/harddisk-decision.md §7.3
400003-00A_ACB-4000A_Users_Manual_Oct85.ocr.txt (177 KB) OCR text layer for the above (per the convention at the top of this file), generated 2026-10-12 at 120 dpi with tools/ocr_pdf.sh. Good enough to grep — §2.3.1’s “ID field of cylinder 6” is legible — but the OCR mangles part numbers (ACB-4000 renders as ACB-4909/ACB-4990 in places), so read the PDF for anything load-bearing
ACB_4000_and_5000_Disk_Controllers_OEM_Manual_Preliminary_May83.pdf (2.6 MB) preliminary OEM manual (ACB-4000 + ACB-5000)
ACB4000A_Schematic_ARD_2025.pdf (2.4 MB) board schematic
firmware/ six 8 KiB controller ROM images + an Intel-HEX build (ACB-4000A-rev-B.hex)
pictures/ seven board photographs

The adaptec/*.img software above is the host side (the MD-10 driver); this folder is the controller side. The personality needs both: the host side is what it must answer, the controller side is what it must be.

diagnostics/ — Osborne 1 confidence tests (2026-10)

File Label comment
DIAG_2.04.IMD 2.04 / diagnostics
OCC1_DIAG_2.1.IMD occ diag / ver 2.1 / 12/31/84

Source: bitsavers bits/Osborne/Osborne1/_floppy_images/. These are the known-good boot images the validation matrix (#29) needs.

nuevo/ — Nuevo Electronics double-density upgrade (2026-10)

File What it is Sourced from
Nuevo_Electronics_DD_Upgrade_Manual.pdf (13 MB) DD upgrade manual bitsavers bits/Osborne/Osborne1/ (13 810 755 B, exact match)
Nuevo_Electronics_DD_Upgrade_Manual.ocr.txt (919 lines) OCR of the manual — the BIOS 1.4→1.5 difference list (DPBs for IBM/Xerox/DEC removed, 80-column presence check, jump table at E100h), the 380K double-sided claim and the single/double-density system-track rules, all quoted in docs/dd-upgrade-formats.md derived in-project with tools/ocr_pdf.sh … 300
OS1NUEVO.IMD Osborne 1 with DD and 80 col mods boot disk, re-imaged 2026-10-10 with Disk-Utilities derived in-project
photos/nuevo-80-column-board.jpg 80-column board, REV. D, serial 4348 — the first hardware view of the product. Its own NEC D780C-1 (Z80), a Nuevo SY6545A-1 CRTC, 2 × SRM2016C25 SRAM, a socketed CHARACTER GENERATOR ROM dated 6-29-84, a 12.288 MHz oscillator, and the optional BAUD generator. Devices, and what they imply, in docs/o1-upgrade-boards.md §”The Nuevo boards themselves” (#73, #24) Mike Loewen, private Facebook group, 2024-10-14 — post
photos/nuevo-double-density-board.jpg DD board, serial 20011 — DAL DENS silk, a 34-pin IDC header with a STD DRIVE jumper, and a discrete write-precompensation/separator chain (74LS195A, SN74LS14N ×2, SN74LS00N ×2, SN74LS02N) plus a 16-pin 9216-family device (U1). Same analysis as above same post
photos/nuevo-80-column-and-double-density-boards.jpg Both boards in one frame — the shot the 2024-10-14 post is built around same post

The photographs are third-party (Mike Loewen, a private group) rather than in-project like pictures/, and they are the only hardware views of either Nuevo product — there is no Nuevo schematic in the archive. Neither board bears on Drive C: both are host-side (video, and floppy encoding).

Removed 2026-10-10: NUEVO151.BIN. It was a byte-identical duplicate of tools/emulator-setup/roms/monrom-rev1.51-12.ud11 — same 4096 bytes, SHA-256 aec8f5be127fdff5d30cae0b7f45ee48d13a8c1d5745a262b944b9ba67778906, CRC32 298da402 — verified with cmp before deletion. It carried no provenance of its own, so the tracked emulator-set copy, which has a recorded source, is the one we keep. The BIOS strings are unchanged by this: COPYRIGHT 1983, OSBORNE COMPUTER CORP. / COPYRIGHT 1984, NUEVO ELECTRONICS CORP.

Nuevo shipped two products — a DD upgrade and an 80-column board — and neither matches Osborne’s own implementation of that function; see #73.

OS1NUEVO.IMD is the Nuevo 80-column system disk: its system area carries CBIOS vers 1.5 and the text “THIS IS AN EIGHTY (80) COLUMN DISPLAY FOR THE OSBORNE ONE”. It does not currently boot under MAME — it prints the CBIOS banner and stops (measured over 260 emulated seconds; the pristine upstream copy behaves the same), which blocks validating the Nuevo video path in the emulator. See #79 and docs/mame-emulation.md §4.

osmosis/ — Osmosis Computer upgrades (2026-10)

Source: bitsavers bits/Osborne/Osborne1/Osmosis_Upgrades/ — all seven files byte-for-byte identical.

Osmosis_CPM_Disk_Emulator.pdf, Osmosis_80_Column_Board.pdf, Osmosis_SSDD_Upgrade.pdf, OSMO-EMU.TD0, OSMOS-DD.TD0, plus two disk-label photos. Osmosis likewise shipped two products (a CPM disk emulator and an 80-column board). The 1983 CPM Disk Emulation System is direct prior art for this project — see #73 and inspired-by/.

gotek/FF/ — working Gotek configuration (2026-10)

FF.CFG + IMG.CFG — the project’s own Gotek configuration for the O1: interface = shugart, pin02/pin34 = nc, and an IMG.CFG default of 40 cylinders × 1 head × 16 × 128 B, FM, 125 kbit/s — i.e. O1 SSSD. Captured from the physical Gotek, so these are primary sources, not downloads. Used by #62 and #20.

roms/ — ROM images (2026-10)

Canonical inventory, CRC32s and the third-party ROM survey are tracked in #70. Files here come from two sources:

swp-p88.rom is 8088 code, not Z80: the SWP co-processor board’s own ROM. The board and interposer card are common to every host machine (only the driver varies), so this is the 8088-side firmware of the CoPower-88 family — #3/#4/#5.

OZROM-1E/OZROM_1E_Manual.ocr.txt — OCR text layer for the OZROM manual (image-only scan; generated 2026-10-10 at 300 dpi with tesseract, per the OCR convention above). Good enough to grep for the technical appendices — which is how the IEEE-488 removal and the 3 KB video window were found (#81) — but not reliable for exact code or pin detail.

OZROM-1E/OZROM_1E_Utilities.IMD — the OZROM 1E Utilities disk, the software half of the OZROM package, which we did not previously have. ImageDisk 1.19, label ozrom; 40 cyl × 10 × 256 B FM (O1 single density), 22 of 40 tracks in use. Its CP/M directory is complete and readable: 20 files, all Micro Management 1984, with READ.ME, CRCKLIST.CRC and the docs for TIMER, CLKPATCH, DVORAK and TD. Condition: whole — all twenty files complete. 290 of the 400 sectors read back as uniform 0xE5, and that is CP/M format fill (free blocks), not damage: every file extracts at the size its directory entry gives, no file holds an interior run of 0xE5 longer than a byte, and READ.ME reads as continuous prose across all five of its sector boundaries. The “most files are truncated or holed” reading this row used to carry was our own sector-skew bug (#170), and it is what a search for a complete copy was filed on (#168, since closed). It is a data disk, not a system disk — the boot tracks carry a stub whose message is “There is no CP/M system on this disk.” Provenance: supplied by the project owner as a local file (2026-10-11); the original source URL is not yet recorded, so treat the chain of custody as open. Analysis: docs/ozrom-1e.md §10.2 and §10.5. Used by #81.

shipped-software/ — the shipped application library (2026-10)

Fourteen ImageDisk images of the disks that shipped with (or were published for) the Osborne 1. The ImageDisk label comment is quoted because it is the disk’s own identity. Source: Dave Dunfield’s ImageDisk archive (dunfield.classiccmp.org, which rotates its directory to discourage deep links) — d/osborne1.zip and d/o1ddsys.zip, byte-identical to these copies.

File Label comment
O1CPM.IMD Osborne-1 CP/M System Disk / Double Density
O1CBMB.IMD Osborne 1 - CBASIC/MBASIC / Double Density
O1SCALC.IMD Osborne 1 - SuperCalc 1.12 / Double Density
O1WSMM.IMD Osborne 1 - WordStar/MailMerge 2.26 / Double Density
OS1SYSD.IMD CP/M-2.2 system disk for Osborne 1 w/ DD drives / SSDD 1024 byte sector
OS1SYSS.IMD CP/M 2.2 System Disk for Osborne 1 / SSSD 256 byte sector
OS1BASIC.IMD CP/M 2.2 Basic Disk for Osborne 1
OS1DBASE.IMD dBase II for the Osborne 1 on sysgened disk
OS1DIAS.IMD CP/M 2.2 Diagnostics Disk for Osborne 1
OS1MCAL.IMD CP/M 2.2 Mocro-Call Disk for Osborne 1 (Communications)
OS1MDM7.IMD Modem 740 communications for Osborne 1
OS1UTLS.IMD CP/M 2.2 Utilities Disk for Osborne 1
OS1XUTLS.IMD Extended utilities for Osborne 1 w/ 80 column video card
OS1WRDST.IMD CP/M 2.2 WordStar 3.0 Disk for Osborne 1

The corpus for the image library (#20), the updated boot disk (#28) and the validation matrix (#29). Dunfield also publishes a 2005 single-density set (O1CPMS.IMD, O1CPMU.IMD, …) which is deliberately not copied here.

source-code/ — Osborne’s own firmware source (2026-10)

Path What it is Sourced from
BIOS-144-source/ (18 files) The original Osborne ROM source: rom144.asm (DOUBLE DENSITY ROM REV 1.44, Roger W. Chapman, 2/4/1983), occbio05..95.asm (OCC CBIOS Rev 1.41), occram15/25.asm, occtxt6.ast (ACT80 assembler equates), bios141.gen, release.txt, tran.txt. Basis for #72. github.com/BrettHallen/Osborne_1 ROM/Source_Code_v144/ (all 18 file sizes match)
ROM-13-source/ (21 files) The ROM 1.3 source, extracted from the disk below (#205): occrom03.asm (the master, which LINKs the rest), occrom13..93.asm, occbio03..83.asm (CBIOS rev 3), plus occram2.asm and occtxt6.ast, which are not on the ROM disk and come from OSUTLSRC.IMD. The 8-inch disk is 77 tracks, track 0 FM 26×128, tracks 1–76 MFM 8×1024, CP/M 2.2 with skew 3; read with tools/research/o1imd8.py. See docs/rom-13-reconstruction.md. OSROM13.IMD + OSUTLSRC.IMD below
BIOS-200-source/ (3 files) ROM 2.0’s monitor source (#247): rom2.diff, the change to BIOS-144-source/rom144.asm as a unified diff — HCBOOT, the two re-lettered boot keys, the boot screen and IMSG’s own slot — with rom2-screen.asm and the folder’s own README. There is deliberately no rom2.asm: a second copy of a 4,655-line vendor file would drift from the first silently, and the point of this monitor is that it is the shipped 1.44 with new code in the free tail. tools/research/rom2asm.sh applies the diff, assembles with the vendor’s ACT80 and checks that nothing below the new code moved. See docs/o1-rom-2.0.md §4. in-project (#247)
OCCCBIOS.IMD, OCCROM.IMD, OCCUTIL.IMD, OSBIOSSR.IMD, OSBROM.IMD, OSROM13.IMD, OSUTLSRC.IMD OCC engineering source disks (1981–82): CBIOS, ROM rev B, utility programs, ROM 1.3. The ImageDisk comments carry the OCC part numbers (4d2007-00 occbi00a.asm, rev b 9/12/81). All seven are 8-inch disks in the same 77-track FM/MFM format, and OCCROM.IMD additionally carries a shipped rev-B ROM image (OCCROM0B.COM, 4096 B) and its listing — the control material for #205. bitsavers

Licensing: release.txt is Osborne’s 1985 letter permitting the First Osborne Group to reprint the schematics and the BIOS listing, on condition that any money goes to the non-profit FOG and that the trademark/copyright notices are retained. FOG is long defunct and this material is already public in several places — this copy came from a public GitHub repository. The notice is retained verbatim; if the terms ever need re-examination, that is the file.

act-assembler/ — SORCIM ACT80, the assembler the firmware source needs (2026-10)

The Osborne firmware source in source-code/ is written for SORCIM’s ACT (“Assembly Code Translator”), not for any assembler we had. This is that assembler, kept as a safety copy because rom144.asm reassembles with it to the shipped ROM byte for byte (#203, docs/o1-rom-source.md).

Path What it is Sourced from
ACT80.COM (13 312 B) SORCIM ACT80, version 3.0 — 8080/8085 + Z80. md5 4f76a911a66ddc8e1e02a1676c232d45. The build the 1.44 reassembly was done with. cpmarchives.classiccmp.org — cpm/Software/rlee/S/SORCIM/ACT Cross Assemblers/
3.2/ACT80.COM (13 312 B) SORCIM ACT80, version 3.2, serial 279. md5 cf73590466736fd1960e829b5f2221aa. Held so the version question can be re-checked against a local file rather than the mirror: it assembles rom144.asm to output byte-identical to 3.0’s. same mirror, cpm/Software/rlee/L/LOOSECPM/119/ — 073 and 087 hold the same bytes
3.2/ACT80-eca-bundle.COM (13 312 B) A third, different 3.2 build, the one that circulated inside a third-party software bundle. md5 88256731a9399dee68a29054865d072c. It reports SORCIM 808x Assembler ver 3.2 in its listing banner, but carries no SORCIM copyright or serial line at all, and it differs from the SORCIM 3.2 in 13 112 of 13 312 bytes — a redistribution rather than the release, kept as evidence of how many binaries claim to be 3.2. same mirror, cpm/Software/rlee/D/DIGITAL RESEARCH/ECA/ (it is named ACT80.COM there; renamed here so both 3.2 copies can sit side by side)
ACT65.COM, ACT68.COM, DOCOP65.ASM, DOCOP68.ASM The same kit’s 6502 and 6800 translators and their mnemonic listings. Kept with the kit rather than dropped — they are 30 KB, and a hand-picked subset invites the question of what was left out. same
DOCOP80.ASM (8 192 B) Every ACT mnemonic, in alphabetical order — the dialect reference, and the file to diff against if a future source is rejected. same
READ.ME, INSTALLA.SUB The kit’s own file list, and its SUBMIT script for moving the kit from B: to A:. same
IPARAMS.ACD, UTIL.ACD, ACD-EG.ASM ACT’s LINK include files, and an example program using them (LINK IPARAMS.acd). same

Versions. The mirror holds nine copies of ACT80.COM in three distinct builds, and all three are kept here — so the version question can be re-checked against a local file rather than the mirror:

Build md5 Copies in the mirror In-repo
3.0 4f76a911… S/SORCIM/ACT Cross Assemblers/, M/MORROW/OUTSIDE SOFTWARE/ACT/, L/LOOSECPM/{079,109,262}/ the kit above
3.2, serial 279 cf735904… L/LOOSECPM/{073,087,119}/ 3.2/ACT80.COM
3.2, no banner 88256731… D/DIGITAL RESEARCH/ECA/ 3.2/ACT80-eca-bundle.COM

3.5e — the version bios141.gen names — is not in the mirror, and it does not need to be, because the version does not change the answer: 3.0 and the SORCIM 3.2 assemble rom144.asm to byte-identical output, both re-checked against the in-repo copies (#203). Either one reproduces the shipped ROM:

O1_ACT80=research/act-assembler/3.2/ACT80.COM tools/research/rom144-reassemble.sh

The third build is a caution rather than a spare. On the same disk, with the same command, it returns to the A> prompt immediately and prints nothing — no banner, no no ERRORs line, no hex file — where both SORCIM builds assemble the source cleanly. Whatever it is (a relocation for one system size, a bundle installer’s variant, or a patched copy), it is not a drop-in ACT80, and it is kept as evidence of how many different binaries claim to be version 3.2.

Licensing: READ.ME is SORCIM’s own 1980 notice — “This disk contains software proprietary to SORCIM. It may NOT be copied other than for use by the purchaser.” It is retained verbatim and recorded here rather than quietly dropped. The kit has been publicly mirrored since 2011, and this copy exists to read the firmware source the project already holds; if the terms ever need re-examining, that file is where to start.

boot-screen/ — the authored ROM 2.00 boot-screen art (2026-10, in-project)

Not a sourced document: the maintainer’s own drawing of ROM 2.00’s boot screen, authored in the machine’s graphics characters. A design source for #83, kept here for the same reason pictures/ is — it exists nowhere else.

File What it is Sourced from
boot-screen/o1next-art-1ng.txt the 1NG art: 125 × 28 cells, one character per cell, SHA-256 a92bf28e81b54c58f59c48ab16f090e2d26c4e8d3705ed0cfcda7dd10bb1f9ba. Read it in the osborne1-art-* font from tools/romfont.py — character code minus 65 is the glyph index, the same sui 65 the ROM’s own loader does authored 2026-10-14
boot-screen/o1next-art-1ng.png the same art rendered with the ROM’s own glyphs, for reading —
boot-screen/o1next-art.txt the O1NEXT art: 153 × 27 cells, with “osborne 1 next rom 2.0 2026” and “press return to begin” nested in the X’s counter as text rather than graphics (sui 65 cannot reach them — p is 0x70). SHA-256 a3617203212e193651a8dae44b3fdb030f0f5051fe629250ac2ccd0abdacd42a authored 2026-10-14
boot-screen/o1next-art.png the same art rendered with the ROM’s own glyphs —

The measured geometry, the kerning, the byte cost of each piece, and why the long version (O1NEXT, 149 columns) does not fit the 128-column video buffer and is being redrawn narrower are all in boot-screen/README.md.

pictures/ — machine photographs

IMG_4856.jpeg … IMG_4871.jpeg — disassembly/mainboard/upgrade-board photos of the project’s own machines (mainboard, double-density upgrade, ScreenPac, CoPower-88). Taken in-project for #9; these are primary sources, not copies. Further photo sets are tracked in #39.

Provenance

The two long-standing provenance gaps are closed: the Drive C manual and the RT-60A manual/TD0 both come from bitsavers’ bits/Osborne/Osborne1/ (Drive_C/ and Osborne_1_Real_Time_Clock/), as does OCC1_HARDDISK.IMD and the two diagnostic disks (_floppy_images/). Four rows that had been recorded as in-project work or as unknown URLs were corrected on 2026-10-10.

Both remaining TODOs are closed (2026-10-10, #74):

No provenance TODOs remain.