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.

Status: Partially solved. Four 80-column implementations are identified, and each one’s attachment, width selection and detection is read from its own material; what is left is one measurement (§”How software finds the board”) and Osmosis’s technical manual, which the archive does not hold.

Four 80-column upgrades for the Osborne 1 are known, and no two of them attach, select a width or announce themselves the same way. That is the fact a video personality (#24) has to be built around, and it is why “80 columns” is not one interface:

Board What it actually is
OCC SCREEN-PAC (1983) an add-on board inside the machine, hardware only — the stock BIOS already carries the width routine, so no ROM changes
The Research Associates’ TRA (1983) an external video unit that reaches the monitor through the IEEE-488 port, with the printer moved to a parallel port on the TRA
Nuevo (1984) a second video system: a replacement monitor ROM, a CBIOS patched on disk, and its own Z80, CRTC, video RAM and character generator on the board
Osmosis (1983) a Z80 interposer board — the CPU moves onto it and it takes the CPU’s socket — with a terminal-style escape-sequence interface and a supplied monitor ROM

Nuevo Electronics and Osmosis each shipped two Osborne 1 upgrades — a double-density board and an 80-column video board — and neither follows the pattern OCC used for the same two functions. That matters twice over: it is prior art for this project’s interposer, and it means a video personality (#24) cannot assume the ScreenPac’s interface.

This doc covers the video side. The DD side is in docs/dd-upgrade-formats.md and docs/research-gaps.md gap #11.

“Executive” is a different machine, and treating it as an Osborne 1 is the one mistake that corrupts every conclusion below. In Osborne’s world the Osborne Executive is the O2-class machine, not a variant of the 1, so upgrades, software and advertisements aimed at it are not Osborne 1 evidence — for example the $595 double-sided-drive upgrade advertised in The Portable Companion #14, PDF p. 47, which is an Executive product. The three machines this project cares about are the Osborne 1, the Osborne 1 with OCC’s own upgrades, and clones of them (docs/research-gaps.md gap 12).

The four implementations at a glance

  OCC SCREEN-PAC (1983) TRA (1983) Nuevo (1984) Osmosis (1983)
What it is add-on board, hardware only external video unit with its own parallel printer port replacement monitor ROM + patched CBIOS + own char-gen Z80 interposer board + supplied monitor ROM
How it attaches board inside the machine (2F00040 service manual §6.12) IEEE-488 port (P3) and an external monitor swap the monitor ROM; patch the CBIOS with DDT Z80 moves onto the board, board into the Z80 socket
ROM change none — MAME models osborne1sp as a clone none mentioned — vendor-supplied driver software Nuevo 1.51 monitor ROM + BIOS 1.5 objects supplied ROM marked “Osborne 1.44 GM”
Widths 52 / 80 / 104 80 and 100, windows into the same 128×32 format 80 80 (and back to 52)
Selected by BIOS variable — poke E1A5H, call E4E5H the driver, after a “one-time, automatic disk configuration process” patched CBIOS entry (02C9 → JP 02D4) escape sequences — ESC 8 / ESC 5 / ESC F
Detected by software 2400H bank 2, bit 2 LOW — (not documented in the ad) BIOS 1.5 reads a port the O1 does not decode — (no register; the probe is a command)
Horizontal scroll 32-column blocks soft scrolling — —
Character generator stock late — its own (adds underline via row 9) stock 1.44

Sources for each row are quoted below. “—” means the material in the archive does not say; it is not a claim that the feature is absent. The TRA column is the exception: that board is not in the archive at all, so its row comes from its advertisement alone (below).

OCC SCREEN-PAC — hardware only, driven by the stock BIOS

The one that needs no ROM: docs/o1-memory-io-map.md §2a has the full statement. The parts that make it the odd one out are that the stock BIOS already contains the width-switching routine (E1A5H / E4E5H), that the board announces itself at 2400H in bank 2, and that the SCREEN-PAC’s own 2 KiB ROM is the same image as the stock late character generator (docs/o1-rom-variants.md) — so the capability genuinely is not in a ROM.

The Research Associates (TRA) — 80/100 columns, over the IEEE port

The only one of the four that reaches the screen through a different connector — and the only one the archive does not hold. An advertisement in The Portable Companion #5 (February/March 1983, PDF p. 98 = printed 84; extract research/technical-articles/TPC-Issue05-1983-pp98-tra-80-100-column-video-and-winchester-ads.pdf) announces “80 and 100 column video” from The Research Associates (Forest Hill, Dunstable, Massachusetts) as “an add-on to the Osborne 1”, providing:

  • an 80×24 window and a 100×24 window “into a 128 character by 32 line video format” — the same 128-wide buffer this project already models (docs/o1-memory-io-map.md §2a), so TRA widens the window exactly as the others do;
  • soft scrolling, reverse video and blinking, composite and TTL video monitor 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;
  • “a one-time, automatic disk configuration process”, with the driver software supplied by the vendor.

Two things make it worth recording. First, the interface is the IEEE port, not the video bus — so a video personality (#24) cannot assume an 80-column upgrade lives on the video side of the machine, and the IEEE port becomes contended between a video device, Drive C (#6/#84), the RT-60A (#18) and Trantor storage. Second, its on-board printer port is the same “duplicated port” pattern as the Drive C panel (docs/drive-c-protocol.md): the add-on re-presents an existing connector. Neither its width-switching mechanism (no ROM or BIOS patch is mentioned — a driver plus a disk configuration step) nor its presence detection is documented in the ad, and unlike SCREEN-PAC and Nuevo it is not in the archive: the ad is all we hold.

Nuevo — a new monitor ROM, and a CBIOS patched on disk

Nuevo’s video upgrade is software, and the archive holds the installer. The system disk research/nuevo/OS1NUEVO.IMD (Osborne DD: 40 tracks × 5 × 1024 B, 250 kbps MFM) contains 19 files, and two of them are the installers — not SUBMIT scripts despite the extension, but DDT patch scripts run through XSUB:

$ cpmcp -f osborne1 OS1NUEVO.IMD 0:SYS-80C.SUB -
B:XSUB
B:DDT B:MOVCPM.COM      ; patch the CP/M system image in memory
IBIOS15.OBJ             ; insert Nuevo's BIOS 1.5 object
R1F00                   ; ...at 1F00
IBIOS15BM.OBJ           ; and the second module
R29E0                   ; ...at 29E0
S2C9                    ; set the byte at 02C9
C3                      ;   to C3 (JP)
D4                      ;    D4 02 = JP 02D4  <- the 80-column hook
02
.
FA10,A17,00             ; clear the width/variable block
G0
SAVE 43 MOVE-CPM.COM    ; 43 pages = the new system
MOVE-CPM 59 *           ; write it to the system tracks

So the Nuevo mechanism is: patch 02C9 to jump into new code at 02D4, carrying a replacement BIOS (1.5) and its variables. SYS-DDEN.SUB is the same script with the DD patches added — four sites set to 64 (100 decimal, the width value used for the wide display elsewhere) — plus an ERA AUTOST.COM.

Two more files patch an application rather than the system:

$ cat DB23PTCH.SUB
XSUB
DDT B:DBASE.COM
S130 / 01        ; patch dBASE's screen handling
S154 / 50        ; 50 hex = 80 columns
S395A / C9       ; RET
G0
SAVE 70 B:DBASE.COM

That is byte-for-byte the same three-byte patch Osmosis’s manual gives for dBASE (below). Two independent vendors arrived at the same fix, because what needs patching is dBASE’s assumptions, not the board.

Plus the ROM side: Nuevo 1.51 is an OCC ROM carrying Nuevo’s copyright and patches (docs/o1-rom-variants.md), and the Nuevo char-gen differs from stock — it forces row 9 to 0xFF when bit 7 of the character code is set, which is the underline attribute made real in the font.

The Nuevo product line, from its own advertising

The Portable Companion #15 (Jan/Feb 1985, PDF p. 39) advertises four upgrades, and the 80-column one is not just a wider window:

80 COLUMN UPGRADE — Increase your screen size to a full 80 columns (& still be able to scroll horizontally if you choose). Excellent resolution, with a choice of six cursor styles and reverse video capability for easy viewing. Optional baud rate generator provides 16 baud rates from 50 to 19.2K baud. $250.00 (add $30 for the optional baud rate generator)

Three things worth taking from that:

  • “Six cursor styles” — Nuevo’s video board treats the cursor as a feature, which is the same circuit the block-cursor modifications (#24) were chasing on the stock machine and on the SCREEN-PAC. A personality that renders only the O1’s underline cursor is rendering the worst of the available options.
  • Horizontal scroll in 80 columns — the same window-sliding behaviour the SCREEN-PAC gives, here described as a user choice rather than a side effect.
  • A baud generator to 19.2K — well beyond the 9600 Hogan’s modification reaches on the stock board (docs/o1-memory-io-map.md §2a), and an option rather than a cut track. It is the other end of #21’s range.

The disk products from the same advertisement are in docs/dd-upgrade-formats.md.

The Nuevo boards themselves (2024 photo drop)

No Nuevo schematic is in the archive, so both products have only ever been known through their software. Two field photographs change that: they are the first hardware views of either Nuevo board. Source: Mike Loewen, private Facebook group, 2024-10-14 (post); archived as research/nuevo/photos/.

80-column board — silk NUEVO ELECTRONICS CORPORATION © 1984, 80 COLUMN UPGRADE REV. D, serial 4348. It is not a video tap; it is a self-contained video subsystem:

Device What it is
NEC D780C-1 (PI9455-138) a Z80 — the board has its own CPU
Nuevo SY6545A-1 (8325) a 6545/6845-class CRTC, so the board makes its own video timing
SRM2016C25 ×2 2 K × 8 SRAM — about 4 KB of video RAM
socketed CHARACTER GENERATOR ROM, 88332 SINGAPORE, dated 6-29-84 the board’s own font, under a Nuevo printed label
NYMPH 12.288 NMP122 12.288 MHz oscillator (dot clock)
74S157 ×3, 74LS194A, 74LS161B, 74LS273, 74LS374B ×2, 74LS166, 74LS174, 74LS86B, 74LS74A, 74LS20, 74LS10, 74LS02 ×2, 74LS00, 7406 ×2 glue: multiplexers, counters, shift registers, latches, drivers
VR1, Y2 video-level trimmer, crystal
J1–J6, J9–J12 configuration jumper blocks
P2, P3 (10-way edge fingers), P4 host/display interface
BAUD section, 14-pin P2111-2 the optional baud-rate generator the advertisement prices at +$30

Two consequences, and both corroborate what the software already said:

  • The 80-column mode is a second video system, not a wider window. Its own CPU, CRTC, video RAM and font are why the advertisement can sell “six cursor styles” and horizontal scroll as features — those are properties of a real CRTC. It is also why docs/o1-rom-variants.md §”one late image, five names” concludes Nuevo’s 80-column design is not the OCC one: the SCREEN-PAC widens the existing video path, Nuevo builds another one.
  • The socketed CHARACTER GENERATOR chip is dated 6-29-84 — the date in MAME’s character_generator_6-29-84.14, the Nuevo-era char-gen already tracked in docs/o1-rom-variants.md. So the font the board carries is an image the project holds, not an unknown ROM.

Double-density board — silk NUEVO ELECTRONICS CORPORATION © 1984 MADE IN USA, DAL DENS (dual density), serial 20011. This one is an encoding board, not a controller:

Device What it is
STD DRIVE silk; J1/J2; 34-pin IDC header (pins 1–34) the drive-cable side of the interface (the O1’s own 34-pin floppy header is P8)
WRITE PRE COMP silk; 74LS195A (U4), SN74LS14N ×2 (U3, U5), SN74LS00N ×2 (U2, U6), SN74LS02N (U7) a discrete write-precompensation delay line (shift register + Schmitt inverters), and the data separator
U1, 16-pin DIP marked VD9216 / CA00 / 8402 unresolved. 9216 is the number Standard Microsystems uses for its FDC9216 floppy data separator, which would make U1 the separator and put the precompensation in the LS chain — but the marking needs a higher-magnification re-read before either reading is asserted
RN1, RN2; P4 resistor networks; 4-pin power

The likely division of labour — that the DD upgrade supplies the MFM data separator and write precompensation the stock machine lacks, and leaves the controller alone — fits the rest of the archive: the controller is MB8877 ≡ WD1793 on the mainboard, and the separator is a separate board on the drive side rather than part of it (docs/o1-mainboard-schematic.md §2, sheets 9/9 and 2/3, U8 and U12–U22). That is an inference, not a finding — Nuevo’s own installation instructions are the missing piece.

Does any of this bear on Drive C? No. Both boards are host-side — video, and floppy encoding — while Drive C is an IEEE-488 peripheral that draws power from the EXT. VIDEO connector and speaks over the GPIB port (docs/drive-c-protocol.md). The photo drop is recorded here because it is the first hardware view of either Nuevo product, not because it changes the Drive C contract.

The 80-column board’s own feature list, from the shipped software

OS1NUEVO.IMD also carries AUTOST.COM — a two-instruction CP/M program (LD C,9 / CALL 5, print a string, exit) whose string is the product’s own feature list. It is the autostart banner: the 80-column installer keeps it (SYS-80C.SUB) and the DD-only installer erases it (SYS-DDEN.SUB), so it ships only with the video product. The same file is on the disk twice, as AUTOST.COM and AUTOST.BAK, byte-identical.

THIS IS AN EIGHTY (80) COLUMN DISPLAY FOR THE OSBORNE ONE.

FEATURES: (1) EIGHTY COLUMN STANDARD DISPLAY. (2) PROGRAMMABLE BAUD RATE GENERATOR (16 RATES, 50 TO 19.2K BAUD). (3) FULL SCREEN REVERSE VIDEO CAPABILITY (JUMPER SELECT). (4) CURSOR PROGRAMMABLE AS UNDERLINE OR REVERSE VIDEO BLOCK. (5) CURSOR PROGRAMMABLE FOR FAST, SLOW, OR NO BLINK. (6) REVERSE VIDEO BLOCK CAPABILITY (INSTEAD OF UNDERLINE; JUMPER SELECT). (7) COMPOSITE VIDEO OUTPUT FOR EXTERNAL MONITOR (NO MORE ADAPTERS). (8) SIMPLE USER INSTALLATION. — AUTOST.COM, research/nuevo/OS1NUEVO.IMD (extracted with cpmcp -T raw -f osborne1; the manual’s BIOS 1.4→1.5 list item 3 is the “autostart command space”, so the file name is the feature)

This is the product describing itself, and it splits hardware from software cleanly. Only item (1) is the CBIOS patch. Items (2)–(7) are properties of the board’s own circuitry: the CRTC gives the programmable cursor (4, 5), the J1–J6 / J9–J12 jumper blocks give the reverse-video options (3, 6), the BAUD section and 14-pin P2111-2 header give (2), and the board’s own video output gives (7). That is what “a second video system, not a wider window” means in practice — and it is why the advertisement can sell cursor styles and horizontal scroll as features while the software delivers exactly one of the eight.

Osmosis — a Z80 interposer with an escape-sequence interface

Osmosis’s 80-column board is the one that matters most to this project, because of how it attaches. From research/osmosis/Osmosis_80_Column_Board.pdf §3:

Locate the monitor ROM (fig 2.1). This ROM can be carefully removed … Now take the ROM we supplied … Fit the ROM we supplied (marked Osborne 1.44 GM) … Locate the Z80 processor … Remove the 40 pin Z80 … Fit the Z80 into the blue 40 way header on the Osmosis 80 column board … Fit the Osmosis 80 column board onto the Osborne PCB by inserting the 40 pins … into the socket on the Osborne PCB from which you have just [removed the Z80].

The CPU is moved onto the board and the board takes the CPU’s socket. That is a Z80 bus interposer — the same technique this project’s interposer uses (#11/#12) — which is why #73 calls Osmosis direct prior art. It is also why the board can be ROM-based: it ships its own monitor ROM.

Its interface is a terminal’s, not the BIOS’s. From §4, the applications guide:

B3. Press escape (ESC), 8 and return. The screen should now be showing a full 80 columns. … B5. Try pressing escape (ESC), 5 and return to take you back to 52 columns.

and, for software:

B7. … The board will react to a command such as: 10 PRINT CHR$(27)+"8"

So Osmosis switches width with escape sequences (ESC 8 = 80 columns, ESC 5 = 52, ESC F = auto-CR/LF) — the shape of a real terminal’s control codes, reachable from BASIC, rather than a memory variable. The manual also notes that the board “has been well proven in Europe”.

The manual’s own dBASE instructions are the same patch Nuevo’s disk applies:

(c) Enter S130 (ret) … Enter 01 (ret) … (e) Enter S154 (ret) … Enter 50 (ret) … (g) Enter S395A (ret) … Enter C9 (ret)

The escape-sequence set (Appendix E)

Appendix E is the complete control interface — every code is ESC plus one character. It is a display-type table that tesseract renders as noise (the .ocr.txt interleaves the columns), so it is hand-transcribed from the PDF here. * in the first column is the manual’s own mark for “unique to the OS/5 80-column OSBORNE 1”:

OS/5 only ASCII Hex Dec Effect
✱ - 2C 44 sets chars to inverse
✱ , 2D 45 sets chars to normal
  ( 28 40 sets full intensity
  ) 29 41 sets half intensity
✱ * 2A 42 sets 3/4 intensity
✱ + 2B 43 sets 1/4 intensity
✱ . 2E 46 normal character set
✱ / 2F 47 alternate character set
✱ 0 30 48 select default 24-line screen with up/down 48-line scroll
✱ 1 31 49 reserved
✱ 2 32 50 reserved (hand-annotated “NO scroll”)
✱ 3 33 51 reserved (hand-annotated “scroll 9 lines only”)
✱ 4 34 52 reserved
✱ 5 35 53 display 52 characters
✱ 8 38 56 display 80 characters
✱ C 43 67 reserved
  E 45 69 insert a line
✱ F 46 70 enable 80-column auto-CR/LF
  G 47 71 end graphics
✱ H 48 72 disable 80-column auto-CR/LF
✱ I 49 † 73 † invert screen
✱ N 4E † 78 † normal screen
  R 52 82 † deletes a line
  S 53 83 † insert a character
  T 54 84 † deletes to end of line
✱ U 55 85 † reserved
✱ V 56 86 † reserved
✱ W 57 87 † deletes char at cursor
✱ X 58 88 † reserved
✱ Y 59 89 † sets 60 Hz operation
✱ Z 5A 90 † sets 50 Hz operation
✱ ? — — reserved (unreadable in the original)
  g 67 103 † start graphics
  l 6C 108 start underline
  m 6D 109 stop underline

The source table has two defects. Both are recorded rather than repaired, because they are what the original prints:

  • † the Dec column is damaged. From R (52H) to Z (5A) the original prints 8- — the page edge cuts the column — and it prints 183 for g where 67H is 103. The values marked † are the correct decimal for the hex shown, not a reading of the original.
  • One row disagrees with itself. The ASCII column prints I (invert screen) and N (normal screen); the Hex and Dec columns print 61H/97 and 62H/98, which are a and b. Every other row’s three columns agree, and the table is otherwise in ascending hex order — 49H/4EH would sit exactly where 61H/62H are printed, so the hex is the likelier typo. That is inference from the ordering, not a reading, and which byte the board decodes is not settled by this manual; treat ESC I / ESC N as the intent and verify against the board.

Note what the * marks mean: (, ), E, G, R, S and T are not board-unique — they are codes a standard system already reacts to. The manual’s closing warning about them is the detection story, and it is covered in “How software finds the board” below.

The whole Osmosis product line, from their own advertising

The Portable Companion #11 (October 1983), PDF p. 32 = printed p. 30, is a full-page OSMOSIS advertisement — “The Next Step Forward” — and it gives the boards product numbers, which the manual we hold does not:

Product What it is Price
OSMOS 1 Double Density, “184k per drive using existing drives”, “fully compatible with all Osborne software” — “Install it yourself and save $165 on Osborne’s charge” $195
OSMOS 1.5 single-sided double density now on existing drives, upgradable to 370k per drive later with new drives $335
OSMOS 2 double-sided double density, 370k per drive $895
OSMOS 3 double-sided double density double tracking, 750k per drive, “an incredible 1.5MB of portability” $1050
OSMOS 4 disk formatting and conversion — “read, write and format over 20 other manufactures CP/M formats” $225
OSMOS 5 the internal 80-column board — this section’s subject $250
OSMOS 6 telex: prepare telexes in WordStar, timed send, “automatic transmission/reception in background while running any standard CP/M software”, log and queue $1195
OSMOS 7 Winchester disks, 5¼ / 10½ / 15¾ / 21 MByte units, “a 42 Megabyte max capacity” $2475–$3650
OSMOS 8 instrumentation — analog/digital, digital/analog, digital/digital cards —
OSMOS 10 Olivetti ET121 to Osborne RS-232 KSR interface, 2K print buffer $500

with a bundle — “OSMOS 1, 4 AND 5 ALL FOR $385” against $670 separately, “offer extended by popular demand until December 31, 1983” — and a house claim that matters for the interposer work:

All OSMOSIS upgrades are plug-in. There is no soldering or cutting. Our products are supplied as kits for existing units and require no modifications to the Osborne computer. No specialized knowledge is required.

That is the design philosophy this project’s interposer adopts, and the opposite of the SCREEN-PAC’s soldered fly-leads (2F00040 §6.12). Vendor address for the record: 430 Pacific Avenue, San Francisco, CA 94133, (415) 864-6372.

The 80-column board carries a hard-disk interface. The OSMOS 5 entry reads in full:

An INTERNAL 80 column board with lots of additional features and NO alteration required to existing software:

  • Full width screen > 80 columns on internal and external video
  • Normal/Inverted video · Underline · ⅔ and ½ intensity
  • Additional graphics characters for screen form design
  • Includes a Hard Disk interface for future plug-in expansion
  • At a very low price of only $250

So the video board was also the doorway to a hard disk — “future plug-in expansion”, i.e. the interface ships before OSMOS 7 (the Winchester product) is something you can buy. The board this project treats as a bus-interposer template is therefore a storage interposer too, not only a wider screen, and that is a second reason to read its manual closely for #11/#12.

Note the attribute list as well: normal/inverted video, underline, and ⅔ / ½ intensity. The stock machine exposes underline and dim as attribute bits (docs/o1-memory-io-map.md §2a/§3); Osmosis advertises two intensity steps. The mapping between the two is not established — relevant to #24.

What this means for this project

  1. A video personality (#24) needs four interfaces, not one. The SCREEN-PAC’s E1A5H/2400H, Nuevo’s patched CBIOS, Osmosis’s escape sequences and the TRA’s IEEE-port driver plus disk configuration are four different contracts. Emulating the window is not enough; the selection mechanism — and, for the TRA, the connector — differs per board.
  2. Osmosis is prior art for the interposer, not just for video (#11/#12): a board that takes the Z80’s socket and drives the bus is the shape being designed here.
  3. The application-patch problem is universal. WordStar and SuperCalc needed settings on the SCREEN-PAC; dBASE needed a three-byte patch on both Nuevo and Osmosis. Any 80-column story for this project has to answer for software that assumes 52 columns — the practical reason the O1’s upgrades shipped with application notes.
  4. The 128-character row is the common ground. All four widen the window, not the memory (docs/o1-memory-io-map.md §2a); only the control interface differs.

How software finds the board

Three of the four answer “is the board there?” in three different ways, and only one of them is a register — the TRA’s advertisement documents no detection at all.

Board How software detects it Status
OCC SCREEN-PAC read 2400H in bank 2: bit 2 LOW = board fitted documented (user guide p. 13, docs/o1-memory-io-map.md §2a)
Osmosis no detection register — the interface is commands, so a program probes by sending an OS/5-unique escape and watching the screen the manual’s own warning is the answer; its paid Technical Manual is where “the CP/M interface” is documented
Nuevo BIOS 1.5 reads a port the Osborne 1 does not decode; if it is not zero (i.e. no board), the 80-column configuration is skipped read out of the two CBIOS images (#173); the one step needing a board is named below

Osmosis: the probe is the interface

The closing note to Appendix E is the answer, stated as a warning:

Commands which are unique to the OS/5 80-column OSBORNE 1 are annotated with an * symbol. Use of these commands on standard systems may cause spurious data to be displayed on the screen; the program should not be otherwise affected.

That is the contract. There is no flag to read; a program that wants to know whether the board is there sends a *-marked code and looks at the screen — and the manual is telling the programmer that doing so on a stock machine is safe but visible, which is a design decision rather than an omission. It is the opposite of the SCREEN-PAC’s passive status bit, and it means a video personality (#24) has to model command effect, not a register, to be Osmosis-compatible.

Nuevo: the check is a read of an undecoded port (#173)

The DD manual says BIOS 1.5 does the check (quoted in docs/dd-upgrade-formats.md), and we hold CBIOS 1.5 in two forms: the DDT objects the installer inserts (BIOS15.OBJ, BIOS15BM.OBJ) and the relocated system the disk actually boots — the system tracks of OS1NUEVO.IMD. The latter confirms the manual’s item 1: its sign-on reads

Osborne 1 Computer
CP/M vers 2.2
CBIOS vers 1.5

against the factory disk’s Osborne Computer System / 59k CP/M vers 2.2. The 59k is gone, exactly as the 1.4→1.5 list says.

The check is the first thing the cold boot does that 1.4 does not. Both CBIOSes open identically — stack, one CALL, five variable loads — and then part:

  1.4 (O1CPM.IMD) 1.5 (OS1NUEVO.IMD)
cold boot E4ED–E513, 38 bytes E469–E4BE, 85 bytes
after the five variable loads XOR A / LD (0004H),A / INC A 47 new bytes, then those same three

and those 47 bytes open with a guard:

E488  IN   A,(04H)      ; read port 04H
E48A  AND  1FH
E48C  JR   NZ,E4B7H     ; not present -> skip the configuration

Port 04H is not a device on the Osborne 1. Its I/O decode uses only A0/A1 (docs/o1-memory-io-map.md §4), so a write to 04H is an alias of the bank-2 latch — but a read of it is answered by nothing on the machine, and a floating TTL bus reads back non-zero. On a stock machine the guard therefore takes the skip; with the board fitted, the board has only to pull those five bits low. That is the manual’s behaviour exactly: “If not present, the configuration routine for the 80 Column Upgrade will be skipped.”

What the guard protects is a video configuration, and that is the rest of the evidence:

  • it reads the config byte at BIOS variable E1A6H — the address immediately above E1A5H, the width variable the SCREEN-PAC user guide documents (docs/o1-memory-io-map.md §2a), so Nuevo kept OCC’s variable layout rather than inventing one;
  • it then drives the BIT 9 latch with OUT (06H),A — 06H aliases 02H, the latch that maps bank 3, the video/dim bank, into F000–FFFF (docs/o1-memory-io-map.md §4). The factory BIOS never touches that latch: its only bank selects are ports 00H and 01H.

The consistency check that lifts this above a guess: BIOS 1.5 ships on both the DD-only disk and the 80-column disk — the installer inserts the same BIOS15.OBJ either way — and it is precisely a shared BIOS that needs the check, because the DD customer has no video board for the configuration routine to talk to. That is the manual’s stated reason, verbatim.

The one step that needs a board: that IN A,(04H) reads zero with the card fitted. Everything above is read out of the two images. If that read is not zero with the card fitted, the guard is something else and this subsection is wrong — so it is recorded as the evidence supports it, not as established fact.

Open

  • Nuevo’s presence check — the mechanism is read out of the images (#173); the remaining step is one measurement, IN A,(04H) with the board fitted.
  • Osmosis’s technical manual (“theory of operation; documentation on the CP/M interface; cursor addressing”) is a paid document the archive does not hold. If a detection register exists, that is where it is — but the user manual’s own warning suggests there is not one.
  • The DD halves of both vendors are covered, in docs/dd-upgrade-formats.md: Nuevo’s boot-time auto-configuration via NEWSETUP.COM and its 2 x 2 / 4 x 4 formats, and Osmosis’s OSMOS 1–3 density line plus the standalone OSMOS 4 conversion product.

Sources