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: Solved. The Executive manual is a direct port of the Osborne 1 one and the hardware is one design for both machines; the Executive’s own loader software (ECL/ECN/ECU) is not in the archive.

The Executive Drive C is the same box as the Osborne 1’s Drive C — same hardware, same IEEE-488 contract — sold against a different host. The Executive runs CP/M+, so its software set is ECL/ECN/ECU in place of DCL/DCN/DCU and there is no DFD.SPR to substitute for a CCP (§2, §4). The manual settles four things: the mechanical, electrical and performance appendices are word for word the same, apart from the connector the unit draws power from and a 0.1 s better load time (§1); the CP/M parameter table carries the same numbers, which reconciles the two books’ different “useful capacity” figures (§3); DFD.SPR is a replacement CCP rather than the transport, which is why the Executive needs none (§4); and the product’s own printer-cable appendix corroborates the P3 pinout we transcribed from the Technical Manual, line for line (§6).

Source: research/drive_c/Drive_C_Executive_Users_Manual.pdf, OCR layer at Drive_C_Executive_Users_Manual.ocr.txt. Drive C:, 1690 65th Street, Emeryville CA. First edition July 1984, second printing 9/84; © 1984 Drive C:, a joint venture. 111 pages, image-only scan, OCR’d at 300 dpi per the research/README.md convention. The Osborne 1 manual already in the archive is Drive_C_Users_Manual.pdf — second edition, June 1984, same publisher.

Provenance: supplied by Jim Rayburn from a private Facebook group, posted 2024-01-06 — two posts, …/2063148960719887 and …/2023247321376718; recorded in research/README.md.

Bibliographic note: the cover is stamped PRELIMINARY, while every running header in the body reads “(final draft)” — so the PDF is a draft/review copy, not a production print. Its content is complete: the appendices are all present and internally consistent (§3).

1. The headline: the hardware is one design

§6-2 (Executive) and §7-2 (Osborne 1) are the same appendix. Verbatim, apart from one line:

Specification Osborne 1 manual §7-2 Executive manual §6-2
Dimensions 5.6 × 5.9 × 1.0 in (14 × 15 × 2.5 cm) identical
Weight 1 lb (0.45 kg) identical
Shipping weight 3 lb (1.4 kg) identical
Operating temperature +5 to +60 °C (+40 to +140 °F) identical
Storage temperature −20 to +100 °C (−5 to +212 °F) identical
Power 11 to 14.5 VDC at 0.3 A max (2.5 W nominal, 384K unit) identical
Source “the O1 EXT. VIDEO connector” “the Executive EXT. VIDEO connector”
Latency none (solid state, no moving parts) identical
Max data throughput 55,000 B/s (445,000 b/s) identical
Random-sector access ≈40 µs (software access time) identical
Load a 32K .COM and run it 1.1 s max 1.0 s max

The only substantive differences are the connector the unit draws power from and a 0.1 s better load time on the Executive — which is a host difference, not a Drive C one.

This is the useful result for #84. The Drive C is not two products; it is one box with two software personalities. The docs/drive-c-protocol.md contract is therefore the contract for both, and a replacement built to it serves either machine. It also means the P3 work in docs/o1-ieee488-port.md §6 transfers wholesale — the Executive’s port is the same port.

The power figure is worth quoting precisely because it is the one number a reproduction has to beat: 2.5 W nominal for the 384K unit, from a 11–14.5 V rail. A Pico plus flash plus SRAM is far inside that.

2. The software is not the same

The two manuals describe the same product in the same words, so the diff is clean. Everything below is a real difference:

  Osborne 1 Executive
Host operating system CP/M 2.2 CP/M+ (CP/M Plus), OCC BIOS 1.0/1.1
Loader DCL.COM ECL
Rename the unit DCN.COM ECN.COM
Utility program DCU.COM ECU.COM
QuickPac installer DCQINS.COM ECQINS
CCP replacement DFD.SPR — 3K (192K) / 4K (384K) none
Utility options S space, O optimize, buffer, no-menu, E erase S space, I install/un-install, buffer, D file delete, no-menu, E erase
System utilities appendix COPY, SETUP, SYSGEN, MOVCPM COPY, COPYSYS, SETUP
Floppy orientation left / right upper / lower (the Executive stacks its drives)
ZCPR appendix §7-8 absent
CP/M tutorial section Section 4 absent
Useful storage capacity 187 KB / 376 KB 190 KB / 380 KB
TPA RAM, no print buffer 1152 B 768 B
TPA RAM, with print buffer 2176 B 786 B
TPA RAM, with TurboPac 3172 B 1024 B
Disk overhead, 384K unit 8K min, 20K with options 4K min, 16K with options

Two of these are large enough to be worth stating plainly.

The Executive runs CP/M+, and that is why the driver differs at all. CP/M+ is a different system with a different BIOS and a different memory map, so the Executive’s loader has to patch a different host. Everything the loader does is host-side; none of it touches the wire.

The Executive variant has no DFD.SPR, and uses a fraction of the TPA RAM. Both follow from the first point, and §4 explains why.

3. The CP/M parameters are identical — and the capacity figures reconcile

§6-3 (Executive, headed “CP/M+ PARAMETERS”) and §7-3 (Osborne 1, “CP/M PARAMETERS”) carry the same table, with the same numbers:

Parameter 192K unit 384K unit TurboPac
Useful storage capacity 190 KB 380 KB 10, 20 or 33 MB
Block size 1K 2K 8K (modifiable)
Directory entries 64 128 512 per logical drive
Directory space 2048 B 4096 B 16384 B per drive
Logical sectors 1536 3072 varies
Reserved tracks 0 0 612, landing area
Logical sectors / track 2 2 32
Bytes per sector 128 128 256

The three rows from Reserved tracks down are printed once, centred across the two RAM-disk columns — the 192K and 384K units share them — while the TurboPac column carries its own values. The OCR loses that alignment; read column-wise it would put 32 sectors per track and 256-byte sectors on the 384K unit, which the DPB DCL.COM installs refutes (SPT = 2, 128-byte records, docs/drive-c-protocol.md §8.1).

The 384K column is our reverse-engineered DPB, field for field. Block size 2K is BSH = 4; 128 directory entries is DRM = 127; 4096 bytes of directory is AL0 = C0h (two 2K blocks). And the capacity falls out exactly: DSM = 191 means 192 blocks of 2K = 384 KB total, of which the directory takes 2, leaving 190 blocks = 380 KB usable — the Executive’s own figure.

That in turn explains why the two manuals quote different “useful” capacities for the same unit, which had looked like a documentation error:

  192K unit 384K unit
Raw capacity 192 KB 384 KB
less the directory −2 KB −4 KB
= Executive’s “useful” 190 KB 380 KB
less DFD.SPR −3 KB −4 KB
= Osborne 1’s “useful” 187 KB 376 KB

Both books are right. The Osborne 1 figure is net of the CCP replacement the Osborne 1 needs and the Executive does not — which is a neat independent confirmation of §4.

4. DFD.SPR is a replacement CCP, not the transport

DFD.SPR is a substitute console command processor — a CP/M program that CP/M reloads on every warm boot in place of the Osborne 1’s CCP.COM. It is not the transport and it is not device firmware: the transport is DCL.COM’s (docs/drive-c-protocol.md §3). The Osborne 1 manual’s software appendix says so in the vendor’s own words, and the research/README.md row for the FOGHORN review says the same.

The Osborne 1 manual settles it, in the vendor’s own words and in more detail than the review had:

The DFD.SPR file is absolutely necessary to Drive C operation. It’s located in user area 15. DO NOT ATTEMPT TO ERASE DFD.SPR FROM USER AREA 15. This file is loaded each time you type ^C and each time a program exits to CP/M. … The Drive C software automatically loads the DFD.SPR file to replace the O1 Digital Research CCP which would normally be loaded from the drive A: floppy disk after each ^C.

So the picture is:

  • DCL.COM is the loader and carries the transport. It announces itself as DRIVEC version 2.08, carries DCN version 2.07 inside it, and creates DCN.COM on the unit when it runs.
  • DFD.SPR is the command processor that CP/M reloads on every warm boot in place of CCP.COM. It is a CP/M program, not device firmware — which is exactly why its committed listing “sweeps through zero blocks”: it is a CCP image, not a driver.
  • It must be present on logical drive A: during initial installation, and thereafter lives on the unit in user area 15. Its name “has no significance, but DCL always checks for its existence each time DCL is run.”
  • It costs 3K on a 192K unit and 4K on a 384K unit — a number the review did not give, and the one that makes §3’s capacity arithmetic close.

Why the Executive has none of this. The Executive runs CP/M+, whose CCP is a different animal; the O1’s trick of substituting a custom CCP on warm boot has no counterpart there. That is the whole reason the Executive variant is 4K lighter on the unit and uses 384 fewer bytes of TPA without a print buffer (768 vs 1152), and 1390 fewer with one (786 vs 2176).

What this changes for #84. A replacement does not need to reimplement DFD.SPR: it inherits it. The CCP that ships with the product already talks to whatever driver is installed, so a box that satisfies drive-c-protocol.md §7 gets the command layer for free, and no disassembly of DFD.SPR is needed — docs/drive-c-protocol.md §10 item 4 records that.

5. The disk and RAM budget, from the manuals

§7-4 (Osborne 1) and §6-4 (Executive) give the unit’s overhead directly. This is the passage that answers whether a fixed print buffer costs RAM-disk space — it does:

  Osborne 1   Executive  
Function 192K 384K 192K 384K
Directory space 2K 4K 2K 4K
DFD.SPR 3K 4K — —
DCN.COM / ECN.COM 1K opt 2K opt 1K opt 2K opt
DCU.COM / ECU.COM 10K opt 10K opt 10K opt 10K opt
minimum 5K 8K 2K 4K
with options 16K 20K 13K 16K

DCL.COM also creates a second system file the disk-space table does not list — the manual’s screen illustration shows DCLOADED.SYS and DRIVEC.SYS on the unit, and notes that DRIVEC.SYS is “a SYSTEM file and will not be displayed by DIR. To check for the DRIVEC.SYS file, use XDIR or STAT.” (Those two names come from a poor OCR of a screen dump, so treat the spelling as provisional; what is solid is that there are two, they are hidden system files, and DCL creates them.)

RAM. The O1 needs 1152 bytes of the 64K, plus 1024 more if the print buffer is installed — and the buffer’s RAM “is not used unless you install the Buffer at the time you run DCL.COM”. The manual even names the consequence: applications that use all of the TPA (it cites DeskTop Accountant) may need the print buffer removed to load.

Print buffer, in the unit’s own memory:

  • Fixed Buffer Capacity: user-selectable 16K to 128K in 16K steps.
  • Dynamic Buffer Capacity: up to the maximum space available on the unit.
  • TurboPac spool buffer: 128K fixed on the 384K unit, not installed on the 192K unit.

That last line is the 192K/384K TurboPac split, and it is also the explanation for the print buffer’s disk cost: a fixed buffer is carved out of the RAM disk, so a 384K unit running TurboPac is 256K cache + 128K fixed print buffer — which is precisely what the Osborne 1 manual says in its feature list.

6. The printer connector is P3, re-presented — and it checks our pinout

Both manuals carry the same appendix (§7-5 / §6-5) giving the cable from the unit’s PRINTER/HARD DISK/8088 connector to a Centronics printer. That connector is a 26-pin card edge — the same edge as P3 — and its pin assignments match our transcribed P3 table (docs/o1-ieee488-port.md §6, §6.2) line for line:

Drive C 26-pin edge Signal (manual) Our P3 table Centronics
1, 3, 5, 7 data bits 0–3 DIO1–DIO4 (odd pins) 2, 3, 4, 5
2, 4, 6, 8 data bits 4–7 DIO5–DIO8 (even pins) 6, 7, 8, 9
11 Output Strobe DAV 1
13 NRFD NRFD (12, PE — must be cut)
15 Printer Busy NDAC 11
19 SRQ SRQ 13 (SLCT OUT)
12, 14, 16, 18, 20, 22, 23, 24 ground the same eight grounds 19–30

Four things this buys us:

  • The interleave is confirmed by a second, independent source. Odd pins are data bits 0–3, even pins are 4–7 — the vendor’s own cable table, not just Figure 4.1.2.
  • The ground set is confirmed exactly — pins 12, 14, 16, 18, 20, 22, 23, 24, the same eight our table lists.
  • The dual-purpose pin 15 is confirmed. It is NDAC on the bus and Printer Busy in the printer personality — precisely the “one pin, two meanings” constraint docs/o1-ieee488-port.md §6.2 already recorded, now corroborated by the product that has to live with it.
  • The Drive C watches pin 13 (NRFD). The manual is emphatic that a cable carrying Centronics pin 12 (Paper Error) to pin 13 “will interfere with normal Drive C operation if connected and MUST be removed”. So the box must leave NRFD harmless as well as NDAC — a constraint for docs/drive-c-design.md §2.3.

One discrepancy between the two tables. Our P3 table has pin 10 as REN (CB2), while the Drive C’s cable table lists pin 10 as ground (tied to Centronics pin 30). Either the Drive C’s cable grounds a line it does not need, or the two connectors are not bit-identical on that pin. It does not affect anything a replacement has to do — REN is unused by every personality we care about — but the two tables disagree and that should be noted rather than assumed away.

7. What the Executive manual does not have

Worth listing, because absence is evidence about how the product was maintained:

  • No DFD.SPR, and no equivalent. CP/M+ needs no CCP substitution (§4).
  • No ZCPR appendix. The O1 manual’s §7-8 covers running ZCPR with Drive C; the Executive’s appendix list stops at §6-6.
  • No CP/M tutorial. The O1 manual’s Section 4 teaches CP/M; the Executive manual renumbers everything up by one and drops it, so its Section 4 is the RAM-disk/print-buffer/utility section.
  • No MOVCPM rule. The O1’s §7-1 warns that MOVCPM “must be run BEFORE running the Drive C Loader, DCL.COM” or it reports a “Synchronization Error”. The Executive’s §6-1 has no equivalent — CP/M+ is configured with SETUP, and GENCPM-style rebuilding is a different mechanism.
  • No fourth system utility is named. The Executive’s §6-1 opens “There are four programs supplied with your Executive” and then documents three: COPY, COPYSYS and SETUP. The O1’s §7-1 names all four of its own (COPY, SETUP, SYSGEN, MOVCPM). Whether the Executive’s fourth was dropped from the draft or simply never written is not determinable from this copy.

8. What this changes

For #84, the Executive variant is not a second design to build. It is the same box with a different loader, and the contract in docs/drive-c-protocol.md is the contract for both. The reproduction target does not change.

It removes work rather than adding it. DFD.SPR is a CCP, it ships with the product, and a replacement inherits it — so no disassembly of it is needed (§4).

It closes the fixed-print-buffer question. A fixed buffer is carved out of the unit’s RAM disk, and the manuals give the budget table (§5). The 384K TurboPac split — 256K cache + 128K fixed print buffer — is stated outright.

It corroborates the P3 pinout from a second vendor document (§6), including the interleave, the ground set and the dual-purpose pin 15.

It supplies the mechanical, electrical and performance envelope (§1): 5.6 × 5.9 × 1.0 in, 1 lb, 11–14.5 VDC at 0.3 A max and 2.5 W nominal for the 384K unit, 55,000 B/s throughput, ≈40 µs random-sector access. Those are now targets rather than unknowns in docs/drive-c-design.md.

What is still open

  1. How the box resolves banks 0–2 (drive-c-protocol.md §10 item 2) — unchanged by any of this, still the sharpest question, still not a blocker.
  2. Bus loading (drive-c-design.md §7 item 2) — unchanged.
  3. The Executive driver software. The manual tells us what the Executive’s loader is called (ECL) and what it does, but we do not have it. If one surfaces, the interesting comparison is why the Executive driver needs 768 bytes of TPA where the Osborne 1’s needs 1152 — the answer is in how each patches its host, and it would tell us how much of the driver is host adaptation versus device work.
  4. DCLOADED.SYS and DRIVEC.SYS (§5) — two hidden system files that DCL creates on the unit and the disk-space table does not itemise. Their names come from a poor OCR of a screen illustration, so even the spelling is provisional.
  5. Pin 10 (§6) — REN in our P3 table, ground in the Drive C’s cable table.

Sources