The PET Breeds: Teachers' PET, SuperPET/SP9000, MDS 6500, 4032P, The PET Register, PET 200

Teachers' PET

.jpg Image of the Teachers' PET (61K, courtesy Stephen)

aka NOT the Educator 64!
Introduced 1978?
Hardware, Graphics and Sound Standard 8K "Chiclet" PET 2001 with tape recorder, and presumably with original BASIC version.
Eventual Fate Released as branded, unknown release number.

MDS 6500

.jpg Image of Co-Branded 2040 (55K; courtesy Leo LaFlamme, from a George Page picture)

aka Microcomputer Development System
Introduced ?
Hardware, Graphics and Sound Specially modified PET 2001-32N with built-in software and matching 2040 disk drive badged "Commodore Semiconductor Group." (In fact, the backplate of George's unit still says 2001-32N.)
Eventual Fate Unknown. Apparently only 500 were made.

Comments
What is this thing? What was it built for? The annotation "Commodore Semiconductor Group" alleges that this might have been customized for integrated circuit design, but no one seems to know how.

SuperPET "MicroMainFrame"

.jpg Image of the SuperPET (22K; courtesy Marko, Andreas and George)
.gif Image of the SuperPET (24K)
.gif Image of a SuperPET/B128 Ad (167K) -- shows some sample software. See index page for credits.

aka SP9000, CBM 9000 (NOT CBM 900!), MMF 9000, MicroMainFrame
Introduced 1981?
Hardware Modified 8000-series (probably mostly 8032s) PET with true RS-232 port (@ 9600 baud), 96K RAM, and a number of languages (apparently Cobol, BASIC, APL, Pascal and Fortran) written in Waterloo Systems Language (see the microWAT below). Sold in Germany as MMF (MicroMainFrame) 9000. Twin 6502 and 6809 CPUs. Machines sold in Italy had 134K.
Graphics and Sound Identical to the 8000-series PETs. Had four built in fonts.
Eventual Fate Released, mostly as a development/programming machine. Virtually unknown outside the academic sector/development sector in those days.

Comments
The SuperPET was at least partially designed by the University of Waterloo (Canada); see its history at the end of this comment block. It was usually a development tool, where programmers could send/receive their work over modem and exchange data with remote mainframes for development.

According to Hans Franke, the SuperPETs have a daughterboard that contains the 6502 and 6809, 64K RAM, a 6551 ACIA, a ROM socket (see below) and a replacement Kernal ROM with 6809 instructions for when the 6809 was active; the daughterboard replaced the 6502 by connecting to its socket. Hans notes that there are also some boxes that are labeled MMF9000 but don't have a 6809 (so they're really just 8096s with extra RAM and not SuperPETs at all: love that Commodore naming scheme).

The additional 64K on the board is mapped into main memory through a LIM EMS-like strategy to a 4K window at $9000-$9FFF. I/O registers $EFFC-D select one of the 16 4K tracts in the expanded RAM and map it to $9000. Additionally, the daughterboard ROM socket can be banked into the $9000 range through $EFFE-F, probably to allow the 6502 to bootstrap itself after the 6809 has ceded control. The daughterboard can be hardwired to select the 6809 automatically, or a register can toggle between the 6502 and 6809 (the 6502 goes into reset when the 6809 goes down, so a bootstrap is necessary); the 6502 and 6809 all share the same address, data and control lines, so both CPUs all see the same things except ROM: each unit sees the ROM written for that unit only, so the 6809 isn't running 6502 Kernal instructions and vice versa. This is in contrast to the 8096, which is actually an add-on for the 8032 providing two software-selectable 32K RAM banks. The RAM could even be write-protected (!), according to Andre Fachat.

A pesky annoyance on some SuperPET boards is the 6702, a mysterious chip that Commodore refused to document. Probably some form of enable chip to prevent knock-off hacks from being used instead of buying the board from Commodore, the 6702 appeared in I/O space connected through the RAM data bus. Early SuperPET units required the 6702 on a separate PCB to be connected to the 6809 daughterboard.

Andre Fachat remembers an advertisement linking the CBM 9090 hard drive with the SuperPETs.

Dave Dunfield and Heinz Wolter provide us with information on the microWAT, the SuperPET's probable ancestor, which is a pure 6809-based system consisting of several boards on a backplane (typically a CPU card, 48 to 96K of bank-switched RAM, 64K of bank-switched ROM and an IEEE-488 bus card). These bank-switched ROMs then bootstrapped the machine monitor and the various Waterloo Structured interpreted languages (themselves written in a C-like portable development dialect called WSL "Waterloo Systems Language" -- these are apparently the same language ROMs as in the SuperPET, with WSL being used to abstract the hardware). Unlike the PETs, microWATs did not have intrinsic video capability and were originally intended for dumb Volker Craig terminals. U Waterloo developers D.D. Cowan and J.W. Graham later took the microWAT architecture and modified a PET (again, probably an 8032) to accomodate the extra RAM (in this case, 64K) and 6809 CPU, as well as an RS-232 interface; the result was the original SuperPET.

As a consequence of their ancestry, the SuperPET's Waterloo languages run on the 6809 and completely independently of the 6502, using the PET hardware only for I/O. In fact, there are no options to make a 6502-based object at all!

As for the microWAT itself, it lived on as an educational tool and 6809 assembly trainer at the U Waterloo CS labs.

4032-P

The Commodore Cash Register

aka PET Register, PET 3220
Introduced Never officially; development started late 1980.
Hardware Specially modified 8032 with cash register pad, integrated monitor and printer.
Graphics and Sound Identical to original unit.
Eventual Fate Approximately 50 units made. Eventually deployed to four beta test sites (locations?); project scuttled mid-1982.

Comments
Thanks to Ric Rainbolt for almost this entire entry.

Calling the CCR a modified 8032 doesn't really do it justice. The custom motherboard featured 32K of battery-backed CMOS RAM instead of the normal DRAM, and instead of one 4K ROM, it had seven. The unit carried a built-in 9" green screen alongside an Epson 522 retail receipt printer, and a 160-plus key Point of Sale keyboard. The extra ROMs were based on the stock Kernal, but additionally carried on-board routines for driving the monitor and printer and maintaining the CMOS RAM. The reason for the battery-backed RAM, of course, was to make the machine resistant to power failures and allowing it to resume normal operation when power was restored without data loss. The internal printer and the other PoS hardware were driven by the "Magic Board," an internal daughterboard with a 6504? CPU and its own ROM and interfacing logic, bolted to the inside roof of the CCR's heavy steel case.

Ric also notes that the CCR was one of the first CRT-based PoS systems (of course, nowadays, CRT PoS devices are everywhere). The CCR's firmware supported inventory management, retail PoS, shipping and receiving, and a battery of built-in sales reports; it even could use laser scanners, which were a novelty at the time the CCR was released, and Mike Shartiag says a scale interface was available as well. All the functions of the CCR were accessed by a ROM-based cursor-selectable menu interface, which in 1980 was unheard of on a Point of Sale-level device.

The board annotation "Universal Dynamic PET" is intriguing but actually only proves that the unit is an 8000-series, since all Fat Forties and 8000-series PETs use this same abstractable motherboard.

The CCR was eventually deployed to four retail stores which were using it as their only PoS terminals (apparently the devices performed very well in this capacity), but the project was killed off by Tramiel in 1982 and rumour has it that the rights were eventually sold to a Japanese PoS concern. Nevertheless, the Register never did resurface ... or did it?

PET 200

Introduced 1979?
Hardware Identical to the 8032-SK ("separated keyboard"), with new rounded case.
Graphics and Sound Identical to the 8032.
Eventual Fate Released in Europe.

Comments
This very sexy pet (no Penthouse jokes, please :-) is an unusual rounded model, basically a rebadged 8032-SK. Per's unit here is fully functional and showing the standard BASIC 4.0 screen. Marko Mäkelä described it and the B series as "the best looking computers I've ever seen"; his patent search revealed the designer of the casing to be one Ira Velinsky, Plainfield, New Jersey. This is, of course, the same Velinsky who designed the B/P series and other cases for Commodore, obviously the form factor on which those later high-profile systems were based.

PET 8296-GD

Introduced 1980?
Hardware, Graphics and Sound Identical to the PET 8296 with built-in high resolution graphics expansion board and 8250 disk drive (see below).
Eventual Fate Known released in North America.

Comments
Thanks to Steve Gray.

The PET 8296-G is a PET 8296 with a built-in 8250 disk drive; the PET 8296-GD is a PET 8296-G with an HRE ("HiRes Emulator", Assy#324890) board. These boards were based on Commodore Germany's HSG ("High Speed Graphik", Assy#324402) board with the Thomson EF9365/6 chip. The HSG came in 512x512 and 512x256 monochrome display variants with its own RAM, but the HRE only modifies the built-in 8296 video and has no video RAM of its own. Steve notes that the board attaches to the CRTC and CHRGEN chips on the 8296 board and came with extended BASIC ROMs and CAD software written specifically for it.