Sothius' Home

-

Museum

-

Commodore

-

Commodore 8296

-

additional pictures

Commodore 8296
Additional pictures. Note that loading the page could take a while...
(pictures copyright by M.A.Grundke)

serial number: C= commodore 8296 SER.NO.WG 15083

Commodore 8296 opened up (from behind): the first interesting detail is that the 8296 (i.e. its upper part, onto which the CRT display is mounted) opens up to the front, unlike the older PETs (e.g. CBM 3016) did, although there's no appearant reason for that. Another difference to these older models is a metal pin (right) to support the upper part while the machine is opened up, making maintenance easier. The case's left part is dominated by the mainboard (see an overview below), and the power supply to its right. The PSU itself is very similar to that of the older machines, since power consumption and case proportions didn't change much; PETs with integrated peripherals, however, came with an actively cooled, more powerful PSU (such as the 8296's bigger brother, the Commodore 8296-D). Also notable in that context is that the 8296-D series made use of the same mainboards as the 8296s, which reduced overall production cost

ASSY NO.324645 REV.B board overview: beginning at the machine's front (which is the board's back from that point of view), you can see the 25 pin SUB-D keyboard connector (at location J13). To its right, the machine's piezo-electric speaker (at MT1, responsible for the 8296's poor sound output - but what did you expect from a 1984 office workstation?). Below the keyboard connector and a 74LSxx at UE1, one of the two R6520AP 'PIA' is located at UE2 (made by Rockwell, like the other one). Further below, there's four more logic devices (three 74LSxx and an identical MOS 65245) at UE3, UE4, UF1 and UF2, with the 6 pin CRT connector to their right. Below these are two MOS 8700 located at UE5, UE6, the BASIC ROM (UE7) and the three function ROM sockets (UE8, UE9, UE10). Socket UE8 is occupied by a 901474-04 screen editor. Left of the function ROM sockets, you can see the internal expansion port (2x 50 pin connectors at J4 & J9), and the machine's second cassette port at J6 (which is also different from the older machines, since it's accessible from the outside). Back to the CRT connector mentioned earlier: right from there, surrounded by lots of 74LSxx logic, resides the MOS 901447-10 character generator (UC5), and below, the MOS 6545 CRTC (UC9) - no doubt, this area generates the video signal. Right from the CRTC are two 74LS257APC (UB10 & UB11), responsible for the machine's 80 column output (but can apparently be configured for 40 columns, see the closeup below). Right from there, the massive voltage regulator's heatsink hits the eye, and above that, you can see the machine's RAM bank. Getting back to the CRTC, there's the MOS 6502 CPU below it (UC11), and further down the machine's I/O area, consisting of a MOS 6522 'VIA' (UC12), the second R6520 (UC13) and three MC3446N (UD9, UC14, UC15) IEEE488 transceivers. On the lower edge, finally (from left to right), the 8296's interfaces: cassette port (J3, with a reset-switch above it), user port (J2) and IEEE-bus (J1)

board silkscreening, part 1: (C) COMMODORE 80/ 40 COLUMN CPU 64/ 128K ASSY NO.324645 REV.B MADE IN W.GERMANY Like the 'Dynamic PET boards' introduced in some older machines, this board can easily be configured for different RAMs, ROMs and 40/ 80-column operation. Commodore had appearently learned that such a standardized but modular board can greatly reduce production cost of the whole series, and makes maintenance and replacement much easier. However, if you have a look at some later machines of the 64- or 264-series, maybe they didn't learn that lesson well enough... Earlier 8296(-D), by the way, were based on older revisions of this board (ASSY 324644 and ASSY 324645 REV.A, although I don't know the differences between them)

board silkscreening, part 2 (lower side of the board): FABRICATION NO 324646-01 ARTWORK NO 324647-01 B

The machine's MOS 6502 CPU, here in the 'B'-version which can be clocked with up to 3 MHz (of course, not in the PET-series - they were all clocked at 1 MHz, but I assume that the slower types' production was simply stopped in favour of the faster ones). The MOS 6502 and its descendants were used in all 8 bit Commodore computers; invented by Chuck Peddle, whose company MOS Technology had joined forces with Commodore in 1976, these processors were the foundation on which the microcomputer revolution was built. The 6502, for example, also was the heart of the famous Apple ][, a best-seller and major Commodore-competitor. Similar to today's companies, however, not all wanted to just buy MOS-CPUs; Zilog had their own plans of dominating the microprocessor-market with their Z80. Looking at the facts today, we know that only few popular machines used the Z80, like AMSTRAD's CPC-series (whose prototype was, ironically, based on a 6502 as well, but was changed because Locomotive BASIC wasn't available for MOS-CPUs) and Sinclair's ZX-series

The Commodore 8296's RAMs, 128 KBytes organized as 16 pcs. 64Kx1 RAMs. Interesting is that this machine has a 'mixed' memory configuration (i.e. 8 pcs. NEC D4164C-2 and 8 pcs. Fujitsu MB8264A-12) - since there are no soldering traces, they all must have been there from the fab (but as mentioned earlier, the board can also be configured for 64K operation). The whole amount of memory could not completely be accessed under BASIC; for that purpose, Commodore offered the alternative operating system LOS96 ('loadable operating system for 96K') for their bigger PETs, which could access 96 KBytes. Furthermore, the MOS 6502 with its 16 bit address bus cannot address all that RAM without 'tricks' - bankswitching is used here, where memory is divided into 16K banks which can individually be selected. Also, the ROMs 'overlay' part of the memory, since they also have to be in the CPU's address range to operate. To access the RAM there, the ROMs can temporarily be disabled, memory blocks can be moved etc. Other architectures (like ATARI's XL/ XE-series, for example) use dedicated chips for these functions. In the PETs, they are, among others, provided by the two MOS 8700 PLAs discussed below

Two of the 8296's ROMs: MOS 901474-04, residing in one of the function ROM sockets (UE8), is the screen editor for BASIC 4-machines with business keyboard (i.e. without graphical characters printed to the keycaps) and 80 column/ 50Hz operation. There are other dedicated versions for 40 column-, 60Hz- or CRTC-less machines, and for either normal or business keyboard variants, or localized keyboard layouts. MOS 324746-01 is a 23128-type ROM (16 KBytes), which replaces four ROMs used in older machines: 901465-20 (BASIC4 $C000), 901465-21 (BASIC4 $D000), 901465-22 (BASIC4 KERNAL) and 901465-23 (BASIC4 $B000), reducing production cost and improving reliability. Compared to the simpler BASIC 2.0, BASIC 4.0 features many additional disk-related commands for professional work (which can also be found in later CBM basic versions, such as BASIC 3.5 and BASIC 7.0)

The machine's character generator ROM, MOS 901447-10, containing the standard ASCII charset in this case (which makes sense, since the keyboard has the ASCII layout as well). They are also dedicated versions with country-specific characters, e.g. for Hungary or Scandinavia. I assume that there have also been 8296 with german DIN-keyboard, which must have contained different character generator and screen editor ROMs (although I've never seen a DIN 8296). By the way, this type of character ROM was used in both BASIC 2 and BASIC 4 machines

The MOS 6545-1 CRTC ('cathode ray tube controller'), which generates the video-output on all later PET-series motherboards (together with some additional logic ICs). It is identical to Motorola's MC6845 and can also be used to output wonderful color-graphics (but not in the PET-series, where it only shows characters created by the character generator). Very useful, furthermore, is the CRTC's ability to take up to 8 KBytes from the machine's RAM and use it as video RAM - earlier machines, like the CBM 3016 (which didn't even have a CRTC - all of its functions were 'hardwired' in discrete logic on the mainboard), required dedicated SRAM for that purpose

As you can guess from the board silkscreening, these two chips have to be exchanged in order to configure the CRTC for either 40- or 80-column operation - extremly useful when you needed the board as replacement for an older 40-column PET (e.g. from the 4000-series). Well, before doing that yourself: that's not all! Resistors at JC1, JC3 and JC5 have to be removed and put at locations JC2, JC4 and JC6, in order to change video timings for 40 column-operation. Since I never tried it myself, I take no responisibility for a damaged PET - do at your own risk!

The machine's piezo-electric speaker, making that wonderful noise at power-up! It's not good for anything else than a beep, but better than nothing - early PETs even came without one. Who needs sound, anyway?

This Rockwell R6520AP at location UE2 is one of the 8296's PIAs ('peripheral interface adaptor'). The R6520 is identical to the MOS 6520; however, I don't know what the 'AP'-suffix means (maybe Rockwell-specific). This fellow here is mainly responsible for keyboard I/O, and thus located at the board's front, near the keyboard connector. But it also supports the MOS 6522 at location UC12 in controlling the machine's two cassette ports (see below)

Shown here are the 8296's second PIA (another Rockwell R6520AP, at UC13) and the MOS 6522 VIA ('versatile interface adaptor'). The PIA's only purpose is to control the IEEE-488 interface bus, i.e. the three MC3446N signal drivers/ receivers used for that (described further below). The VIA controls the machine's user port and the two cassetteports (together with the first PIA already mentioned)

Here are two MOS 8700 PLA ('programmable logic array', also referred to as FPLAI and FPLAII in this design). These gate arrays provide logic equations for several addressing functions, such as RAM/ ROM address decoding, which is required for RAM/ ROM selection and bankswitching in general. A side effect of that is that they can also be used to maintain compatibility with the older 32k machines (e.g. by disabling the expansion RAM). Even more important in this context (but not shown here) is a Texas Instruments TBP18S030N PROM at location UC2, which provides the timing for all RAM access operations

The three already mentioned Motorola MC3446N (only two shown here, at UC14 & UC15; the last one is located at UD9). They are used as line drivers and receivers for the IEEE488 bus. From the schematics, I assume that the one at UD9 acts as the signal driver, and the other two are used as receivers. The IEEE488 interface bus is also referred to as GPIB ('general purpose interface bus') and HPIB ('Hewlett Packard interface bus', since they invented it). It was (and for some applications still is) a major industry standard for controlling and programming machines; before introducing the CBM serial bus with the VIC-20, Commodore used IEEE488 for connecting printers and diskdrives to all their PET-series machines. In fact, CBM's later serial bus was some sort of stripped down IEEE488, which was most probably dropped because there usually was no need for 15 devices (computer plus 14 peripherals - original HP standard even allowed 31)

A view on the machine's internal expansion port (2x 50 pin connectors at J4 & J9), and the second cassetteport (J6) before it. As mentioned above, the cassettport is somewhat special. It's a remnant of the older machines' integrated datasette, for which the internal connector was used. But in contrast to them, the 8296's 2nd cassetteport can also be accessed from the outside (you can see the opening in the case, which can also be used for expansion port cables, provided they fit through it). As for the expansion port, it contains signals from all major parts of the machine (CPU, RAM, ROM, CRTC, I/O-area); so theoretically, expansion devices can be graphic cards, CPU- or RAM expansions or else. But I've never seen such a device, and I don't know if Commodore ever produced one. Above expansion you can see the already discussed ROMs and PLAs, as well as the other two unoccupied function ROM sockets at UE9 & UE10. As far as I know, these sockets were present in all PET-series machines, and could be used for additional commercial or homebrewn software (in this case, UE8 is used for the BASIC4 screen editor). Interestingly, the concept of function ROMs was picked up again in the C128 and the 264-series, giving the user (and Commodore, of course) an easy way of extending a machine's functionality

Sothius' Home

-

Museum

-

Commodore

-

Commodore 8296

-

additional pictures