Zorro II expansion card which gives your Amiga the following:
2 x serial ports (DB9) up to 614400 baud.
1 x parallel port. (DB23)
1 x MIDI port.
1 expansion port
Additional details:
Rev 4
Prototype board never released (#011). A missing trace on the rear of the PCB layout (jumper trace present). Socketed PAL chip and clock.
Rev 6
First production board. All possible connectors, sockets, and potential components, no PCB layout mistakes.
Rev 7
Cost reduced. The MIDI option was deemed not to be popular, but the serial ports were very popular. Unnecessary components not included. This includes J4, Rom socket, MIDI header on the card, and 37 0-Ohm jumpers in front of the connectors. The MIDI connector was still functional in software if a 10-pin header was added.
Previously Undocumented 'Reserved' Jumpers:
J3 - Hardware AutoConfig Disable (for testing/debug use only) - Shorted, the board will not AutoConfig, and any downstream boards are not sent the AutoConfig signal on the Zorro bus.
J4 - ROM Disable - If the U5 ROM socket was installed, any boot ROM (like a GVP FastROM) could be placed there to activate other boards at boot time. However, the GVP I/O extender never had a 'boot ROM'. All drivers were put in the Expansion drawer, and activated with Binddrivers during startup.
J6 - Modified a parallel port signal between the PC and Amiga types.
There were no products made for the Option connector, though it contains most, if not all, of the important signals from the DPRC or serial/parallel port (pin out was never documented). Standoff holes were provided at the bottom of the card for option card stability.
Manual Mistakes:
The GVP I/O Extender serial port, as included on the GVP A2000 G-Force '040 board, is numbered as serial port 0. When followed by a GVP I/O extender in a Zorro II slot, the first two ports on the first I/O Extender are numbered serial port 1 and 2, on the gvpser.device. The manual incorrectly states that port 1 in the sequence is not used. Subsequent I/O Extender serial ports on additional cards would be numbered 3 & 4, 5 & 6, etc. Stacked parallel port interfaces would also increase in ther port numbering by 1.
Note: The GVP G-Force '040 board has the same kind of hardware (but a VL16C551 chip with only 1 serial port). The software knows there is no port or connector space for the missing port, and correctly skips the unconnected hardware. The board had the wiring for another port, but external no header space.
Picture note:
The Rev 4 card is a prototype that was never shipped. It is much like the Rev 6, but has a missing trace on the rear (jumper wire visible). This card has a 20Mhz oscillator, and is functional, but it does not adversely impact the serial/parallel performance. The oscillator was soldered in on production boards.
Any oscillator value is valid (for serial/parallel operations) up to the speed rating of the VLSI I/O chip on the PCB as long as the clock frequency value is updated in the Expansion drawer's driver Tooltype. The (default) 7.372800 clock value only matters to MIDI function/calculations. All boards must have the same clock frequency, or irregular results may occur on some communications.
BOM details from internal GVP documents can be found here: https://archive.org/details/gvp-io-extender-bom