This is the current setup, with two cards and a backplane in a make-shift rack. A unsupported set of cards tends to have more problems and I don't need that. And it looks better too :-). The 3D-printed parts are available.
The processor card uses a Hitchi 6309 processor. Not that I need als those extra instructions yet, but as a CMOS part is uses less power. The part I didn't install is the TTL-RS232c level conversion. This is for now a desktop system directly connected to a PC via USB, making USB-serial adapters the obvious, compact solution.
The meory map is mainly RAM, with only $F800-$FFFF used for the monitor ROM and $F000-$F3FF for I/O. The space in between is Monitor workspace RAM. This is in 'system' mode, it is all RAM in 'user' mode.
For the current usage, just one serial port with a fixed BAUD rate, it needs only a limited set of components. The BAUD rate is fixed at 9600 BAUD, later to be changed to the more common 19200 BAUD used by FlexNet . Like the CPU board, the serial interface is configured for direct TTL serial to a USB-TTL adapter.
The essential signal lines are all connected between the slots, with the I/O-range signals are not. Here the DIV4 (for the FDC) and DIV3 (for the serial card) are patched for the two slots with a connector.
Both cards have their data bus tri-stated between read/write operations. There are no terminators present on the bus, leaving those eight lines in a high impedance state. It works, but I am not sure this is a desirable situation.
It took a long time to get the FDC properly adjusted. The procedure in the datasheets is
not complete enough for my situation. The new
KiCAD schematic from Robert Offner
helped a lot with concise adjusment values and pins. And the hint that there was an issue with value of
the trimmer capacitor.
With the 65pF trimmer capacitor, 500 kHz was out of reach. The lowest possible frequency was 550
kHz. This means the maximum capacitence was too low, so an extra capacitor was added in parallel
with the trimmer.
The schematic suggests a series capacitor, but this reduces the maximum capacitance and I needed more,
not less.
The original boards are designed with with another CAD program, and not quite up to modern standards. My main complaint is that the grounded pins are completely embedded in the ground plane, creating a very low thermal impedance. With a less than powerful soldering iron, you cen solder them, but don't expect to clean the hole afterwards.
Another complaints is the labeling is not very informative. Nothing that can be solved with proper documentation, but that isn't available either. The KiCAD schematics from Robert Offler are a step in the right direction however.
The issues I have with getting Flex running are mainly caused by my own chosen route of getting there, however.