7.9 KiB
rc2014-compat
A respin of the RC2014 ecosystem. I don't have any opposition to building things out of SMD parts (in some ways, I find it easier to work with), so some boards might go that way to the extent that parts availability allows. Also, I'd like to keep all board sizes within the 100x100mm limit within which PCB fabs offer their cheapest rates (like $2 double-sided at JLCPCB, for instance).
Available boards will be listed below as I add them. KiCad 6 is required to
work with the designs. Make sure to check out the repo with git clone --recursive ..., as the different boards are in their own repositories
now, and each of them in turn pulls in other repositories for KiCad
libraries.
rc2014-template
I took the board file from here and fleshed it out into a full template for use with KiCad. It supports the extended signals (such as NMI, WAIT, and the second clock), but not 16-bit CPUs...see below for 16-bit support.
Boards built from this template will fit backplane16pcie (see below) or any RC2014-compatible backplane. Some backplanes might be 40 pins wide; make sure you get pin 1 where it belongs, or bad things will happen.
rc2014-template16
The above, but 16-bit.
Boards built from this template will fit backplane16pcie (see below) or any extended RC2014-compatible backplane. Again, watch out for 40-pin-wide backplanes. Many backplanes have only the base 39-pin connector; the 10-pin extension will be left flapping in the breeze if you use one of those.
6502-cpu
The Apple II got me through high school and college; writing software for it was the first computer-related activity that made me money (small amounts in retrospect, but it was a start). I'm far more familiar with the 6502 than the Z80, so it'd be nice to be able to work with one here.
This board is a re-creation of this board, recaptured in KiCad and laid out to fit within the template provided above. The Tindie page provides these resources, which I probably ought to include in this repo somehow:
A long-term goal would be to take the dialect of Microsoft BASIC Apple used in the II and modify it to work here, perhaps with improved graphics from a TMS9918 or one of its successors. The Apple II's monitor would also be nice to have, but while Apple open-sourced Integer BASIC and Microsoft open-sourced its BASIC, I'm less sure about the Apple II monitor. (The Apple I monitor was open-sourced...that's the "WozMon" that's easily located.)
One significant change from the original design: the I/O page can be changed from the default $C0 if desired by cutting and soldering the 8 jumpers near the 74HCT688. This happens to also be the page the Apple II uses for I/O, so I'm inclined to leave it alone. :)
z80-cpu
Basically the same as the RC2014 Z80 module, but with a 10-MHz LQFP-44 CPU and a crystal oscillator to run it at that speed.
backplane16pcie
A modular backplane that provides 4 slots (compatible with 16-bit CPUs) with 20-mm spacing between them on a 100x100mm double-sided PCB, with USB-C power input, a power-on reset circuit (with reset button), two 4-pin TTL UART connectors, and an idiot light that indicates it's receiving power. Connectors of the type normally used for 4-lane PCI Express are on opposite sides of the board to allow easy daisy-chaining...just slot one board into another.
The UART pin labeling for TX and RX is intentionally swapped (the RX line is labeled "TX" and vice versa). When plugging in your cable, connect TX to TX and RX to RX.
This project previously included backplane designs that used right-angle pin headers and sockets for inter-board connect, but the PCI Express connectors are more readily available, and cheaper too (one of them is free, as it's part of the PCB design). Those other designs have been removed, as they're redundant.
v9958-video
A video card built around the Yamaha V9958, a more powerful successor to the TI TMS9918A with support for 80-column text, up to 512x384 graphics resolution, and up to 192K of dedicated memory that can be installed in 64K chunks. (As cheap as RAM is now, you might as well max it out.)
I/O configuration is fairly flexible. You can allocate either two addresses to it (like the 9918A, which will restrict access to the 9958's added features) or four (for access to all features), depending on your needs. Four addresses at $98-$9B is the MSX-compatible configuration that I've seen another 9958 board use, while two addresses at $BE-$BF might run well-behaved ColecoVision games. Want a multi-monitor setup? Install multiple cards, all on different addresses.
68b50-dual-serial
Two serial ports, one card. One is connected to the TX/RX lines of the backplane for a TTL-level console port, while the other drives a 9-pin serial port through a MAX232 to control printers, modems, and other RS-232 devices. Bitrates of 115.2, 38.4, 19.2, and 9.6 kbps are independently selectable with jumpers, and by using the 6850's internal bitrate divider, rates down to 300 bps can be set up if needed. The card's base I/O address is also jumper-selectable.
This board has its own 7.3728-MHz crystal oscillator, making it independent of CPU speed. (I'm aiming for 10 MHz with both of my CPU boards.)
paged-ram-rom-1mb
This started out as a copy of the schematic from the RC2014 GitHub. I brought it into KiCad 6, replaced the "rescued" symbols from the original with more current ones in the KiCad library, redid the connections in what I think is an easier-to-follow style (global labels instead of buses), and re-did the PCB layout from scratch (since that wasn't included). I also swapped out SMDs for through-hole parts and replaced the DIP flash chip with a socketed PLCC.
All of this freed up enough board space for a second SRAM chip, so I added one to bump the RAM up to 1 MB. Banks 0-31 select ROM (flash) and banks 32-95 (previously 32-63) select RAM. 16K chunks of either ROM or RAM can be paged into any of four slots in the memory map: $0000-$3FFF, $4000-$7FFF, $8000-$BFFF, $C000-$FFFF. Reset maps ROM page 0 into all slots.
I was having some trouble puzzling out how the I/O decoding worked in the original schematic. I found this page that described how the board worked, and once I had redone the schematic with the newer symbols, it became clear what was going on. In particular, the newer symbols for the 74LS139 have the selection pins labeled A1 and A0 instead of A and B. Looking back now at the function table on page 1 of the TI datasheet, I see that the "select" column headings are in reverse alphabetical order (B, then A). That must've escaped me while I was trying to puzzle out the schematic.
new-board.sh
Instead of setting up the templates within KiCad, you can add new boards within this project with this script. The first parameter is the name of the board; the second (optional) parameter is "16" to use the 16-bit template (otherwise, use the 8-bit template).