Defcon:Blog Keklkakl blog blag

Building a CP/M compatible system on perf-board

With a fresh success of building a computer from a kit, I decided to implement an existing design on perfboard, instead of drawing a design of my own right-away for breadboard building. Considering that the RC2014 is closely based on the work of Grant Searle, it was natural for me to take another design of his, and adapt to the 8-slot backplane: the "fully functional CP/M machine using only 9 chips" design.

I broke the design down to separate functional blocks, and built each as a separate card. By splitting the build into these distinct parts, I was able to build one card, and verify functionality of that card using the RC2014 before continuing to the next. All the cards were built on 100mm wide perfboard, using a direct soldered wire approach.

• RAM/ROM card
• SIO/2 serial IO card
• 8-bit IDE adapter for CF card
• CPU board with onboard Clock and Reset
• YM2149F sound card

I chose to make the RAM/ROM as a single combined card, because I did not want to split the ROM-disable function needed for CP/M up across two cards. It also allowed me to reduce the potential part count (by 1 IC). Because of the close relationship between the RC2014 design and the one I was implementing, I could simply use the ROM content from the RC2014 for verifying the ROM/RAM using the CPU and Serial I/O board from the kit.

 

Grant's CP/M design uses a different UART from the BASIC design. The BASIC design uses a 65B50, while the CP/M design uses a SIO/2. So this was the next card to be built. To verify this one, I burned a new ROM containing the "monitor" for the complete CP/M-design. This monitor also includes an 8k BASIC version, so it was easy to verify the SIO/2 board without having a CF-storage approach available.

 

The original breadboard design page talks a bit about how to wire up a Compact Flash card, and how to hand-solder the SMD connector used for CF-cards. I did really not want to deal with that connector directly, and gambled on being able to use an IDE-to-CF adapter, considering the design really uses CF in 8-bit IDE mode anyway. My "IDE adapter" is basically a couple of OR-gates, an '138 address decoder, some resistors and a 40-pin male header connector. And it worked! With a short IDE cable between my "adapter" and a commercial IDECF-adapter, I was able to prep the CF card, load software onto it, and boot to a lovely "A>" prompt!

 

I could have rested here, but decided to whip up a new CPU board, as that, plus the clock generator, now was the only cards from the RC2014 kit that was in use. I integrated the clock, and also added an auto-reset onto the CPU card.

With these cards on the backplane, I now had a working Z80 based system, that I had needed to do some desgin-work for myself to adapt it to multiple cards. It was even a working CP/M system, with bonus ROM-based BASIC. To further verify that I did indeed have a CP/M compatible system, I grabbed a copy of Zork 1, I think it was from http://www.retroarchive.org/cpm/games/games.htm, and loaded it onto the CF-card using Grant's DOWNLOAD.COM utility. I believe any fully CP/M compatible system should have a copy of Zork 1 on it. And mine does, and it works 😀 I have used Zork as a demonstration of the system so much now, that my friends and colleagues have started to call my homebrew CP/M computers "the Zork machines" 😀

The final card I built for the RC2014 bus, was a YM2149F synthesizer sound chip adapter. It required some fairly weird address decoding logic, so I spent quite a while on the design of this relatively simple card. After laying it out on perfboard and soldering up the wiring, I had a working sound card. I do believe I am starting to get a hang of this... What I still don't have as good a hang of, is large-scale assembly projects. So, I have tested the YM-card extensively from BASIC, and I have some general test-routines for it written in Assembly. But I have not made any proper songs using it, and I am quite far from implementing a Tracker or Player of any kind.

What I have used the system for, however, is getting to know how to code z80 Assembly for CP/M. It's a neat little operating system. A task as "small" as reading a file from disk has you digging quite deep into the system itself. As a result you end up with a very deep understanding of how the operating system internals work. And being such a small operating system, it is quite possible for a person to understand the entire operating system, after a reasonably short time studying it.

I created a final board for the system. By using a very cheap Chinese strip-board, I created a barebones version of the backplane. This freed up the RC2014 backplane to go back to being used with the original kit, and also made the computer a completely homebrew build. After this, the electronics part of the perfboard project was done. With such great results on the electronics side, I figured it deserved a nice place to live, and not become yet another stack of project parts in one of my bins of such. So I built a wooden cabinet for it, with some 3D printed parts to hold the cards in place. The case is made from beech wood, with an inlaid aluminium front.

This post is part of a series documenting my journey in making my own homebrew computer, the posts can be found on this link.

My first working homebrew was a kit: RC2014

After having solved the mystery of counterfeit parts, I felt ready to start attacking my master plan of building my own Z80 based computer. I wired up a CPU test circuit and used with great success. I then wired up a simple system on a set of cheap breadboards, consisting of the CPU, a ROM chip and a slow clock generator. With LEDs on the address bus, I punched up some simple assembly that NOP'ed for a while, did a JMP, NOP'ed some more, and then JMP'ed back to start. This was assembled using TASM under Dosbox, and loaded into the ROM using my Minipro programmer. When I applied power to the circuit, it started ticking along, and did just what I expected it to do!

Success! With great spirit, I next wired up a far more "complex" circuit. I added a few support logic gates, a 74LS138 4-to-8 decoder and a 74LS373 8-bit latch, and moved the LEDs from the address bus to the outputs of the latch. With the added components, I had a kind-of 8-bit digital output. With a simple counter code typed and assembled, I had partial success. The circuit and code worked. And then sometimes it didn't. And sometimes when it didn't work, it would start working after some time.

With the bad experiences from fake chips, I started suspecting that my cheap breadboards were causing problems. So I moved the circuit over to one of my "known good" breadboards, part of an old digital trainer device. Now the circuit and code worked most of the time, but still sometimes not. I was getting frustrated. I tore down my wiring, checked and verified my circuit, wired it back up, and got the same result. Was there something that I had completely misunderstood? Was there something very wrong with my method?

Eventually, I decided to stop faffing about. I decided that the best way forward, would be to start building a known good design. With that, I would be able to verify that I am able to pull this off, and I would be able to use it to test and verify what I was designing from scratch. So, enter the RC2014!

It was now late december 2015, perfect timing for a "christmas gift to self" kind of a purchase. In my research for options and ideas, I had looked at Spencer Owen's RC2014 several times. It is a dead simple design, and cheap as well. Perfect match. Off to Tindie I went, my order placed on saturday the 19. december. I was very impressed when the kit arrived in my mailbox on the 23rd! I brought the kit with my on my Christmas holiday, and by the 27th it was completed. I connected up a TTL-USB adapter, and powered the thing up. To much joy, the kit sprung to life! Grant Searle's BASIC monitor greeted me with Microsoft BASIC for Nascom (modified by Searle and Owen).

This was back when the only RC2014 kit available was the "Classic" version. My kit was the "Full monty" set, with all components, and the 8-slot backplane. The kit was genuinely well put together. The PCB's are simple, but cleanly presented. Components were well packaged, and sorted. The instructions were simple, but clear and concise. The overall packaging was small! As for ease-of-build, there are many connections to solder, and get right in a kit of this type, so I would not recommend it to a beginner. But even with the simple pen-style soldering iron I keep at mum's home, I was able to complete the job cleanly in a reasonable amount of time. So I'd call it an intermediate kit easily. I can happily recommend Spencer's kits, and the new, updated versions of RC2014 kits look nice indeed. I'm tempted to buy another one, to be honest 🙂

I now had my first working Z80 based homebrew computer! And with an 8-slot backplane, I had 3 slots available, as the original kit uses 5 slots: CPU, ROM, RAM, Serial I/O and Clock generator. With a simple bus format using Single row connectors, it should be easy to make expansion cards. I dug out a perf-board, added the needed right-angle single-row header pins, and soldered up the exact circuit I had tried on my breadboard. I plugged it into the backplane, and tested it using OUT statements in BASIC. It worked! Another great success, but at the same time somewhat of a mystery to me. Why did it not work on breadboard, when it worked fine in a more complex system? I decided to test removing the RAM and Serial cards, and load my test-ROM onto the ROM card. Suddenly, the I/O board was behaving erratically again. Thanks to RC2014, I now had proof that my hardware was working, and my software was bad. The kit had actually already accomplished what I wanted: verify that I could assemble a working system from someone else's design, and verify that I could design working hardware myself.

All of this made me make a change of plans. Instead of going back to the breadboard, I decided to try to implement Grant Searle's Minimal CP/M system using home-built cards for the RC2014 backplane, before proceeding with my own design. The process of this is the topic of a post of it's own.

I have since (much later, really) dug out the wonky test-code, and corrected my mistakes. I can't remember what exactly was wrong, but I seem to remember it was embarrassing. I have not tried wiring it up on breadboard again, as all my Z80 CPUs have been busy running on other systems 🙂

This post is part of a series documenting my journey in making my own homebrew computer, the posts can be found on this link.