Building a computer (Omega MSX)

One thing I’ve wanted to do for a very long time is to build a kit computer. Originally I planned to build a Spectrum 128k clone. I’ll probably still do that in the future, but as I had a Spectrum already, there was less pressure to get another speccy. I kept pushing it back.

Then I saw the Omega MSX.

The Omega is an MSX 2 “remake” with some compatibility with the MSX 2+ standard. It’s an open source project and the boards can be ordered from any PCB fab of your choice. I ordered some boards last year, as well as a big order from an e-bay seller who was selling “all the silicon” components as a bundle. I ordered the special 64 pin oddly spaced socket for the video chip from AliExpress. Just before Christmas I ordered the rest of the components from two different suppliers. This was mostly passives, connectors and sockets. Incidentally, if any suppliers want to sponsor me, reach out to me!

Once I had all the components, (Bar three. More on that later) it was time to star assembling. This was remarkably simple. I simply worked from lowest to highest, starting with the diodes, then resistors, then resistor arrays, sockets, capacitors and finally connectors. I only had a few mishaps along the way.

• Firstly, one of the DRAM did not want to go in to the socket and I accidently snapped off one of the legs. A quick patch with a bit cut off one of the resistor arrays and it dropped right in.
• Secondly, I bent a pin on the Yamaha sound chip. I didn’t even notice this until final inspection. Simply bending it out was enough to fix the issue.
• Thirdly, and this only came up in testing, I had the SRAM and EPROM switched. Neither was the exact brand as specified in the BOM. They were substitutions, so this is my own damned fault. Swapping the two chips fixed this final main board issue.

I was testing and inspecting continuously as I assembled things. Mostly I was just making sure there were no shorts between 5v and GND. I checked after the resistors and sockets went in, and then again after the capacitors went in, and again, once more, once the chips were in. At no point did I get a short, so that made me confident enough to actually try the system.

On my first power up, with I got a black screen. This was good! This meant that enough signal was being generated that my screen was synching to a valid video signal. This usually means that most of the system is executing. Occasionally on startup, I’d also get a few seconds of garbage on screen. More signs that it was mostly working. After some hair tearing, I finally caught on to the issue with the SRAM and EPROM being swapped. Swapping them back and we had success! The MSX logo came up and it booted to a command prompt.

At this point, the second round of issues came up, and they were all with the keyboard.

• Firstly, I’d accidently ordered the wrong resistor array for the keyboard. Initially I just bodged in a set of the appropriate resistors and made my own array. However, while doing some cleaning of a storage container full of components, I later discovered that I did have the correct array. Ironically, it was a part I’d incorrectly ordered for a completely different project. Hang on to your spare parts, folks. You never know when they’ll be useful.
• Secondly the LEDs were super dim. Guess what? A 470kΩ resistor is not the same as a 470Ω resistor. I’d used resistors with a thousand times the resistance. No wonder the resistors were dim! A quick substitution and we were good to go.
• Finally I’d got some of the LEDs in backwards. In my defense, the boards were marked poorly for polarity on these components.

I will mention some of my experience building this board:

• There are a lot of capacitors! Over 90. I think I went a bit cross-eyed installing them. Finding where they all went was time consuming. There were two I couldn’t find after several minutes of searching. In the end, I opened the PCB in KiCAD and used the built in search function to find where they were supposed to go.
• Ditto for 74 series “jellybean” logic. I ended up having to use KiCAD for finding one of the locations.
• There were three components that I simply could not locate. I have them on back order. Thankfully they are only the audio out RCA socket, the RTC battery holder and the printer port connector, none of which will slow me down.
• The instructions are more than sparse. They’re practically hiding. I hit several issues where I had to guess. Of course, after the build, I found the following site which has excellent instructions. They’re in Spanish. Use an auto translator. You’ll be fine.
• I haven’t been able to get RGB video working yet. I still have a little more work to try before I give up. SVideo is awesome, though.
• Compiling the BIOS seems completely undocumented. I just grabbed a “pre rolled” BIOS instead.
• The JED files for the GALs were difficult to find. They’re on the GitHub page. They’re just not obvious. All the instructions talk about using a converter to convert a text file to a JED. Unfortunately the software doesn’t work under 64 bit Windows. Just look for the JEDs.
• The keyboard assembly for the spacebar is horrible. You have to bend your own stabiliser bar. Mine is still not right. It works but it’s just not “right”.
• There’s this one Sony chip with a strange width. I ended up using two pieces of machined pin strip socket.
• The stabilisers for the keyboard vexed me for a long time. Firstly, you need to slide the smaller part into the larger part. Then you clip the stabiliser bar into a pair of the stabilisers. It goes in the middle hole. It clips in at the edge. The stabilisers then clip in. You put the smaller side with the lip into the larger hole and the side with the split goes into the smaller hole.
• The LEDs on the keys go through the keys. Don’t get this wrong.