Start of the project

For many years now I've been thinking about building an old 1970s style computer, but I've never got around to doing it. Recently I came across a number of sites where others have gone a step further, by building computers not from the old microprocessors but from 74xx MSI chips.

This has got me a thinking, it is one thing to bolt a few VSLI components together according to manufacturers data sheets, and it is another to create your own computer from the raw logic gates (or transistors if you have enough space and time).

So this project, which I fear will be very long, is all about designing and maybe building a reasonable 1970s style mini computer from the raw logic gates.

The nice thing about rolling your own is that you can choose what your machine will be like. This is the tough part, as you have to decide between nice to have and need to have. For this first shot I'm going to try and limit myself to need to have.

I've already decided upon the operating system, it will be IPS. In another life I fly spacecraft (AMSAT), and IPS is the operating system we use on the Phase 3 and Phase 5 spacecraft. It has a small footprint (ideal for 1970s computers), and needs only a small amount of work to customize for different processors.

For more info on IPS look here:

http://www.amsat.org/amsat/projects/ips/index.html

Obviously as I am designing the processor from scratch and know the target operating system, the new design can be optimized from the start to match.

Back in 2002 Lyle Johnson and I designed a processor called the AM1601, which was a candidate to be the processor for the AMSAT IHU-3 (Integrated Housekeeping Unit). At that time AMSAT had been using 1802 chips from the 1970s, and radiation harden ones had become harder and harder to find. The AM1601 was a software implementation, designed to be burnt onto FPLAs. The idea being that we could keep the design fixed, and burn to the latest FPLA when required. Anyway, it didn't happen, we decided in the end to buy a huge batch of ARM chips instead, yes we were hooked on ARM before it took over the world.

We did, well Stacey Mills did, a lot of radiation testing on the ARM and it is incredibly radiation tolerant straight out of the box, which is a good thing when you're using it in orbit. Anyway that is by the side, the point is that my new processor will not be an AM1601, as reviewing the user manual we wrote in 2002 it seems like it was too complicated, and as I said above I'm only going to implement what is needed not the nice to have. That said, there are elements of the AM1601 design which will filter into the new processor, the AM1601 design process was also very educational.

Details of the abortive AM1601 project can be found here:

http://www.amsat.org/amsat/projects/ips/Am1601.html

So now I'm going to sit down and decide what I need to have in the processor. One interesting thought is that by building from raw gates you do not have to split between the processor and the rest of the system, ... they can be one and the same. I'll stop on that one 'cos it is a bit of a mind shift.