GM Petrosian

The III-S Glasgow prototype had been a success. It worked perfectly, running the program at twice the computing speed of the Mephisto 'S' module, and thirteen times faster than the original Mephisto III computer. But having been conceived as a prototype, I found that it was not easy to build a nice housing for it, nor to adapt it into the drawer of a Mephisto board.

Luckily, Wolfgang Zugrav offered to finance the design of a new PCB, adopting all the improvements experienced with the prototype, to use the housing of some Briketts that he had defective. Most of these Briketts had the bottom part damaged (due to corrosion because of battery leakage), and they were also without cover or with damage or absence of some other part. But the important thing was only to take advantage of the top cover, since the bottom one could be built by me through 3D printing.

So I took measurements, and started designing new PCBs. I had to vary the layout of the components as well as the type of some of them: SMD, surface mounted device, to save space. And I also modified the power circuit by a switched one of higher efficiency and smaller size.

The height of the new Brikett was important, as the idea was to fit it into the drawer of a Mephisto ESB, so it could only be 10 mm higher than the original Brikett. I created a 3D model of the bottom cover, and I made some simulations.

I then sent the new design of the PCBs to the factory. Since I had made so many layout changes with respect to the prototype, I ran the risk of having made a mistake, but fortunately this was not the case.

I noticed this when I assembled the first unit...

...after connecting the board and power, and verifying that everything worked properly.

The bottom cover was made with a 3D printer:

I also had to improve the look of the reused top covers, by washing and polishing them with a Tamiya compound:

Final assembly:

And tests with the computer completely ready:

Later I made more units for reliable people, which allowed to recover the initial investment of the project, since for only one or two units the expenses were very high.

Hardware specifications:

Programs:

Performance testing:

Tests carried out with the Glasgow program, after playing 1.f2f3 in analysis mode.
 The number of positions calculated after exactly 6 minutes is shown.

Units produced:

Only 16 units have been manufactured, numbered from #01 to #16.

Finally a few pictures (courtesy of Heiko Berger) of one of the units in its final placement:

To conclude, I would like to comment that I have been able to carry out this whole project thanks to the motivation and economic support of my good friend Wolfgang Zugrav, and the use of Free Software in all phases of the project: