Saturday, February 4, 2017

Working on small flaw in disk tool, also repairing an old vacuum tube oscilloscope

Been distracted by a number of things for the last few weeks, but time to get back to work and back to posting.


In a previous testing session with the Alto, I had brought the disk tool to attempt to archive several cartridges plus write a 'games' image to an available cartridge. For reasons that are still not clear, the tester did not smoothly and rapidly move through the cylinders when attempting to read packs, nor was the written cartridge able to boot.

We also found power supply problems, with the supply dropping with apparent thermal outages jn spite of its low demand and low temps. During one of those dropouts, the first 12 sectors (one cylinder on one head) was overwritten with garbage, so that our main boot cartridge was unusable.

I brought the tool back to debug it in my lab, found that it seemed to work flawlessly here, smoothly seeking through the pack and reading with no errors. The power supply did not act up either. We decided to transport the entire setup, including the Diablo drive itself, for the next test session with the Alto in Marc's lab.

The tool operated smoothly on my drive in our session yesterday, but the Smalltalk cartridge I brought did not boot. We did some extensive diagnostic collections and determined that the first sector (cyl 0, head 0, sector 0) had been properly written with improper data.

That is, some gibberish had been formatted into proper header, label and data records in that sector, given proper checksums and thus read without error, but the resulting words in memory were random, not boot code.

I have to dig through the logic of my WriteEntireCartridge transaction, which I suspect is not setting up the memory address correctly to access the appropriate part of the RAM for this sector, but the logic to increment sector, head and cylinder clears things up for all the remaining sectors.

It also means that I should be able to write just the first sector with an individual WriteSector transaction, hopefully giving us usable Smalltalk and Games images. I have left the drive and other gear at Marc's lab, but should be able to stop by to verify my quick fix (rewrite just sector 0) sometime soon.

Ken Shirriff brought his ethernet tool for testing, as well. He was able to collect good information by using the file transfer protocol of the Alto to drive various packets out on the Ethernet port. He receives a substantial portion of the packet correctly and can now dig into the anomalies before the next session.


A friend is loaning an old 1950s oscilloscope to the DGM for use at an upcoming game developers conference (GDC) where they will recreate a very early video game, Tennis for Two, which displays on a scope such as this.

It had a block capacitor, two faulty tubes, and noisy controls, at least as far as I have taken it so far. The bad parts are replaced and it is powered up, but the controls have not yet been cleaned. I need to get this working sufficiently with an X-Y drive to display the game, so more work is needed.

No comments:

Post a Comment