Herb Lin <LIN@Mit-Mc.ARPA> (01/19/85)
I have bought a Macrotech MI-286 from Gifford Computer Systems, whose problems I described in a previous note. Their recommended cure was to upgrade my hard disk controller board from a Morrow HDCA-4 to a Compupro Disk2. However, a conversation with the techs at Macrotech resulted in confusing information: Macrotech tells me that the problem is most likely in the Compupro RAM 16 boards that I am using; these are alleged to have a wrong pull-up resistor in R1. They said that the fix would be to put the correct resistor there, and all would be well. They also said that other Compupro memory boards did not have the same problem, and that they knew of people with Morrow controllers running Macrotech boards without difficulty. I tried something else, with their concurrence. I replaced my 4 RAM 16 boards with a RAM 22 board, and tried running the Macrotech board with that configuration. Here I had a very strange experience -- the system worked fine in the morning, but then I shut it down. In the afternoon, I powered up again, and found that the system no longer worked. Closer investigation revealed that my hard disk directory had been trashed in a very subtle way, and Gifford says that their other customers with Morrow controllers and Macrotech boards have reported similar trashing problems. I have decided to take Gifford's solution, since they were generous in offering me some trade-in value for my Morrow controller, *and* assuring me of refund if their solution did not work. More as it develops. herb lin
Jerry E. Pournelle <POURNE@Mit-Mc.ARPA> (01/24/85)
fascinating. One of the real difficulties is that there are KNOWN bugs in the B-step revs of the 80286 chip; so many that CompuPro refuses to sell a 286 board except for "development", and warns users. Macrotech and Gifford think they have fixed or wired around all the problems, and perhaps they have, although your strange experiences make one wonder. Dual processors were a great idea in their time. After some thought, though, I am converting my system to a new CompuPro 80286/287 board (as of now they are shipping the C-step 80286 chip, which doesn't have the problems of the b step) and a Z-80 SPUZ slave Z-80 p[rocessor. This will run both processors at once, and if fed a z-80 .com file that runs and the 286 is available for other simultaneous work. Seems more advanced than a dual processor. But I will also keep a vanilla z-80 around to write books with until I'm sure... jep