Building the CPU280 and working with it (rightaway). By Helmut Jungkunz, ZNODE #51, Munich, Germany Having read all these articles in Eight Bits & Change and TCJ, I was surprised, how much work there is obviously in- volved in building the YASBEC and getting it to work. After all, CPU boards like the YASBEC have been built here in Germany since 1986 with exactly the same capabilities. Now, the idea was, to have a clever singleboard computer, both usable for any sort of industrial purpose and for semi-commercial use as well. Therefore the solution was obvious: the INIT process had to come from a ROM-based startup program, whatever final purpose one was aiming at. A most intelligent system would naturally have to be fast(!) Thus, the CPU 64180 was skipped, since, during the time of the development of the CPU280, the highest production specification rate was 12 MHz with this processor. And that would still only be considered equal to a Z80. No big achievement there. So after tideous test (and hilarious errors on the production side from ZILOG!!!!), a mask was found, that proved to be fairly stable, i.e., several so-called production masks proved to produce too many unwanted semiconductive crosslinks etcetera. This last Z280 CPU then was used as the centre of the new design. A few modifications and workarounds had to be done by Tilmann and a couple of friends, to make HD-drives writeable and to produce reliable performance for daily use. Driven at 12 MHz, ZILOG's maximum specification for that release, the Z280 would be equivalent to 20 MHz of Z80 power, plus allow the use of it's outstanding addressing capabilities and other extras, like the built-in cache etc. To be able to use standard ECB-elements, like floppy controllers and other peripherals, 6 MHz external bus rate was considererd adequate. This also helped to keep costs down. I want to try to explain the ECB board concept a little: All signals that are used between computer components have to be led in a very careful way, minimizing RF generation and sensitivity, on a minimum distance and with as little contact crossing as possible, to keep the signals "clean". This is why a so-called bus-system is used in different computer environments, be it cables or circuit boards. Mounting several plugs in parallel and routing the leads as narrow as possible and with the best grounding achievable, finally led to a concept of a motherboard across the bus line and several system and peripheral boards beeing plugged in a right angle into that bus-board. Keeping the dimensions of the boards narrower than 3 inches allowed for rackmountable units, suitable for rough, industrial surroundings: the ECB bus system. This ECB concept had been used by Tilmann for several years on a 64180 based computer, so he wanted to keep working peripheral boards usable for the CPU280. For instance, by that he was able to still use his static RAM-disk and an EPROM burning device. The board looks like this: ------------------------------------------------------- ! ! ! ! ------------------------------------------------------- CPU280 board with two serial I/Os on it, one for a serial terminal (you can use your old stuff!!!) and one for either a printer or a modem. A 765 compatible floppy controller with full pinout for 8" floppies can be used with either large floppies or any combination of 5 1/4" with 3" and 3.5" floppies. Yes, also High-Density! A good 1MB of RAM helps to give you a sound RAM-disk of 768K net, big enough for most people. Nevertheless, it is as easy as replacing the socketed chips by bigger ones to get as high as 4MB (minus the 300k for the system stuff)!! The user gets all the hard-to-get parts and the board itself, two EPROMS, ODD and EVEN, containing a full blown BIOS with CP/M Plus (!), a printed board-layout with all the what-goes-wheres and a short description of the first steps after soldering. All he needs are some floppy drives, a few (very cheap) electronic spares and a soldering iron, a terminal capable of 9600 8 N 1 and a cable to hook it up, plus a power supply to drive everything, 12V 2A and 5V 3A is typically enough. Any PC power supply will do. Bringing up the system on the CPU280. When I first had everything soldered together, I was really afraid. Would I break something? Would everything have been in vain? But, when I switched on the machine, a few garbled letters appeared on my screen. After a short phone call, I found out, I had simply overlooked the fact, that the SETUP in the DALLAS clock module's extra 64-byte non-volatile RAM was containing invalid data. All I had to do is shortcircuit the suitable jumper and - voila! I saw a memory count in a speed that makes me always feel sick, when I have to go back to my MS-DOS machine! Then, a small menu appears, allowing for setup of terminal and AUX characteristics like baudrate and handshake and so on. I set the terminal to 19200 8 N1 plus handshake and got a real swift output. After that, I told the system, what floppy drives there were, two DS 80 Track Highdensity drives and one 3.5 DS 80 Highdensity drive, plus one optional external 3" drive, that I use every now and then. Also, I accepted automatic summertime-followup and requested drive 0 to be used on bootup, to be able to execute automatic startups. I entered the drive search chain for CP/M Plus and checked the time and date entries, but they were just fine. When I finally saved this and entered 0 for reboot, I held my breath . . . CP/M Plus was up and running!!! The disk that came with it contains an intelligent ID code, that tells the system, which parameters to use. I was able to read the floppy rightaway. Then I could see a file containing the FORMAT MANAGER, which was the reason I bought the CPU280 in the first place. After extracting it, I looked and found more than 30 quite common formats ready to use! Next thing for me was to take Z3PLUS to the machine and see it running. Yes! No problem, except that some versions of LSH seem to be intolerant to some CP/M 3.1 BDOSes. From the CPU's system disk I noticed a file called PIPRSX.RSX. The documentation (in a file) told me, that this was a special patch to PIP, so that the file creation time of a file could be preserved, while copying with PIP in CP/M Plus. After 2 days of soldering (nights) and a week of organizing all the parts, I had the basic system running. At first, I was a bit disappointed about the relatively slow disk performance, there are faster BIOSes, yes, but then I agreed to Tilmann's arguments of absolute data integrity, which resulted in a double verify. After switching off the copy verifies in ZFILER and other copy programs, I found the speed to be equivalent to disk I/O on my 386SX. Development on the CPU280 hasn't stopped. Tilmann Reh shortly after its completion introduced a card with a Hercules graphic chip on it, including a parallel terminal and a port to hook up any old IBM PC keyboard! When I saw the performance of it, I at first thought, somebody played a trick on me. The video output is about 10 times as fast as on my 16MHz 386Sx with ANSI.SYS!! Excellent graphic capabilities are known to all the IBM PC users, I don't have to mention that. But, that wasn't enough, the cracks were putting their heads together again !!! And there we have it: -------------------------------------------------------------- ! ! ! ! -------------------------------------------------------------- the IDE controller board, including a parallel printer port! Any AT-BUS (IDE) fixed disk can be used, but we decided to buy CONNERs, since they have the full set of commands available, so that by an external command "IDEMOT OFF", I can silence the motor on the harddisk completely, until I undo this by "IDEMOT ON" or by effectively accessing the harddisk itself. Not all seeks respond in disk usage (same on the floppies), since directory buffering is used. I have been working my CPU280 almost everyday for at least 2 hours upto 8 hours and more. I found it a reliable little machine, that by it's clever design, not only allows for disk exchange with almost impossible disk formats, such as the COMMODORE C128's format on the 1581, produced by the JUGGLER program. It also allows to carry disk ID information on most systems that do not use a format ID themselves, like the AMSTRAD PCW for example. Then you just take out an ALTOS disk and mount a KAYPRO IV (C128) and immediately after that you use OSBORNE 1 and EPSON QX10 disks without calling the FORMAT MANAGER once! Just logging in the disk does the trick! Pure magic!! Talking about the C128 1581 MFM formats, they use head IDs of 1/0 instead of 0/1 and start reading the disk data from the backside (of course). On the other hand, 778 K or 794 K are a good reason to use such formats. But not only C128, also SHARP MZ computers and strange ROBOTRON machines from the former GDR, our eastern "pardner", have floppy formats that are hard to analyze and next to impossible to produce without a wonderful little thing on my desk called CPU280. Thank you, Tilmann, for putting so much work into a machine that has so far only very little to do with devices, that some people create "for the money". You made it "for the people". And they know it. Helmut Jungkunz