Fehlersuche P3

    Gestern ist ja leider mein Alphatronic P3 kaputt gegangen (wie hier erwähnt: UCSD Pascal für die P3).

    Ich habe mich also auf die Fehlersuche begeben und den Fehler inzwischen zumindest eingegrenzt:


    Wenn ich den Rechner ohne Speicherkarte und Floppy Controller starte, so erscheint meist korrekt der MOS Startbildschirm:

    Reset4001 4000

    CAAP-3603 1B76 4000 40FF


    Beim Einsetzen von zwei 16 KB SRAM Karten aus der P2 (mit Basisadresse 8000 und C000) kommt ebenfalls meistens der obige Startbildschirm (manchmal bleibt der Rechner allerdings mit der Meldung Reset hängen).

    Nach der MOS Meldung können die üblichen MOS Befehle eingegeben werden. Beschreiben des Bildschirmspeichers ab $3000 bzw. der SRAM Karten ab $8000 und $C000 funktioniert (Fxxxx,yyyy,zz und Dxxxx,yyyy liefern die gewünschten Ergebnisse).


    Allerdings kann ich den Speicher direkt auf der CPU Karte nicht beschreiben. Auch wenn ich die originale 64 KB DRAM Karte, ein drittes SRAM Modul (ab $4000) oder den Floppy Controller einsetze, so bleibt der Rechner mit der Meldung Reset hängen bzw. versucht den Boot-Prozess durchzuführen und hängt dann mit blinkender Reset Meldung).


    Offenbar hat die Bus-Adressierung auf der CPU Karte (möglicherweise auch die Logik zur Umschaltung zwischen 48K und 64K Modus) etwas abbekommen. Damit habe ich zumindest eine erste Spur, weitere Tests werden nun wohl etwas mühsamer. Entweder ich mache mich mit dem Oszilloskop auf die Suche, oder ich versuche eine Fehlereingrenzung mit einfachen 8085 Maschinensprachprogrammen.

    Einmal editiert, zuletzt von gpospi ()

    Also es geht um deine P3 - ok.


    Wenn alle RAM- Karten entfernt vom Steckrahmen sind, dann aber mit dem Floppycontroller im Reck - klar OHNE Disk im Drive - wie ist dann die Mosanzeige?


    Es sind nicht klar aus deinen Aufzählungen der EIN / AUS Karten - ob gleichzeitig, oder je einzeln für die Karten probiert wurden.


    Komisch ist aber warum ein Teiltext auf dem Display angezeigt als „BLINKEN“ wurden.

    Hallo gpospi !

    Deine erste Tat war doch:

    Zitat von gpospi

    Wenn ich den Rechner ohne Speicherkarte und Floppy Controller starte, so erscheint meist korrekt der MOS Startbildschirm:

    Reset4001 4000

    CAAP-3603 1B76 4000 40FF

    Ist die Anzeige vom MOS auch identisch (wie in deinem Foto) - wenn jetzt der FD-Controller im Rahmen steckt?

    (ich meine bei der P3 /P30 wird ja ein AUTOBOOT versucht vom Drive 1 - ohne Diskette könnte aber eine andere MOS Meldung erfolgen?)

    ODER hängt der RECHNER schon ohne RAM und nur mit dem FD-Controller?????

    Ja der Rechner hängt auch ohne RAM (nur mit Floppy Controller) mit blinkender Reset Anzeige, egal ob eine Disk eingelegt ist oder nicht.


    Auch beim Drücken der Reset Taste wenn nur CPU und Display / Tastaturcontroller eingesetzt sind wird Reset nicht immer sauber abgearbeitet (dh. mit der korrekten MOS Meldung), sondern führt manchmal auch nur zur Anzeige von "Reset".


    Ich vermute, dass irgendwas in der Reset/Adressierungslogik nicht korrekt funktioniert, werde wohl diverse Bauteile auf der CPU Karte durchmessen müssen...

    Ich werde natürlich auch wieder beginnen, die Karten einzeln in meinem P2 Rechner zu testen. Zumindest bei RAM, CPU und Floppy Controller ist das ja grundsätzlich möglich. Damit sollte ich zumindest klar sehen, ob nur die CPU Karte ein Problem hat.

    Kurzes Update: Habe alle passenden Komponenten der P3 erfolgreich in mein P2 System transplantiert (800 Kb FDD, FDD Controller, CPU Karte). Damit wird das "Allover" Kopierprogramm erfolgreich geladen (dieses arbeitet grundsätzlich im 48 KB Modus). Sobald ich aber die 64 KB RAM Karte einbaue hängt das System.


    Erkenntnis daher:

    a) Die 64 KB RAM Karte ist offenbar defekt (Helwie hatte ja neulich ebenfalls einen defekt auf diesem Teil).

    b) Die P3 funktioniert nicht mit 3*16 KB SRAM Karten in meinem P3 Chassis (sehr wohl aber im P2 Chassis). Die Backplane ist bei P3 ja anders (Keyboard Anschluss, Interrupt-Leitungen und evtl. die 64 KB Adressierung).


    Ich muss also wohl die 64 KB RAM Karte näher untersuchen (und eventuell die P3 Backplane). Vielleicht ist da auch ein Kondensator hinüber. Insgesamt hoffentlich eine machbare Analyse-Aufgabe ;)

    Einmal editiert, zuletzt von gpospi ()

    Gefällt mir 3

    Hello gpospi


    Definitely the memory wiring is different. I don't have the 64K configuration, but the 48+16 one. Months ago I found that both cards "talk" between each other using slot C. the 16KB board doesn't send the data bus values into the memories, they are receiving them from the 48K board and then are latched. There are lots of ICs in that board for only 16K... that must be the logic responsible for banking. I found another memory IC other than the '373, which is a '74 (d-type flip-flop). I traced its output to the connector (which feeds some gates before reaching its end). I suggest you to find such flip-flop in your 64K board.


    I'm glad you found the defective board. Thinking you had a defective CPU card, I started to disassemble CAAA08 last night. My P3 is also broken, even after a year... not because I couldn't but because convenience. Its unpopulated backplane is used for electrical tests for my own boards before connecting them into a working system (preventing wrecking the systems that work fine). The second reason is reverse engineering. Not long after receiving it I took care of documenting the backplane but the cards still remain to be documented.


    Regards,

    Jaume

    When I tried to list all retro systems I have at home, the "The message is too long, must be under 500 characters" error appears! :lol:

    Gefällt mir 1

    I tried to do some further debugging, but I did not really reach any conclusion. Firstly I wanted to find out if the CPU card as such is really ok. In my current test setup (P3 backplane, P3 KBD+Video controller, single SRAM card at 8000hex, P3 FDD controller) the reset cycle often doesn't complete properly. In the correct case, the computer display is:

    The computer did the power on memory test, failed immediately (as no board is present at address 4000hex). Hence the MOS version and stack pointer address are shown, afterwards the MOS prompt appears.


    Nevertheless, quite often I just get:

    There is just the text "Reset" with a block cursor at the end. Sometimes some garbage characters appear at the screen as well. I assume this is a consequence of the reset cycle not executing fully correct. I might analyze the Reset timing/capacitor behaviour later, but at the moment I am fine (considering the CPU card not as root of my problem).


    My next test was to find out if the "onboard mechanisms" of the CPU card work properly. Unfortunately the P3 doesn't have any general purpose RAM on the CPU card. According to the system manual, the memory map is as follows:

    In this case, "frei" (empty) means really empty (i.e. there is no RAM). However I could put some small machine language test programs into the SRAM card at 8000hex and write data to the video RAM or do some other simple stuff. Also MOS "F" and "D" commands work for video RAM and in the area "Variablen / Stack" (i.e. this area can be partially filled). Hence it seems that the addressing logic of the CPU card is overall working.


    Finally I did some analysis on the 64 KB RAM card:

    • Check capacitors: They seem to be ok, no obvious short circuit (resistance is always 300-600 Ohm).
    • Insert SRAM card into the computer: Computer shows "Reset", spins up the FDD (but doesn't position the head and read data) and finally "Reset" begins to flash.
    • Putting the SRAM card into the computer without inserted RAM controller chip: Screen remains black, computer doesn't start at all.

    Hence I would obviously need a deep-dive into the RAM card, but I don't really know where to start...

    Which problems do you have with your P3? You have the 48K+16K RAM configuration in your P3?

    Hallo jlopez and gpospi , netmercer and other USER ;


    Have a find function in „Triumph Adler“ thread with „backplane“ startet!


    I think we should document the mechanism for bank switching with all the many P2 / P2 / P30 and sks KISS users.


    reverse engineering :

    I think we should document the mechanism for bank switching with all the many P2 / P2 / P30 and sks KISS users.


    Back then I was only concerned with the software. Unfortunately, I had therefore not collected any real circuit diagrams from sks / TA as paper.


    Or is the work too much?

    Regards Helmut


    Zitat von gpospi

    Which problems do you have with your P3? You have the 48K+16K RAM configuration in your P3?

    I can't recall if it started at all... at least a drive was running free and the other had an IC that seemed to be blown. But at this moment won't work for other reasons... Thanks for your interest.


    Zitat von helwie44

    I think we should document the mechanism for bank switching with all the many P2 / P2 / P30 and sks KISS users.

    :thumbup:





    Zitat von helwie44

    Or is the work too much?

    Cumbersome for a single person. Nevertheless, I could gradually post captures of exposed boards so nets could be generated. Having proper schematics is becoming a necessity for repairs...


    If you want, I could start a new thread on this subject as soon as the full board is exposed.


    Regards,

    Jaume

    When I tried to list all retro systems I have at home, the "The message is too long, must be under 500 characters" error appears! :lol:

    Gefällt mir 2

    YES jaume - something like that is an enormous amount of work.


    If I still had some time on the side, I would have a "still running P2U and / or a KISS

    (with a threaded MIMIK (hardware patch!! from 1981 ** ) for BANK selection).

    I have a SCOPE and an 8 bit (replica) logic analyzer.



    Post-Comment:

    Unfortunately my friend Bernd S. (cp / m signature BS) made the hardware patch at the time.

    That was around 1981 at the UNI Hamburg in the department of computer science. After a few years he went to UNI Weimar in the IT department. But unfortunately died a few years ago.

    Hallo gpospi


    if you need some help for searching the error, perhaps I can give you some hints.

    Last month Aquarius and I have checked his defective BOSS Z80. There were problems with the RAM, so the machine crahes after power on. Details you can find here.


    Olypia Boss with Z80 doesn't work


    We had also no circuit diagrams of the BOSS., So Aquarius found out a lot of connection of the ICs and in combination with a disassembly listing of the BIOS we could write some small test programms to find out the position of the defective RAM.


    For this purpose I wrote test programs, which are working without any RAMs.

    Problem was, to find a way to communicate. On the BOSS there was a serial interface controller (RS232 asynchron) we could use to send information to the user.


    If you think about drwaing a circuit diagramm taken from the board, I can say you, that is really a hard work. To find out the RAM-Bankswitching mechanisum it's better to disassemle the BIOS and make a small testprogram to check the switching.


    For the testprogram burning into the EPROM, Aquarius had found a good solution with an NVRAM. It is quick erasable and programable.


    Regards, PAW

    Yes I have seen your work on the BOSS, nice debugging done! I still need to make a plan for debugging my P3. My main problem is at the moment that I can't interrupt the start process - when I insert the 64 KB RAM card the machine always tries to boot and stops with the blinking "Reset" message (but doesn't jump into the MOS monitor). Hence I would indeed need to disassemble the ROM and change the boot process (probably experiments could be done in the MESS emulator first). Alternatively I may try to force the system into the MOS monitor via some hardware changes (e.g. keeping the RAM card in Reset state until the CPU has done the initialization, fails to find the RAM and goes into MOS). As soon as I can enter MOS, I can use the embedded commands to test the RAM (addressing/chips), run simple test programs for the 48 KB/64 KB bank switching, etc.