• Faulty 8568?

    From Daniel Karlsson@daniel.j.karlsson@telia.com to comp.sys.cbm on Wednesday, July 09, 2003 20:20:27
    From Newsgroup: comp.sys.cbm

    Hi all!

    I upgraded my VDC RAM a couple of days ago to 64 kB in my flat 128.
    Everything worked fine and afterwards I ran the C128D factory test to realy test the VDC RAM. This program divides the VDC RAM in four blocks
    ($0000-$3FFF, $4000-$7FFF, ...) and tries to write first a value of $FF to every place in block 1, tries to read it, tries to write $AA and read it and the same with $55 and $00, the same with block 2, 3 and 4 wherafter it tries
    to execute a block write and a block copy in block 1 and starts over again. Everything worked just fine here to. Happy as I was I thought that I would
    test the VDC RAM in my DCR to, just for the fun of it. It was now that I was
    a bit suprised. The VDC chip seems to miss roughly every 100.000 write (the "old" value is read after the write, i.e. if the VDC have tried to write
    e.g. $55 it reads $AA). Of course I thought that this was a sign of a 8568
    soon to break down but now I've heard someone say that this actually is a "feature" in _all_ 8568. Now is this right, i.e. is it pointless to exchange the VDC for another one or should I try to get hold of another VDC to fix
    the problem. I might ad that when an error occur it occures in both U23 and
    U25 and in all memory locations whitch seems to rule out faluty RAM chips.

    Regards
    /djk


    --- Synchronet 3.18b-Win32 NewsLink 1.113
  • From Cameron Kaiser@ckaiser@floodgap.com to comp.sys.cbm on Wednesday, July 09, 2003 16:35:22
    From Newsgroup: comp.sys.cbm

    "Daniel Karlsson" <daniel.j.karlsson@telia.com> writes:

    The VDC chip seems to miss roughly every 100.000 write (the
    "old" value is read after the write, i.e. if the VDC have tried to write
    e.g. $55 it reads $AA). Of course I thought that this was a sign of a 8568 >soon to break down but now I've heard someone say that this actually is a >"feature" in _all_ 8568.

    Early VDC chips had major problems with timing collisions and would fail
    memory writes at unacceptable levels due to the VDC's slightly different timing. Even now, it would not be surprising to hear that writes continue
    to be glitchy (in fact, what *is* surprising that the glitched writes are
    that rare -- I would have expected a much higher failure rate :-P).

    --
    Cameron Kaiser * ckaiser@floodgap.com * posting with a Commodore 128
    personal page: http://www.armory.com/%7Espectre/
    ** Computer Workshops: games, productivity software and more for C64/128! **
    ** http://www.armory.com/%7Espectre/cwi/ **
    --- Synchronet 3.18b-Win32 NewsLink 1.113
  • From Nicolas Welte@welte_spam@freenet.de to comp.sys.cbm on Thursday, July 10, 2003 15:54:02
    From Newsgroup: comp.sys.cbm

    Daniel Karlsson wrote:
    a bit suprised. The VDC chip seems to miss roughly every 100.000 write (the "old" value is read after the write, i.e. if the VDC have tried to write
    e.g. $55 it reads $AA). Of course I thought that this was a sign of a 8568 soon to break down but now I've heard someone say that this actually is a "feature" in _all_ 8568. Now is this right, i.e. is it pointless to exchange the VDC for another one or should I try to get hold of another VDC to fix
    the problem. I might ad that when an error occur it occures in both U23 and U25 and in all memory locations whitch seems to rule out faluty RAM chips.

    I tested this on my machine, and it passes the complete test several times without failures. When I have time I'll run the test on the other two DCRs I have, and on my spare 8568.

    Nicolas


    --
    Email address is valid for replies (requires Re: in the subject) <--
    - See my Commodore hardware projects at http://people.freenet.de/x1541 -
    - Visit the German X1541 Shop at http://sta.c64.org/x1541shop_ger.html -

    --- Synchronet 3.18b-Win32 NewsLink 1.113