Image Map Image Map
Page 14 of 20 FirstFirst ... 4101112131415161718 ... LastLast
Results 131 to 140 of 193

Thread: PET 2001-32N - Diag works in UD9, but with Kernal it fails to start

  1. #131
    Join Date
    Jun 2012
    Location
    UK - Worcester
    Posts
    3,218

    Default

    Yes, thanks for prompting me. I did have a few thoughts.

    This is all just "thinking out aloud" though for the time being...

    With Frank's board in place (i.e. with the PET ROM and RAM being replaced) the only thing that should be operational is the I/O sub-system from $E800 to $EFFF. Everything else should be bypassed.

    Not knowing how Frank's board works is a bit of a disadvantage at the moment. Perhaps you could enlighten is?

    Is a ROM and/or RAM access brought out to the external world of the PET (on the address/control bus for example) but just ignored when the data returns? If this is the case, some prevailing problem with the PET ROM or RAM could be corrupting the I/O space causing your strange problems with the cursor signal possibly.

    If, however, the access doesn't come to the PET, then we should be able to find if something is being enabled by accident possibly?

    Possibly look at the outputs from UD2 (74154) and see if any output other than /SELE is being activated.

    The only thing I could suggest next would be to remove the 6502 CPU and Frank's board and then test the address and data bus through from the CPU socket to the I/O subsystem one line at a time to ensure that everything looks OK. In particular, I am thinking of the address buffers (UB3 and UC3). Perhaps a NOP generator for this?

    If you have the kernel ROM running off Frank's board and my PETTEST in the EDIT ROM socket and this is not working, there has to be something awry with the ROM or screen selection somewhere that should be locatable.

    Dave

  2. #132

    Default

    Quote Originally Posted by daver2 View Post
    Yes, thanks for prompting me. I did have a few thoughts.

    This is all just "thinking out aloud" though for the time being...

    With Frank's board in place (i.e. with the PET ROM and RAM being replaced) the only thing that should be operational is the I/O sub-system from $E800 to $EFFF. Everything else should be bypassed.

    Not knowing how Frank's board works is a bit of a disadvantage at the moment. Perhaps you could enlighten is?

    Is a ROM and/or RAM access brought out to the external world of the PET (on the address/control bus for example) but just ignored when the data returns? If this is the case, some prevailing problem with the PET ROM or RAM could be corrupting the I/O space causing your strange problems with the cursor signal possibly.

    If, however, the access doesn't come to the PET, then we should be able to find if something is being enabled by accident possibly?

    Possibly look at the outputs from UD2 (74154) and see if any output other than /SELE is being activated.

    The only thing I could suggest next would be to remove the 6502 CPU and Frank's board and then test the address and data bus through from the CPU socket to the I/O subsystem one line at a time to ensure that everything looks OK. In particular, I am thinking of the address buffers (UB3 and UC3). Perhaps a NOP generator for this?

    If you have the kernel ROM running off Frank's board and my PETTEST in the EDIT ROM socket and this is not working, there has to be something awry with the ROM or screen selection somewhere that should be locatable.

    Dave
    Hi Dave,
    the version of the RAM/ROM board that he have, only access the original PET bus from $E800 to $E8FF. ROMs can be selected in 4K ranges (for example, $F000 to $FFFF can be accessed on the motherboard and all the rest from the daughterboard. Any combination of the 4K ranges from $9000 to $FFFF can be selected by moving jumpers).
    The board has obviously a 74LS245 transceiver on the data bus, so all on-board accesses are isolated from the main PET databus.
    Of course if something on the PET board (like RAM/ROMs) conflicts with the I/O accesses on the databus, the board can't help.

    Frank

  3. #133

    Default

    Quote Originally Posted by iz8dwf View Post
    Hi Dave,
    the version of the RAM/ROM board that he have, only access the original PET bus from $E800 to $E8FF. ROMs can be selected in 4K ranges (for example, $F000 to $FFFF can be accessed on the motherboard and all the rest from the daughterboard. Any combination of the 4K ranges from $9000 to $FFFF can be selected by moving jumpers).
    The board has obviously a 74LS245 transceiver on the data bus, so all on-board accesses are isolated from the main PET databus.
    Of course if something on the PET board (like RAM/ROMs) conflicts with the I/O accesses on the databus, the board can't help.

    Frank
    Thanks Frank.

    So, Dave... As a recap, when first installing the daughterboard and the 7407/7417 that had a leg or two that had deteriorated to intermittent contact, the PET booted to BASIC perfectly, though it has that odd spurious signal (only with BASIC running) hitting the PIA, apparently causing the fast-cursor flash. From there, the longer the PET has run, the worse it has become, until we are back where we are now. It's pretty clear that TIM is invoked immediately and the display is 32Gs followed by 32 random characters that don't update followed by 32Gs, etc., when I have the petdiag ROM switched in. When I boot from a custom ROM that I wrote for KERNAL that just iterates over incrementing characters through the VRAM space, it follows that same pattern (32 chars ok, 32 not), with some intermittent "junk" now and then even in the 32 good characters.

    It very much seems that there are so many marginal components on the board that every time I fix one thing, three more symptoms appear. I've actually been eyeing the buffer chips with suspicion, especially with the 32 character chunks of good and bad in the diag ROM.

    When I switch in BASIC KERNAL with your DIAG in the 2k EDIT slot, it does not appear to even get as far as clearing the screen.

    I appreciate the assistance from both of you (and everyone else who has pitched in).

  4. #134

    Default

    Quote Originally Posted by iz8dwf View Post
    Hi Dave,
    the version of the RAM/ROM board that he have, only access the original PET bus from $E800 to $E8FF. ROMs can be selected in 4K ranges (for example, $F000 to $FFFF can be accessed on the motherboard and all the rest from the daughterboard. Any combination of the 4K ranges from $9000 to $FFFF can be selected by moving jumpers).
    The board has obviously a 74LS245 transceiver on the data bus, so all on-board accesses are isolated from the main PET databus.
    Of course if something on the PET board (like RAM/ROMs) conflicts with the I/O accesses on the databus, the board can't help.

    Frank
    oh, and obviously goes into the motherboard databus also for $8000-$8FFF (because of the video ram).

    Frank

  5. #135

    Default

    Quote Originally Posted by dhoelzer View Post
    Thanks Frank.

    So, Dave... As a recap, when first installing the daughterboard and the 7407/7417 that had a leg or two that had deteriorated to intermittent contact, the PET booted to BASIC perfectly, though it has that odd spurious signal (only with BASIC running) hitting the PIA, apparently causing the fast-cursor flash. From there, the longer the PET has run, the worse it has become, until we are back where we are now. It's pretty clear that TIM is invoked immediately and the display is 32Gs followed by 32 random characters that don't update followed by 32Gs, etc., when I have the petdiag ROM switched in. When I boot from a custom ROM that I wrote for KERNAL that just iterates over incrementing characters through the VRAM space, it follows that same pattern (32 chars ok, 32 not), with some intermittent "junk" now and then even in the 32 good characters.

    It very much seems that there are so many marginal components on the board that every time I fix one thing, three more symptoms appear. I've actually been eyeing the buffer chips with suspicion, especially with the 32 character chunks of good and bad in the diag ROM.

    When I switch in BASIC KERNAL with your DIAG in the 2k EDIT slot, it does not appear to even get as far as clearing the screen.

    I appreciate the assistance from both of you (and everyone else who has pitched in).
    I had a thought and decided to try booting without the 6520s populated... It still goes straight to TIM. Removed the 6522 as well, the same.

  6. #136
    Join Date
    Feb 2009
    Location
    Southern California, USA
    Posts
    2,712

    Default

    Quote Originally Posted by dhoelzer View Post
    It's pretty clear that TIM is invoked immediately...
    Does the Machine Language Monitor work? Is there a C* in the upper right corner for Call to Monitor or a B* for Break to monitor?

  7. #137
    Join Date
    Jun 2012
    Location
    UK - Worcester
    Posts
    3,218

    Default

    Back in post #135 when you removed the 6520's and 6522 - what was the configuration of the rest of the machine? Were you using Frank's board or 'real' ROMS from the PET and (if so) which ones?

    Dropping into TIM and/or the inability to start up my PETTESTER and/or strange addressing issues with the screen smack of address line problems to me somewhere. If this is the case (and your problem turns out to be an address line issue somewhere) this could also affect the I/O sub-system when running with Frank's ROM/RAM board.

    32 bytes (decimal) = $20 (hexadecimal) so this may be a good place to start?

    If we assume that the power supply and oscillator circuits are OK for the time being, could I suggest we proceed as follows:

    1. Remove the 6502 CPU (and Frank's board), power up the beast and post a photograph of the screen. Does the 'power on garbage screen' 'look' right?
    2. Have you got (or can you make, beg, buy, borrow or steal) a NOP generator. This should allow you to checkout the address lines, address buffers and 'high level' address decoding circuits.

    Dave

  8. #138

    Default

    Quote Originally Posted by dave_m View Post
    Does the Machine Language Monitor work? Is there a C* in the upper right corner for Call to Monitor or a B* for Break to monitor?
    Yes, though it's unusable with the screen and fast-flash cursor.

  9. #139

    Default

    Quote Originally Posted by daver2 View Post
    Back in post #135 when you removed the 6520's and 6522 - what was the configuration of the rest of the machine? Were you using Frank's board or 'real' ROMS from the PET and (if so) which ones?

    Dropping into TIM and/or the inability to start up my PETTESTER and/or strange addressing issues with the screen smack of address line problems to me somewhere. If this is the case (and your problem turns out to be an address line issue somewhere) this could also affect the I/O sub-system when running with Frank's ROM/RAM board.

    32 bytes (decimal) = $20 (hexadecimal) so this may be a good place to start?

    If we assume that the power supply and oscillator circuits are OK for the time being, could I suggest we proceed as follows:

    1. Remove the 6502 CPU (and Frank's board), power up the beast and post a photograph of the screen. Does the 'power on garbage screen' 'look' right?
    2. Have you got (or can you make, beg, buy, borrow or steal) a NOP generator. This should allow you to checkout the address lines, address buffers and 'high level' address decoding circuits.

    Dave
    Good morning, Dave-

    From 135, yes, Frank's board was still socketed. I tried PETDIAG, full BASIC 4 ROMs, your EDIT ROM with BASIC 4 Kernal, my own KERNAL replacement.

    PETDIAG runs, blocks of characters are wrong and run in blocks of 32.
    BASIC drops straight to TIM
    Your diag doesn't start (or attempt to clear the screen)
    Mine just iterates characters across the VRAM (from when I was building VRAM replacements) and runs, still with blocks of 32 characters good/bad.

    I'll build a NOP generator this morning in addition to socketing a 6502 directly.

    Beast is right.

  10. #140
    Join Date
    Jun 2012
    Location
    UK - Worcester
    Posts
    3,218

    Default

    >>> Beast is right.

    It builds character !

    Dave

Tags for this Thread

Bookmarks

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •