Re: Loading basic through the serial port
Posted: Mon Oct 31, 2011 2:45 pm
Hi Phil,
The 2114 chips are organized as 1K x 4 bits, so two chips form each 1K block, each chip spanning the entire 1K. The 1K boundaries span xx00-xx3F, xx40-xx7F, xx80-xxBF, and xxC0-xxFF. So all the errors are in a single 1K block, from 4400-44FF.
The RAM on the 527 is organized as 3 x 8K banks, which can be placed anywhere in the memory space by jumpering the appropriate outputs of a 74LS138 3-to-8 decoder to the /CE lines of the three banks. By default, the board is jumpered to place the banks at 0000-1FFF (bank 1), 2000-3FFF (bank2), and 4000-5FFF (bank3).
So, the failure is in the third RAM bank, second 1K block. From the 527 schematic, each bank has its own 74LS138 decoder to select the chips. Remember that on a RAM board, it doesn't really matter in what order the CE lines are mapped to the chips, as long as each chip gets its CE signal. So the board layout designer may route the chip selects for the cleanest layout. In this case, the third RAM bank generates chip enable signals CE16-CE23, but looking at the schematic, the second 1K block ($4400-$47FF) activates CE22, (rather than CE17, as you may have guessed), so that would indicate a problem with the chips at E7/F7.
Try re-seating those chips to see if that fixes the problem. If not, try replacing the chips. Next, try removing the chips, and checking the values at $4400-$47FF. Inspect the sockets for damage, then inspect the traces for small breaks or shorts. A break in a trace to the chip would affect that bank. A short may affect all the banks (except a short involving the CE line), so would be less likely. Try resoldering the pins on these chips. You can try to use a logic probe to detect a pulse on the CE pin to that bank (pin8 of E7 and F7), and at the decoder (pin 14 of E9).
Do you by any chance have an oscilloscope? Not necessary, but it could come in handy.
Finally, I'm basing the above on the schematic for the 527 board, but you indicate you have a 427 board. I couldn't find that in my archive or in Mark S's board archive (http://osi.marks-lab.com/boards/boards.html). If you have a 427 board, would you mind taking a picture of it and posting it here so we can see the differences? If it's a different board, then most of the above would likely still apply, but you may have to figure out the address decoding scheme by looking at which address lines go into each decoder, and which chip enable lines go to which RAM chips. It would likely be similar to the 527 board, but routing may not be identical. About 10 minutes worth of work with an ohmmeter.
Best regards,
Dave
The 2114 chips are organized as 1K x 4 bits, so two chips form each 1K block, each chip spanning the entire 1K. The 1K boundaries span xx00-xx3F, xx40-xx7F, xx80-xxBF, and xxC0-xxFF. So all the errors are in a single 1K block, from 4400-44FF.
The RAM on the 527 is organized as 3 x 8K banks, which can be placed anywhere in the memory space by jumpering the appropriate outputs of a 74LS138 3-to-8 decoder to the /CE lines of the three banks. By default, the board is jumpered to place the banks at 0000-1FFF (bank 1), 2000-3FFF (bank2), and 4000-5FFF (bank3).
So, the failure is in the third RAM bank, second 1K block. From the 527 schematic, each bank has its own 74LS138 decoder to select the chips. Remember that on a RAM board, it doesn't really matter in what order the CE lines are mapped to the chips, as long as each chip gets its CE signal. So the board layout designer may route the chip selects for the cleanest layout. In this case, the third RAM bank generates chip enable signals CE16-CE23, but looking at the schematic, the second 1K block ($4400-$47FF) activates CE22, (rather than CE17, as you may have guessed), so that would indicate a problem with the chips at E7/F7.
Try re-seating those chips to see if that fixes the problem. If not, try replacing the chips. Next, try removing the chips, and checking the values at $4400-$47FF. Inspect the sockets for damage, then inspect the traces for small breaks or shorts. A break in a trace to the chip would affect that bank. A short may affect all the banks (except a short involving the CE line), so would be less likely. Try resoldering the pins on these chips. You can try to use a logic probe to detect a pulse on the CE pin to that bank (pin8 of E7 and F7), and at the decoder (pin 14 of E9).
Do you by any chance have an oscilloscope? Not necessary, but it could come in handy.
Finally, I'm basing the above on the schematic for the 527 board, but you indicate you have a 427 board. I couldn't find that in my archive or in Mark S's board archive (http://osi.marks-lab.com/boards/boards.html). If you have a 427 board, would you mind taking a picture of it and posting it here so we can see the differences? If it's a different board, then most of the above would likely still apply, but you may have to figure out the address decoding scheme by looking at which address lines go into each decoder, and which chip enable lines go to which RAM chips. It would likely be similar to the 527 board, but routing may not be identical. About 10 minutes worth of work with an ohmmeter.
Best regards,
Dave