Inspection:
The 4K board was in better condition than the main board, but still has some very bad solder joints. It had been configured for on-board 7805 regulation using 8V from the bus, but that regulator was previously removed and the board reconfigured for using 5V from the main board regulator. This board draws 600 mA and so it is within the capabilities for the main board regulator to support it, The PC lands in the regulator area had partially lifted due to excessive used in the modifications done to it. The electrolytics were replaced and the board cleaned up.
Cleanup:
All the poor soldering was corrected and we retained the last configuration for using 5V from the main board.
Testing:
I configured the board to start @ 0x0000 and attempted to enter the simple Q test program. This failed. Using an oscilloscope I started checking signals going to the board and to the RAM array and discovered that the MWR signal was not getting through to the RAM array. Further testing revealed that a driver, A4 in the 75C902 hex buffer, had failed. I bypassed this driver with a single 74LVC1G125 driver (see photos) and was then able to enter in a program successfully. I sourced a few 74C902 (hard to find) from EBay and replaced this bad one. I then wrote a memory test that stores various combinations of data in the memory array and verifies them. I will post this in another BLOG post. The test program writes alternating ones, zeros, and walks a one through memory, then walks a zero through memory, and finally writes all ones and zeros. The test was looped for 24 hours with no failures.
