ALU Error
Previous tests suggest that the ALU could be generating an all-1s condition when it should be generating all-zeros.
ALU Board is an Undocumented Version
The ALU board in this 600 machine is coded 6192 and this board number is not found in the known Wang schematics, these show the 600 series ALU to be a 5964 board. Checking another 600 machine in the collection showed that it had a 5964 ALU. Comparing these boards showed:
- minor trace routing changes
- the Z-bus destination decoder on the 6192 board had wiring associated with destination ‘7’, while on the 5964 destination ‘7’ is unused and forms an ALU no-op code.
The new destination 7 wiring is brought to the ALU edge connector but there is no connection on the machine backplane, so destination 7 remains a no-op in the 600. Discussion with other Wang experts suggests that the 6192 board was actually for the Wang 1200 computer, where the destination 7 code would have been used. Since the 6192 was based on the 5964 it was back-compatible and late-production 600 machines may have used a 6192 to allow production of 5964 boards to cease.
Microcode mini-ROM
miniROM emulator
Another useful DoPECC-developed test instrument is the Microcode mini-ROM. This is a ROM emulator that contains two words of microcode that are easily set to any code by a set of rotary switches. One word is active at all even ROM addresses and the other at all odd addresses. This allows microcodes of one or two words to be rapidly set up and executed.Testing the ALU and A-bus Selector
The ALU board was hooked up to the logic analyzer and the mini-ROM set to repeat a single ALU operation. The A-bus was seen to have incorrect values for some operations. Monitoring the A-bus data selector, a 74151 1-of-8 selector, showed:
- incorrect response to its disable input - generating a high rather than low output
- incorrect selection of some inputs
- inconsistent true and complemented outputs
All of which suggests a malfunctioning 74151 chip.
Testing and Replacing the 74151
The 74151 was removed, testing on a generic chip multi-tester reported an error, this tester can correctly identify a good ‘151.
The chip was replaced with NOS 74LS151. Compatibility between TTL and LSTTL is generally good, the reduced output drive of LSTTL is not relevant here.
The chip did not show any physical evidence of damage or deterioration and its failure mode was rather unusual. It had not failed completely, instead it produced incorrect outputs for only some combinations of inputs and there was no obvious relationship to any one pin or function. Peculiar.
ALU Repaired
Repeating the single-instruction tests with the miniROM showed that the ALU now functioned correctly.
