BOARD NO:     0113
GAME:         Scramble
MANUFACTURER: Konami
DATE:         1981
HARNESS:      KONAMI-S

SYMPTOM (05/05/96)
-------
(1) No sound.
(2) Starfield not flashing.

SOLUTION
--------
(1) Digital noise from the speaker suggested that the amplifier was on. Using a pulse injector on all six sound channel outputs in turn from the two AY-3-8910's produced tones at the speaker for each channel. This proved that the analogue audio post-processing and amplification circuits were all broadly working. Looking at the six channel outputs on an osciloscope proved that there was no output from the AY-3-8910's. Sound commands are presented to the audio processor by the video processor via a peripheral bus atached to one of the AY-3-8910, and signalled by an interrupt. Valid sound codes were being generated by the video processor at the AY-3-8910 and an interrupt signal was being generated at the audio Z80 but this interrupt was never acknowledged by the audio Z80 (INT always low). Changing all the socketed components had no effect (the two AY-3-8910's, the Z80 and all the EPROMS). All the system clocks were running and all the chip-selects and write strobes were pulsing. All the address and data bus pins were pulsing. Using an oscilloscope, the address buffers 3H, 3F, 5F and 4F were verified as working correctly as was the data buffer 5G. The control buffer 4H was working correctly. The EPROM decode chain (4J, 5K, 5J, 3K and 4E) all seemed to be working properly as did the RAM decode chain. The AY-3-8910 decode chain (4A, 5A, and 6A) also seemed to be working correctly. Valid address and data were present on all the devices. Atatching a logic analyser to the Z80 address, data and control signals suggested the processor had crashed (execution from RAM, invalid opcode fetches, execution from non-existent address space and generally incomprehensible nonsense). The RAM was the only component that hadn't been checked, and the analyser indicated a possible stuck bit in the RAM but this was by no means conclusive proof given the state the processor had got itself into. Changing the 2114 at 6D had no effect but changing the 2114 at 6C fixed the sound completely (phew!).
(2) The starfield was being displayed but not changing. The starfield psuedo-random counter circuit was running according to a logic probe but the selector circuits (8P and 9P) were static. The selector circuit was fed from an NE555 timer at location 10P, and there was no output from this device. Using a pulse injector set to 0.5pps on the NE555 output pin brought the starfield to life. The two resistors on the device were measured as indicated. Changing the 10uF time constant capacitor had no effect. Changing the NE555 device itself fixed the game.