Editing Acorn Archimedes A3020 Troubleshooting Guide (section)

== POST fault codes (floppy-LED flash decode) ==

When POST fails, the fault is a hex value flashed on the floppy LED in eight groups of four flashes (long flash = 1, short = 0, most-significant nibble first). The documented fault and status bits are:<ref name="retrokit">[https://www.retro-kit.co.uk/page.cfm/content/Common-RISC-OS-POST-failure-errors/ "Common RISC OS POST failure errors"], Retro-Kit.</ref>

{| class="wikitable styled-table" style="width:100%; text-align:left;"
|+'''RISC OS POST fault bits (A30x0 / A4000 / A5000)'''
! Bit !! Meaning !! Likely cause on an A3020
|-
| 0x00000200 || ROM checksum failure || Corrupt or badly seated RISC OS ROM; check links LK10/LK11/LK12/LK14
|-
| 0x00000400 || MEMC CAM mapping fault || Fault in the ARM250 memory controller or its DRAM
|-
| 0x00000800 || MEMC protection fault || As above
|-
| 0x00001000 || IOC register test failure || I/O controller section of the ARM250
|-
| 0x00004000 || VIDC (Virq) timing failure || Video timing; often a VIDC-enhancer/clock issue, not always fatal
|-
| 0x00008000 || Sound (Sirq) timing failure || Sound/interrupt timing in the ARM250
|-
| 0x00010000 || CMOS clock/calendar unreadable || PCF8583 RTC or its tracks killed by battery leakage
|-
| 0x00020000 || RAM control-line failure || Battery leakage has broken a RAM control-line track — very common
|-
| 0x00040000 || Long RAM test failure || Faulty DRAM device
|}

Status bits accompany the fault: 0x1 power-on, 0x10 ARM3 fitted, 0x40 integrated I/O controller present, 0x100 CMOS checksum error.<ref name="retrokit" /> A '''RAM control-line failure (0x20000)''' on a corroded board is the classic battery-damage symptom; inspect and buzz out the RAM control tracks near the ARM250 and the battery.<ref name="retrokit" />

'''To clear a CMOS fault''', hold '''Delete''' at power-on until a black screen with a red border appears, which resets the configuration to defaults.<ref name="retrokit" />