Acorn Archimedes A3020 Troubleshooting Guide

This guide gives fault diagnosis for the Acorn Archimedes A3020 (ARM250 "Heron E" board, part 0194,500). It follows the module-level fault-finding of the Service Manual, with test points and values from the Technical Reference Manual circuit diagrams. Because the ARM250 integrates the ARM2 core, MEMC1a, VIDC1a and IOC into one package, processor-family faults are isolated to that single device rather than to separate chips.

⚠️ Mains warning

The A3020's power supply is built onto the main PCB, so the board carries mains voltages. Disconnect the mains lead before removing the cover, and discharge the PSU primary bulk capacitor before probing the left-hand (power) end of the board. After any internal work, re-apply the earth-continuity and DC-insulation (class 1) safety tests.

Acorn A3020 "Heron E" main PCB circuit diagram, sheet 1 of 3 (drawing 0194,500/C), showing the ARM250, the RISC OS ROMs, the PCF8583 battery-backed RAM/RTC, reset circuitry, the IDE and floppy interfaces and the filtered +5 V supply. Sheets 2 and 3 are also on the wiki.

Normal power-on sequence

A healthy A3020 runs a short power-on self test (POST) before booting RISC OS. The screen border steps through a colour sequence as the test proceeds, and any fault is reported by flashing the floppy-drive LED. Knowing the normal sequence tells you how far the machine gets before it stops:

1. Brief ROM and RAM check, IOC and VIDC initialised (border colour change).
2. Extended memory test (a few seconds; longer with 4 MB).
3. Check for an ARM3 (there is none on a standard A3020).
4. Memory-size message, then RISC OS boots to the desktop.

A soft reset (the Reset key, or re-entry to the reset vector) skips POST and starts RISC OS immediately; a power-on reset runs the full POST. If a machine boots from the Reset key but not from cold, suspect the power-on-reset circuit or a marginal PSU.

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:

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. 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.

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.

Power and PSU diagnosis

Most "dead A3020" faults are in the on-board PSU or its feed. Work through the supply before anything else, with the mains lead disconnected except where a live reading is explicitly required.

A3020 supply rails and test points

Rail / point	Nominal	Feeds / note
+5 V	5.0 V (accept ≈4.9–5.1 V)	Logic (ARM250, RAM, ROM, I/O), floppy, hard disc
+12 V	12.0 V (accept ≈11.7–12.3 V)	Hard disc motor; video connector SK1 pin 12 (SCART switching)
Floppy power	pin 1 = +5 V, pins 2–3 = 0 V	White lead = +5 V
IDE 4-way power	pin 1 = +12 V (yellow), pin 4 = +5 V (red), pins 2–3 = 0 V	Hard-disc feed
PSU primary	≈4 Ω continuity	Open circuit = blown transformer thermal fuse

The tolerances above are typical electronics practice; Acorn's module-level manual treats the PSU as a whole and does not publish tight rail tolerances.

Procedure:

1. Completely dead (no green power LED, no Caps-Lock LED response): check the mains-plug fuse, then the internal fuse FS1. If a new fuse blows on switch-on, the PSU section is faulty.
2. Measure the PSU primary continuity with the machine unplugged — about 4 Ω is healthy; an open circuit means the transformer's internal thermal fuse has blown.
3. With power applied, measure +5 V and +12 V. A single missing or low rail points to the PSU section.
4. Runs then cuts out: thermal shutdown — check ventilation and load; repeated shutdowns mean a faulty supply.
5. On a machine that is dead but the fuses are good, inspect the transformer pin carrier: a cracked carrier can detach a coil wire and kill all DC output — a documented A3020 fault, repairable by resoldering.

Dead machine with power present

If the rails are good but the machine is dead, the fault is on the logic side of the board. Check the following, escalating from the simple to the scope-level:

• ROMs: confirm the two RISC OS ROMs are seated with no lifted pins and links LK10/LK11/LK12/LK14 correct for the fitted ROM size.
• Battery corrosion: inspect and buzz out the tracks around the battery and under the floppy connector — broken RAM control lines here give the 0x20000 POST fault.
• Clocks: with an oscilloscope, confirm the oscillators are running (see the clock table below). No CPU clock means the machine cannot start.
• Reset: the reset line should sit high in normal running and pulse low at power-on. A reset stuck low halts the ARM250.
• Bus activity: with a healthy clock, reset released and rails present, the ARM250 should drive the address bus (address lines toggling as it fetches from ROM) and the DRAM should show RAS/CAS activity. If the clock, reset and rails are all present but the ARM250 produces no bus activity, the ARM250 itself is the suspect — a whole-device replacement.
• Do not leave the board powered with no clock: loss of DRAM refresh can corrupt or damage the RAM.

A3020 clocks — expected oscilloscope readings

Signal	Frequency	Where
CPU / memory clock	12 MHz	ARM250; derived from the master oscillator
Master oscillator	72 MHz (crystal X4)	Divided down to the CPU, IOC and pixel clocks
IOC reference	8 MHz	I/O timing
VIDC pixel clock	24 / 25.175 / 36 MHz	Selected by monitor sync polarity (crystal X5 = 25.175 MHz)
Floppy controller clock	24 MHz (crystal X7)	82C711
RTC crystal	32.768 kHz (crystal X2)	PCF8583 real-time clock

Clock frequencies are from the TRM circuit diagram.

Video faults

A very common "dead" report is actually a monitor-type mismatch: POST sets a 15 kHz mode, so a VGA or multisync monitor may show an unsynchronised or blank picture at power-on even though the machine is healthy.

Symptom	Probable cause	Action
No picture, machine clearly boots (floppy/sound active)	Monitor-type mismatch, or monitor/cable	Try a 15 kHz/multisync monitor; force the monitor type (hold Delete at power-on to reset CMOS, then reconfigure)
No picture, machine does not boot	Logic/PSU fault	Use the power and dead-machine procedures above
Corrupt or striped display	DRAM fault	Run the memory test; a single failed device gives a repeating pattern
Wrong or shifted geometry only	Monitor adjustment / mode	Not a board fault

The analogue RGB output is 0.7 V peak-to-peak into 75 Ω at the 15-pin video connector; if the machine boots (confirmed by disc and sound activity) but there is no output at the connector, suspect the VIDC section of the ARM250 or the video output components.

Storage faults

• Hard disc not detected: check the 40-pin-equivalent IDE cable and the 4-way drive power; reseat both. On a battery-corroded board, check the IDE tracks near the connector — the drive commonly dies from leakage damage rather than its own failure.
• Drive spins but will not boot / read errors: ageing 2.5-inch drive (stiction or surface faults). Replace it, keeping any replacement within the ~512 MB RISC OS 3.1 usable limit.
• Floppy will not read: clean the head and try a known-good disc before suspecting the controller. If the floppy is completely dead on a corroded board, check the tracks under the connector (which sits over the battery).

Battery, clock and configuration

• Clock wrong / configuration lost / drops to the Supervisor prompt: flat or leaked NiCd cell not holding CMOS. A healthy cell reads roughly 3.4–3.6 V at rest; below about 3.6 V it drains too quickly. Replace the cell (see the Acorn Archimedes A3020 Maintenance Guide).
• "Not enough memory in module area" or random CMOS resets: same cause — a weak cell.
• 0x10000 POST fault (CMOS unreadable): the PCF8583 RTC or its tracks have been destroyed by leakage; repair tracks and replace the chip.

Keyboard, mouse and ports

• Dead keyboard or mouse with a working display: the keyboard interface and quadrature mouse run through the I/O in the ARM250. Try the keyboard/mouse functional test; a dead keyboard points to the keyboard, its cable or the I/O.
• Keyboard gives random or multiple characters after cleaning: the membrane springs are not fully seated in the keycaps.
• Parallel or serial faults: run the external-port tests; the parallel port is a frequent casualty of battery leakage, so check its tracks.
• Expansion fails only with a card fitted: clean the mini-podule or network edge connector and check for bridged contacts.

Boots to the Supervisor prompt instead of the desktop

If the machine starts to a text screen at the Supervisor * prompt (or a BASIC > prompt) instead of loading the RISC OS desktop, the desktop itself is almost certainly fine — the machine has lost its configured startup because the CMOS settings are corrupt or have been cleared.

• To reach the desktop now: type *Desktop and press Return (at the Supervisor prompt the leading * is optional, so Desktop also works). The desktop loads normally for the rest of the session.
• Why it happens: the setting that tells RISC OS to start the desktop at power-on is held in battery-backed CMOS RAM. A flat or leaked backup battery loses or corrupts that setting, so the machine falls back to the Supervisor prompt. It usually appears alongside a wrong clock, lost configuration, or the CMOS POST faults (checksum error, or "CMOS unreadable" on a battery-corroded board).
• Permanent fix: replace the backup battery (see the Acorn Archimedes A3020 Maintenance Guide) so the configuration is retained, then restore a known-good configuration — hold Delete at power-on until the screen clears to reset CMOS to its defaults (which re-enable desktop startup), and set anything machine-specific again with *Configure. With a healthy battery fitted, the machine boots straight to the desktop again.

If *Desktop reaches the desktop but the machine still drops to the Supervisor on every cold start even after a CMOS reset, the battery is not holding charge — fit a new cell.

Common fault catalogue

• Boots to the Supervisor * prompt, not the desktop — corrupt/cleared CMOS from a flat battery; type *Desktop to enter RISC OS, then replace the battery and reset CMOS.
• Completely dead — blown plug fuse or FS1, cracked transformer pin carrier, or a faulty PSU section.
• Boots from Reset key but not from cold — power-on-reset circuit or marginal PSU.
• Random resets / cuts out when warm — PSU thermal shutdown or ageing PSU electrolytics.
• Dead floppy, dead IDE, corroded parallel port — battery leakage; repair tracks, replace the cell.
• RAM control-line failure (0x20000) — broken RAM control track from battery leakage.
• No picture but machine boots — monitor-type mismatch; reset CMOS and reconfigure.
• Corrupt display / crashes — DRAM; run the memory test.
• Clock/config lost — flat or leaked NiCd cell.
• No combination of CPU/memory/video works — the ARM250 itself; whole-device replacement.

Related pages

• Acorn Archimedes A3020
• Acorn Archimedes A3020 Maintenance Guide
• Acorn Archimedes A3020 Capacitor Guide
• Acorn Archimedes A4000 Troubleshooting Guide — three-box sibling
• Acorn A3010, A3020 and A4000 Service Manual · Acorn A3010, A3020 and A4000 Technical Reference Manual