Jump to content

Acorn Archimedes A3010 Troubleshooting Guide

From RetroTechCollection

This guide gives fault diagnosis for the Acorn Archimedes A3010 (ARM250 board, part 0195,200; early "Adelaide" board 0194,000). It follows the module-level fault-finding of the Service Manual, with test points and values from the Technical Reference Manual circuit diagrams. The ARM250 integrates the ARM2 core, MEMC1a, VIDC1a and IOC into one package, so processor-family faults are isolated to that single device.[1][2]

โš ๏ธ Mains warning

[edit | edit source]

The A3010'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 power (left-hand) end of the board. Re-apply the earth-continuity and DC-insulation (class 1) safety tests after any internal work.[1]

Acorn A3010 "Adelaide" main PCB circuit diagram, sheet 1 of 3 (drawing 0194,000/C), showing the ARM250, RISC OS ROMs, PCF8583 battery-backed RAM/RTC, the two joystick ports, the DRAM-size link table and the filtered +5 V supply. Sheets 2 and 3 are also on the wiki.

Normal power-on sequence

[edit | edit source]

A healthy A3010 runs a short power-on self test (POST) before booting RISC OS, stepping the screen border through a colour sequence and flashing any fault on the floppy-drive LED.[1] The sequence tells you how far the machine gets:

  1. Border purple: brief ROM and RAM test, IOC and VIDC initialised.
  2. Border blue: extended memory test (up to about 12 s, depending on RAM).
  3. Border purple: ARM3-present test (a standard A3010 has none).
  4. Black with a memory-size message, then RISC OS boots.

A soft reset (the Reset key) skips POST and starts RISC OS immediately; a power-on reset runs the full POST. A machine that boots from the Reset key but not from cold points to the power-on-reset circuit or a marginal PSU.[1]

On a VGA or multisync monitor the picture may be unsynchronised during POST because the test uses a 15 kHz mode; this alone is not a fault.[1]

POST fault codes (floppy-LED flash decode)

[edit | edit source]

A POST failure is flashed on the floppy LED as a hex value (eight groups of four flashes, long = 1, short = 0). The documented fault and status bits are:[3]

RISC OS POST fault bits
Bit Meaning Likely cause on an A3010
0x00000200 ROM checksum failure Corrupt or badly seated RISC OS ROM; check links LK10/LK11/LK12/LK14
0x00000400 MEMC CAM mapping fault ARM250 memory controller or DRAM
0x00000800 MEMC protection fault As above
0x00001000 IOC register test failure I/O section of the ARM250
0x00004000 VIDC (Virq) timing failure Video timing / clock
0x00008000 Sound (Sirq) timing failure Sound/interrupt timing
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
0x00040000 Long RAM test failure Faulty DRAM device

To clear a CMOS fault, hold Delete at power-on until a black screen with a red border appears, resetting the configuration to defaults.[3]

Power and PSU diagnosis

[edit | edit source]

Most "dead A3010" faults are in the on-board PSU or its feed. Diagnose power first, with the mains lead disconnected except where a live reading is required.[1]

A3010 supply rails and test points
Rail / point Nominal Feeds / note
+5 V 5.0 V (accept โ‰ˆ4.9โ€“5.1 V) Logic (ARM250, RAM, ROM, I/O) and floppy
+12 V 12.0 V (accept โ‰ˆ11.7โ€“12.3 V) Video connector SK1 pin 12 (SCART switching)
Floppy power +5 V and 0 V Drive feed
PSU primary โ‰ˆ4 ฮฉ continuity Open circuit = blown transformer thermal fuse

The tolerances are typical electronics practice; Acorn's module-level manual does not publish tight rail tolerances.[1]

Procedure:

  1. Completely dead (no green power LED, no Caps-Lock LED): check the mains-plug fuse, then internal fuse FS1. A new fuse that blows on switch-on means the PSU section is faulty.
  2. Measure PSU primary continuity unplugged โ€” about 4 ฮฉ is healthy; open circuit means the transformer thermal fuse has blown.[1]
  3. With power applied, measure +5 V and +12 V. A missing or low rail points to the PSU section.
  4. Slow or unreliable power-on / out-of-spec 5 V: ageing PSU electrolytics โ€” recap the PSU section. The fault worsens under added load (for example a fitted memory upgrade).[4]
  5. Runs then cuts out: thermal shutdown โ€” check ventilation and load.

Dead machine with power present

[edit | edit source]

If the rails are good but the machine is dead, work the logic side of the board:[1][5]

  • ROMs: confirm the two RISC OS ROMs are seated with no lifted pins and their size links correct.
  • Battery corrosion: inspect and buzz out the tracks around the battery and RTC โ€” broken RAM control lines here give the 0x20000 POST fault.
  • Clocks: with an oscilloscope, confirm the oscillators are running (see the clock table).
  • Reset: the reset line should idle high and pulse low at power-on; a reset stuck low halts the ARM250.
  • Bus activity: with clock, reset and rails present, the ARM250 should drive the address bus and the DRAM should show RAS/CAS activity. No bus activity with all of those present makes the ARM250 itself the suspect โ€” a whole-device replacement.
  • Do not leave the board powered with no clock; loss of DRAM refresh can damage the RAM.
A3010 clocks โ€” expected oscilloscope readings
Signal Frequency Where
CPU / memory clock 12 MHz ARM250
VIDC pixel clock 24 / 25.175 / 36 MHz Selected by monitor sync polarity (crystal X5 = 25.175 MHz)
IOC reference 8 MHz I/O timing
Floppy controller clock 24 MHz (crystal X7) 82C711-type controller
RTC crystal 32.768 kHz (crystal X2) PCF8583 real-time clock

Clock frequencies are from the TRM circuit diagram.[2]

Video faults

[edit | edit source]
Symptom Probable cause Action
No RGB picture, machine boots (floppy/sound active) Monitor-type mismatch, monitor or cable Try a 15 kHz/multisync monitor; reset CMOS (hold Delete at power-on) and reconfigure the monitor type
No picture, machine does not boot Logic or PSU fault Use the power and dead-machine procedures
No TV picture but RGB works UHF modulator Run the UHF-modulator test; check the modulator and its feed
Corrupt or striped display DRAM fault Run the memory test; a single failed device gives a repeating pattern

The analogue RGB output is 0.7 V peak-to-peak into 75 ฮฉ at the 15-pin connector; if the machine boots but there is no output there, suspect the VIDC section of the ARM250 or the video output components.[2]

Sound faults

[edit | edit source]

Crackle, pops or intermittent sound are usually the audio hybrid card beside the mini-podule slot: its pins fracture with age (worsened by any nearby battery corrosion), and its surface-mount electrolytic capacitors degrade.[6] Resolder or reseat the hybrid's connector first; if the crackle remains, recap the hybrid's SMD electrolytics (see the Acorn Archimedes A3010 Capacitor Guide). A faint power-on pop is normal on some units.[6]

Storage faults

[edit | edit source]

The A3010 has no internal hard disc, so storage faults are floppy-only:

  • Floppy will not read: clean the head and try a known-good disc before suspecting the controller; run the disc functional test.
  • Floppy completely dead on a corroded board: check the tracks around the battery and the floppy connector.[5]

Joysticks, keyboard, mouse and ports

[edit | edit source]
  • Joystick not read: the two 9-pin ports are unique to the A3010 and are read through the I/O; run the joystick functional test and check the port and its buffer (the schematic shows the ports feeding line-buffer logic).[2]
  • Dead keyboard or mouse with a working display: the keyboard interface and quadrature mouse run through the I/O in the ARM250; run the keyboard/mouse test.
  • Keyboard gives random characters after cleaning: the membrane springs are not fully seated in the keycaps.
  • Parallel or serial faults: run the external-port tests; check the connector and the I/O.
  • Expansion fails only with a card fitted: clean the mini-podule edge connector and check for bridged contacts.

Boots to the Supervisor prompt instead of the desktop

[edit | edit source]

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 A3010 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

[edit | edit source]
  • 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 โ€” blown plug fuse or FS1, or a faulty PSU section.
  • Slow/unreliable power-on, out-of-spec 5 V โ€” ageing PSU electrolytics.
  • Boots from Reset key but not from cold โ€” power-on-reset circuit or marginal PSU.
  • Crackle / intermittent sound โ€” audio hybrid card (pins or SMD electrolytics).
  • No RGB but machine boots โ€” monitor-type mismatch; reset CMOS and reconfigure.
  • No TV picture, RGB works โ€” UHF modulator.
  • RAM control-line failure (0x20000) โ€” broken RAM control track from battery leakage.
  • Clock/config lost โ€” flat or leaked NiCd cell.
  • Corrupt display / crashes โ€” DRAM; run the memory test.
  • No combination of CPU/memory/video works โ€” the ARM250 itself; whole-device replacement.
[edit | edit source]

References

[edit | edit source]
  1. โ†‘ 1.0 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 Acorn A3010, A3020 and A4000 Service Manual, Acorn Computers Ltd โ€” hosted on this wiki. Dead-machine procedure, PSU-fail and main-PCB-fail routines, the 5 A mains fuse and internal fuse FS1, thermal shutdown, +5 V/+12 V rails, drive-power links LK1/LK2, ROM/DRAM links, reset behaviour and the safety tests.
  2. โ†‘ 2.0 2.1 2.2 2.3 Acorn A3010, A3020 and A4000 Technical Reference Manual, Acorn Computers Ltd โ€” hosted on this wiki. Main-PCB ("Adelaide") circuit diagrams: crystals, clocks, PCF8583 RTC, joystick ports, reset circuit and per-designator values.
  3. โ†‘ 3.0 3.1 "Common RISC OS POST failure errors", Retro-Kit.
  4. โ†‘ "A3010 PSU electrolytics / slow to power on", Stardot forums.
  5. โ†‘ 5.0 5.1 "Dead A3010", Stardot forums.
  6. โ†‘ 6.0 6.1 "A3010/A4000 audio hybrid crackle", Stardot forums.