Editing Acorn Archimedes A3000 Troubleshooting Guide

<templatestyles src="Template:StyledTable/styles.css" />

This guide gives component-level diagnosis for the '''[[Acorn Archimedes A3000]]''' (board 0180,000 / 0280,000). The procedures, test points, IC-pin references and signal expectations below are taken from the [[Acorn A3000 Service Manual]] and the [[Acorn A3000 Technical Reference Manual]] circuit diagrams.<ref name="sm">[[Acorn A3000 Service Manual]], Acorn Computers &mdash; hosted on this wiki. Source for the "system failure" procedure (RAS/CAS on R129&ndash;R133, clocks on IC41 pin&nbsp;19 / IC37 pins&nbsp;1&ndash;2, RAS on DRAM pin&nbsp;5, reset on IOC IC13 pin&nbsp;29, RST to IC44 pin&nbsp;44 / IC37 pin&nbsp;9, 24&nbsp;MHz on LK28 / IC47 / X3, data bus on R49&ndash;R65, interrupts on IC37 pins&nbsp;7&ndash;8 and IOC IC13 pins&nbsp;50&ndash;51), the Test ROM memory test, the DRAM physical-address map, the repetitive-reset oscilloscope procedure, and the +5&nbsp;V tolerance.</ref>

== ⚠️ Mains warning ==

The A3000 has a mains-powered PSU inside the case. Disconnect the mains lead before removing the cover. Live measurements on the logic board are referenced to 0&nbsp;V and are low voltage, but the PSU primary side is at mains potential.<ref name="sm" />

== ⚠️ Do not run a dead machine with no clock ==

The Service Manual carries a specific warning: if the computer is powered on for more than a few seconds with '''no system clock''', all four ARM-family chips (ARM2, MEMC1a, VIDC1a, IOC) and the DRAM can be damaged, because the DRAM loses refresh. When chasing a dead machine, confirm the 24&nbsp;MHz clock early and do not leave an unclocked board powered.<ref name="sm" />

[[File:Acorn Archimedes A3000 circuit diagram (1 of 2).png|center|thumb|680px|Acorn A3000 main PCB circuit diagram, sheet 1 of 2 (A3000 Technical Reference Manual). Sheet 2 of 2 is also on the wiki. Use these to locate the test points named below. (Source: The Centre for Computing History.)]]

== Power-up sequence ==

On power-on the PSU brings up +5&nbsp;V; the 24&nbsp;MHz oscillator (X3/IC47) runs; IOC (IC13) releases a delayed reset; the ARM (IC37) fetches RISC OS from the ROMs (IC14&ndash;IC17) through MEMC (IC44); VIDC drives the display; and RISC OS reaches the desktop. A failure at any stage gives a dead machine, no/garbled video, or a hang.

== System failure (dead machine) ==

Work through the Service Manual sequence. Keep power-on time short until the clock is confirmed (see the warning above).

# '''Reset''': look for a delayed reset pulse at power-on on '''IOC IC13 pin 29'''. Check that '''RST''' driving '''IC44 pin 44''' and '''IC37 pin 9''' is not stuck high.
# '''24&nbsp;MHz master clock''': check for 24&nbsp;MHz on '''LK28'''. If absent, check '''IC47 pins 3, 4, 5 and 6''' and replace IC47; finally replace the crystal '''X3'''.
# '''DRAM refresh''': check '''RAS and CAS''' on '''R133 and R129&ndash;R132'''; check for a '''RAS''' signal on '''pin 5 of every DRAM'''.
# '''Processor clocks''': check '''IC41 pin 19''' and '''IC37 pins 1 and 2''' for clock signals; check clocks on '''IC44 pin 67''' and '''IC41 pin 19'''.
# '''Address activity''': hold RESET (press the reset key, or short SKT3 pins 15&ndash;16) and watch the address lines increment on '''IC29 pins 12&ndash;19, IC30 pins 12&ndash;19 and IC31 pins 12&ndash;15''' (A0/A1 do not toggle). If none toggle, check the PHI&nbsp;1 clock at its source '''IC44 pin 66''' and on R135, and that addresses reach those devices; replace IC29/IC30/IC31, or the ARM (IC37) if no addresses appear.
# '''MEMC address lines''': with RESET held, check the address lines on MEMC (IC44) and LA2&ndash;LA6 / LA16&ndash;LA21 on IOC (IC13).
# '''Data bus''': probe R49&ndash;R65, R115&ndash;R128 and R87; none should be stuck high, low or mid-rail. Removing one of these resistors isolates the DRAM bank from the CPU to trace shorts. Also check IC43 pins 12&ndash;19 and IC46 pins 12&ndash;19 on the BDATA bus.
# '''ROMs''': check data and address activity on all four ROMs (IC14&ndash;IC17), especially after a ROM swap &mdash; a mishandled screwdriver can lift legs or damage tracks. MEMC will not enable the ROMs while RESET is held.
# '''Interrupts''': FIQ and IRQ on '''ARM IC37 pins 8 and 7''' must not be stuck low (IRQ pulses low, FIQ sits high). Check the source on '''IOC IC13 pins 50 and 51'''; trace further on IOC interrupt inputs pins 30&ndash;42 (pins 30, 31, 42 are active-high).

<templatestyles src="Template:StyledTable/styles.css" />
{| class="wikitable styled-table" style="width:100%; text-align:left;"
|+'''Dead-machine quick reference'''
! Check !! Test point !! Expected
|-
| Master clock || LK28 (then IC47 pins 3&ndash;6, X3) || 24&nbsp;MHz present
|-
| Reset || IOC IC13 pin 29; IC44 pin 44; IC37 pin 9 || Delayed pulse at power-on; RST not stuck high
|-
| DRAM RAS/CAS || R133, R129&ndash;R132; DRAM pin 5 || RAS/CAS toggling; RAS on every DRAM
|-
| ARM clocks || IC37 pins 1&ndash;2; IC44 pins 66&ndash;67; IC41 pin 19 || Clocks present; PHI 1 at IC44 pin 66
|-
| Address bus (RESET held) || IC29/IC30 pins 12&ndash;19, IC31 pins 12&ndash;15 || Lines increment (not A0/A1)
|-
| Interrupts || IC37 pins 7&ndash;8; IOC IC13 pins 50&ndash;51 || IRQ pulses low, FIQ high; not stuck
|}

== Test ROMs and the memory test ==

For "failure to initialise" (apparently dead) machines, Acorn supplied a set of four '''Test ROMs''' that replace the RISC OS ROM set (IC14&ndash;IC17 → test ROMs 1&ndash;4). Provided the ARM, MEMC and VIDC work, the test ROMs auto-boot a menu-driven memory test.<ref name="sm" />

* The top line shows the repeating sequence '''0123456789'''; every four digits is one 32-bit word. Missing or corrupt digits indicate a fault.
* Memory size is cycled (0.5/1/2/4&nbsp;MB) with the '''M''' key. The 0.5&nbsp;MB test takes 3&ndash;4&nbsp;s; the 4&nbsp;MB test about 29&nbsp;s.
* Errors are reported as: <code>AT ADDRESS &nnnnnnnn WROTE &pppppppp READ xxxxxxxx</code> (data read back shown in binary, LSB rightmost). The test stops after 8&ndash;9 errors; repair the first error first, as errors cascade.
* If video is dead, select option&nbsp;1 to send the test output to the parallel printer port instead of the screen.
* The test ROMs also flag a short-to-0&nbsp;V on the CMOS control lines '''C0/C1''' at power-up.

Because screen memory starts at physical address '''&2000000''', the corruption pattern on screen identifies the failed device:

<templatestyles src="Template:StyledTable/styles.css" />
{| class="wikitable styled-table" style="width:70%; text-align:center;"
|+'''A3000 DRAM physical-address map'''
! Physical address !! DRAM ICs
|-
| &2000000 || IC21, IC20
|-
| &2000001 || IC22, IC23
|-
| &2000002 || IC24, IC25
|-
| &2000003 || IC26, IC27
|-
| (repeats to &20FFFFF) || 1&nbsp;MB bank
|}

== Repetitive-reset test (no display from the test ROMs) ==

When the test-ROM menu will not initialise, the Service Manual's repetitive-reset procedure exercises the machine under a forced reset so the bus can be scoped:<ref name="sm" />

# Connect a square-wave or (preferably) negative-going pulse generator at about '''10&nbsp;kHz''' to the reset line via SKT3 pin 13.
# Set the border colour to white. The IOC output-select signals should then appear in a defined order, verifying IC13 (IOC) decoding &mdash; e.g. '''nB/W''' low at IC44 pin 63 / IC37 pin 84, the SVPMD lines at IC44 pin 52 / IC37 pin 10, and the IOC chip-selects and CO/C1 lines on IC13 pins 56&ndash;67.
# Equipment needed: an oscilloscope and a signal/pulse generator.

== Video faults ==

<templatestyles src="Template:StyledTable/styles.css" />
{| class="wikitable styled-table" style="width:100%; text-align:left;"
! Symptom !! Probable cause !! Action
|-
| No display, machine otherwise alive || VIDC, video DAC or monitor || Confirm RISC OS boots (sound/disc activity); check the analogue RGB at the monitor connector; verify VIDC clocks
|-
| Corrupt/striped display, repeating pattern || A failed DRAM (IC20&ndash;IC27) || Run the test ROMs; the on-screen corruption and the address map identify the device
|-
| No display and dead machine || Clock/reset/CPU || Use the system-failure sequence above; confirm 24&nbsp;MHz first
|}

== Peripheral and I/O faults ==

* '''Floppy''': if a drive will not read, clean the head and try a known-good disc before suspecting the VL1772 controller; check the 34-way interface and the 4-pin power.
* '''Keyboard/mouse''': handled by the Intel 8051 keyboard controller over a reset/acknowledge protocol; a dead keyboard with a working display points to the 8051 or its connection. The mouse is powered from +5&nbsp;V and returns quadrature X/Y (REF/DIR) signals.
* '''Podules''': clean the mini-podule and external podule edge connectors; a machine that fails only with a card fitted usually has a dirty or bridged connector.
* '''Lost configuration / wrong clock''': flat or leaked CMOS battery &mdash; see [[Acorn Archimedes A3000 Maintenance Guide]].

== Common fault catalogue ==

* '''Dead, +5&nbsp;V present''' &mdash; missing 24&nbsp;MHz clock (LK28/IC47/X3) or stuck reset (IC13 pin 29 / IC44 pin 44 / IC37 pin 9). Do not leave powered without a clock.
* '''Won't initialise''' &mdash; ROM socket/track damage (IC14&ndash;IC17) or DRAM; fit the test ROMs.
* '''Corrupt display''' &mdash; a DRAM device; identify via the test-ROM pattern and address map.
* '''Random resets / out-of-spec +5&nbsp;V''' &mdash; PSU module; check ripple, recap or replace.
* '''Lost time/config, corrosion bottom-left''' &mdash; leaked CMOS battery; clean, repair tracks, replace cell.
* '''Expansion only fails with a card''' &mdash; dirty podule edge connector.

== Related pages ==

* [[Acorn Archimedes A3000]]
* [[Acorn Archimedes A3000 Maintenance Guide]]
* [[Acorn Archimedes A3000 Capacitor Guide]]
* [[Acorn A3000 Service Manual]] · [[Acorn A3000 Technical Reference Manual]]
* [[Acorn Archimedes A4000 Troubleshooting Guide]]

== References ==

<references />

{{Navbox-AcornComputers|state=collapsed}}

[[Category:Acorn Computers]]
[[Category:Troubleshooting Guides]]