Psion Series 3a Troubleshooting Guide
Fault-finding for the Psion Series 3a. Most of it applies to the Psion Series 3, Psion Series 3c and Psion Series 3mx as well; model-specific differences are called out.

Psion published a detailed SIBO Hardware Reference describing the architecture, the supply rails and the power states, which makes systematic diagnosis possible on these machines.[1]
⚠️ Before you start
editStep 1: the four usual suspects
edit- Batteries. Fit known-good alkaline AAs and a fresh CR1620. Check the AAs under load, not just open-circuit.
- The battery lead. See below — this is the single most under-diagnosed Series 3 fault.
- The screen ribbon. Vertical lines on the display mean the cable, not the LCD.
- The hinges. A crooked back edge or a button bar that flexes out means a broken hidden hinge.[3]
Step 2: understand the power states
editA SIBO machine has four power states, and knowing which one it is stuck in narrows the fault fast.[1]
| State | What is powered | What it means |
|---|---|---|
| Standby | Main oscillator and processor shut down; ASIC9 and all memory powered; the real-time clock running | The normal "off" state. Data is retained. |
| Backup | As standby, but Vcc1 comes from the backup cell and switching to operating is inhibited | The main cells have failed or are absent. A machine in backup state will refuse to switch on — that is by design, not a fault. |
| Idle | Processor powered, oscillator running, processor clock stopped | Waiting for an interrupt. The bus is quiet apart from screen refresh. |
| Operating | Everything running | Normal use. |
A machine that will not switch on with good-looking batteries is very often sitting in the backup state because the main supply is not actually reaching the board — which brings you straight back to the battery lead.
Step 3: the supply rails
edit| Rail | Function | Symptom when missing or wrong |
|---|---|---|
| Vcc1 | ASIC9 and all memory. Present in every state. 3.3 V nominal on ASIC9 machines | Nothing works, and data is lost. Present-when-off but absent-when-on means the main converter has failed. |
| Vcc2 | The processor and the expansion port. 5 V, off in standby | Machine appears dead but retains data — it never leaves standby. |
| Vcc3 | LCD module and display drivers | Blank display although the machine otherwise runs and beeps. |
| Vee1 | LCD bias — the contrast rail. Negative, from a flyback converter, set by a DAC in software | No usable contrast at any software setting. Note there is no contrast potentiometer on these machines — if the software control does nothing, the rail has failed. |
| Vcc5 | 5 V to the external SSD slots | SSD packs not recognised at all. |
| Vh | +16 V Flash programming voltage to the SSD slots | Flash SSDs read but cannot be written or formatted. Reads do not use this rail. |
That last row is the cleanest diagnostic split in the whole machine: a Flash pack that reads fine but will not format points at the Vh converter, not at the pack.
Symptom table
edit| Symptom | Likely cause | Action |
|---|---|---|
| Eats batteries at a phenomenal rate | The battery lead. The lead between the mainboard and the battery contacts is twisted every time the lid opens; the wires fatigue and short together | Remove the screws behind the batteries and re-seat the two-pin connector — often a permanent cure. If the wires are damaged, replace them.[4] |
| Appears to lose the battery connection intermittently | Same fault, earlier stage | As above.[4] |
| Will not switch on at all | Machine held in the backup state because the main supply is not reaching the board; flat cells; battery lead | Confirm the AAs are good under load, then go to the battery lead. |
| Vertical lines across the display | Deterioration of the screen ribbon cable running between the keyboard half and the screen | Replace the cable, or fit a donor lid. See Psion Series 3a Hinge Repair for the separation procedure.[5] |
| Display blank, machine otherwise alive | Contrast set out of range, or Vcc3/Vee1 failure | Adjust contrast in software first. Contrast is a software-controlled DAC, so there is no wheel to turn.[1] |
| Button bar does nothing | Its ribbon has come adrift from the lower PCB, usually after an impact | Push the ribbon back into its connector with a paperclip, with the lid half closed.[4] |
| Back edge looks crooked when opening; button bar flexes out | Broken hidden hinge nipple | See Psion Series 3a Hinge Repair.[3] |
| Lid loose or detached on one side | Visible hinge sheared at the 8 mm neck, from over-opening | Epoxy repair or donor lid.[3] |
| Backup battery reported flat but measures fine | Contact pressure | A small piece of paper between the cell and the top contact is the documented fix.[4] |
| SSDs not recognised in either slot — "not ready" | On a 3mx: broken SSD door microswitches. The plastic tips snap off, almost always from prising the motherboard out with the SSD doors closed | Re-glue the broken tips, or recreate a missing tab from a scrap of ABS. Both switches must be repaired — fixing one is not enough.[2] |
| SSDs not recognised — 3c | The 3c senses the doors with PCB tracks and a small metal connector on the door hinge, not microswitches | Clean the tracks and the contact.[2] |
| One slot works, the other does not | Slot contacts, or that slot's door sensing | Swap a known-good pack between slots to confirm before opening anything. |
| Flash SSD reads but will not format | The +16 V Vh converter | See the rail table above.[1] |
| RAM SSD has lost its contents | Flat button cell inside the pack | Replace the pack's own cell. |
| Flash SSD reports itself full when it looks empty | Normal | Flash packs do not reclaim space until the whole pack is reformatted.[5] |
| "Memory corruption" after a RAM upgrade | Incompatible SRAM devices | Original Sony parts are specified; at least one attempt with Hitachi 628128 devices produced exactly this message.[6] |
| Machine resets or hangs unpredictably | Watchdog NMI, or a marginal supply | The watchdog counts at 4 Hz and raises an NMI if not reset — about 0.75 s with interrupts disabled. A machine that keeps resetting is often a supply that is sagging under load.[1] |
| Loses the time when the batteries are out | Normal on a Series 3; a fault on a 3a | The original Series 3 relies on the CPU waking periodically, so with no main batteries the time is lost. The 3a has a hardware 32-bit RTC and should keep time regardless.[1] |
| Keys need a hard press | Dirty membrane or pads | Wash the membrane, clean the pads. See Psion Series 3a General Maintenance. |
| A whole row or column of keys dead | Keyboard matrix line | The matrix is scanned by driving each column in turn and reading the rows, so a dead row is one line and a dead column is another.[1] |
| Sticky, glue-like case | 3c only — the rubberised soft-touch coating degrading | Strip with high-purity isopropyl alcohol. See Psion Series 3a General Maintenance. |
| Corrosion, green tracks | Leaked AA cell | Neutralise and clean. See Battery Explosion, Capacitor or Corrosion Damage. |
What is realistically repairable
edit| Part | Repairable? | Notes |
|---|---|---|
| Battery lead | Yes | Re-seat or rewire. The highest-value repair on the machine. |
| Hinges | Yes | Epoxy, with or without a steel reinforcing strip.[3] |
| Button bar ribbon | Yes | Often just needs reseating.[4] |
| Screen ribbon | Yes, with a donor lid | The cable is the failure, not the panel. |
| SSD door microswitches (3mx) | Yes | Superglue the broken tips; recreate a missing tab from ABS.[2] |
| Keyboard membrane | Yes | Washable. |
| SRAM / ROM | Yes, for SMD-capable hands | Spare SRAM pads exist on 256 KB boards.[6] |
| ASIC9 | No | A Psion custom device. A failed ASIC9 means a donor board. |
Related pages
edit- Psion Series 3a · Psion Series 3a General Maintenance · Psion Series 3a Hinge Repair
- Psion Series 3 Troubleshooting Guide · Psion Series 3c Troubleshooting Guide · Psion Series 3mx Troubleshooting Guide · Psion Siena Troubleshooting Guide
- Battery Explosion, Capacitor or Corrosion Damage · Recommended Tools
References
edit- ↑ 1.0 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 Psion PLC, SIBO Hardware Reference, version 2.10, 3 February 1995, in the Psion Documentation Project. Source for the SIBO architecture, ASIC1/ASIC2/ASIC9, memory decoding, interrupt allocation, the watchdog, the NMI sources, the four power states and every supply rail described here.
- ↑ 2.0 2.1 2.2 2.3 Richard Zealley, "Psion Series 3mx SSD door microswitch repair", Retro Coding, 25 May 2020. Source for the SSD door microswitch failure, its cause, the symptom, and the fact that both switches must be repaired.
- ↑ 3.0 3.1 3.2 3.3 David MacKay, "Psion 3a Hinge breakages". Source for the hinge locations, failure modes and symptoms.
- ↑ 4.0 4.1 4.2 4.3 4.4 Psion Series 3/3a community FAQ material and the comp.sys.psion newsgroup archives. Source for the "dreaded battery lead problem" and its fix, the backup-cell contact shim, and the button-bar ribbon reseating trick.
- ↑ 5.0 5.1 "Psion Series 3", Wikipedia. Source for the three documented clamshell problems: hinge breakage, button-bar failure and screen-cable deterioration producing vertical lines.
- ↑ 6.0 6.1 David Kimberlin-Wyer, "Psion Series 3a Memory Upgrade", primrosebank.net. Source for the disassembly procedure, the board layout, the spare SRAM pads and the incompatible-memory result.