Jump to content

Atari ST Troubleshooting Guide: Difference between revisions

From RetroTechCollection

Josh (talk | contribs)

5 user groups
6,663 edits
undo
VisualWikitext
Created shared troubleshooting guide covering 520ST and 1040ST
ย 
No edit summary
ย 
(3 intermediate revisions by the same user not shown)
Line 1: Line 1:
{{Infobox guide
This guide covers fault diagnosis and repair for the '''Atari [[Atari 520ST|520ST]]''', '''[[Atari 520STM|520STM]]''', '''[[Atari 520STF|520STF]]''', '''[[Atari 520STFM|520STFM]]''', '''[[Atari 1040ST|1040ST]]''', '''[[Atari 1040STF|1040STF]]''', and '''[[Atari 1040STFM|1040STFM]]''' computers. These machines share common motherboard architectures (including the C070789-001 used in both 520STFM and 1040STFM) and the same custom IC set, so troubleshooting procedures are largely identical. Differences between the 520ST and 1040ST (principally RAM population and PSU type) are noted where relevant.
|title=Atari ST Troubleshooting Guide
|image=
|applies_to=[[Atari 520ST]], [[Atari 1040ST]]
|skill_level=Intermediate to Advanced
|tools_required=Multimeter, oscilloscope (recommended), logic probe, soldering equipment, spare ICs for swap-testing
}}


This guide covers fault diagnosis and repair for the '''[[Atari 520ST]]''' and '''[[Atari 1040ST]]''' computers, including the STF and STFM variants. The 520ST and 1040ST share the same custom chip set (GLUE, MMU, Shifter, DMA), the same support ICs (MC68000, MC68901, MC6850 ACIAs, WD1772, YM2149F), and in many cases the same PCB (the C070789-001 board was used for both 520STFM and 1040STFM). Troubleshooting procedures are therefore identical for both models, with the exception of RAM faults, where the 520ST has eight DRAM chips and the 1040ST has sixteen.
For the Atari STE range ([[Atari 520STE|520STE]] / [[Atari 1040STE|1040STE]]), see [[Atari STE Troubleshooting Guide]].


For general preventive maintenance procedures, see [[Atari ST General Maintenance]].
== Prerequisites ==


== Before You Begin ==
Before beginning troubleshooting:
* A digital multimeter (DMM) capable of DC voltage, continuity, and resistance measurement is essential.
* An oscilloscope is helpful for diagnosing clock and timing issues but is not strictly required for most common faults.
* An Atari diagnostic cartridge is a valuable tool โ€” it provides hardware-level testing via the cartridge port and can output diagnostics over the serial port even when the display is not functioning.<ref name="diagcart">{{cite web |url=https://www.atari-wiki.com/index.php?title=Basic_Troubleshooting_-_Diagnostic_Cartridge_basics |title=Basic Troubleshooting - Diagnostic Cartridge Basics |publisher=Atari Wiki |access-date=2026-03-27}}</ref>
* A known-good external floppy drive, spare TOS ROMs, and spare RAM chips are useful for substitution testing.


# '''Check the PSU first.''' The majority of Atari ST faults trace back to a failing or out-of-spec power supply. See [[Atari ST General Maintenance#Power Supply Maintenance]] for voltage testing procedures. Do not proceed with motherboard diagnosis until you have confirmed +5 V is within 4.75โ€“5.25 V under load.
== Boot Sequence Overview ==
# '''Re-seat all socketed ICs''' โ€” CPU, GLUE, MMU, Shifter, DMA, ROMs, and DRAM (if socketed). Oxidised contacts in chip sockets are an extremely common cause of apparent hardware failure on 35+ year-old machines.
# '''Inspect for obvious damage''' โ€” cracked traces, burned components, electrolyte leakage from capacitors, corrosion around ICs.


== Bomb Error Codes ==
Understanding the ST boot sequence helps pinpoint where a failure occurs. This sequence is the same for the 520ST and 1040ST:


When the Motorola 68000 encounters a hardware exception that TOS cannot handle, it displays a row of small bomb icons on screen. The number of bombs corresponds to the 68000 exception vector number and indicates the type of fault:<ref>{{cite web |url=https://www.atari-forum.com/viewtopic.php?t=15500 |title=TOS ERROR # and Bombs: What they mean |publisher=Atari-Forum |access-date=2026-03-27}}</ref>
# '''Power-on:''' PSU provides +5V and +12V rails. The 68000 is held in reset by the GLUE chip until voltages stabilise.
# '''Reset release:''' GLUE releases the 68000 RESET line. The CPU reads the initial SSP (Supervisor Stack Pointer) from address $000000 and the initial PC (Program Counter) from address $000004, which points into TOS ROM space ($FC0000โ€“$FEFFFF).
# '''TOS initialisation:''' TOS performs a RAM sizing/test routine. On TOS 1.00โ€“1.04, this is a brief test. The screen remains blank (white or black depending on the monitor type) during this phase.
# '''GEM desktop:''' If all tests pass and no bootable floppy is detected, the GEM desktop appears. If a bootable floppy is present, the boot sector is loaded and executed.
ย 
'''Failure at each stage produces characteristic symptoms:'''
* No power LED โ†’ PSU failure or blown fuse
* Power LED on but no display โ†’ CPU not running, clock failure, or ROM space fault
* White screen (colour monitor) โ†’ CPU running but stuck in early TOS code; likely RAM fault, ROM fault, or MMU/GLUE failure
* Green screen with corrupted icons โ†’ Partial boot; likely bad RAM chip, video fault, or corrupted TOS
* Desktop appears but crashes โ†’ PSU instability, intermittent RAM fault, or software issue
ย 
== Power Supply Diagnosis ==
ย 
The PSU is the most common source of problems in ageing ST machines. The 1040ST has an internal switched-mode PSU; the 520ST/520STM uses an external supply.
ย 
=== Voltage Rails ===


<templatestyles src="Template:StyledTable/styles.css" />
<templatestyles src="Template:StyledTable/styles.css" />
{| class="wikitable styled-table"
{| class="wikitable styled-table"
! Bombs !! Exception !! Typical Hardware Cause
! Rail !! Wire Colour (1040ST internal header) !! Nominal Voltage !! Acceptable Range !! Notes
|-
| 2 || Bus Error || Failed or missing hardware on the bus. RAM fault, ROM fault, bad custom chip (GLUE/MMU not decoding addresses correctly), or failed peripheral IC. Also triggered if attempting to access unmapped address space.
|-
| 3 || Address Error || 68000 attempted a word or long-word access at an odd address. Usually a software bug, but can indicate RAM corruption (bit flip changing a pointer value) or a failing MMU generating incorrect addresses.
|-
| 4 || Illegal Instruction || CPU fetched an invalid opcode. ROM corruption, RAM fault in program code area, or a failing CPU. Can also occur with incorrect TOS ROM installed.
|-
|-
| 5 || Zero Divide || Division by zero. Almost always a software issue, but can be triggered by RAM corruption changing a divisor value.
| +5V || Red || 5.00V || 4.75V โ€“ 5.25V || Primary logic supply; most critical rail. Below 4.75V causes instability; above 5.5V risks IC damage.
|-
|-
| 6 || CHK Instruction || CHK instruction range check failed. Typically software.
| +12V || Blue || 12.0V || 11.0V โ€“ 13.0V || Floppy drive motor, RS-232 transceivers. The machine will operate without it but floppy and serial will not function.
|-
|-
| 7 || TRAPV || TRAPV overflow trap. Typically software.
| GND || Black || 0V || โ€” || Common ground reference.
|-
| 8 || Privilege Violation || User-mode code attempted a supervisor-only instruction. Software issue or RAM corruption.
|-
| 9 || Trace || Trace exception (single-step). Should not occur during normal operation.
|-
| 10 || Line A Emulator || Unimplemented Line A opcode. Software or ROM issue.
|-
| 11 || Line F Emulator || Unimplemented Line F opcode. Software or ROM issue.
|}
|}


'''Key point:''' 2 bombs (Bus Error) and 4 bombs (Illegal Instruction) on boot are the most common indicators of hardware faults. If bombs appear immediately at power-on, the fault is almost certainly hardware โ€” ROM, RAM, or a custom chip.
On the 520ST with external PSU, measure at the 7-pin DIN power connector: +5V (pins 1, 2), +12V (pin 6), GND (pins 3, 4, 5). Some external supplies also provide โˆ’12V (pin 7) for the RS-232 transceivers.


== Power Supply Faults ==
'''Testing procedure:'''
# '''Do NOT test a switched-mode PSU disconnected from the motherboard''' (or a dummy load). Without a load, output voltages may be dangerously high or the PSU may not start at all.<ref name="exxospsu">{{cite web |url=https://www.exxosforum.co.uk/atari/last/psu/index.htm |title=The LaST Upgrade - Atari PSU Repair |publisher=exxos |access-date=2026-03-27}}</ref>
# With the machine assembled and powered on, measure the +5V rail at the motherboard power header (red wires to black wires).
# If the voltage is below 4.9V under load, the PSU capacitors likely need replacement.
# Check for AC ripple on the +5V rail if you have an oscilloscope. A healthy recapped PSU should show less than 200mV of ripple. More than 500mV indicates failing capacitors and will cause instability.<ref name="exxospsu"/>
ย 
=== Common PSU Symptoms (Both 520ST and 1040ST) ===


<templatestyles src="Template:StyledTable/styles.css" />
<templatestyles src="Template:StyledTable/styles.css" />
Line 54: Line 58:
! Symptom !! Possible Cause !! Recommended Action
! Symptom !! Possible Cause !! Recommended Action
|-
|-
| Machine completely dead, no power LED || Blown mains fuse in power cable or IEC socket; broken power switch; failed internal PSU fuse; dead PSU || Check mains fuse. Test continuity through power switch (on/off positions). Check internal PSU fuse. If all fuses are good and switch has continuity, the PSU main switching circuit has likely failed โ€” recap or replace the PSU.
| No power at all (no LED) || Blown mains fuse in cable or PSU; failed power switch (1040ST); open fusible resistor in PSU; failed external PSU brick (520ST) || Check mains fuse. On 1040ST: check PSU internal fuse and continuity through power switch. On 520ST: test external PSU output voltages with DMM.
|-
|-
| Power LED on but no video, no drive activity || +5 V rail absent or severely low; +12 V present (powers LED circuit on some models) || Measure +5 V at the motherboard power connector. If absent or below 4.0 V, the PSU is faulty. Recap or replace.
| Power LED flickers || PSU attempting to start but shutting down (overcurrent protection) || Likely shorted component on motherboard or PSU. Disconnect motherboard from PSU and test PSU with dummy load (1040ST) or try a different external PSU (520ST).
|-
|-
| Machine boots but crashes randomly || +5 V rail sagging under load, excessive ripple || Measure +5 V with an oscilloscope if possible. Ripple above 200 mV peak-to-peak indicates failing PSU capacitors. Recap the PSU. See [[Atari ST Capacitor Replacement Guide]].
| Random crashes during use || Low or noisy +5V rail || Measure +5V under load. Recap PSU if voltage below 4.9V or ripple exceeds 300mV.
|-
|-
| +5 V reads above 5.5 V || PSU over-voltage, typically from failed feedback components or dried-out capacitors || '''Do not connect to motherboard.''' Recap the PSU. If voltage remains high after recapping, check the feedback optocoupler and voltage divider resistors in the PSU. Some PSUs have an adjustment pot โ€” adjust with a dummy load only.
| Screen dims when floppy drive activates || Poor peak power delivery on +5V rail (1040ST internal PSU) || Recap PSU. On Mitsumi SR98 PSUs, the biasing resistors R201/R202 may need adjustment โ€” replace R202 with 1Kฮฉ and R201 with 1.2Kฮฉ, then readjust the voltage trimmer to 5.00V.<ref name="exxospsu"/>
|-
|-
| +12 V reads very low (<9 V) or very high (>15 V) || Transformer winding ratio issues or failed series diode || On some PSUs a series diode on the +12 V line drops ~1 V. Check for a failed diode (open or shorted). If the transformer secondary is damaged, the PSU needs replacement.
| Audio noise/hum || Ripple on +5V rail coupling into YM2149 audio output || Recap PSU. Check ground connections.
|-
| Intermittent floppy read errors || Unstable +12V rail (floppy motor speed variation) or noisy +5V (logic errors) || Measure both rails under load. Recap PSU.
|}
|}


== Boot Sequence Failures ==
== Display / Video Troubleshooting ==


The normal 520ST/1040ST boot sequence is:
The video subsystem (Shifter, GLUE, and related circuitry) is identical across the 520ST and 1040ST.


# Power-on reset (GLUE generates /RESET signal)
=== No Display at All ===
# CPU begins executing from ROM (vector table at $000000 points to TOS entry point in ROM space $FC0000โ€“$FEFFFF)
# TOS performs basic hardware initialisation (RAM test on TOS 2.06 only; earlier TOS versions do not test RAM at boot)
# GEM desktop loads and displays the green (or grey, depending on TOS version) desktop with drive icons


<templatestyles src="Template:StyledTable/styles.css" />
<templatestyles src="Template:StyledTable/styles.css" />
Line 78: Line 81:
! Symptom !! Possible Cause !! Recommended Action
! Symptom !! Possible Cause !! Recommended Action
|-
|-
| No video output at all (blank screen, no sync signal on monitor) || GLUE not generating clock/sync signals; dead CPU; dead MMU; failed Shifter; no +5 V; ROM completely dead || Confirm +5 V is present. Check for 8 MHz clock on CPU pin 15 (CLK). If no clock, suspect GLUE (C025915) or the 32 MHz oscillator (which is divided to produce the 8 MHz CPU clock and 2 MHz bus clock). If clock is present but CPU /HALT is asserted (pin 17 active low), the CPU has taken a double bus fault โ€” suspect ROM or MMU.
| No display, no power LED || PSU failure || See Power Supply Diagnosis section above.
|-
|-
| White or grey screen, no desktop || CPU is running but display initialisation has failed. Typically ROM, RAM, or Shifter fault. || Re-seat ROMs. Swap-test ROMs with known-good set. Check DRAM โ€” a failed DRAM chip can prevent TOS from initialising the display. Swap-test the Shifter (C025914) if available.
| Power LED on, completely blank screen || CPU not running; no clock signal; or monitor cable/connection fault || Verify monitor cable is connected and monitor is set to correct input. Check 32 MHz crystal oscillator with oscilloscope. Reseat CPU, Shifter, GLUE.
|-
|-
| Screen shows "garbage" or random characters at boot || Shifter malfunction; RAM fault causing corrupted video data; MMU generating incorrect video DMA addresses || Swap-test Shifter. Run RAM diagnostics (use a diagnostic cartridge or boot from a diagnostic floppy). Check MMU (C025912).
| Power LED on, white screen only (colour monitor) || CPU running but failing during early TOS initialisation. Common causes: RAM fault, TOS ROM fault, MMU failure, GLUE failure. || Reseat all socketed ICs. Try known-good TOS ROMs. Try with diagnostic cartridge โ€” if cartridge menu appears, motherboard is partially functional and fault is likely in RAM or TOS ROMs. If no cartridge output, fault is in ROM address space decoding (GLUE/MMU).<ref name="diagcart"/>
|-
|-
| 2 bombs immediately on boot || Bus error during TOS initialisation โ€” hardware not responding at an expected address || Re-seat all socketed ICs. Check ROM sockets (corroded contacts cause bus errors when ROM addresses fail). Check DRAM. On the C070789 board, verify that the RAM configuration jumpers (if any) match the installed RAM (8 chips for 520ST, 16 chips for 1040ST).
| Power LED on, white screen only (monochrome monitor) || As above, plus: mono detect pin not being grounded (faulty cable/connector) causing Shifter to output colour mode to mono monitor || Check pin 4 on the 13-pin DIN connector. It should be grounded for monochrome mode.
|-
|-
| 4 bombs immediately on boot || Illegal instruction โ€” CPU executing corrupt data from ROM or RAM || Verify correct TOS ROM set is installed (matched pair, correct HI/LO orientation). Re-seat ROMs. If ROMs are correct, suspect RAM corruption in the low memory area (vector table region $000000โ€“$000400).
| Display appears briefly then goes blank || PSU voltage droop under full load; thermal fault in Shifter or GLUE || Measure +5V rail under load. If voltage is stable, suspect thermal failure in Shifter โ€” use freeze spray to isolate.
|-
| Machine boots to desktop but freezes within seconds || RAM fault (intermittent); PSU voltage marginal; thermal fault in custom chip || Run RAM diagnostics. Check +5 V under load. If the fault appears after warm-up, try cooling individual chips with freeze spray to isolate a thermal intermittent (common with GLUE and MMU).
|}
|}


== Video Troubleshooting ==
=== Display Corruption ===
ย 
The 520ST and 1040ST output video through the 13-pin DIN monitor connector and (on STFM models) through an RF modulator. The Shifter chip (C025914) generates the video signal, the GLUE chip (C025915) generates sync timing, and the MMU (C025912) generates the video DMA addresses to fetch pixel data from RAM.
ย 
=== Monitor Detection ===
ย 
The ST detects whether a colour or monochrome monitor is connected via pin 4 of the 13-pin DIN connector (active low = monochrome). The SM124 monochrome monitor grounds this pin; colour monitors (SC1224, SC1435, or a SCART-adapted TV) leave it floating or high.
ย 
If the wrong monitor type is detected, the ST will boot in the wrong resolution and the display will be unviewable. This is a common issue with homebrew monitor cables that have incorrect wiring.


<templatestyles src="Template:StyledTable/styles.css" />
<templatestyles src="Template:StyledTable/styles.css" />
Line 105: Line 98:
! Symptom !! Possible Cause !! Recommended Action
! Symptom !! Possible Cause !! Recommended Action
|-
|-
| No picture on colour monitor, but SM124 mono works fine || Colour video signals (R, G, B on DIN pins 7, 8, 10) not reaching monitor; Shifter analogue output failure; bad cable || Test with a known-good cable. Measure RGB signals at the DIN connector with an oscilloscope (should see ~0.7 V peak-to-peak analogue video). If signals are present at the DIN but not at the monitor, the cable is faulty.
| Vertical or horizontal bars/stripes || Shifter failure; bad connection between Shifter and data bus; failing RAM affecting video memory area || Reseat Shifter chip. Check Shifter retaining clips for adequate tension. If problem persists, try known-good Shifter from donor board.
|-
| No picture on SM124 mono monitor || Monochrome output (DIN pin 1) dead; mono detect pin (4) not grounded || Check that pin 4 is grounded by the monitor cable. Measure mono video signal at DIN pin 1. If absent, suspect Shifter.
|-
|-
| Picture rolls vertically || Incorrect sync signal; wrong vertical frequency for the monitor || Verify the monitor matches the region (PAL 50 Hz or NTSC 60 Hz for colour; ~71.2 Hz for mono). If using a modern TV or capture device, confirm it supports the ST's non-standard sync rates.
| Wrong or garbled colours || Shifter output fault; bad solder joint on RGB output path; monitor cable fault || Check 13-pin DIN cable for continuity on R, G, B pins (6, 7, 10). Check monitor connector for bent pins. Reseat Shifter.
|-
|-
| Colour fringing, blurry picture on RF output (STFM) || RF modulator failing or mistuned || The RF modulator has a tuning coil โ€” adjustment may improve the picture. However, RF output quality on the ST was never high. Using the RGB monitor output is strongly recommended.
| Screen jitter or rolling || Sync signal fault; HSync or VSync not being generated correctly by GLUE || Check HSync (pin 9) and VSync (pin 12) on monitor port with oscilloscope. If absent, suspect GLUE chip. Check 32 MHz crystal.
|-
|-
| Distorted or shifted display, wrong colours || Shifter fault; video RAM corruption || Swap-test Shifter. Run RAM diagnostics. Check for cold solder joints on the Shifter socket or DIN connector.
| Ghosting or wavy lines || PSU ripple; poor grounding; RF interference from internal PSU (1040ST) || Recap PSU. Check ground connections.
|-
|-
| Every other 16-pixel group is blank or garbled || MMU address generation fault โ€” video DMA is reading incorrect memory locations || Swap-test MMU (C025912). This specific symptom (16-pixel-wide columns of missing data) is characteristic of an MMU fault because the MMU generates the word-aligned video DMA addresses.
| "Jail bars" (thin vertical lines on solid colours) || Impedance issues on the video output path; common on STFM boards || Sometimes improved by replacing decoupling capacitors near the Shifter. A known characteristic of the ST video circuit rather than a definitive fault.
|}
|}


== RAM Fault Diagnosis ==
=== RF Modulator Issues (FM Models โ€” 520STFM / 1040STFM) ===
ย 
The 520ST uses eight 41256 (256Kร—1) DRAM chips providing 512 KB. The 1040ST uses sixteen of the same chips providing 1024 KB. On some board revisions, higher-density chips (1Mร—1 or 256Kร—4) are used in smaller quantities.
ย 
=== Symptoms of RAM Failure ===


<templatestyles src="Template:StyledTable/styles.css" />
<templatestyles src="Template:StyledTable/styles.css" />
Line 128: Line 115:
! Symptom !! Possible Cause !! Recommended Action
! Symptom !! Possible Cause !! Recommended Action
|-
|-
| 2 bombs on boot || One or more DRAM chips returning incorrect data during TOS initialisation || Run a RAM diagnostic (e.g., from a diagnostic cartridge โ€” the cartridge ROM executes without needing working RAM in the main bank). Note which addresses fail; cross-reference with the board layout to identify the faulty chip.
| No RF output, RGB works fine || RF modulator failure; modulator not connected || Check modulator solder connections. A composite video mod bypassing the modulator is a common upgrade.
|-
|-
| Machine detects less RAM than expected (e.g., 1040ST shows 512 KB in system info) || Failed DRAM chips in the upper bank; MMU configuration issue; solder bridge or trace break || Check DRAM chip continuity. On unified boards (C070789), verify that all 16 DRAM positions are populated for 1 MB, or 8 for 512 KB.
| RF output is snowy/noisy || Modulator mistuned; weak signal || Adjust tuning slug on modulator. Clean modulator connections.
|-
|-
| Random crashes during use, especially in memory-intensive applications || Intermittent DRAM fault; marginal cell that passes at cold but fails when warm || Run extended RAM diagnostics that cycle for several hours. Use freeze spray to isolate the failing chip โ€” spray one chip at a time while the test is running to see if the errors shift or stop.
| RF output with herringbone pattern || Ground loop or interference || Ensure proper grounding. Consider composite/RGB output instead.
|-
| Corrupted graphics (random pixels, flickering areas) || Video RAM area corrupted โ€” the ST uses main RAM for video (screen memory typically at top of RAM) || Same as above. The specific DRAM chips responsible for the video area can be determined from the board layout and the screen memory base address.
|}
|}


== Custom Chip Failures ==
== RAM Fault Diagnosis ==
ย 
RAM faults are a common cause of white screens, crashes, and corrupted displays on both the 520ST and 1040ST.
ย 
=== RAM Architecture ===


The four custom Atari ASICs are the most difficult components to replace (no longer manufactured). Diagnosis is typically done by swap-testing with known-good chips from a donor machine.
* '''520ST/520STM/520STF/520STFM (512 KB):''' Typically 16ร— 41256 (256Kร—1) DRAMs arranged as two banks of 8, or 8ร— 44256 (256Kร—4) DRAMs on later boards. RAM is soldered on most revisions.
* '''1040ST/1040STF/1040STFM (1 MB):''' 32ร— 41256 or 16ร— 44256, depending on board revision. On the shared C070789-001 board, the 1040 simply has more RAM positions populated than the 520. RAM is soldered.


=== GLUE (C025915-38A) ===
The MMU handles row/column address multiplexing (MAD0โ€“MAD9), RAS (Row Address Strobe), and CAS (Column Address Strobe) signals to the DRAM array.


The GLUE chip handles address decoding, interrupt management, and sync/timing generation. GLUE failures can produce a wide range of symptoms:
=== Diagnosing RAM Faults ===


<templatestyles src="Template:StyledTable/styles.css" />
<templatestyles src="Template:StyledTable/styles.css" />
{| class="wikitable styled-table"
{| class="wikitable styled-table"
! Symptom !! Possible Cause !! Recommended Action
! Symptom !! Likely Cause !! Diagnostic Steps
|-
| White screen at power-on || One or more RAM chips completely dead; multiple data lines stuck || Use diagnostic cartridge with serial output โ€” it will report stuck data lines and the address of the failing location.<ref name="diagcart"/> If no diagnostic cartridge, desoldering RAM to test in isolation may be necessary.
|-
|-
| No video sync at all (monitor cannot lock on) || GLUE not generating H-sync / V-sync || Check 32 MHz master oscillator output. If present, swap-test GLUE.
| Machine detects less RAM than installed (e.g., 512 KB on a 1040ST) || Failed RAM chip(s) in upper bank; MMU configuration register mismatch || Boot with diagnostic software and run memory test. Check the MMU bank configuration register at $FF8001.
|-
|-
| No interrupts firing (keyboard dead, no cursor blink, system hangs after boot) || GLUE interrupt priority encoder or IPL output fault || Swap-test GLUE.
| Random crashes during use || Intermittent RAM failure; marginal cell that fails under thermal stress || Run extended RAM test software (YAART is recommended). If crashes worsen as the machine warms up, use freeze spray on RAM chips to isolate the failing device.
|-
|-
| Bus errors at specific address ranges || GLUE address decode fault โ€” peripheral select signals (ROM, I/O, etc.) not asserting correctly || Swap-test GLUE.
| Corrupted graphics (random pixels, garbled text) || RAM fault in the video display area (ST-RAM is shared between CPU and video) || Specific corrupted patterns can indicate which data bit is failing. A single stuck bit produces a repeating pattern every N pixels. Use diagnostic software to identify the failing bit.
|-
|-
| Machine works intermittently, especially after warming up || Thermal fault in GLUE die || Apply freeze spray to GLUE while fault is present. If machine recovers, GLUE has a thermal defect.
| Bomb screen (row of bombs at top of display) || Software crash; the number of bombs indicates the 68000 exception type || While not always a RAM fault, persistent bomb screens at random addresses suggest RAM corruption. Run memory diagnostics.
|}
|}


=== MMU (C025912-38) ===
=== 68000 Exception Types (Bomb Count) ===


<templatestyles src="Template:StyledTable/styles.css" />
<templatestyles src="Template:StyledTable/styles.css" />
{| class="wikitable styled-table"
{| class="wikitable styled-table"
! Symptom !! Possible Cause !! Recommended Action
! Bombs !! Exception !! Common Hardware Cause
|-
| 2 || Bus Error || GLUE timeout โ€” accessed unmapped address space. Often caused by bad ROM contact, failed GLUE, or bus contention.
|-
| 3 || Address Error || 68000 attempted word/long access at odd address. Usually a software bug, but can indicate RAM corruption of pointer values.
|-
|-
| No boot, CPU /HALT asserted || MMU not generating DTACK or RAS/CAS for RAM, causing double bus fault || Swap-test MMU.
| 4 || Illegal Instruction || CPU fetched invalid opcode. Corrupt ROM, failing RAM in code area, or bus noise.
|-
|-
| Partial or corrupted display (16-pixel columns missing) || Faulty video DMA address generation || Swap-test MMU.
| 5 || Division by Zero || Almost always a software issue.
|-
|-
| RAM test fails in a pattern (e.g., every other bank) || Faulty RAS/CAS line for specific DRAM bank || Check RAS/CAS signals with logic probe or oscilloscope. If lines from the MMU are dead or stuck, replace MMU.
| 8 || Privilege Violation || Software attempted supervisor-mode instruction in user mode.
|}
|}


=== Shifter (C025914) ===
== Custom IC Failure Diagnosis ==
ย 
The same custom IC set is used across all 520ST and 1040ST models:


<templatestyles src="Template:StyledTable/styles.css" />
<templatestyles src="Template:StyledTable/styles.css" />
{| class="wikitable styled-table"
{| class="wikitable styled-table"
! Symptom !! Possible Cause !! Recommended Action
! IC !! Failure Symptoms !! Diagnostic Steps !! Notes
|-
| '''Shifter''' (C025914) || No display output or grossly corrupted display; no video sync signals on RGB pins; vertical/horizontal bars || Check for video output on monitor port pins with oscilloscope. Reseat Shifter. If socketed, try known-good replacement. || On boards where Shifter is in a metal can with retaining clips, check clip tension. The Shifter is the most commonly failing custom IC.
|-
| '''GLUE''' (C025915) || Machine does not boot at all; no bus activity; bus error (2 bombs) on every access; no video sync; interrupts not generated || Check with logic probe for activity on AS, DTACK, and address bus lines. If CPU is running but getting no DTACK, GLUE is likely faulty. || GLUE generates all chip select signals โ€” a GLUE failure can mimic almost any other chip failure.
|-
| '''MMU''' (C025912 / C100109) || White screen; reduced RAM detected; video DMA not fetching data (display shows content of address $000000 repeating) || With diagnostic cartridge serial output, check if RAM can be written and read. If all reads return the same value, MMU is likely not multiplexing addresses correctly. || On the 1040ST, the full 1 MB RAM requires correct MMU bank selection โ€” a partial MMU failure may cause the 1040ST to report only 512 KB.
|-
|-
| No video output (but system otherwise boots โ€” floppy drive activity is normal, keyboard LEDs respond) || Shifter not serialising video data or analogue output stage dead || Swap-test Shifter.
| '''DMA''' (C025913) || Floppy drive not working (no data transfer despite motor spinning); ACSI hard disk not responding || Floppy motor/head select are controlled via YM2149, not DMA. If drive motor spins but no data is read, and WD1772 appears responsive, the DMA chip may be at fault. || Relatively uncommon failure. Check WD1772 first.
|-
|-
| Wrong colours, colour palette corrupted || Palette register fault in Shifter || Swap-test Shifter.
| '''WD1772''' || No floppy drive response; drive seeks endlessly; read/write errors on all disks || Verify floppy drive itself works (try known-good drive). Check drive select and motor-on signals via YM2149 Port A. If drive motor runs and head moves but data is never transferred, WD1772 is suspect. || Socketed on most boards; can be directly swapped.
|-
|-
| Display works in mono but not in colour (or vice versa) || Partial Shifter failure affecting one video mode || Swap-test Shifter.
| '''YM2149F''' || No sound output; parallel printer not working; floppy drive select lines not toggling (drive won't spin); serial RTS/DTR not working || Test each function independently: sound on monitor port; printer data on DB-25; floppy activity. If multiple YM2149-controlled functions fail simultaneously, the chip is likely bad. || The YM2149 controls many non-obvious functions. A failed YM2149 can make the floppy appear completely dead.
|}
ย 
=== DMA (C025913) ===
ย 
<templatestyles src="Template:StyledTable/styles.css" />
{| class="wikitable styled-table"
! Symptom !! Possible Cause !! Recommended Action
|-
|-
| Floppy drive does not read or write (drive spins but no data transfer) || DMA chip not transferring data between WD1772 and RAM || Verify WD1772 is functioning (check WD1772 signals with logic probe). If WD1772 appears to operate but data does not appear in RAM, swap-test DMA.
| '''MC68901''' (MFP) || System timers not running (no cursor blink); keyboard not responding; RS-232 serial not working || If machine boots to desktop but cursor doesn't blink and keyboard doesn't respond, MFP timer or interrupt system is likely faulty. || Socketed on most boards. Try reseating first.
|-
|-
| ACSI hard disk not detected or produces read errors || DMA fault on the ACSI channel || Swap-test DMA. Note: on some boards, a separate ACSI DMA IC may be used.
| '''MC6850''' (ACIA ร—2) || Keyboard ACIA failure: no keyboard/mouse input. MIDI ACIA failure: no MIDI I/O. || Swap the two 6850s to determine which is faulty (keyboard vs MIDI). || Two physically identical chips; one handles IKBD, the other MIDI.
|-
|-
| Data corruption on disk reads/writes || DMA transferring incorrect data || Swap-test DMA.
| '''MC68000''' (CPU) || Machine completely dead despite good PSU; random and unpredictable behaviour; consistent wrong data on bus || Check for clock signal (8 MHz) on CLK pin. Check that RESET is being released. If clock present and reset releases but no bus activity, CPU is likely faulty. || CPU failure is relatively rare. Verify 32 MHz crystal and clock chain before condemning the CPU.
|}
|}


== Floppy Drive Troubleshooting ==
== Floppy Drive Troubleshooting ==
The floppy subsystem is the same across 520ST (with external drive) and 1040ST (internal drive). The WD1772 controller and DMA chip are on the motherboard; drive select and motor control are handled by the YM2149.


<templatestyles src="Template:StyledTable/styles.css" />
<templatestyles src="Template:StyledTable/styles.css" />
Line 203: Line 202:
! Symptom !! Possible Cause !! Recommended Action
! Symptom !! Possible Cause !! Recommended Action
|-
|-
| Drive does not spin at all || +12 V absent (drive motor needs 12 V); floppy cable disconnected or faulty; WD1772 not issuing motor-on signal || Check +12 V at the floppy power connector. Check ribbon cable continuity. If +12 V is present and cable is good, check WD1772.
| Drive motor does not spin || YM2149 not asserting drive select/motor on; bad cable; drive motor failure || Check YM2149 Port A bits for drive select and motor on. Check ribbon cable (1040ST) or DIN cable (520ST external). Try known-good drive.
|-
|-
| Drive spins but cannot read any disks || Dirty or worn head; head alignment out of spec; WD1772 fault; DMA fault || Clean the drive head. Try a known-good disk. Try a different drive. If the replacement drive also fails, suspect WD1772 or DMA.
| Drive seeks but cannot read || Dirty heads; misaligned heads; failing WD1772; bad data cable || Clean heads with IPA. Try multiple known-good disks. Try known-good drive. If external drive also fails, suspect WD1772 or DMA.
|-
|-
| Drive reads some disks but not others || Head partially dirty; ageing drive with marginal alignment; disks themselves may be degraded || Clean head. Try the disks in another machine. If the drive consistently fails on disks that work elsewhere, the drive mechanism is likely worn.
| Drive seeks to track 0 repeatedly (clicking) || Drive cannot find track 0 sensor; head mechanism stuck; belt failure || Check track 0 sensor. Clean and lubricate head carriage rails. Replace drive belt if stretched.
|-
|-
| Write operations fail (verify errors) || Write-protect tab set; head dirty; WD1772 write precompensation issue; DMA fault || Check write-protect tab. Clean head. Try a different drive. If the replacement drive also fails on write, suspect DMA or WD1772.
| Drive reads but cannot write || Write-protect switch stuck or misread; WD1772 write circuit fault || Check write-protect sensor. Try a different disk with the tab in the correct position.
|-
| Intermittent read errors || Dirty heads; ageing media; PSU instability; thermal WD1772 issue || Clean heads. Try fresh disk. Check PSU voltages. If errors appear only after warm-up, suspect thermal WD1772 failure.
|-
| Very slow boot (30+ seconds) || TOS searching for bootable disk on drive A with no disk inserted || Not a fault. Insert a non-bootable disk or upgrade to TOS 1.04 for improved boot behaviour.
|}
|}


== Keyboard and Mouse Troubleshooting ==
== Audio Troubleshooting ==
ย 
The keyboard assembly contains an HD6301V1 (or compatible, e.g., HD6303) microcontroller that scans the key matrix, reads the mouse quadrature signals, and reads the joystick ports. It communicates with the mainboard via the keyboard ACIA (MC6850) at 7812.5 baud.


<templatestyles src="Template:StyledTable/styles.css" />
<templatestyles src="Template:StyledTable/styles.css" />
Line 220: Line 221:
! Symptom !! Possible Cause !! Recommended Action
! Symptom !! Possible Cause !! Recommended Action
|-
|-
| Keyboard completely dead (no keys respond, mouse does not move cursor) || Keyboard ribbon cable disconnected or damaged; keyboard ACIA (MC6850) failed; HD6301 controller in keyboard assembly failed || Re-seat the keyboard ribbon cable. Check for +5 V at the keyboard connector (the keyboard assembly is powered from the mainboard). If +5 V is present and cable is good, swap-test the keyboard ACIA on the mainboard. If ACIA is good, the HD6301 in the keyboard assembly may be dead.
| No sound at all || YM2149 failure; bad solder joint on audio output path; monitor cable not carrying audio || Check pin 1 (audio out) on 13-pin DIN. If no signal, check YM2149. If signal present, check cable and monitor.
|-
|-
| Some keys do not work || Membrane trace break; dirty contacts; faulty key switch || See [[Atari ST General Maintenance#Keyboard Maintenance]].
| Distorted or crackling sound || Failing electrolytic capacitors in the audio output path || Replace audio-path caps. See [[Atari ST Capacitor Replacement Guide]].
|-
|-
| Keys repeat or produce wrong characters || HD6301 ROM corruption (very rare); keyboard matrix short || Check for conductive debris under the keys. If the matrix scan is producing incorrect results, the keyboard PCB may be damaged.
| Constant hum or buzz || PSU ripple; ground loop with monitor || Recap PSU. Try a different monitor.
|-
|-
| Mouse works but keyboard does not (or vice versa) || Partial HD6301 failure; ACIA serial link error || The HD6301 handles both keyboard and mouse โ€” a partial failure can affect one but not the other. Swap-test with a known-good keyboard assembly.
| Sound from only some programs || Software compatibility issue with specific TOS version || Not a hardware fault.
|}
|}


== MIDI Port Troubleshooting ==
== Keyboard Troubleshooting ==
ย 
Applies equally to 520ST and 1040ST (same keyboard design and IKBD controller):


<templatestyles src="Template:StyledTable/styles.css" />
<templatestyles src="Template:StyledTable/styles.css" />
Line 235: Line 238:
! Symptom !! Possible Cause !! Recommended Action
! Symptom !! Possible Cause !! Recommended Action
|-
|-
| MIDI OUT does not transmit || MIDI ACIA (MC6850) failed; MIDI OUT buffer/driver circuit fault; broken solder joint on MIDI DIN connector || Check the MIDI ACIA (there are two MC6850s on the board โ€” one for keyboard, one for MIDI). Swap the two ACIAs to test. Check for +5 V at the MIDI ACIA. Inspect solder joints on the MIDI OUT DIN connector.
| Keyboard completely dead || Keyboard ribbon cable disconnected/damaged; keyboard ACIA (MC6850) failure; IKBD controller (HD6301) failure || Check ribbon cable. Swap the two MC6850 ACIAs to test keyboard ACIA. If still dead, suspect IKBD controller.
|-
| Some keys not working || Degraded keyboard membrane; dirty contacts; cracked membrane trace || Clean membrane contacts with IPA. Inspect membrane for cracks. Replace membrane if damaged.
|-
|-
| MIDI IN does not receive || Opto-isolator (typically 6N138 or CNY17) on MIDI IN circuit failed; MIDI ACIA fault || The MIDI IN circuit uses an opto-isolator for electrical isolation as per the MIDI specification. Test or replace the opto-isolator. Swap-test MIDI ACIA.
| Key repeat stuck || Shorted key on membrane; IKBD controller glitch || Check for debris under affected key. Power-cycle. Inspect membrane for short circuits.
|-
|-
| MIDI timing is erratic (notes delayed, jitter) || Excessive interrupt latency; software issue; less commonly, ACIA clock fault || The ST's MIDI timing is generally excellent (hardware ACIA driven at 31.25 kbit/s from a fixed clock). If timing is poor, check for software conflicts. If the problem persists with different software, check the 500 kHz ACIA clock signal.
| Mouse not working, keyboard works || Mouse port connector fault; faulty mouse; IKBD issue || Try a different mouse. Check DE-9 port for bent pins. Clean connector.
|}
|}


== Serial and Parallel Port Troubleshooting ==
== Using the Diagnostic Cartridge ==


<templatestyles src="Template:StyledTable/styles.css" />
The Atari diagnostic cartridge is an essential repair tool for both 520ST and 1040ST machines:
{| class="wikitable styled-table"
ย 
! Symptom !! Possible Cause !! Recommended Action
* It runs hardware diagnostics independent of TOS ROMs and can test RAM, keyboard, display timing, floppy drives, and other subsystems.
|-
* If the machine cannot produce a display, the diagnostic cartridge outputs test results over the RS-232 serial port. Connect a null modem cable to a PC running a terminal emulator at 9600 baud, 8N1.<ref name="diagcart"/>
| RS-232 serial port does not communicate || MC68901 MFP USART fault; SN75188/SN75189 (or MC1488/MC1489) line driver/receiver failed; incorrect baud rate or handshaking settings || Check software configuration first. RS-232 requires proper handshaking โ€” many issues are cable/null-modem related. If configuration is correct, check the line driver and receiver ICs (these are standard, cheap, and easy to replace). The MFP is less likely to fail but can be swap-tested.
* During boot, it sends a memory test report over serial showing written values, read values, and failing addresses โ€” allowing identification of specific stuck data lines.
|-
* '''If the diagnostic cartridge menu does not appear on screen OR via serial:''' The fault is likely in ROM address space decoding (GLUE) or the cartridge port edge connector. Clean the connector and check for proper contact.
| Parallel port does not print || Software driver issue; data lines not driven; failed output buffer || Check with different software. Measure data lines (active during a print operation) at the DB-25 connector. Some boards buffer the parallel port through an LS244 or LS245 โ€” check this chip if data lines are dead.
* Modern alternatives include the exxos multi-diagnostic cartridge with Atari diagnostics, EmuTOS, and YAART memory testing on a single cartridge with rotary selector.
|}


== Diagnostic Tools ==
== Diagnostic Flowchart ==


Several diagnostic aids are available for the 520ST and 1040ST:
A suggested order of operations when confronted with a non-working 520ST or 1040ST:


* '''Diagnostic cartridges:''' ROM-based test suites that execute from the cartridge port, bypassing the need for working main RAM. Useful for testing RAM, ROM, and basic I/O.
# '''Check PSU:''' Measure +5V and +12V. If out of spec, recap PSU (1040ST) or replace/test external PSU (520ST) before proceeding.
* '''Diagnostic floppy disks:''' Software-based test suites (e.g., "STD TEST" and similar). Require a working boot path (ROM, basic RAM, floppy drive).
# '''Reseat all socketed ICs:''' This alone resolves a significant percentage of faults after decades of thermal cycling.
* '''Oscilloscope:''' Essential for PSU ripple measurements, clock verification, and signal integrity checks on bus lines.
# '''Try diagnostic cartridge:''' If it works, use it to test RAM, keyboard, floppy, and timings.
* '''Logic probe:''' Useful for quick checks on chip select lines, clock signals, and bus activity.
# '''Check TOS ROMs:''' Swap with known-good set if available. White screen is often caused by ROM issues.
* '''Freeze spray:''' For isolating thermal intermittent faults โ€” spray individual components to cool them and observe whether the fault changes.
# '''Check RAM:''' Use diagnostic cartridge serial output or substitution testing.
# '''Check clock:''' Verify 32 MHz crystal and derived clocks (8 MHz CPU, 2 MHz YM2149, 500 kHz ACIA) with oscilloscope.
# '''Check custom ICs:''' If all of the above are good, fault is likely in Shifter, GLUE, MMU, or DMA โ€” substitution testing is the most reliable method.


== See Also ==
== Related Pages ==


* [[Atari 520ST]]
* [[Atari 1040ST]]
* [[Atari ST General Maintenance]]
* [[Atari ST General Maintenance]]
* [[Atari ST Capacitor Replacement Guide]]
* [[Atari ST Capacitor Replacement Guide]]
* [[Atari 520ST]]
* [[Atari STE Troubleshooting Guide]]
* [[Atari 1040ST]]


== References ==
== References ==
<references />
<references/>


{{Navbox-AtariVintageComputers}}
{{Navbox-AtariVintageComputers}}


[[Category:Atari Troubleshooting Guides]]
[[Category:Atari Troubleshooting Guides]]
[[Category:Atari Systems]]

Latest revision as of 17:22, 29 March 2026

This guide covers fault diagnosis and repair for the Atari 520ST, 520STM, 520STF, 520STFM, 1040ST, 1040STF, and 1040STFM computers. These machines share common motherboard architectures (including the C070789-001 used in both 520STFM and 1040STFM) and the same custom IC set, so troubleshooting procedures are largely identical. Differences between the 520ST and 1040ST (principally RAM population and PSU type) are noted where relevant.

For the Atari STE range (520STE / 1040STE), see Atari STE Troubleshooting Guide.

Prerequisites

[edit | edit source]

Before beginning troubleshooting:

  • A digital multimeter (DMM) capable of DC voltage, continuity, and resistance measurement is essential.
  • An oscilloscope is helpful for diagnosing clock and timing issues but is not strictly required for most common faults.
  • An Atari diagnostic cartridge is a valuable tool โ€” it provides hardware-level testing via the cartridge port and can output diagnostics over the serial port even when the display is not functioning.[1]
  • A known-good external floppy drive, spare TOS ROMs, and spare RAM chips are useful for substitution testing.

Boot Sequence Overview

[edit | edit source]

Understanding the ST boot sequence helps pinpoint where a failure occurs. This sequence is the same for the 520ST and 1040ST:

  1. Power-on: PSU provides +5V and +12V rails. The 68000 is held in reset by the GLUE chip until voltages stabilise.
  2. Reset release: GLUE releases the 68000 RESET line. The CPU reads the initial SSP (Supervisor Stack Pointer) from address $000000 and the initial PC (Program Counter) from address $000004, which points into TOS ROM space ($FC0000โ€“$FEFFFF).
  3. TOS initialisation: TOS performs a RAM sizing/test routine. On TOS 1.00โ€“1.04, this is a brief test. The screen remains blank (white or black depending on the monitor type) during this phase.
  4. GEM desktop: If all tests pass and no bootable floppy is detected, the GEM desktop appears. If a bootable floppy is present, the boot sector is loaded and executed.

Failure at each stage produces characteristic symptoms:

  • No power LED โ†’ PSU failure or blown fuse
  • Power LED on but no display โ†’ CPU not running, clock failure, or ROM space fault
  • White screen (colour monitor) โ†’ CPU running but stuck in early TOS code; likely RAM fault, ROM fault, or MMU/GLUE failure
  • Green screen with corrupted icons โ†’ Partial boot; likely bad RAM chip, video fault, or corrupted TOS
  • Desktop appears but crashes โ†’ PSU instability, intermittent RAM fault, or software issue

Power Supply Diagnosis

[edit | edit source]

The PSU is the most common source of problems in ageing ST machines. The 1040ST has an internal switched-mode PSU; the 520ST/520STM uses an external supply.

Voltage Rails

[edit | edit source]
Rail Wire Colour (1040ST internal header) Nominal Voltage Acceptable Range Notes
+5V Red 5.00V 4.75V โ€“ 5.25V Primary logic supply; most critical rail. Below 4.75V causes instability; above 5.5V risks IC damage.
+12V Blue 12.0V 11.0V โ€“ 13.0V Floppy drive motor, RS-232 transceivers. The machine will operate without it but floppy and serial will not function.
GND Black 0V โ€” Common ground reference.

On the 520ST with external PSU, measure at the 7-pin DIN power connector: +5V (pins 1, 2), +12V (pin 6), GND (pins 3, 4, 5). Some external supplies also provide โˆ’12V (pin 7) for the RS-232 transceivers.

Testing procedure:

  1. Do NOT test a switched-mode PSU disconnected from the motherboard (or a dummy load). Without a load, output voltages may be dangerously high or the PSU may not start at all.[2]
  2. With the machine assembled and powered on, measure the +5V rail at the motherboard power header (red wires to black wires).
  3. If the voltage is below 4.9V under load, the PSU capacitors likely need replacement.
  4. Check for AC ripple on the +5V rail if you have an oscilloscope. A healthy recapped PSU should show less than 200mV of ripple. More than 500mV indicates failing capacitors and will cause instability.[2]

Common PSU Symptoms (Both 520ST and 1040ST)

[edit | edit source]
Symptom Possible Cause Recommended Action
No power at all (no LED) Blown mains fuse in cable or PSU; failed power switch (1040ST); open fusible resistor in PSU; failed external PSU brick (520ST) Check mains fuse. On 1040ST: check PSU internal fuse and continuity through power switch. On 520ST: test external PSU output voltages with DMM.
Power LED flickers PSU attempting to start but shutting down (overcurrent protection) Likely shorted component on motherboard or PSU. Disconnect motherboard from PSU and test PSU with dummy load (1040ST) or try a different external PSU (520ST).
Random crashes during use Low or noisy +5V rail Measure +5V under load. Recap PSU if voltage below 4.9V or ripple exceeds 300mV.
Screen dims when floppy drive activates Poor peak power delivery on +5V rail (1040ST internal PSU) Recap PSU. On Mitsumi SR98 PSUs, the biasing resistors R201/R202 may need adjustment โ€” replace R202 with 1Kฮฉ and R201 with 1.2Kฮฉ, then readjust the voltage trimmer to 5.00V.[2]
Audio noise/hum Ripple on +5V rail coupling into YM2149 audio output Recap PSU. Check ground connections.
Intermittent floppy read errors Unstable +12V rail (floppy motor speed variation) or noisy +5V (logic errors) Measure both rails under load. Recap PSU.

Display / Video Troubleshooting

[edit | edit source]

The video subsystem (Shifter, GLUE, and related circuitry) is identical across the 520ST and 1040ST.

No Display at All

[edit | edit source]
Symptom Possible Cause Recommended Action
No display, no power LED PSU failure See Power Supply Diagnosis section above.
Power LED on, completely blank screen CPU not running; no clock signal; or monitor cable/connection fault Verify monitor cable is connected and monitor is set to correct input. Check 32 MHz crystal oscillator with oscilloscope. Reseat CPU, Shifter, GLUE.
Power LED on, white screen only (colour monitor) CPU running but failing during early TOS initialisation. Common causes: RAM fault, TOS ROM fault, MMU failure, GLUE failure. Reseat all socketed ICs. Try known-good TOS ROMs. Try with diagnostic cartridge โ€” if cartridge menu appears, motherboard is partially functional and fault is likely in RAM or TOS ROMs. If no cartridge output, fault is in ROM address space decoding (GLUE/MMU).[1]
Power LED on, white screen only (monochrome monitor) As above, plus: mono detect pin not being grounded (faulty cable/connector) causing Shifter to output colour mode to mono monitor Check pin 4 on the 13-pin DIN connector. It should be grounded for monochrome mode.
Display appears briefly then goes blank PSU voltage droop under full load; thermal fault in Shifter or GLUE Measure +5V rail under load. If voltage is stable, suspect thermal failure in Shifter โ€” use freeze spray to isolate.

Display Corruption

[edit | edit source]
Symptom Possible Cause Recommended Action
Vertical or horizontal bars/stripes Shifter failure; bad connection between Shifter and data bus; failing RAM affecting video memory area Reseat Shifter chip. Check Shifter retaining clips for adequate tension. If problem persists, try known-good Shifter from donor board.
Wrong or garbled colours Shifter output fault; bad solder joint on RGB output path; monitor cable fault Check 13-pin DIN cable for continuity on R, G, B pins (6, 7, 10). Check monitor connector for bent pins. Reseat Shifter.
Screen jitter or rolling Sync signal fault; HSync or VSync not being generated correctly by GLUE Check HSync (pin 9) and VSync (pin 12) on monitor port with oscilloscope. If absent, suspect GLUE chip. Check 32 MHz crystal.
Ghosting or wavy lines PSU ripple; poor grounding; RF interference from internal PSU (1040ST) Recap PSU. Check ground connections.
"Jail bars" (thin vertical lines on solid colours) Impedance issues on the video output path; common on STFM boards Sometimes improved by replacing decoupling capacitors near the Shifter. A known characteristic of the ST video circuit rather than a definitive fault.

RF Modulator Issues (FM Models โ€” 520STFM / 1040STFM)

[edit | edit source]
Symptom Possible Cause Recommended Action
No RF output, RGB works fine RF modulator failure; modulator not connected Check modulator solder connections. A composite video mod bypassing the modulator is a common upgrade.
RF output is snowy/noisy Modulator mistuned; weak signal Adjust tuning slug on modulator. Clean modulator connections.
RF output with herringbone pattern Ground loop or interference Ensure proper grounding. Consider composite/RGB output instead.

RAM Fault Diagnosis

[edit | edit source]

RAM faults are a common cause of white screens, crashes, and corrupted displays on both the 520ST and 1040ST.

RAM Architecture

[edit | edit source]
  • 520ST/520STM/520STF/520STFM (512 KB): Typically 16ร— 41256 (256Kร—1) DRAMs arranged as two banks of 8, or 8ร— 44256 (256Kร—4) DRAMs on later boards. RAM is soldered on most revisions.
  • 1040ST/1040STF/1040STFM (1 MB): 32ร— 41256 or 16ร— 44256, depending on board revision. On the shared C070789-001 board, the 1040 simply has more RAM positions populated than the 520. RAM is soldered.

The MMU handles row/column address multiplexing (MAD0โ€“MAD9), RAS (Row Address Strobe), and CAS (Column Address Strobe) signals to the DRAM array.

Diagnosing RAM Faults

[edit | edit source]
Symptom Likely Cause Diagnostic Steps
White screen at power-on One or more RAM chips completely dead; multiple data lines stuck Use diagnostic cartridge with serial output โ€” it will report stuck data lines and the address of the failing location.[1] If no diagnostic cartridge, desoldering RAM to test in isolation may be necessary.
Machine detects less RAM than installed (e.g., 512 KB on a 1040ST) Failed RAM chip(s) in upper bank; MMU configuration register mismatch Boot with diagnostic software and run memory test. Check the MMU bank configuration register at $FF8001.
Random crashes during use Intermittent RAM failure; marginal cell that fails under thermal stress Run extended RAM test software (YAART is recommended). If crashes worsen as the machine warms up, use freeze spray on RAM chips to isolate the failing device.
Corrupted graphics (random pixels, garbled text) RAM fault in the video display area (ST-RAM is shared between CPU and video) Specific corrupted patterns can indicate which data bit is failing. A single stuck bit produces a repeating pattern every N pixels. Use diagnostic software to identify the failing bit.
Bomb screen (row of bombs at top of display) Software crash; the number of bombs indicates the 68000 exception type While not always a RAM fault, persistent bomb screens at random addresses suggest RAM corruption. Run memory diagnostics.

68000 Exception Types (Bomb Count)

[edit | edit source]
Bombs Exception Common Hardware Cause
2 Bus Error GLUE timeout โ€” accessed unmapped address space. Often caused by bad ROM contact, failed GLUE, or bus contention.
3 Address Error 68000 attempted word/long access at odd address. Usually a software bug, but can indicate RAM corruption of pointer values.
4 Illegal Instruction CPU fetched invalid opcode. Corrupt ROM, failing RAM in code area, or bus noise.
5 Division by Zero Almost always a software issue.
8 Privilege Violation Software attempted supervisor-mode instruction in user mode.

Custom IC Failure Diagnosis

[edit | edit source]

The same custom IC set is used across all 520ST and 1040ST models:

IC Failure Symptoms Diagnostic Steps Notes
Shifter (C025914) No display output or grossly corrupted display; no video sync signals on RGB pins; vertical/horizontal bars Check for video output on monitor port pins with oscilloscope. Reseat Shifter. If socketed, try known-good replacement. On boards where Shifter is in a metal can with retaining clips, check clip tension. The Shifter is the most commonly failing custom IC.
GLUE (C025915) Machine does not boot at all; no bus activity; bus error (2 bombs) on every access; no video sync; interrupts not generated Check with logic probe for activity on AS, DTACK, and address bus lines. If CPU is running but getting no DTACK, GLUE is likely faulty. GLUE generates all chip select signals โ€” a GLUE failure can mimic almost any other chip failure.
MMU (C025912 / C100109) White screen; reduced RAM detected; video DMA not fetching data (display shows content of address $000000 repeating) With diagnostic cartridge serial output, check if RAM can be written and read. If all reads return the same value, MMU is likely not multiplexing addresses correctly. On the 1040ST, the full 1 MB RAM requires correct MMU bank selection โ€” a partial MMU failure may cause the 1040ST to report only 512 KB.
DMA (C025913) Floppy drive not working (no data transfer despite motor spinning); ACSI hard disk not responding Floppy motor/head select are controlled via YM2149, not DMA. If drive motor spins but no data is read, and WD1772 appears responsive, the DMA chip may be at fault. Relatively uncommon failure. Check WD1772 first.
WD1772 No floppy drive response; drive seeks endlessly; read/write errors on all disks Verify floppy drive itself works (try known-good drive). Check drive select and motor-on signals via YM2149 Port A. If drive motor runs and head moves but data is never transferred, WD1772 is suspect. Socketed on most boards; can be directly swapped.
YM2149F No sound output; parallel printer not working; floppy drive select lines not toggling (drive won't spin); serial RTS/DTR not working Test each function independently: sound on monitor port; printer data on DB-25; floppy activity. If multiple YM2149-controlled functions fail simultaneously, the chip is likely bad. The YM2149 controls many non-obvious functions. A failed YM2149 can make the floppy appear completely dead.
MC68901 (MFP) System timers not running (no cursor blink); keyboard not responding; RS-232 serial not working If machine boots to desktop but cursor doesn't blink and keyboard doesn't respond, MFP timer or interrupt system is likely faulty. Socketed on most boards. Try reseating first.
MC6850 (ACIA ร—2) Keyboard ACIA failure: no keyboard/mouse input. MIDI ACIA failure: no MIDI I/O. Swap the two 6850s to determine which is faulty (keyboard vs MIDI). Two physically identical chips; one handles IKBD, the other MIDI.
MC68000 (CPU) Machine completely dead despite good PSU; random and unpredictable behaviour; consistent wrong data on bus Check for clock signal (8 MHz) on CLK pin. Check that RESET is being released. If clock present and reset releases but no bus activity, CPU is likely faulty. CPU failure is relatively rare. Verify 32 MHz crystal and clock chain before condemning the CPU.

Floppy Drive Troubleshooting

[edit | edit source]

The floppy subsystem is the same across 520ST (with external drive) and 1040ST (internal drive). The WD1772 controller and DMA chip are on the motherboard; drive select and motor control are handled by the YM2149.

Symptom Possible Cause Recommended Action
Drive motor does not spin YM2149 not asserting drive select/motor on; bad cable; drive motor failure Check YM2149 Port A bits for drive select and motor on. Check ribbon cable (1040ST) or DIN cable (520ST external). Try known-good drive.
Drive seeks but cannot read Dirty heads; misaligned heads; failing WD1772; bad data cable Clean heads with IPA. Try multiple known-good disks. Try known-good drive. If external drive also fails, suspect WD1772 or DMA.
Drive seeks to track 0 repeatedly (clicking) Drive cannot find track 0 sensor; head mechanism stuck; belt failure Check track 0 sensor. Clean and lubricate head carriage rails. Replace drive belt if stretched.
Drive reads but cannot write Write-protect switch stuck or misread; WD1772 write circuit fault Check write-protect sensor. Try a different disk with the tab in the correct position.
Intermittent read errors Dirty heads; ageing media; PSU instability; thermal WD1772 issue Clean heads. Try fresh disk. Check PSU voltages. If errors appear only after warm-up, suspect thermal WD1772 failure.
Very slow boot (30+ seconds) TOS searching for bootable disk on drive A with no disk inserted Not a fault. Insert a non-bootable disk or upgrade to TOS 1.04 for improved boot behaviour.

Audio Troubleshooting

[edit | edit source]
Symptom Possible Cause Recommended Action
No sound at all YM2149 failure; bad solder joint on audio output path; monitor cable not carrying audio Check pin 1 (audio out) on 13-pin DIN. If no signal, check YM2149. If signal present, check cable and monitor.
Distorted or crackling sound Failing electrolytic capacitors in the audio output path Replace audio-path caps. See Atari ST Capacitor Replacement Guide.
Constant hum or buzz PSU ripple; ground loop with monitor Recap PSU. Try a different monitor.
Sound from only some programs Software compatibility issue with specific TOS version Not a hardware fault.

Keyboard Troubleshooting

[edit | edit source]

Applies equally to 520ST and 1040ST (same keyboard design and IKBD controller):

Symptom Possible Cause Recommended Action
Keyboard completely dead Keyboard ribbon cable disconnected/damaged; keyboard ACIA (MC6850) failure; IKBD controller (HD6301) failure Check ribbon cable. Swap the two MC6850 ACIAs to test keyboard ACIA. If still dead, suspect IKBD controller.
Some keys not working Degraded keyboard membrane; dirty contacts; cracked membrane trace Clean membrane contacts with IPA. Inspect membrane for cracks. Replace membrane if damaged.
Key repeat stuck Shorted key on membrane; IKBD controller glitch Check for debris under affected key. Power-cycle. Inspect membrane for short circuits.
Mouse not working, keyboard works Mouse port connector fault; faulty mouse; IKBD issue Try a different mouse. Check DE-9 port for bent pins. Clean connector.

Using the Diagnostic Cartridge

[edit | edit source]

The Atari diagnostic cartridge is an essential repair tool for both 520ST and 1040ST machines:

  • It runs hardware diagnostics independent of TOS ROMs and can test RAM, keyboard, display timing, floppy drives, and other subsystems.
  • If the machine cannot produce a display, the diagnostic cartridge outputs test results over the RS-232 serial port. Connect a null modem cable to a PC running a terminal emulator at 9600 baud, 8N1.[1]
  • During boot, it sends a memory test report over serial showing written values, read values, and failing addresses โ€” allowing identification of specific stuck data lines.
  • If the diagnostic cartridge menu does not appear on screen OR via serial: The fault is likely in ROM address space decoding (GLUE) or the cartridge port edge connector. Clean the connector and check for proper contact.
  • Modern alternatives include the exxos multi-diagnostic cartridge with Atari diagnostics, EmuTOS, and YAART memory testing on a single cartridge with rotary selector.

Diagnostic Flowchart

[edit | edit source]

A suggested order of operations when confronted with a non-working 520ST or 1040ST:

  1. Check PSU: Measure +5V and +12V. If out of spec, recap PSU (1040ST) or replace/test external PSU (520ST) before proceeding.
  2. Reseat all socketed ICs: This alone resolves a significant percentage of faults after decades of thermal cycling.
  3. Try diagnostic cartridge: If it works, use it to test RAM, keyboard, floppy, and timings.
  4. Check TOS ROMs: Swap with known-good set if available. White screen is often caused by ROM issues.
  5. Check RAM: Use diagnostic cartridge serial output or substitution testing.
  6. Check clock: Verify 32 MHz crystal and derived clocks (8 MHz CPU, 2 MHz YM2149, 500 kHz ACIA) with oscilloscope.
  7. Check custom ICs: If all of the above are good, fault is likely in Shifter, GLUE, MMU, or DMA โ€” substitution testing is the most reliable method.
[edit | edit source]

References

[edit | edit source]
  1. โ†‘ 1.0 1.1 1.2 1.3 Basic Troubleshooting - Diagnostic Cartridge Basics, Atari Wikiโ€”link(accessed 2026-03-27)
  2. โ†‘ 2.0 2.1 2.2 The LaST Upgrade - Atari PSU Repair, exxosโ€”link(accessed 2026-03-27)


Retrieved from "https://wiki.retrotechcollection.com/index.php?title=Atari_ST_Troubleshooting_Guide&oldid=10260"