IBM System/23 Datamaster Capacitor Replacement Guide

This guide documents capacitor diagnosis and replacement for the IBM System/23 Datamaster (machine types 5322 all-in-one and 5324 tower). The Datamaster uses a linear power supply very similar in concept to the IBM 5120 — heavy mains transformer, bridge rectifier, large bulk filter electrolytics, and series-pass linear regulators producing the five rails (+5 V, +12 V, +24 V, −5 V, −12 V) plus the Power Good signal. After 40+ years all electrolytics should be considered out of specification, and the memory daughter card bypass capacitors are flagged specifically by community restorers as a top-of-list failure point.

Important Caveat

edit

Per-board exact capacitor values for the 5322 / 5324 are NOT transcribed into any community-accessible source. The PSU schematic appears in SY34-0171-0 (IBM 5322 Computer Service Manual, December 1980) at page 78 and the parts list is in the same document, but a per-component BoM is not separately mirrored online. This guide therefore documents the typical linear-PSU practice that applies to recapping the Datamaster, with categories of capacitor and representative value ranges, and flags the memory daughter card decoupling cap problem that is specifically called out in community references.[1] Each restorer should pull the PSU board, identify each cap in situ by its printed markings, and replace with same-or-higher voltage and same capacitance.

Safety Warning

edit

The 5322 PSU contains mains-rectified bulk capacitors that hold a lethal charge after power-off. The integrated CRT carries lethal high voltage on the flyback transformer and anode. Before any work:

  1. Power off and unplug the mains lead.
  2. Wait at least 30 seconds.
  3. Discharge each PSU bulk filter capacitor through a 1 kΩ / 5 W resistor.
  4. Discharge the CRT anode to chassis ground via a high-voltage probe.
  5. Verify both with a multimeter.

The Memory Daughter Card Bypass-Cap Problem — Primary Triage

edit

The Datamaster's memory daughter cards carry SMD or small through-hole decoupling capacitors on the +5 V supply to the TMS4132 stacked-die DRAMs. Community restorers (Bits Passats, VCFed restoration accounts) report that a shorted decoupling cap on the RAM rails will brick the system — the +5 V rail is pulled out of tolerance, parity errors trigger the 8085's TRAP, and the system halts before completing POST.

Recommended first-test triage before any other cap work:

  1. Power off, unplug, discharge.
  2. Pull the memory daughter cards.
  3. On each card, identify the decoupling caps on the +5 V to ground rails (typically small through-hole or radial electrolytics, ~10–47 µF / 10–25 V).
  4. Test each in-circuit with a multimeter set to diode mode: black probe to ground, red probe to rail. A healthy cap reads open / high resistance; a shorted cap reads close to 0 Ω.
  5. If a short is found, remove the cap entirely as a first triage — the daughter card can be tested without the bypass cap fitted, and replacement caps can be soldered in once the rest of the system is verified.

This procedure is specifically endorsed by Bits Passats: "remove them [shorted decoupling caps] as a first triage step before recapping."

Linear PSU Topology

edit

The 5322 PSU is a single-board linear supply with:

  • Mains input section — input fuse, X2 line suppression capacitor (replace if RIFA-branded), mains transformer.
  • Rectifier section — bridge diodes on the secondary windings.
  • Bulk filter section — large axial-can aluminium electrolytics smoothing the rectified DC for each rail.
  • Series-pass regulator section — bipolar pass transistors on heatsinks for +5 V, +12 V, +24 V, −5 V, −12 V.
  • Power Good logic — comparators monitoring each rail; assertion of Power Good drives the system Reset line and the 8085 Reset In pin.
  • Output filtering — smaller electrolytics at each rail output to the motherboard connector.

5322 Main PSU Bulk Filter Capacitors — Primary Replacement Target

edit

The bulk filter capacitors after the bridge rectifier are by far the most likely failure point on a 40+ year old linear PSU. Representative values (not IBM-published; verify printed values before ordering):

5322 PSU bulk filter capacitor (representative)
Value Voltage Type Position Quantity (approx)
4700–10000 µF 16 V Axial-can aluminium electrolytic, 105 °C +5 V rail bulk after bridge 1
2200–4700 µF 35 V Axial-can aluminium electrolytic, 105 °C +12 V rail bulk 1
4700–10000 µF 35 V Axial-can aluminium electrolytic, 105 °C +24 V rail bulk (8-inch floppy spindle / head load) 1
1000–2200 µF 16–25 V Axial-can aluminium electrolytic, 105 °C −5 V bias rail 1
1000–2200 µF 25–35 V Axial-can aluminium electrolytic, 105 °C −12 V RS-232 rail 1

Lead spacing is fixed by the PSU board layout. Use radial-to-axial adapter leads or modern axial replacements with the correct lead pitch if needed.

5322 Series-Pass Regulator Capacitors

edit

Around each linear regulator's input and output:

5322 PSU series-pass regulator capacitors (representative)
Value Voltage Type Position Quantity (approx)
47–470 µF 16–35 V Aluminium electrolytic, 105 °C Regulator input / output for each rail 8–12
0.1–1 µF 50 V Film or tantalum bypass (typically still good — inspect only) Regulator bypass at each pass transistor 8–12

Mains Suppression Capacitor

edit

Inspect the X2 mains suppression capacitor on the primary side of the PSU. If it is:

  • RIFA-branded (the brand most associated with venting failure on 1981-era IBM PSUs), or
  • Cracked, bulging, or showing fluid leakage, or
  • Producing a "fish" odour during operation,

replace immediately with a modern X2-class 0.1 µF / 275 VAC capacitor. The fish odour is the RIFA polymer impregnation venting and is a fire hazard.

Memory Daughter Card Capacitors

edit

The 32 KB and 64 KB memory daughter cards carry decoupling caps on the +5 V supply to each TMS4132. Representative cap list:

Memory daughter card capacitors (representative)
Value Voltage Type Position
10–47 µF 10–25 V Aluminium electrolytic or tantalum +5 V bulk decoupling at the card edge connector
0.1 µF 25 V Ceramic disc Per-DRAM decoupling (one per TMS4132)

The bulk electrolytic decoupling caps are the documented failure target. The per-DRAM ceramic caps rarely fail and can be left in place unless evidence of shorting is found.

CRT Deflection / Flyback Board

edit

The 9-inch (or possibly 12-inch — see verification note in the Maintenance Guide) CRT deflection and flyback board carries small electrolytics around the vertical / horizontal oscillator and high-voltage rectifier filtering. Typical values:

CRT deflection / flyback board capacitors (representative)
Value Voltage Type Position
1–10 µF 25 V Aluminium electrolytic Vertical / horizontal oscillator bypass
22–47 µF 35 V Aluminium electrolytic Vertical deflection driver
100–470 µF 35 V Aluminium electrolytic +12 V deflection bulk
0.01–0.1 µF 1.6–2 kV Ceramic disc (HV) Snubber on flyback collector

The HV ceramic caps rarely fail. Aluminium electrolytics on the deflection board can leak and corrode the board over time — inspect under magnification.

8-Inch Floppy Drive Logic Board

edit

The built-in 8-inch floppy drives (typically Shugart-class) carry their own logic board with electrolytics:

8-inch floppy drive logic board (representative)
Value Voltage Type Position
47 µF 16 V Aluminium electrolytic Spindle motor driver
22 µF 16 V Aluminium electrolytic Sector buffer
10 µF 25 V Aluminium electrolytic Head amp
4.7 µF 16 V Tantalum IC bypass

If a drive becomes unreliable warm but stable cold, recap the drive logic board with 105 °C low-ESR equivalents.

ROS Module Sockets

edit

The motherboard's ROS module sockets do not carry their own electrolytic caps. Decoupling for each ROS module is provided by ceramic caps on the motherboard, which rarely fail. ROS module failure (Mostek MK36000 mask ROM degradation) is documented in IBM System/23 Datamaster Troubleshooting Guide but is not a capacitor failure.

Optional 5246 / 5247 PSUs

edit

If you have an external IBM 5246 Diskette Unit or 5247 Winchester Disk Unit attached, each has its own linear PSU with its own bulk filter caps. Recap separately using the same approach — bulk filter caps after the bridge are the primary target.

edit

For all electrolytic replacements, choose:

  • Manufacturer — Panasonic FR / FM / FC, Nichicon HE / HZ (post-2007 date codes), Rubycon ZLH / ZLJ / YXJ, United Chemi-Con KZH / KZE. Avoid general-purpose Chinese-brand electrolytics for PSU rebuild.
  • Temperature rating — 105 °C even where the IBM original was 85 °C (small price premium, much longer life).
  • Voltage rating — equal to or higher than the original.
  • Capacitance — equal to original (do not overshoot — increased capacitance can stress the bridge rectifier).
  • Lead spacing — verify before ordering; many original 5322 caps are axial, but modern replacements are typically radial.

Recap Procedure

edit
  1. Discharge the PSU bulk capacitors; verify with a multimeter.
  2. Discharge the CRT anode if working on the deflection board.
  3. Slide the PSU sled out from the rear of the chassis.
  4. Remove the PSU board from the sled.
  5. Photograph the board from both sides at high resolution. Record every cap's location, value, polarity, lead pitch.
  6. Desolder each electrolytic with solder wick on each lead. Limit each desolder cycle to 5–7 seconds at no more than 350 °C.
  7. Clean each pad with solder wick.
  8. Fit replacements matching the silkscreen polarity (− on cap stripe to − on silkscreen).
  9. Solder both leads from the underside. Inspect for clean fillets. Trim leads flush.
  10. Reassemble. Verify rails on the bench with a multimeter under a 1–2 A resistive load before refitting to the chassis.

ATX Bench Test Alternative

edit

For initial board-level diagnostics, a modern ATX PSU can substitute for the original 5322 supply, provided:

  • The +24 V rail is supplied from a separate brick (ATX does not include +24 V).
  • The Molex pinout matches the 5322's motherboard connector.
  • The CRT board is disconnected (the CRT cannot run from an ATX supply).
  • The 8-inch floppy drives are disconnected (they need +24 V at significant current).

This is a bench-test arrangement only, not a permanent replacement.

Post-Recap Verification

edit
  1. Power on with no diskette in either drive and no peripherals attached.
  2. Probe each rail at the motherboard power connector — verify within tolerance.
  3. Verify the service probe shows code 01 (CPU self-test) within ~100 ms of Power Good assertion.
  4. Verify the probe code increments through 02, 04, 05, 06, 07.
  5. Verify the CRT initialises after test 07.
  6. Verify all subsequent PID-1200 tests pass.
  7. Insert a system diskette and verify the BASIC prompt appears.

If any test fails after recap, re-inspect the polarity of every replaced cap before suspecting another fault — reversed polarity is the most common error.

Polarity Reference

edit
 
Polarity reference for IBM motherboard tantalum and aluminium electrolytic capacitors. Match the silkscreen "−" to the cap stripe. The convention applies to the 5322 PSU and memory daughter cards. (Image: minuszerodegrees.net)

When Not to Recap

edit

If the 5322 powers on cleanly, all rails are within tolerance, PID-1200 passes all tests (01 through 3E), and there is no visible cap failure, the caps are within tolerance. Given the machine's age (~45 years), planned recap before any extended use is strongly recommended to avoid bulk filter cap failure damaging downstream circuitry — particularly the 8085 and 8275 which are now relatively scarce.

Always recap if:

  • Any visible cap failure (bulged top, leaked electrolyte) on PSU, memory daughter cards, deflection board, or drive logic.
  • PSU smoke, fishy odour or audible whine.
  • Rails out of tolerance.
  • RIFA-branded X2 mains suppression cap present (preventive replacement).
  • System unstable when warm but stable when cold.
  • Repeated memory parity errors (TRAP halts) — almost always a daughter card bypass cap.
edit

References

edit


  1. Bits Passats, "IBM System/23 Datamaster" — Power Supply notes.