IBM RS/6000 Maintenance Guide

Revision as of 11:52, 22 May 2026 by Josh (talk | contribs) (Deep technical RS/6000 page rewrite with verified sources, embedded images, full cap lists)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)

This guide documents preventive maintenance for the IBM RS/6000 family — IBM's POWER / PowerPC / POWER3 / RS64 RISC UNIX workstation and server line, machine types 7011 / 7012 / 7013 / 7015 / 7020 / 7025 / 7026 / 7043 / 7044 / 7248. Procedures are common to the family where possible; submodel-specific procedures are called out where they differ.

Safety Warning

edit

All RS/6000 PSUs contain mains-rectified bulk capacitors that hold a lethal charge after power-off. Before any work inside any PSU shell:

  1. Power off and unplug the mains lead.
  2. Wait at least 30 seconds.
  3. Discharge the bulk capacitor through a 1 kΩ / 5 W resistor.
  4. Verify with a multimeter.

Rack-class RS/6000s (7015, 7026 H-series) may carry dual or redundant PSUs and may also be wired to −48 V DC instead of mains (especially R24 frames in telco service).[1] Confirm the input topology of the specific system in front of you before any service work.

Identifying Your RS/6000

edit

The machine type and submodel are printed on the rear-panel label (e.g. "Type 7012-320", "Type 7043-150 — 43P-150"). The machine type maps to a service guide:

RS/6000 service-guide mapping
Machine type IBM service guide
7011, 7012, 7013, 7015 SA38-0531 series (POWER1/POWER2 era)
7020 / 7007-N40 N40-specific HMR (rare)
7025 (F30/F40/F50/F80) SA38-0532 series + Redbook SG24-5143
7026 (H10/H50/H70/H80) SA38-0535 series + Redbook SG24-5143
7043 (140 / 150 / 240 / 260) SA38-0512 series + Redbook SG24-5144
7044 (170 / 270) SA38-0538 series
7248 (100 / 120 / 133) IBM 7248 HOWTO + per-board reference

The machine-type prefix also tells you the era:

  • 70xx — Micro Channel POWER1/POWER2 (1990–1996).
  • 7248 — PReP / Carolina board (1995–1997).
  • 7025 / 7026 / 7043 / 7044 — CHRP / PCI / Open Firmware (1996–2000).

Opening the System

edit

7011 (pizza-box desktop)

edit
  1. Power off, unplug, discharge PSU.
  2. Loosen the two thumbscrews at the rear top.
  3. Slide the cover back ~10 mm and lift off.
  4. The planar is on the chassis floor; PSU on the left; drive cage at the front.

7012 / 7013 (deskside)

edit
  1. Power off, unplug, discharge.
  2. Open the rear access panel (two captive screws).
  3. The CPU MCM card and memory cards mount in a vertical riser at the rear.
  4. Drive cage runs vertically up the front.
  5. PSU is at the top or rear, depending on submodel.

The MCM module on POWER2 7012-39H / 397 is large, heavy and surrounded by heatsink mass. Do not lift the MCM out unless you have the IBM extraction tool — the module pins are fragile and the seating force is high.

7015 (rack)

edit

Standard 19" rack handling rules. CPU drawer typically slides out on chassis rails; ground yourself before any work; never disconnect rear power without locking the rack rails first.

7025 / 7026 / 7043 / 7044 (PCI tower / rack)

edit
  1. Power off, unplug, discharge.
  2. Remove side panel (single thumbscrew or latch on most submodels).
  3. Planar mounts horizontally; PSU at top or top-rear; drives stacked vertically.
  4. CPU and memory cards on carrier cards that plug vertically into the planar.

7248 (Carolina PReP)

edit
  1. Power off, unplug, discharge.
  2. Remove the cover (two screws at the rear).
  3. Carolina motherboard is in the bottom of the chassis. ISA cards on the left riser; PCI cards on the right.

Inspecting the Planar

edit

The Micro Channel era 7012 / 7013 planars carry surface-mount aluminium electrolytic capacitors of the same vintage and chemistry as the IBM PS/2 Model 70 / 80. They leak. Inspect every SMD electrolytic on the planar before any other work — see IBM RS/6000 Capacitor Replacement Guide for the full procedure.

Later RS/6000 planars (7025 / 7026 / 7043 / 7044) carry through-hole and SMD tantalum decoupling plus through-hole aluminium electrolytics near the PSU input. These age more slowly than the PS/2-era SMD aluminium electrolytics but still warrant inspection.

Regular Cleaning

edit
  • Soft brush and low-pressure compressed air for the planar, the riser, PCI cards, the drive cage and the PSU vents.
  • Hold any fan blades by hand if using compressed air.
  • Clean SCSI ribbon cable connectors and edge connectors. Original IBM SCSI cables age and develop intermittent contact in the IDC connectors.
  • Clean MCA card edge fingers (POWER1/POWER2 era) with a soft eraser or deoxidising contact cleaner.
  • Inspect the planar for any signs of leaked electrolyte around SMD or through-hole aluminium electrolytics.

PSU Voltage Checks

edit

Probe the PSU output rails with a multimeter while the system is powered on. RS/6000 PSUs supply the standard ATX-like rails plus, in many cases, a separate +3.3 V rail for the CPU MCM:

Typical RS/6000 PSU rail tolerances
Rail Acceptable range Notes
+5 V +4.75 V to +5.25 V I/O logic
+12 V +11.4 V to +12.6 V Drive motors
+3.3 V +3.15 V to +3.45 V POWER2 / PowerPC / POWER3 CPU rails (on PSUs that supply it)
−12 V −11.4 V to −12.6 V RS-232 / legacy
−5 V −4.75 V to −5.25 V Legacy (omitted on some later PSUs)

The exact per-machine PSU rating is in the corresponding IBM service guide. Documented values:

  • 7026-H70 — 750 W PSU, FC 6290.[2]
  • 7015 R10 / R20 / R21 CPU drawer — 0.29 kVA typical, 200–240 V AC.
  • 7015 R24 — 0.685 kVA typical, 200–240 V AC or −48 V DC.
  • Per-machine watts for 7011, 7012, 7013, 7043 — refer to SA38-0531 / SA38-0512.

NVRAM / RTC Battery

edit

The RS/6000 family uses Dallas Semiconductor TimeKeeper modules on the planar to hold real-time clock, system NVRAM and boot configuration. The internal lithium cell fails after 10–20 years, producing:

  • Lost boot list (SMS boots from CD instead of disk after every power-cycle).
  • Clock at epoch (1 Jan 1970 or similar).
  • Lost machine serial number / model code (rare but possible if the part holds the IPL ROS / VPD).
  • Solid LED panel halt at NVRAM tests (typically 200 / 202 / 203 on the LED panel — see IBM RS/6000 Troubleshooting Guide).

Common Dallas modules in the family include:

  • DS1287 / DS1287A — 24-pin DIP, used on early MCA 7012 / 7013 planars (same family as PS/2 50/70/80).
  • DS1385 / DS1387 — DS1385 + integrated battery/crystal in an encapsulated module.
  • DS1644 — non-volatile TimeKeeping RAM.
  • DS1742W — 32-pin nonvolatile timekeeping RAM, 5 V or 3 V variants.
  • DS1746 / DS1746P — 128 KB × 8 NVSRAM + RTC, Y2K-compliant.

Replacement options:

  1. Direct replacement with a fresh-date Dallas/Maxim part (NOS or refurbished). DS12887+, DS1387 with cell replaced, DS1746P+.
  2. Modern drop-in board replacements such as the Glitch Works GW-1742-1 (DS1742 substitute), GW-1387-1 (DS1387 maintainable repair board) and Necroware nwX287 (DS12887 replacement using a CR1225 + modern SRAM + RTC chip).[3][4]
  3. DIY: cut the Dallas plastic case open, find the internal cell, replace with an external CR2032 holder. Electrically valid but messy.

Replacement procedure:

  1. Power off; discharge PSU bulk capacitor.
  2. Identify the Dallas DIP on the planar — typically near the RTC crystal or near the boot ROM.
  3. Desolder with solder wick at no more than 350 °C. Limit each cycle to 5–7 seconds — RS/6000 planars are no more forgiving of repeated heating than PS/2 planars.
  4. Fit a socket (recommended) before fitting the replacement Dallas / Glitch Works / Necroware module.
  5. Power on; enter SMS (F1 or F4); re-enter boot list, IP / SLIP config, date/time.

After replacement, run SMS → Utilities → Update System Vital Product Data to write the machine serial back into NVRAM if the previous module had lost it.

Capacitor Health

edit

Through-hole tantalum capacitors and aluminium electrolytics on the planar, drive logic boards, GXT graphics cards and PSU all age. Full procedure: IBM RS/6000 Capacitor Replacement Guide.

Specific known capacitor failure points:

  • 7012 / 7013 planar SMD aluminium electrolytics — same generation as PS/2 70/80 plague.
  • 7013 500-series PSU — will refuse to stay on if the cooling fan signal is lost. Aged secondaries cause fan stutter which causes PSU shutdown.[5]
  • POWER GXT graphics card decoupling — SMD tantalum on the GXT800P / GXT3000P boards can fail short and pull down the +5 V rail.

Drive Maintenance

edit

SCSI Spindle Stiction

edit

Original IBM SCSI drives in long-stored RS/6000s are prone to spindle stiction. Symptoms: drive does not spin up; faint "thunk" on power-on but no rotation; SCSI BUS RESET timeouts.

Field fix (for one-time data recovery only):

  • Power off.
  • Pull the drive sled.
  • Gently rotate the drive case 45–90° around its spindle axis to free the heads.
  • Re-install. Power on.

Image the drive immediately to a modern SCSI-to-USB or SCSI2SD setup. The drive cannot be trusted to spin up reliably again.

SCSI Cable Replacement

edit

The original IBM Centronics-50 / HD-50 SCSI cables in early RS/6000s develop intermittent contact in the IDC connectors. Replace with modern shielded cables (SCSI-3 cables are backward-compatible with SCSI-2 drives).

SCA-Tray Drives (7026 / 7044)

edit

Hot-swap SCA-tray drives in the 7026 H-series and 7044-270 require the drive carrier to be fully inserted in the bay to register on the bus. Aged carrier latches can cause intermittent drive disappearance — clean the SCA connector with contact cleaner.

Reference Diskette / Reference CD

edit

Pre-CHRP RS/6000s (7011 / 7012 / 7013 / 7015) used per-machine Reference Diskettes for SETUP, similar in concept to the PS/2 line. CHRP RS/6000s (7025 / 7026 / 7043 / 7044) use the built-in SMS utility on Open Firmware — no Reference Diskette required.

The 7248 (PReP) uses a per-board firmware that boots to a text setup screen at POST.

Cooling System

edit

Restoration points:

  • All RS/6000s with the original PSU fan are 30+ years old by now; the sleeve bearings are due for replacement. Listen for fan rattle or stutter on power-on.
  • The 7013 500-series and 7015 rack systems require positive airflow to keep the MCM heatsink within thermal limits. A failed CPU fan on these can cause CPU overtemperature shutdown — visible in the LED panel as a thermal-class error.
  • The 7044-270 PSU has redundant fans; one fan failure does not stop operation but is logged in the service processor diagnostic log.

Connector Care

edit
  • Keyboard / mouse — RS/6000 desktops use either PS/2 Mini-DIN-6 ports (later submodels) or older AT-style 5-pin DIN (early submodels).
  • Serial — DB-9 male on most submodels (RS-232C).
  • Parallel — DB-25 female on submodels with parallel port.
  • SCSI — Centronics-50 (early), HD-50 / HD-68 (later), SCA-2 80-pin (hot-swap).
  • Ethernet — AUI on early submodels; 10Base-T / 100Base-TX RJ-45 on later submodels.
  • Graphics — 13W3 connector (IBM GXT family; same as Sun) on most RS/6000 GXT cards; 15-pin DSUB on 7248 integrated graphics.

The 13W3 connector requires a 13W3-to-15-pin VGA adapter for use with modern monitors.

edit
  • Philips #2 screwdriver and T15 Torx.
  • Anti-static strap.
  • Digital multimeter.
  • IPA + foam swabs.
  • Soldering iron with fine tip + solder wick.
  • Hot-air rework station for SMD electrolytic removal on 7012 / 7013 planars.
  • USB microscope for inspecting SMD cap leakage.
  • SCSI2SD or BlueSCSI for replacement of original SCSI drives.
  • 13W3-to-VGA adapter for monitor work.
  • Spare Dallas DS12887+ or Glitch Works / Necroware NVRAM replacement.
  • Per-machine IBM Service Guide PDF (SA38-0531, SA38-0512, etc.).
edit

References

edit