IBM PS/2 Model 50 Capacitor Replacement Guide
This guide covers capacitor diagnosis and replacement on the IBM PS/2 Model 50 (machine type 8550) โ the original 8550-021 and the Model 50 Z, 8550-031 and 8550-061. It covers the planar, the 94 W power supply, the diskette drive and the fixed-disk drives.
Safety Warning
[edit | edit source]The 94 W supply is a mains-rectified switching supply. Its primary side sits at mains potential and its reservoir capacitors hold a charge that can kill after the machine is unplugged. Recapping it is a reasonable thing to do, but treat it as live-mains work:
- Power off and unplug the mains lead.
- Discharge the primary reservoir capacitor through a 1 kฮฉ / 5 W resistor across its terminals, then confirm with a meter that it reads near zero before touching anything.
- Work one-handed where practical, and do not probe a powered primary side unless you are using an isolation transformer.
There is no CRT in the system unit; the matching IBM 8503, 8512, 8513 and 8514 displays contain their own tube and supply and are serviced separately. See CRT Discharge Procedure if you are working on the display.
The supply is universal and voltage-sensing โ there is no voltage selector switch to set wrongly.[1]
What Actually Fails on a Model 50
[edit | edit source]This is worth stating plainly, because the Model 50 is often lumped in with the PS/2 models that suffer surface-mount electrolyte leakage.
- The planar does not carry surface-mount aluminium electrolytic capacitors. Photographs of both the Model 50 planar (P/N 90X9161 / 15F6969) and the Model 50 Z planar (FRU 90X9533) show surface-mount tantalum chip capacitors and ceramics for supply-rail decoupling. There is no equivalent on these boards of the leaking aluminium cans found on the P70/P75 video board and planar.[2][3][4]
- The diskette drive does. The documented, repeatable electrolyte problem on PS/2 machines of this era is in the 3.5-inch drive's own logic board, where surface-mount aluminium electrolytics go open-circuit or high-ESR and leak onto the pads.[5][6]
- The fixed-disk logic boards do. The integrated drive-plus-controller ("DBA") drives fitted to the 8550-031 and 8550-061 carry the same style of surface-mount electrolytic, and corrosion around their pads has been documented.[6]
- The power supply is a normal 1980s switcher and ages like one, but IBM published no component-level data for it.
IBM never published capacitor values for any part of this machine. No parts list, technical reference, hardware maintenance reference or schematic set gives a capacitance or a working voltage for a planar, adapter or power-supply capacitor: IBM serviced these at FRU level and replaced whole assemblies.[7][8][9] Any table of "Model 50 capacitor values" you find elsewhere is therefore somebody's reading of a particular board, not a manufacturer specification. Read the values off your own board before ordering parts.
Planar Capacitors
[edit | edit source]
The Model 50 and 50 Z planars use surface-mount tantalum chip capacitors (the amber or gold moulded blocks and beads visible across both boards) together with surface-mount ceramics. Tantalums fail short far more often than they fail open, and a shorted decoupling tantalum drags its rail down.[2]
IBM did not mark these positions with values on the silkscreen, and no published designator-to-value map exists for either planar. Do not fit a replacement by reference to a value quoted for a different IBM machine โ the widely-cited "10 ยตF / 16 V marked 106 16V" figure is documented for IBM 51xx (PC, XT, AT) motherboards, which use through-hole bead tantalums, and has never been shown to apply to a PS/2 planar.
Finding a shorted tantalum
[edit | edit source]- Unplug the machine and discharge the supply.
- Remove the planar. Take out the adapter cards, the diskette drive, the fixed disk and the drive riser card; on the Model 50 also remove the fixed-disk adapter and the exhaust fan assembly. The drive-structure removal tool is stowed under the battery/speaker assembly and is used to lift the white plastic snaps holding the drive structure down โ including the one hidden under the speaker. Three screws hold the planar itself, and a centring pin at the rear locates it, so lift the planar straight up rather than at an angle.[10]
- Set a multimeter to its lowest resistance range or to diode test and probe from ground to the rail side of each tantalum. A healthy part settles to a high reading after a brief charging pulse; a shorted one sits near 0 ฮฉ.
- Everything on a given rail is in parallel, so a low reading points at a rail rather than at one part. Lift one end of the suspect capacitor and re-measure before condemning it.
- Replace with an equivalent surface-mount tantalum or a low-ESR ceramic of the same value and equal or higher working voltage, read from the part you removed.
Removal and replacement
[edit | edit source]Leaking or failed surface-mount capacitors contaminate their own solder joints. Jim Shorney's account of PS/2 capacitor work is worth heeding: the joints often refuse to reflow, and the legs pull out of the contaminated solder before it melts. Work at a controlled temperature, add fresh leaded solder and flux to each joint first, and try to desolder rather than twist โ but expect some parts to come away without the solder ever melting. Clean each pad and check it is still bonded to the board before fitting the new part.[4]
Match polarity to the board. A tantalum fitted backwards will usually short on first power-up.
Power Supply (94 W)
[edit | edit source]The Model 50 and 50 Z share one supply: an ASTEC AA13621, IBM P/N 90X9527, IBM FRU P/N 90X9366, rated 94 W maximum continuous output.[11][7] It is the same physical family as the Model 70 supply but with no built-in blower โ on the Model 50 the fan is a separate assembly that plugs into the planar and exhausts through the rear panel.[11]
Opening it: the case is held by security Torx screws, five on the side and two on the top, and opens with a T10 security bit. There is no ultrasonic welding and no glue.[11]
What is published, and what is not
[edit | edit source]| Item | Value | Source |
|---|---|---|
| Rating | 94 W maximum continuous output | IBM announcement letter; IBM HMR |
| Input | 100โ125 V or 200โ240 V, automatically selected, 50/60 Hz | IBM HMR |
| Outputs | +5 V, +12 V, −12 V (there is no −5 V rail) | IBM HMM pinout |
| Internal fuses | Two (F1, F2) โ a dead supply with no outputs usually means one has blown | Ardent Tool |
| Capacitor values | Not published by IBM, and no reverse-engineered schematic of this unit is known to exist |
Note the last row. The 94 W ASTEC unit has not been reverse-engineered and documented the way the tower supply has, so a recap means reading each can before you remove it. Photograph the board, write down every value and working voltage, and order to match.
Recapping the 94 W supply
[edit | edit source]- Discharge the primary reservoir capacitor and confirm with a meter.
- Remove the supply: two screws at the back of the case, one screw at the front inner corner, then pull it off the planar edge connector.[11]
- Open the case (T10 security Torx).
- Photograph the board from both sides before touching anything. Record each electrolytic's value, working voltage, diameter, lead spacing and polarity.
- Replace like for like: same capacitance, equal or higher working voltage, 105 ยฐC rating, low-ESR parts on the secondary side where the originals were low-ESR types.
- Inspect the mains-side X- and Y-class suppression capacitors. These are safety-rated parts and must be replaced with equivalently rated X or Y class parts, never with ordinary film capacitors. Any RIFA-branded part showing crazing of the epoxy should be replaced on sight; they crack, absorb moisture and eventually vent.
- Check both fuses for continuity before reassembly.
- Reassemble and bench-test the outputs under a modest resistive load before refitting it to the machine. The connector is a 50-pin edge connector: pin 1 is −12 V, odd pins 3โ13 are +12 V, odd pins 15โ47 are +5 V, even pins 2โ48 are ground, pin 49 is System Status and pin 50 is Power Good.[11]
IBM's own service limits, measured at the planar connector, are given on the maintenance page; a rebuilt supply that sits outside them has not been fixed.
Diskette Drive Logic Board
[edit | edit source]This is the repair most Model 50s actually need.
PS/2 3.5-inch drives take power and signals through a single interface with no separate power connector, and were supplied by several manufacturers โ Mitsubishi, ALPS, Sony and YE Data โ under the same IBM part numbers (P/N 90X6766, FRU 72X8523 for the card-edge type used on the Model 50). The drives are physically different inside and most parts do not interchange, though they behave identically at the interface.[5]
Al Savage's teardown of failed PS/2 Mitsubishi drives found the cause to be surface-mount electrolytic capacitors on the drive's logic board that had gone open-circuit or very high-ESR. Replacing them recovered the drives. Reported values differ between drive revisions โ one recap account names 22 ยตF, 1 ยตF and 0.22 ยตF parts, another replaced two capacitors with 10 ยตF โ so read the values off your own board.[5][12][6]
Symptoms that point at the drive's capacitors rather than its mechanics:
- Will not read any diskette, including known-good ones written elsewhere.
- Intermittent read and write errors that worsen as the drive warms.
- A 6xx diskette error at POST. Mitsubishi drives in particular are noted for capacitors that produce 6xx electronic errors.[12]
- Brown crust or discoloured, grey, rough solder joints around a can.
Symptoms that are not capacitor faults, and should be chased mechanically first:
- A 162 at POST with the drive otherwise dead. During POST the heads are driven off track 0 and back within a fixed time; a sticking head carriage, a dirty track-0 photo-sensor or a detached head cushion will miss that window and the drive is thrown out of the configuration. The ALPS drives are known for losing the damper cushion glued to the head assembly; the Mitsubishi 355 collects dirt around the read/write head actuator spindle.[5][12]
- Speed or timing complaints blamed on a belt. There is no belt. The Mitsubishi MF355C specification states that the drive uses a DC brushless direct-drive motor coupled straight to the spindle at 300 rpm. Nothing in a PS/2 3.5-inch drive is belt-driven, and any guide that tells you to fit one is wrong.[13]
Because the drive takes its supply through the interface, a drive fault can also be a supply fault โ IBM lists the power supply among the FRUs for unclassified 6xx errors.[7]
Fixed Disk Drives and Adapters
[edit | edit source]The Model 50 family uses three different arrangements, and they are not interchangeable:[14]
| Machine | Drive | Adapter |
|---|---|---|
| 8550-021 | 20 MB MFM / ST-506, FRU 72X8522 | ST-506 adapter in the dedicated adapter slot, FRU 90X8070 (early) or 90X8642 (later) |
| 8550-031 | 30 MB integrated drive-plus-controller, FRU 6128287 | Bare passive riser, FRU 90X9441 |
| 8550-061 | 60 MB integrated drive-plus-controller, FRU 6128294 | Bare passive riser, FRU 90X9441 |
| 8550-021 with the 60 MB upgrade | 60 MB integrated drive-plus-controller | Riser carrying its own BIOS ROMs, FRU 90X9571 |
The units IBM's error codes call "ESDI" on the desktop machines are not true ESDI drives on a separate controller card. They are a hard disk and a Micro Channel controller built as one physical assembly, plugged into a riser that is mostly or entirely passive; the bare 90X9441 riser is an empty board that only routes signals. True ESDI controllers and separate ESDI drives belong to the towers โ the Model 60 and Model 80.[14]
The practical consequences for capacitor work:
- On the 8550-031 and 8550-061, the capacitors that matter are on the drive's own logic board, not on the riser. These are surface-mount electrolytics of the same family as the diskette-drive parts, and corrosion around their pads has been photographed on PS/2 DBA drives.[6] Read the values from your board.
- On the bare 90X9441 riser there is nothing to recap: the board is a passive interconnect. One variant, circuit board 15F8434, carries unpopulated silkscreen outlines for two capacitors that IBM never fitted โ do not read those as missing parts.[14]
- On the 8550-021 ST-506 adapter and the 90X9571 riser, no published capacitor list exists. Inspect, and replace only what is visibly failed or measures bad.
If an integrated drive reports a 104xx error, note which one before reaching for a soldering iron: 010455 and 010456 are controller errors, 010462 a controller seek error, 010464 a drive read error, and 010483 points at the adapter and the system board โ and, for older drives, at spindle stiction rather than electronics.[15][16] For the ST-506 machine, 1702 indicts the adapter specifically while the rest of the 17xx range points first at the drive, then the cable, then the adapter, the planar and the power supply.[15][7]
After a Recap
[edit | edit source]- Bench-test the supply on its own first, under load, against IBM's service limits.
- Refit and start from the Reference Diskette rather than the hard disk.
- Run Automatic Configuration if any adapter changed, then run Advanced Diagnostics (from System Programs, press Ctrl-A at the main menu).[15]
- A successful POST ends in one short beep.[8]
- Leave the machine running a long memory test to shake out marginal decoupling.
If something fails immediately after a recap, check the polarity of every part you fitted before suspecting anything else.
When Not to Recap
[edit | edit source]A Model 50 that starts reliably, POSTs clean, holds its configuration and reads and writes diskettes does not need its planar touched. Blanket-replacing surface-mount tantalums on a working board risks lifting pads for no gain.
Do act if you see:
- Any fluid, crust or crystalline residue near a capacitor, or discoloured, grey or rough solder at its pads.[4]
- A rail outside IBM's service limits, or Power Good never asserting (the green front-panel light stays off).[8]
- Diskette read/write failures or 6xx errors on a mechanically sound drive.
- A supply that whines, smells, or will not start when warm.
- A shorted decoupling capacitor confirmed out of circuit.
Related Pages
[edit | edit source]- IBM PS/2 Model 50
- IBM PS/2 Model 50 Maintenance Guide
- IBM PS/2 Model 50 Troubleshooting Guide
- IBM PS/2 Model 60 Capacitor Replacement Guide โ the tower, which uses a different and better-documented power supply
- Capacitor Failure Symptoms
- Battery Explosion, Capacitor or Corrosion Damage
- Floppy Drive Repair
- Recommended Tools
References
[edit | edit source]- โ IBM Announcement Letter 187-049, "IBM Personal System/2 Model 50", 2 April 1987, reproduced at ardent-tool.com. Source for the 94-watt automatic voltage-sensing power supply, the four 16-bit I/O slots with one occupied by the disk controller, the 1 MB planar memory expandable to 7 MB, the battery-backed clock and the IBM specification of the 1.44 MB diskette drive (300 rpm ยฑ1.5 %, 500 kbps, 80 tracks).
- โ 2.0 2.1 8550 Planar, Ardent Tool of Capitalism. Component-designator map for the Model 50 planar (P/N 90X9161 and 15F6969), including F1, the battery/speaker connector, the MC146818A RTC at ZM47 and the slot layout.
- โ 8550 Z Planar, Ardent Tool of Capitalism. Component-designator map for the Model 50 Z planar (FRU 90X9533), including F1, the battery/speaker header, the MC146818A RTC at ZM8, the single 72-pin SIMM socket and the 85 ns zero-wait-state memory.
- โ 4.0 4.1 4.2 8573 P75 Failing Capacitors, Ardent Tool of Capitalism. Jim Shorney and Peter Wendt on the behaviour of leaking PS/2 surface-mount electrolytics: discoloured, grey or rough solder joints next to the can, joints that will not reflow because the electrolyte has contaminated them, and legs that pull free before the solder melts.
- โ 5.0 5.1 5.2 5.3 8580 Diskette Drives, Ardent Tool of Capitalism, content by Fred Spencer and Al Savage. Source for the PS/2 diskette drive part numbers (P/N 90X6766, FRU 72X8523 for the card-edge drives), for Al Savage's finding that the failing parts are surface-mount electrolytics that go open-circuit or high-ESR, for the diskette-controller dirt problem and for Peter Wendt's list of mechanical causes of a 162 at POST.
- โ 6.0 6.1 6.2 6.3 PS/2 Surface Mount Capacitors, ibmfiles.com. Community teardown source; documents surface-mount electrolytic failure and pad corrosion on PS/2 Mitsubishi diskette drives and on the DBA hard-disk logic boards, and the through-hole-on-SMD-pad replacement technique.
- โ 7.0 7.1 7.2 7.3 IBM, IBM Personal System/2 Hardware Maintenance Manual, S52G-9971-02, October 1994 โ hosted on this wiki as File:IBM PS2 HMM S52G-9971 Oct 1994.pdf. Sections "Power Supply Voltage Check", "System-Unit Power Supplies", "Numeric Error Codes" and the Model 50 / Model 60, 65, 80 parts catalogues.
- โ 8.0 8.1 8.2 IBM, IBM PS/2 Model 50 Hardware Maintenance Reference, part number 64F3989, form number S64F-3989-00, March 1990 โ hosted on this wiki as File:IBM PS2 Model 50 HMR.pdf. Sections 1.2.2 "CMOS RAM", 1.2.7 "Diskette-Drive Controller", 1.2.8 "Keyboard Connector and Pointing-Device Connector", 3.1 "Power Supply" and 3.2 "Power-On Self-Test".
- โ IBM, IBM Personal System/2 Model 50 Technical Reference, May 1988 โ hosted on this wiki as File:IBM PS2 Model 50 Technical Reference May88.pdf. Figure 1-2 "System Board Devices and Features", Figure 1-7 "Physical Specifications" and Section 3 "Real-Time Clock/CMOS RAM".
- โ 8550 and 8570 โ Common Devices, Ardent Tool of Capitalism. Source for the cover removal and locked-case procedures and for the planar removal sequence, including the drive-structure removal tool stowed under the battery/speaker assembly and the planar centring pin.
- โ 11.0 11.1 11.2 11.3 11.4 8550 / 8570 / 756x โ Power Supply, Ardent Tool of Capitalism (Louis Ohland / Tomรกลก Slavotรญnek). Secondary source; it reproduces IBM's own power-supply connector pinout, test-voltage table and FRU data for the 8550. Source for the 50-pin edge-connector pinout, the ASTEC AA13621 identification, the 94 W rating, the security-Torx case fixings, the two internal fuses and the external fan part numbers.
- โ 12.0 12.1 12.2 60, 65 SX, and 80 โ Common Devices, Ardent Tool of Capitalism. Source for the side-cover and front-bezel removal, the dust-related configuration and disk faults under the power-supply fan, IBM retain tip H064300 (1780 on the 8560-041 and 8580-041), the ALPS/Mitsubishi diskette-drive comparison and the Mitsubishi "killer cap" note.
- โ Mitsubishi Electric, 3.5-inch Diskette Drive MF355C Specifications, UGD-0527A, February 1987 โ hosted on this wiki as File:Mitsubishi MF355C 3.5 inch Diskette Drive Specifications (UGD-0527A, Feb 1987).pdf. "The drive uses a DC brushless, direct-drive motor"; the spindle is directly coupled to the motor and rotates at 300 rpm. There is no belt in this drive.
- โ 14.0 14.1 14.2 8550 HDD Adapters, Ardent Tool of Capitalism. Source for the three Model 50 drive adapters (the ST-506 adapters FRU 90X8070 and 90X8642, the bare DBA-ESDI riser FRU 90X9441 and the DBA-ESDI riser with BIOS FRU 90X9571), for the drive FRU numbers and for Peter Wendt's explanation that the desktop "ESDI" drives are integrated drive-plus-controller units rather than true ESDI.
- โ 15.0 15.1 15.2 PS/2 Error Codes and the linked code tables 160โ199, 200โ299, 1500โ2999 and 10000โ11999, Ardent Tool of Capitalism. Secondary source; the tables reproduce IBM's own symptom-to-FRU lists.
- โ Known Problems โ Experience, RETAIN Tips, ECAs, Ardent Tool of Capitalism. Source for Peter Wendt's account of hard-disk stiction and the recovery options, and for IBM retain tip H005662, the adapter installation restrictions and configuration backup/restore caution for the 3270 Connection, Token-Ring and PC Network adapters in the 8550, 8555, 8560, 8565, 8570, 8580 and 9585.