IBM PS/2 Model 80 Capacitor Replacement Guide

IBM PS/2 Model 80 (8580) system unit.

Capacitor work on the IBM PS/2 Model 80 (machine type 8580) is mostly on the power supply and the diskette drive. The 16 MHz and 20 MHz system boards use dipped tantalum capacitors and have no aluminium electrolytics; the 25 MHz board also carries a few small radial electrolytics. IBM's service documents for the Model 80 identify parts at FRU level only and give no capacitor values, and the third-party schematics of system board 90X9152-01 have no parts list. The capacitor values below come from Eric Schlaepfer's reverse-engineered schematic of one supply and from published drive repairs.

Safety

The supply rectifies the mains, and its reservoir capacitors hold a charge after the machine is unplugged. In the 225 W supply P/N 72X6665 they are C5 and C6, each 1000 µF 200 V, connected in series, with seven 220 kΩ balancing resistors in parallel across each (R100-R106 and R108-R114). Before working inside the supply:

1. Unplug the mains lead.
2. Discharge C5 and C6 individually through a suitable resistor across each capacitor's own terminals.
3. Measure both with a meter before touching the primary side.

There is no CRT in the system unit. IBM's displays are separate units; see CRT Discharge Procedure.

System board

The three system boards are built differently, so identify yours before ordering parts. Ardent Tool's photographs of the 16 MHz Type 1 board (FRU 33F8416, P/N 72X7492) and of a 20 MHz Type 2 board show orange dipped tantalum capacitors and surface-mount ceramic chips, with no aluminium electrolytic capacitors. The 25 MHz Type 3 board is built differently. Ardent Tool's photographs of both versions (FRU 34F0022, P/N 33F5571, and FRU 85F0046, P/N 91F7513) show surface-mount moulded tantalums and four small radial aluminium electrolytics: C2 and C4 near the top edge, C9 beside the power connector J16, and one beside the diskette connector J18. Their values cannot be read from the photographs. No published source gives values for any of the system-board capacitors, so read the value off any part you take out. A shorted tantalum pulls its supply rail down and can make the power supply shut down; because every part on a rail is in parallel, lift one end of a suspect part and measure it out of circuit before replacing it. See Capacitor Failure Symptoms.

The common board fault on the Model 80 is dust, not capacitors. Peter Wendt reports that dust drawn in by the supply fan settles on the part of the board holding the 8042, the MC146818A clock, the diskette controller, the NS16550 UART and the clock crystal, and causes lost configuration, a wrong clock and read/write errors. Cleaning is described on the maintenance guide.

Power supply

Supplies fitted to the 8580

Rating	Identification	Component values
225 W	Earlier 8580s. P/N 72X6665 (PEC 3978, red switch); FRU 15F6548 on Ardent Tool's list; a Schrack-built unit labelled P/N 90X9109 and 15F6551	Reverse-engineered schematic of the 72X6665
242 W	Later 8580s	None published
250 W	FRU 57F1600, the replacement part in IBM's 1990 and 1994 parts lists	None published

Supplies with the same IBM part numbers were built by different makers. Al Savage's failed 225 W unit was a Schrack EP071063-A with screwed case halves; Ardent Tool also records a Plessey unit with the same numbers and riveted halves. Read the values off your own board.

The 225 W supply, P/N 72X6665

Schlaepfer's notes describe the 72X6665 as an LLC resonant converter with a separate bias supply for its SG3524 controller. RT1, RT2, R117 and R118 limit the inrush current into the series reservoir capacitors. The bias supply has a triac preregulator that keeps its output about the same on 120 V and 220 V mains. All three outputs are watched for over-voltage, which latches the supply off, and under-voltage, which removes Power Good; the front power light is driven by Power Good, and the fixed-disk light by a signal from the system board.

Electrolytic capacitors in the 225 W supply P/N 72X6665 (from the reverse-engineered schematic)

Designator	Value	Circuit
C5, C6	1000 µF 200 V	Primary reservoir, in series
C20	2200 µF 16 V	120 V/220 V preregulator
C56, C57	10 µF 50 V	Preregulator
C55	470 µF 35 V	Bias supply
C7	470 µF 35 V	SG3524 supply
C13, C14	22 µF 25 V	Power-stage base drive
C34	10 µF 50 V	Power good
C40, C16	4700 µF 10 V	+5 V output filter
C41	470 µF 35 V	+5 V output, after the chokes
C18	4700 µF 16 V	+12 V output filter
C17	470 µF 35 V	−12 V, before the LM7912 regulator
C19	47 µF 25 V	−12 V, after the regulator

The same sheet shows C39 on the +12 V output with no value. The other capacitors on the mains side are 560 pF line-to-earth capacitors at C3, C4, C52 and C53 and unmarked across-the-line capacitors at C1 and C2; replace any of these only with X- or Y-rated safety capacitors of the same value. C8 and C9 (1 µF 250 V) are the power stage's split capacitors, C10 (0.001 µF 1 kV) is a snubber and C26 (0.1 µF 500 V) sits in the preregulator. These values apply to the 72X6665 only.

A documented failure

Al Savage's supply failed with a chirp that did not change when he disconnected the outputs. The cause was a shorted 3300 µF 16 V output filter capacitor, the upper one of an identical pair, on a Schrack board with no component legend; he also found a cold solder joint. That value is not on the 72X6665 schematic, which is a different design.

Recapping the supply

1. Check the power switch first. Peter Wendt reports that about half the failed Model 80 supplies he saw had a broken switch, the red ones more often than the white.
2. Remove the rear drive carrier from the drive support structure, unplug the 4-pin drive power leads and the system board power plug, take out the three screws holding the supply and pull it out.
3. Discharge and measure the reservoir capacitors.
4. Open the case. Al Savage's Schrack unit used security Torx screws (with a pin in the centre) and had a short ground lead through the top cover that must be disconnected; the cover slides about an inch sideways and then lifts off.
5. Photograph both sides of the board and note every capacitor's value, voltage, size and polarity before removing anything.
6. Fit parts of the same capacitance and the same or higher voltage rating, 105 °C, and low-ESR where the original was.
7. Test the supply against IBM's test voltages on the maintenance guide before refitting it.

A crackle from a switched-off Model 80 was discussed on Ardent Tool; Peter Wendt traced similar noises to the switch (the front switch cover acting as an echo chamber), to an insulation fault in the AC line filter, and on another machine to a loose mains lead.

Diskette drive

The 8580's diskette drives take their power through the interface. The -Axx machines use a pin-connector drive; the earlier models use the card-edge drive P/N 90X6766, FRU 72X8523. Al Savage found that all three of his 8580 drives would not read diskettes, and repaired both Mitsubishi units by replacing surface-mount electrolytics that were open-circuit or very high in ESR. On the Mitsubishi MF355W-99M3 he marked five:

Replaced capacitors, Mitsubishi MF355W-99M3 (Al Savage)

Value	Quantity
22 µF 16 V	2
1 µF 50 V	1
0.22 µF 50 V	2

Al Savage's photograph of the MF355W-99M3 logic board with the five failed capacitors marked. (Image: Al Savage, via Ardent Tool of Capitalism)

The ibmfiles.com repair of a Mitsubishi PS/2 drive replaced two surface-mount electrolytics with 10 µF through-hole parts soldered to the pads, so boards differ; read the values off your own drive. In an exchange with Alan Douglas reproduced on Ardent Tool, repairers used surface-mount electrolytics for the 22 µF and 1 µF positions and ceramics for the 0.22 µF positions, since 0.22 µF electrolytics are hard to find, and tantalum chips for the others. The same exchange recommends cutting the old capacitors in two with flush cutters, removing the rubber seal and working on the leg stubs from above, because pulling or overheating a part can lift the pad. Jim Shorney reports that leaked electrolyte contaminates the joints so that the solder will not melt and the leg pulls away first.

Not every drive fault is a capacitor:

• A 162 with the drive dead at POST is usually mechanical. Peter Wendt lists a sticking head carriage, a dirty track-0 sensor and a loose head damper pad.
• Mitsubishi's MF355C specification describes a brushless DC motor turning the spindle directly at 300 rpm, so a drive of that design has no belt.
• Dirt on the diskette controller on the system board, under the supply fan, gives 6xx errors that look like a failed drive.

Fixed disks and memory

IBM's documents list the Model 80's fixed disks, disk adapters and memory cards as FRUs only. Before suspecting capacitors on the disk side, check the cables: IBM retain tip H064300 says to check the fixed-disk cables first for a 1780 on an 8580-041. Memory errors and their procedures are on the maintenance guide.

After the work

1. Test the supply on its own against IBM's test voltages, then refit it.
2. Switch on with the Reference Diskette in drive A, run Automatic Configuration if anything changed, and run the Advanced Diagnostics.
3. Check that the green power light comes on; it is driven by Power Good.
4. Format and verify a diskette if you worked on the drive.

If a fault appears straight after recapping, check the polarity of every capacitor you fitted.

Related pages

• IBM PS/2 Model 80
• IBM PS/2 Model 80 Maintenance Guide
• IBM PS/2 Model 80 Troubleshooting Guide
• IBM PS/2 Model 60 Capacitor Replacement Guide
• IBM PS/2 Model 70 Capacitor Replacement Guide
• Capacitor Failure Symptoms
• Battery Explosion, Capacitor or Corrosion Damage
• Floppy Drive Repair
• Recommended Tools