IBM 5120 Troubleshooting Guide

IBM 5120 Computing System.

Fault diagnosis on the IBM 5120 Computing System follows the method in IBM's maintenance information manual for the machine, which IBM designates the 5110 Model 3. The 5120 is not PC-compatible. Its self-tests live in ROS on its own logic cards, and IBM's diagnostics identify a failing card by its position on the logic board (G2, H2, J2, L2 and so on). It uses the same processor card set as the IBM 5110; for the processor itself see IBM PALM processor.

IBM's field engineers worked from printed maintenance analysis procedures (MAPs) in a separate binder, SY34-0194, starting every call at Start MAP 200. The wiki does not hold that binder. The maintenance manual has the diagnostic charts, the bring-up halt letters, the error code lists and IBM's service tips.

Safety

Mains voltage is present in the power supply with the power switch off. Unplug the line cord from the wall before working on the supply. The internal diskette drives take mains through connectors P4 and P5, which are live whenever the machine is switched on.

IBM warns that if a TV monitor is attached to the video output and has not been properly modified, 110 V can be present on the coax BNC connector and cable shield, and on the frame of the 5120. Do not touch or unplug the coax while the TV monitor is plugged in.

See also CRT Discharge Procedure.

Power and grounding checks

Check the rails before chasing logic faults. IBM's limits, measured at the Y1 socket or at terminals E1 to E14 of power supply board A2, are +5 V 4.6 to 5.5, +8.5 V 7.9 to 9.35, +12 V 11.0 to 13.2, −5 V −4.6 to −5.5 and −12 V −11.0 to −13.2 under load. The full table, with the unloaded limits and ripple, is in the maintenance guide.

Power symptoms

Symptom	Probable cause	Action
Machine shut down by itself and will not restart	The TSR supply (board A2) has tripped on overvoltage (+12 V above +16 V), undervoltage (−5 V less negative than −3 V) or primary overcurrent	Switch off for at least 60 seconds before switching on again. If it trips again, measure each rail.
Supply dead on the bench	The supply shuts down when all loads are removed	Test with the logic connected, or with a suitable load.
Internal drives dead, computer runs	A blown fuse on diskette supply board A1: F2 (+5 V, 3 A), F3 (−5 V, 1 A) or F4 (+24 V, 3 A)	Find the short on the diskette side (drive, control card, cable) before replacing the fuse.
Readings that disagree with each other	Meter error	Check the meter against the +6 V zener reference at G2-S02 (ground at G2-D08) and scale readings by 6 ÷ the reference reading.

For intermittent faults IBM's grounding checks apply. Neutral to ground at the outlet should be less than 2 V AC; hot to neutral and hot to ground should both be about 110 to 120 V AC on a US supply. IBM notes that this does not detect a high-resistance earth. Check that every frame ground connection is clean and tight: on the 5120 these are the main base plate (left rear), the fan bracket to base plate, the power supply, the logic box, the power switch, the AC power box (three), the diskette unit motor and the TV monitor plug. If a line problem is suspected, IBM suggests an isolation transformer, checking that line-side and load-side wires in the AC box do not cross, and looking for other equipment on the same line.

Power-on and the bring-up diagnostic

The bring-up diagnostic runs from the executable ROS card (L2) every time the machine is switched on or RESTART is pressed. It tests the base logic cards, the keyboard and the display. If it completes, LOAD0 (BASIC) or CLEAR WS (APL) appears on the display after 25 seconds; the BASIC/APL switch on the front panel selects the language. A failure lights the PROCESS CHECK light or leaves a bring-up halt code on the display.

As each test runs, a letter is added to line 1 of the display. The halts occur in alphabetical order: a machine stopped at halt F did not reach halt G. If the display shows a jumble of random characters, the halt came before the diagnostic cleared the screen, that is, before halt A.

Bring-up diagnostic halt letters (IBM pp. 4-5 to 4-6)

Line 1 of display	Test	Cards and areas tested
Blank or jumbled	(before halt A)	G2, power, display
A	Bus in bit test	L2, H2, J2
AB	Bus out test	A2, B2, D2, F2, H2, J2, all I/O
ABC	Op code test	J2
ABCD	R/W storage and ROS switching test	H2, J2, L2
ABCDE	Interrupt 1, 2 and 3 tests	A2, B2, D2, H2, A1
ABCDEF	Device address test	B2, D2, F2, H2, J2, all I/O
ABCDEFG	Keyboard test	H2
ABCDEFGH	Storage test (05FF to end) and stuck key test	M2 to N4, H2, keyboard
...HJ	Executable ROS initialize	L2
...HJK	Executable ROS check	L2
...HJKL	Feature ROS check	K4
...HJKLM	Executable ROS check	L2
...HJKLMN	Feature ROS check	K4
...HJKLMNP	ROS content and CRC test	E4, F2
ABCDEFGHJKLMNPQ	Bring-up complete	L2

IBM's letter sequence has no I or O.

Because the bring-up tests do not loop, three of them (op code, read/write storage and ROS CRC) are repeated in the diagnostic control program, where they loop until an error occurs.

Diagnostic control program

The diagnostic control program (DCP) is in ROS and needs no diskette.

To load DCP normal mode after bring-up has completed: press and hold CMD, press HOLD, press − (minus) on the numeric keypad, and release CMD. In normal mode, D displays and A alters any location in read/write storage. To go on to diagnostic mode, press ATTN, hold CMD and press × (multiply) on the numeric keypad. To return to the customer's job, press ATTN, hold CMD and press + (plus).

In diagnostic mode, run a call test by pressing C, the test number and EXECUTE:

DCP call tests

Call	Test	Area tested
0	I/O diagnostic	Diskette read
1	I/O diagnostic	Diskette read
2	Op code test	H2, J2
3	R/W storage test	M2, M4, N2, N4
4	ROS CRC test	E4, F2
5	DCP I/O functions	A2, H2, J2, all I/O

Calls 0 and 1 load the diagnostics supervisor program (DSP) from the diagnostic diskette. The DSP runs the MAP diagnostic integrations (MDIs): 800 printer, 820 asynchronous communications and serial I/O, 821 BSCA, 871 internal diskette, 881 5114 diskette and 890 communications routines. Pressing ATTN once gives the MDI options, twice the DSP menu, and three times returns to DCP diagnostic mode. The diagnostic diskette is not on the wiki.

Error codes

Hardware errors appear on the last line of the display as ERROR xxx yzz, where xxx is the error code, y the device address and zz the subdevice address. Codes 800 to 999 are the CE diagnostic codes. BASIC and APL language errors are in the language reference manuals: the BASIC Reference Manual, SA21-9308, Appendix D, and the APL Reference Manual, SA21-9303, Appendix F.

Error code ranges and device addresses (IBM p. 4-80)

Device	Language	Error codes	Device address	Subdevice address
Printer	APL, BASIC	050–059	5	00
Asynchronous comm / BSC	APL, BASIC	081–099	8	00
Serial I/O	BASIC	002–014	A	02–40
Serial I/O	APL	002–014	0	30–34
Diskette	BASIC	001–045	D	80 drive 1, 40 drive 2, 20 drive 3, 10 drive 4
Diskette	APL	001–045	0	11 drive 1, 12 drive 2, 13 drive 3, 14 drive 4

A few diskette codes that point at hardware:

• 001, diskette timing error: the index pulse must be 166.7 ± 4.2 ms. Look at the drive motor, belt and index sensing.
• 013, 5114 diskette unit not present: the 5114's power switch may be off.
• 016 and 017, CRC errors on a matched ID or on a data read. A data sector is reread ten times before 017 is reported.
• 020 to 023, ID search failure and time-outs on ID, data read and data write.

The 5120 retries diskette and communications operations up to ten times before reporting an error. IBM describes how to disable the retries (set storage locations 00A2 to 00A5 from xxxx to 8xxx) when they are slowing a failing operation down.

Intermittent faults and process checks

The PROCESS CHECK light comes on when a parity error is detected in any of several functional units, and the machine stops at once with the error latched so that it can be traced with a logic probe. With the light on, do not disconnect the I/O cable and do not press RESTART: IBM lists the pin on each card that shows which error line is active, and the FRUs for each, on p. 4-61. After probing, IBM's order is to reseat all cards and cables, check the voltages, check the customer's AC ground, check the fans, and then replace boards.

IBM's tips for making an intermittent fault show itself while a diagnostic runs are to vibrate the machine, cards and connections, to raise the machine temperature by unplugging the blower for no more than 20 minutes, or to cool parts with circuit coolant.

Isolating by removal

Switch off before removing the read/write storage cards (M2, M4, N2, N4); IBM allows the other cards to be removed with power on. Cards and devices that can be removed when they are not part of the failing operation are the A2 I/O cable driver card, the B2 BSCA 1 card, the D2 asynchronous communications / serial I/O card, the APL card E4 (if the failure is in BASIC), storage cards M4, N2 and N4, the internal diskette adapter card, the A4 BSCA 2 card, BSCA attachment cards A4 and C2, the EIA communications interface card C2, the feature ROS card K4, the 5103 printer and the 5114 diskette unit.

The last cabled I/O device must carry a terminator. Without it the machine gives process checks and bring-up diagnostic errors. The 5103 printer has its own internal terminator.

Internal diskette drives

The two internal 8-inch drives each have an ac drive motor, a drive belt and a stepper-driven head/carriage. Run MDI 871 from the diagnostic diskette for a guided test.

• To see whether a fault follows the drive, swap the drives electrically by exchanging the two cable ends plugged into B4 and D4 of the B1 board.
• A drive that does not turn: check the ac drive motor, its capacitor (item 230) and its cable at P4 or P5, and the belt. Discharge the capacitor before handling its terminals.
• No DC to the drives: check fuses F2, F3 and F4 on board A1.
• Do not clean the heads. IBM says the head/carriage assembly is adjusted and tested at the factory and no part of it is to be repaired or cleaned. The service check sets the carriage at cylinder 40 with a timing pin and checks the pointer gap: a 0.51 mm gauge fits snugly and a 0.53 mm gauge does not.
• IBM's own tip for looping the diskette read diagnostic (call 0) while probing is on p. 4-99.

For general 8-inch drive work see Floppy Drive Repair.

Keyboard

During bring-up the keyboard test is halt G and the stuck key test runs with the storage test at halt H. The keyboard's PC board is cleaned with water and a lint-free cloth, and IBM says to clean only the flyplates that are dirty or causing failures. Remove the PC board only with no key pressed, or a flyplate will jump out of its module. The key code chart at item 250 (p. 2-72) gives the code each key produces.

Related pages

• IBM 5120 · IBM 5120 Maintenance Guide · IBM 5120 Capacitor Replacement Guide
• IBM 5120 Maintenance Information Manual · IBM 5120 Computing System Logic Manual
• IBM 5110 Troubleshooting Guide · IBM 5100 Troubleshooting Guide