SGI Indigo Troubleshooting Guide

The rear of an Indigo. The supply fan is behind the grille above the power switch and sockets; if it does not turn, SGI treats the supply as failed (Retro-Computing Society of Rhode Island, CC BY-SA 3.0)

Fault-finding on the SGI IRIS Indigo, R3000 and R4000 models, from the troubleshooting chapter of SGI's IRIS Indigo Workstation Owner's Guide (007-9065-050), with notes from the TechPubs Wiki and owners' repair reports. SGI's procedures stop at the replaceable part: power supply, CPU board or module, graphics board, SIMM, keyboard or mouse. Switch off and unplug both power cables before removing the front cover.

Front LED

When the machine is switched on, the LED on the front turns amber while the power-on diagnostics run for about 20 seconds, then turns green, and IRIX boots.

Power-on symptoms (SGI Table 6-1)

Symptom	System state	Cause
No LED	No power	Power supply failure: check the fan at the rear. Check that the system disk LED flashes during power-on. Possible CPU failure.
Solid amber	–	CPU board or graphics board failure
Flashing amber	Memory diagnostic failure	SIMM failure; check the error messages for the faulty SIMM
No boot tune	No audio	Faulty speaker, faulty CPU board, or volume at or near zero
Solid green	System menu shown on booting	Diagnostics passed
No LED on system disk	System does not boot	Drive not fully in its slot, or drive failure
Diagnostic error messages, no login icons	Diagnostic failure, no graphics	CPU, graphics board or monitor failure

No LED

1. Check that the cables are firmly connected and that the outlet works, and turn up the monitor's brightness.
2. Look at the fan behind the grille above the power sockets at the back, and listen for it. If it is not running, SGI's guide treats the supply as faulty.
3. Take the supply out and check it; see General Maintenance and SGI Indigo Capacitor Replacement Guide.

Blinking amber LED

One or more SIMMs have failed or are not seated. With no message, reseat all the SIMMs; if the LED still blinks and the screen stays blank, SGI's guide says to replace all four SIMMs of the group. The messages are:

• No usable memory found. Make sure you have a full bank (4 SIMMs): reseat the SIMMs, then replace them if the message returns.
• Check or replace: SIMM#, where # names the SIMM that failed, for example A1 or B1: reseat that SIMM, then replace it if the message returns.
• Memory is not usable. Check or replace all SIMMs.: reseat all the SIMMs, then replace them all if the message returns.

Each group must hold four SIMMs of the same capacity, and R3000 and R4000 SIMMs are not interchangeable.

Solid amber LED

The memory has passed but another part has failed.

• A "bing" from the speaker means the graphics board failed its test. Take the board out and refit it with the white ejector levers properly secured. If there is still no message on the monitor, the board is faulty.
• With no "bing", check that the monitor is on and its brightness is up, and that the system disk is fully in its slot, with the green lever below it fully to the right and lined up with the green line on the drive. If there is still no message, SGI's guide points to the graphics board.
• Check or replace: Graphics board.: reseat the graphics board, then replace it if the message returns.
• Check or replace: CPU board: reseat the CPU board with its ejector clips secured, then replace it if the message returns.
• Lines or "measles" on the screen, or a picture that looks wrong, with the cables and brightness checked: graphics board.

Amber LED, no chime and a low +12 V rail

An owner's R4400 Indigo stuck at an amber LED with no chime, no drive activity and a "no keyboard" message, although it gave a picture. The +12 V rail measured about 2.2 V at a drive connector, although the supply gave 12.3 V at its own output. He found a badly cracked solder joint on the +12 V pin of the supply-to-backplane connector, and later a screw on the back of the test supply that stopped it seating fully in the connector. Reflowing the joint and seating the supply fully cured it.

Green LED but the keyboard or mouse does not work

The mouse plugs into the keyboard, which has a socket on each side.

• If the mouse works but the keys do not, switch off, check the keyboard cable and try again. If the keys still do not work, move the mouse to the keyboard's other socket and connect the keyboard to the machine from the side the mouse was on. If both then fail, the keyboard is faulty.
• If the mouse does not work, check both cables, then swap the sides in the same way. If the mouse still does not work it is faulty; if it works after the swap, the keyboard is faulty.

The Indigo needs its own keyboard and mouse; PS/2 devices need an adapter.

No picture

gfx=dead

If the PROM environment shows gfx=dead, the graphics console is not working, which does not necessarily mean the graphics board is faulty. One owner saw gfx=dead (No keyboard) on a machine with no keyboard connected.

An owner of an R3000 Indigo with Entry graphics found the graphics board missing from hinv and IRIX reporting cannot open console "gfx(0)", with a second, replacement board behaving the same way. Forum members suspected the backplane. He found that it has only passive parts apart from the EEPROM, with no electrolytic capacitors, and that the voltages on it were correct. weblacky suggested testing the backplane and its connectors for cracked solder joints with a meter. The thread ends without a fix.

Serial console

The serial ports are 8-pin mini-DIN sockets, wired as follows according to the IRIX serial(7) manual page:

Indigo serial port pins (IRIX serial(7))

Pin	Signal	Pin	Signal
1	DTR, data terminal ready	5	RD, receive data
2	CTS, clear to send	6	RTS, request to send
3	TD, transmit data	7	DCD, data carrier detect
4	Signal ground	8	Signal ground

One forum member says a three-wire null-modem cable should work at 9600 baud, but that on the older 4D systems some very early diagnostic output is lost without the handshaking lines. With the PROM variable console set to d, the power-on diagnostics message and the prompt to press Esc for the PROM also appear on the serial port. One owner found that disconnecting the keyboard and mouse was enough to get the PROM on the serial port; another had to remove the graphics board on his machine.

Can't set tod clock

The machine repeats "Can't set tod clock" while booting. This normally means the clock battery is flat. It is not fatal, and some machines recover after running for a while. The message does not appear once the system is running. Replace the battery; see General Maintenance.

Bad Ethernet address

The system reports a bad Ethernet address, ff:ff:ff:ff:ff:ff. Either the 93C56 EEPROM on the backplane that holds the address has failed, or an IP12 board has been refitted to a machine that has had an IP20 in it, which erases the address. IRIX then reports errors for the ec0 network interface.

The address can be set only from an IP12's PROM, so fit the IP12 board for this step even if the machine will run with an IP20. Following the owner who fixed it:

1. Enter the PROM Command Monitor.
2. Type eaddr followed by the address, without setenv, for example eaddr 08:00:69:12:34:56. The address should begin with 08:00:69, SGI's prefix; one beginning 08:00:06 was rejected as a bad address. Use a value no other machine on the network has.
3. Restart and configure the network in IRIX.

If the chip itself has failed, replace it; it is in a socket, at U1 on one owner's backplane. That owner replaced the chip first, but the address only took once he entered it with the eaddr command. The same owner found that swapping CPU boards had also cleared the PROM password.

Memory controller warning (R4000)

An R4000-family machine may warn at boot that a revision C memory controller is needed for the SIMMs fitted. If hinv shows all the memory and there are no memory errors, the TechPubs Wiki says the warning can be ignored. See SGI Indigo R4000.

Cannot boot from CD-ROM

The IP12's PROM does not recognise optical drives. CD-ROM drives that worked with it had special firmware, set by a jumper, that made them appear as a hard disk. The IP20 works with a wider range of drives.

Power supply faults reported by owners

• A residual-current (GFI) outlet tripping when the Indigo was switched on. The owner opened the supply and blew the dust out. The fault went away when he used a different GFI outlet, and the supply then ran for two weeks without trouble. A forum member worked out that the mains filter's Y capacitors can pass several milliamps to earth, enough to trip the most sensitive outlets.
• On an ITT PEC4044B, corroded rectifier diodes CR207 and CR209, marked FE5D, on the daughterboard; the owner bought FE6D replacements. Another PEC4044B worked again after a shorted diode, CR106, was replaced.
• A PEC4044B whose fan did not run after a recap, with 12.3 V on its 12 V rail; the fan worked in another supply.
• An R4400 supply that failed outright. Its owner notes that it needs a load on the 5 V rail before it will start on the bench.
• A supply that one owner believes he damaged by fitting an R4000 board to a machine with an R3000 supply.

See SGI Indigo Capacitor Replacement Guide.

Diagnostics

IDE tests

The Interactive Diagnostic Environment tests take about 30 minutes, and about 15 minutes more for each extra hard disk. Esc stops them at any point.

1. Choose System Shutdown from the System toolchest, and click Restart in the notifier. If the machine is not responding, press the Reset button behind the front door.
2. When the Starting Up the System notifier appears, click Stop for Maintenance or press Esc. If the System Maintenance menu cannot be reached, SGI's guide says the system is faulty.
3. Click the Diagnostics icon or type 3. If the diagnostics banner does not appear, the disk may be faulty.
4. Wait. The front LED blinks slowly and the screen blanks during the graphics tests. A healthy machine reports "Diagnostic tests completed with no failures detected." A failure is reported as, for example, "ERROR: Failure detected on the CPU module", or names the CPU baseboard, audio board or graphics board.
5. Press Enter to return to the menu, and choose 1 to restart.

Confidence tests

For a keyboard, mouse, monitor, audio or SCSI device other than a hard disk that is not working, run the confidence tests from IRIX: choose Run Confidence Tests from the System toolchest. There are tests for audio, the keyboard, monitor, mouse, quarter-inch and DAT tape drives, floppy drive, CD-ROM drive, dial and button box, digitising tablet, genlock and Spaceball.

Related pages

• SGI Indigo
• SGI Indigo R4000
• SGI Indigo General Maintenance
• SGI Indigo Capacitor Replacement Guide
• SGI Indigo Documentation
• Capacitor Failure Symptoms