VAX-11/780 Troubleshooting Guide

VAX-11/780 booting with a VT220 console (Patrick Finnegan, CC BY 2.0)

Fault-finding on the VAX-11/780 from DEC's VAX Maintenance Handbook (EK-VAXV2-HB-002) and Guide to VAX-11/780 System Troubleshooting (EY-2217E-SG-0001), a field service guide marked for internal use. On a machine being brought back to life, the console subsystem and the power system come first: nothing else can be tested until the LSI-11 console boots and the H7100 supplies all show POWER NORMAL. Power system details are on VAX-11/780 Power System Service.

DEC's method

DEC's guide sets out ten steps for field engineers: define the problem, test, change one thing at a time and record it, and repeat until solved. Its points that matter most to a restorer:

• Keep a system log book with every failure, every unit moved or swapped and the result.
• Return units that turn out not to be faulty to their original positions as soon as each test is over; DEC calls this the most often ignored step.
• A unit put in as a test may itself be faulty, and a system may have more than one fault; treat each separately.
• Declare the problem solved only after the system has run for at least twice the longest time it previously ran between failures.

Console will not start

DEC's console boot check, with the console terminal connected:

1. Switch DC off, then AC off. Push the LSI-11 HALT/ENABLE switch down to HALT.
2. Switch AC on, then DC on. Expect the DC ON LED on the LSI-11 panel to light, the terminal to print 173000 and the @ prompt, and the RUN light to flash.
3. Examine location 173000 (type 173000/); it should hold 000137. Examine 037776 (type 037776/).
4. Push HALT/ENABLE up to ENABLE, make sure console diskette ZZ-ESZAB is in the drive, and type 140200G. This runs the ROM-resident quick check of the console subsystem and, if it passes, boots the console program from the diskette.

If the boot fails, DEC's flowcharts use the ROM diagnostic listing (ESKAA) to locate the failure.

Console start-up faults (DEC console DC ON and Examine 173000 flowcharts)

Symptom	DEC's interpretation
DC ON LED off	Go to the console power flowchart (below)
RUN light flashed, nothing printed	One of the LSI-11 bus signals (BDMR, BREF, BSACK, BDOUT, BRPLY, BDIN, BSYNC, BDMG0, BDAL1) is stuck asserted, or BPOK H is not asserted
RUN light flashed, printout garbled	Check that BDAL0 is asserted and go to the terminal flowchart
A location and @ printed, then nothing	CPU jumpers wrong or LSI-11 CPU faulty
RUN light did not flash	Remove the M9400-YE and power up again; if 173000 is then printed, the cabling or the console interface board (CIB) is at fault
173000/ answered with ?	Cables not correctly mounted, or the CIB not working
Garbled response	Baud rate mismatch between the terminal and the DLV11 interface
No response	Fault in the DLV11, the cable or the terminal's transmitter
037776/ answered with ?	Top of memory bank 1 not responding; check the jumpers against the KC780 print set

Console power

DEC's console power flowchart:

• Check that AC reaches the console box's power supply and that the box is wired for the mains voltage.
• Remove the modules from the box. If DC then comes on, one module is shorting the supply: refit them one at a time until the faulty one is found.
• If the supply gives no DC with its cables to the backplane disconnected, the supply is faulty.

For a terminal that will not talk to the console, DEC's first checks are that the terminal works in local mode and that its baud rate and interface type match the DLV11.

Console power-fail traps

The power monitoring signals of the console's H780 supply and of H7100 supplies 1, 2 and 3 are ORed into the LSI-11's BPOK and BDCOK lines, so a fault in any of them can make the console trap as if power had failed. DEC lists a real loss of power, a false detection, noisy power or monitoring signals, and an LSI-11 fault as causes, and says H7100 AC LO and DC LO should sit at −9.5 V or beyond for noise immunity. To isolate the cause, disconnect the H780 or the H7100 power monitoring signals in turn.

Microcode patches

The console diskette carries the microcode patch file for the CPU. One owner explains that the console checks the version of the CPU's patch logic array against the patch on the diskette and will not start the CPU if they do not match, so a console diskette taken from another machine may not work.

Power supplies

Check the indicators on every H7100:

H7100 indicators (EY-2217E-SG-0001, p. 7-13)

Indicator	Meaning
POWER NORMAL	Power is correct
PLUG IN REGULATOR FAILURE	Fault in one or more of the +5 V, +5 VB or −5 VB, or +12 V regulators
OVER CURRENT	+5 V load of 120 A or more
OVER VOLTAGE	+5 V at 6.2 V or more
POWER INVERTER FAILURE	Main +5 V failure
OVER TEMP	Internal temperature of 150 °F or more

DEC's field fix for a failed supply or regulator was to swap the base supply or the regulator board; the procedure is on VAX-11/780 Power System Service. If a cabinet shuts down by itself, check its blowers and air flow sensors and the controller's overtemperature switch.

Margining

Voltage and clock margins bring out marginal faults. DEC's rules for voltage margins: one voltage at a time, set with a digital voltmeter, never beyond the operating range of the parts supplied, with the diagnostics that exercise the device running, and returned to normal afterwards.

Restoration reports

VCF members restoring 11/780s and a Robotron-built equivalent report missing console processors and diskettes, missing boards, power supplies needing repair and a dead LSI-11 processor module that gave no console prompt. Images of DEC console diskettes have been recovered from files on bitsavers and read under simulation.

Related pages

• VAX-11/780
• VAX-11/780 Maintenance Guide
• VAX-11/780 Power System Service
• VAX-11/780 Documentation