Thursday, August 21, 2014

Power supply capacitor restoration, attachment of expanded storage frame, and completion of inlet voltage change

The last of the inlet voltage switch-over from 208V to 230V was completed. These changes affected the 12V and 48V regulated power and the power that drives the usage meters. I did have to validate the voltage of the internal disk drive to be sure it would work properly with 230V mains.

IBM originally rented these systems and charged by hours of use, providing digital meters that turned only when the processor was actually running code. When the machine was being serviced, the IBM CE would turn a key that recorded time on the alternate CE meter, rather than charging the customer for use of the system by the CE.

The left side of the 1131 is an optional frame, called the 'blister' in the IBM documentation, into which core memory was installed. This was needed for certain models, since the location for core memory in the small (e.g. 4K word) machines was on the right side next to the power supplies but couldn't accommodate larger configurations. This machine has 16KW of memory, which fills the D gate in the blister (8KW per side of a memory gate). If it had been upgraded to the maximum configuration for an 1131, 32KW, the E gate would have been added into the blister.

For shipping, the blister is separate from the main 1131 unit, but the two parts were attached to form a single 90" wide system. Power supply wires and signal ribbon cables were connected to complete this installation task. The signal cables plugged into the B gate in the A2, A3 and A5 connector positions of the C1 compartment, first removing and replacing the frame into which the compartment cover will latch.

Those cables were secured in place on their long run through the machine by holders whose adhesive had failed. Getting these glued into place was a bit of a challenge but with that done, the blister was fully joined to the rest of the machine. The ribbon cables have to be dressed along the gate in such a way that they don't fall in front of the card compartment covers or snag on anything when the gate is swung open or closed.

Most of the large filtering capacitors for the 1131 power supplies are concentrated in the PS compartment beneath the single disk drive in the right side of the machine. However, an additional supply sits in the mid-pack area to the left supporting the 7.5VAC supply that powers all the lamps in the system.

Main PS drawer being removed from 1131
Power supply ready for capacitor reforming and checkout
 Doug Martin and I reformed the capacitors after I had disconnected and removed the main PS unit. We worked on one capacitor at a time, adding a component to limit the current that could flow into the capacitor, then using a variac to gradually bring it up under power. He provided the variac and valuable assistance.

View of the ten capacitors, two transformers and bleeder resistors
These capacitors have not had power applied since 1986, which means they are at great risk for having the electrolytic insulator inside dried out, leaving small short circuits. These short circuits can cascade into catastrophic failures of the capacitor can, exploding and/or damaging the rest of the circuitry.

By limiting the current, we allow the short circuit to melt the insulator, which usually will reflow to heal the flaw if a controlled amount of power heats up that defective spot. Gradually the holes disappear and the capacitor becomes capable of handling the full voltage of the power supply as intended. This is called reforming a capacitor as the paste that forms the insulator between the sandwich of metal layers is restored through the selective, controlled heating.

This main PS unit supplies raw (unregulated) power at -3VDC, +3VDC, and +6VDC. It also supplies regulated 12VDC and 48VDC - although it was a mystery at first since their was no regulator circuitry visible. The two regulated voltages are fed from their own transformer, which is a ferroresonant type. This means that the iron core is designed to become saturated (reach a peak magnetic field) at a relatively low input voltage, thus yielding almost the same output voltage while the input can vary a fair amount.

The design includes a special winding that is connected across a capacitor tuned to the resonant frequency of the core, quenching harmonics and other distortions that can be caused because of the non-linear nature of a ferroresonant transformer. That explains why the 12V and 48V are listed as regulated but the -3, +3 and +6 are designated unregulated and are fed into separate regulator circuits.

In addition to the voltages fed from this capacitor laden main PS unit, there are a few others generated in different places. A separate ferroresonant transformer with its capacitor-winding tank coil sits in the midpack power area along with the regulators for other voltages. This self-regulating transformer produces the 7.8VAC which drives Silicon Controlled Rectifiers (SCRs) which control the lamps on the system. Another transformer produces 51V to power the usage meter of the processor and peripherals. No large electrolytic capacitors are used in these additional power supplies so we had no reforming to do.

Midpack area with three regulators and 7.25VAC supply upper left
Capacitor used with ferroresonant transformer for SCR (lighting) supply

No comments:

Post a Comment