MicroVAX 3100 Power Supply Service

MicroVAX 3100 Model 95 with its cover off; the power supply is the box at the left (Rob Jarratt)

The MicroVAX 3100 range uses the same two power supplies as the VAXstation 3100 and DECstation 2100 and 3100: the H7821-00 in the smaller BA42-A enclosure and the H7822-00 in the larger BA42-B. DEC's owner's manuals describe the supply only as self-sensing, accepting 100–120 V or 200–240 V at 50 or 60 Hz without a selector switch, and publish no schematic or parts list. The repair information below comes from restorers. The full capacitor lists for both supplies, with Malte Dehling's position drawings, are on DECstation 3100 Capacitor Replacement Guide.

Warning: these are mains switching supplies. The primary side carries rectified mains on capacitors rated 250 V and 400 V, which stay charged after the supply is unplugged.

Which supply

Supplies in MicroVAX 3100 systems (Dehling)

Supply	Enclosure	MicroVAX 3100 models
H7821-00	BA42-A	10, 10e, 30
H7822-00	BA42-B	20, 20e, 40, 80, 85, 88, 90, 95, 96, 98

Restorers report exceptions: Antonio Carlini's Model 10 had an H7822-00, and Rob Jarratt's Model 95 supply was made by a different manufacturer from most DEC supplies he had seen. The H7822-00 contains the H7821-00's board plus a second board. The H7821 also provides an isolated −9 V output on pins 13 and 14, which Maciej Rozycki and Tony Duell associate with the built-in ThinWire Ethernet transceiver.

Capacitor failures

• The main board of both supplies carries five 1800 µF 25 V low-ESR electrolytics (Dehling's C1–C5), and the H7822-00's second board four more (C23–C26). Peter Coghlan found all five on an H7821 leaking after several years of storage; the machine would only flash its diagnostic LEDs and twitch its fans. An H7821 in a SCSI disk enclosure stored with it had the same fault.
• Maciej Rozycki identified the leaking parts as Chemi-Con SXF capacitors, one of several lines of that era whose quaternary-ammonium electrolyte works through the seal with age. He notes the H7821 is least affected if the machine is stored the right way up, since the capacitors then face downwards, and recommends replacing them all. Coghlan likewise advises storing machines upright.
• The leaking capacitors can still measure in range: Rozycki, recapping a related H7816, found leaking Chemi-Con parts that still showed normal capacitance and ESR. Inspect for brown residue around the base, which may only be visible once the part is removed.
• RIFA X2 mains filter capacitor. Carlini desoldered a RIFA capacitor (28 × 16 × 13 mm) from his Model 10's supply; it was confirmed as RIFA although it had no visible cracks.

Electrolytic capacitors in the H7821-00 and H7822-00 (Dehling)

Board	Positions	Original part
Main (both supplies)	C1–C5	1800 µF 25 V 105 °C, low ESR
Main (both supplies)	C6, C11	330 µF 25 V 85 °C
Main (both supplies)	C7–C10	10 µF 35 V 85 °C
Main (both supplies)	C12	470 µF 25 V 85 °C
Main (both supplies)	C13, C14	470 µF 250 V 105 °C
Main, daughterboard	C15	39 µF 16 V 85 °C
Main, daughterboard	C16	47 µF 16 V 105 °C
Second board (H7822-00 only)	C20	220 µF 400 V 85 °C
Second board (H7822-00 only)	C21	330 µF 50 V 85 °C
Second board (H7822-00 only)	C22	47 µF 16 V 105 °C
Second board (H7822-00 only)	C23–C26	1800 µF 25 V 105 °C, low ESR

The designators are Dehling's own; the boards have no designators or polarity markings, so photograph them and mark polarity before removing anything. He fits 105 °C parts of the same or higher voltage, low-ESR types for C1–C5 and C23–C26, and 33 µF or 47 µF in place of the 39 µF C15. See DECstation 3100 Capacitor Replacement Guide for his drawings.

Model 95 supply repair

Rob Jarratt's Model 95 was losing memory modules, and its supply's 5.1 V output measured 5.3 V. He replaced the doubtful electrolytics, including the two large smoothing capacitors, one with high ESR and one slightly bulged. One of the pair would not discharge, and sparks from it destroyed the UC3842N PWM controller; a UC3842AN fixed the supply and the discharge problem. After the repair the ripple on the 5 V and 12 V outputs was about 20 mV. His first ripple readings showed spikes that he later put down to poor measurement technique; Brent Hilpert suggested on cctalk that such spikes are ringing from fast switching, damped by the snubber networks around the transformer and rectifiers. Use a short probe ground on the output terminals when measuring ripple.

Testing

Dehling finds an H7821 or H7822 can be tested without a load: the fans run and the voltages are about right, but its LED does not light until there is some load. After a recap, run the machine and check the power-up display and diagnostic LEDs; see MicroVAX 3100 Troubleshooting Guide.

Related pages

• MicroVAX 3100
• MicroVAX 3100 Maintenance Guide
• MicroVAX 3100 Troubleshooting Guide
• MicroVAX 3100 Documentation
• DECstation 3100 Capacitor Replacement Guide
• MicroVAX II Power Supply Guide
• Capacitor Failure Symptoms