Building a Power Supply Shutdown Sequencer for the X68000

The Sharp X68000 arrived in an era when home computers were treated as serious instruments rather than disposable appliances. That ethos still shapes how enthusiasts in Sydney, Melbourne, and Brisbane approach repairs and modifications today. A working X68000 represents hours of careful recapping, belt replacement, and floppy drive alignment, so protecting that investment from a sudden blackout becomes a practical necessity rather than a theoretical nicety.

Australian mains power runs at 230 to 240 volts at 50 hertz, which is generally stable but occasionally delivers nasty transients during summer thunderstorms sweeping in from the ranges west of Brisbane or the hills around Adelaide. Old switched-mode power supplies in vintage machines were not designed with the noise immunity we expect from modern equipment, and a brownout or spike can scramble the host adapter, confuse the SCSI controller, or leave the internal hard drive mid-write. A shutdown sequencer adds a layer of defence by ensuring the operating system has a chance to flush its buffers and the motors in the storage devices have time to park before the rails collapse.

Understanding X68000 power architecture and failure modes

The standard X68000 power supply delivers several rails simultaneously, including plus five volts for logic, plus twelve volts for motors and analog circuits, and a negative twelve volt rail for serial interfaces. There is no soft-start sequencing built into the supply itself, and the host machine has no ATX-style standby capability. When the mains switch is flipped, all rails come up together and when the switch is flipped off, they all fall together within milliseconds.

That collapse is fast enough to corrupt filesystems on MFM and early SCSI drives, and it can leave RAM expansion cartridges with corrupted SRAM contents if the lithium backup battery has drifted below useful voltage. The human floppy interface also dislikes abrupt power removal, as the head can settle on a magnetic surface in a position that damages the media over time. Modern compact flash or SD card replacements running through an SCSI emulator have their own concerns, because they buffer writes internally and only flush to NAND when the host interface goes idle.

Sequencer design principles

A reliable sequencer watches the DC rails and the AC mains, then orchestrates an orderly shutdown when either drops out of specification. The simplest implementations use a large capacitor on the plus five volt rail to hold the logic up long enough for a microcontroller to send a shutdown command to the CPU, then signal a relay to latch off the mains feed. More elaborate designs include brownout detection, battery-backed real-time clock logging of any event, and a soft-key sequence that triggers a clean halt from the keyboard.

The shutdown sequence itself follows a strict order. First, the logic detects that input power is failing or that the user has pressed the shutdown button. Second, an interrupt or NMI is presented to the 68000 CPU, which allows the operating system to halt any in-progress write and flush its disk buffers. Third, the hard drive receives a stop command so the heads park on the landing zone. Fourth, after a measured delay of perhaps three to five seconds, the relay drops the mains feed to the power supply entirely. The total hold-up time required depends on the largest capacitor bank in your particular supply, but a four millifarad reservoir on the logic rail typically gives two full seconds of useful hold-up.

Component selection and Australian sourcing

Sourcing parts locally in Australia is straightforward for most of what a sequencer requires. Jaycar Electronics stocks automotive relays rated at thirty amps at twelve volts, along with electrolytic capacitors, bridge rectifiers, and the ubiquitous ATmega328 microcontroller modules that make prototyping painless. Their stores in Alexandria, Oakleigh, and Springvale cover the eastern seaboard well, and postage to Perth or Darwin usually arrives within three working days.

For the mains side, a forty amp solid state relay with a four to thirty-two volt DC control input gives clean switching without the click and wear of a mechanical contact. The control side runs from the logic rail, while the switched side carries the live and neutral from the Australian power cord to the X68000 supply. Australian wiring standards require double insulation on any mains conductor that runs inside the computer case, and the relay should be mounted on a stand-off so its body cannot contact the chassis. A small fuse on the live conductor, rated just above the maximum draw of the X68000, provides a final layer of protection against a relay failure that welds the contacts closed.

A few specialised parts come from outside the local market. Low-leakage input current sense amplifiers, precision voltage references, and the specific connectors used by the X68000 internal harness are easier to source from international suppliers. When ordering these, check the Australian dollar pricing carefully, because the exchange rate and import duties through Australia Post can add a meaningful premium over the sticker price. Core Electronics in Hobart and Little Bird Electronics on the mainland both offer competitive pricing on breakout boards and development kits if the Jaycar catalogue does not cover a particular component.

PCB layout and assembly considerations

The sequencer board lives inside the X68000 case, so its footprint must fit between the power supply bracket and the right side panel. A two-layer board measuring fifty by seventy millimetres is comfortable in most chassis, leaving room for the relay footprint, the microcontroller, and the support circuitry. Keep high current traces short and wide, with at least three millimetres of copper for the mains paths and generous clearance around any terminal that the user might touch during installation.

Thermal management matters more in Australian conditions than it does in temperate climates. A machine stored in a suburban garage in Perth through February can see ambient temperatures above forty degrees, and the sequencer board will sit in that air stream whenever the case fan pulls warm air past the power supply. Derate capacitor voltages by at least twenty percent and choose components rated to one hundred and five degrees Celsius rather than the usual eighty-five. A small heatsink on the linear regulator that feeds the microcontroller will keep the brain of the sequencer happy during sustained operation.

Assembly is straightforward for anyone who has built a through-hole project before. The microcontroller and its crystal go in first, followed by the support passives, the relay, and finally the screw terminals. Use flux-core solder and inspect every joint with a loupe, because a cold joint on the relay coil will manifest as an intermittent shutdown that is maddening to diagnose later. Once the board is complete, a bench test with a bench power supply set to five volts lets you verify the logic before any mains ever touches the project.

Testing, calibration, and system integration

Bench testing should cover the three main states: normal operation, simulated mains failure, and forced shutdown. For the simulated failure, a variac or a switched mains lead lets you drop the input voltage below the brownout threshold and confirm that the sequencer reacts within the designed hold-up time. A storage oscilloscope on the five volt rail shows the decay curve and confirms that the relay releases only after the rail falls below four point five volts, which protects downstream regulators in the X68000 from running out of specification.

Once the sequencer behaves correctly on the bench, installation in the machine takes about an hour. The mains lead enters the sequencer board, then a short jumper carries the switched mains to the original power supply inlet. A three-wire harness taps the plus five volt standby rail from the supply to power the sequencer logic and to monitor for brownout conditions. Mounting the board on standoffs near the rear panel keeps it accessible for future service and out of the way of the CRT yoke assembly.

Integration with the operating system side requires a small TSR or driver that responds to the non-maskable interrupt raised by the sequencer. Under Human68K, this is a short assembly routine that flushes the disk buffers and halts the processor. With the sequencer and driver in place, a pull of the mains plug results in a clean shutdown rather than a corrupted filesystem, and the machine is ready to power back up the next time you want to use it. While the case is open and the system is stable, it is also a perfect moment to verify other aspects of the display, which is why many owners take the opportunity to run through CRT geometry calibration at the same time.

The completed sequencer becomes a permanent part of the X68000, drawing negligible current when idle and stepping in only when the mains behaves badly. For an Australian enthusiast working with a forty-year-old machine and our sometimes temperamental grid, that small board pays for itself the first time it saves a session of work from a sudden storm-related outage.