The unit is intended for powering high-voltage low-current applications - Geiger tubes, neon bulbs and nixies, photomultipliers, diode junction breakdown voltages, high voltage Zeners measurement, leakage current measurement, charging capacitors for pulsed power applications...
The unit is powered via a 2.1 mm barrel jack from 12 volts. Overcurrent is prevented with a 0.5 amp Polyswitch resettable fuse.
The power for the unit is controlled by a switch and a button in parallel; the former for long-term operation, the latter for when just brief pulses are needed. The operation of the unit is indicated with a red LED, located inside the switch (converted from a 230V neon bulb based one into one based on a LED). Another, little green, LED is on the back panel above the power supply connector; its role is to make the operator aware of the unit being powered.
The oscillator is based on the venerable 555 timer IC.
The oscillator's frequency is controlled by a pot. The lower the frequency, the higher the output voltage but also the higher the ripple.
Another control is via the pin 5, with signal from a voltage feedback. This signal pulls down the pulse width and later also the frequency, lowering the output voltage of the boost converter. This allows a crude stabilization.
The core of the converter is an IRF820 N-FET and a toroid coil, in a standard boost converter topology. This design was chosen to avoid using a transformer. The FET has a snubber across it, to protect it from the transients.
The boost converter can be switched to a multiplied or non-multiplied variant. A physically large switch was chosen for the large distances inside and low chance of arcovers. The non-multiplied variant consists of one STTH112 fast diode (1200V, 1A, 100ns recovery time) and a capacitor, as a classical boost converter; its output voltage reaches up to 600 volts.
The multiplier uses a Cockcroft-Walton generator stage instead, with open-circuit output voltage up to 2100 volts.
The high voltage capacitors are equipped with yellow-marked header pins, for easy discharging during maintenance. These little buggers can keep a nasty little kick for hours.
The output of the converter is connected to a serial resistor for output current limiting. The resistor consists of a 4.7 MΩ pot, and a pair of fixed resistors that can be disabled with a switch.
The pot tends to overheat when set to low values and high output currents. A higher-power variant would be beneficial.
The voltage feedback circuit consists of an adjustable voltage divider (using a 100 MΩ high-voltage resistor composed of 10 MΩ SMD resistors, and a 100 kΩ pot) and an op-amp with adjustable gain, switchable to two levels. The pot allows regulation in a roughly exponential way, with start at 20 or 200 volts (switchable) and end at the maximum possible output voltage. The feedback can be switched either to the output before the limiting resistors, or after them, or disabled entirely.
The device has also a pair of output followers, consisting of adjustable voltage divider (100 MΩ vs 100 kΩ, a trimpot in between to finely set the ratio in between) and an op-amp buffer based on LM2904. Their purpose is to provide the output voltage signal, as normal multimeters end at 1000 volts (the cheaper ones at 600 volts). The followers are set to provide 1 volt for each 1 kilovolt.
The breakout connector on the back allows easy connection of a milliamp meter for monitoring of the output current. A panel switch allows bypassing it.
The connections are located on the back panel to be physically distant from the operator's hands on the front panel controls, reducing the risk of accidentally touching something that should not be touched. (Ouch.)