Overview

The Nixie Tube Propeller Clock's full schematic diagram is available in PDF format, by clicking the thumbnail below.

The hardware of the device consists of three major components (not counting the floppy drive's own circuitry):

1. Power Board — Power supply to the Spinning Board and the floppy drive electronics
2. Power Transfer Unit (PTU) — Interface between the stationary and rotating parts
3. Spinning Board — The entire "clock movement"

Power Board

The Power Board receives 9 V AC from an external power supply, which is full-bridge-rectified and filtered to provide 12 V DC for the floppy drive. It is also the input to the familiar 7805 linear regulator, source of the regulated 5 V DC for the 5V domain of the drive electronics. Prior to NTPClock v3 (see NTPClock Mechanical Design), this was also the power for the Spinning Board, which received it through the slip rings; in the current version, the power is coming through the Power Transfer Unit, supplied from the unregulated 12 V.

To get a better idea about the power requirements, I measured the current draw of the different components of the system. These are:

With all this load, the unregulated 12 V supply drops to 11.5 V, with a 0.8 Vpp ripple; however, this does not prevent the floppy motor from operating as expected. As part of the drive's operation, the motor can be turned off by disconnecting the M.ON ("Motor On") input from GND; I brought this option out to a switch mounted on the drive's face plate.

Power Transfer Unit

The PTU is implemented using the Taidacent TD‑458 kit that consists of a pair of [inductively coupled] coils with their supporting circuitry, based on the XKT‑510 and T3168 chips for the Tx- and Rx-side respectively. The kit accepts 3‑15 V DC on its input and provides 5 V DC at a promised maximum current of 1 A. The two coils are brought to close proximity with a 4 mm gap above the Spinning Board, as described on the NTPClock Mechanical Design page.

With this geometry, the measured short-circuit current on the Rx-side is only 880 mA, and the steady 5 volts begin to deteriorate at around 400 mA. However, this is ample supply for the Spinning Board, with its mere one-third demand.

Spinning Board

The Spinning Board is the heart of the NTPClock, implementing the actual clock functionality. It was built on a prototype board for - yes - prototyping purposes, but since it was working out so well, I saw no reason to replace it with a custom-designed printed circuit board. (The orange wire carries the high voltage - better be careful...!)

The circuit's four major parts are detailed below.

Microcontroller

The MCU picked for this project is the Microchip PIC16F84A, which is a RISC Harvard architecture with a mere 1024 words of flash program memory and 64 bytes of static RAM. Its simplicity and (at the time) high popularity made it an ideal choice for this project. Later, with the addition of the buzzer and the sound engine, it became an exciting challenge to see how far I can push the envelope on this draconically resource-limited processor.

In the NTPClock application, Port A of the PIC is entirely dedicated to the Nixie tube, while Port B is used for receiving signals from the environment, such as the Control button, the 12/24-hr selector jumper and the photodiode required for the stationary display modes; it also hosts the piezo buzzer that emits the sounds.

Nixie Driver

The BCD-encoded digit values from the MCU drive the Nixie tube's cathodes through a 74141 TTL IC. This now iconic DIP chip is a classic BCD-to-decimal decoder, but with a twist: its open-collector outputs are designed to handle high voltage levels that occur in a Nixie circuit. Regrettably, with the (official) obsolescence of Nixies, these chips also got phased out, so until Dalibor Farny decides to fashion them at home as well, their availability remains scarce. Once again, Russia is the last place of refuge, where the equivalent К155ИД1 part is still available from newly discovered stock piles. I had the good fortune to acquire a couple of genuine Texas Instruments 74141's in the beginning of the Nixie-craze, one of which became a cherished part of the propeller clock.

The Nixie tube's decimal point is driven separately, from the MCU's open-drain RA4 output. Needless to say, the PIC does not appreciate high voltage levels very much either, therefore a ZTX458 HV NPN transistor is inserted in-between.

Nixie Tube

The display of choice for the NTPClock is a single CD72P medium-size side-view Nixie tube manufactured by Matsushita. Instead of the typical vacuum-tube-style pins, this Nixie has wire leads, which made it much easier to attach it to the circuit board.

My friend Ray used to tease me that in all these years, only one Nixie tube was consumed. That may be the case, but in reality it acts as continuum Nixies...

High Voltage Generator

One special requirement in devices with Nixie tubes is the generation of the high voltage that ionizes the neon gas inside the tube, resulting in that iconic orange glow. This is typically done with a DC-DC converter operating on the principle of switching-mode power supplies. The required current is very low (2‑3 mA), so the circuit does not need to be bulky. The Maxim MAX771 IC is designed for this particular purpose, but solutions based on the ubiquitous 555 timer chip are also popular, such as Lance Turner's Nixie power supply kit.

The NTPClock, starting out as a prototyping project, opted for an even simpler solution. Most small, battery-operated fluorescent lights employ a garden-variety flyback converter circuit; it is based on a relaxation oscillator that also has a high-voltage winding in its transformer. One such module removed from a fluorescent lamp found its place in the NTPClock.

The output of this circuit is a train of pulses that are one-way rectified and filtered. Since the voltage of the unloaded output can reach up to 400 V, a Zener diode (in reality, two Zeners in series) caps the Nixie's anode voltage at 175 V, which is high enough to reliably ignite the cathodes, but does not cause the infamous ghosting phenomenon (all cathodes being engulfed in an eerie orange haze) when none of cathodes are driven.

Special Mention: Board Position Detection

In order to display a stationary pattern, the MCU needs to be able to determine the Spinning Board's position. This is accomplished with an infrared LED/photodiode pair, which was conveniently repurposed from the drive's original Write Protect cutout detection mechanism. The picture below shows the implemented "mod": the white cylinder is the IR LED, which used to face downward (upward in the original drive orientation) an inch away, while the small black square is the photodiode, de-soldered from the PCB on the opposite side (now discarded). There is a drilled hole in the Spinning Board underneath the photodiode, which was also christened Index Hole, as it is analogous to those in the old floppy discs.

The details of the Index Hole and the related display patterns are explained on the NTPClock Firmware page.