Spinner (Motor)

Every propeller clock needs a propeller, i.e., a rotating contraption that holds the display elements and moves them around on a circular path - which brings us to the first-order question: the question of the spinner. This component is expected to spin the propeller itself; it is typically an electric motor, preferably with a regulated angular velocity. These are not easy to come by, especially in a form that makes them readily available for tinkering. The breakthrough idea for the Nixie Tube Propeller Clock (credited to my humble self) was to use a 5 ¼ inch floppy disc drive turned upside-down. (Being an avid junk collector, I had a few of these lying around.)

These drives have a built-in brushless DC motor, which is designed to spin the floppy disc at a constant speed of 300 or 360 rpm. This made the device ideal for my purposes, as experiments proved that the 6 per second "refresh rate" at 360 rpm was quick enough to bring out the persistence of vision effect. The three evenly spaced small holes on the aluminum platter around the axis were the perfect place to put in mounting screws for the Spinning Board. This platter (a.k.a. "carousel") is attached to the rotating shaft with sticky tape, so it was possible to remove it without any damage.

Propeller (Spinning Board)

The electronics of the Spinning Board is described on the NTPClock Hardware page in great detail. Here we will present an interesting mechanical aspect of the implementation.

As we have learned it from the "spin" cycle of washing machines, a spinning object must be properly balanced, otherwise potentially harmful wobble or vibration can occur. In case of the Spinning Board, symmetry along the longer axis is all but perfect. Not so along the shorter axis, with the high voltage converter being quite heavy. To remedy this imbalance, I added small counterweights on the opposite side of the board in the form of two metal spacers. As for the method of calibration, sometimes lo-tech is just what we need.

Yes indeed - a box of Buckyballs can go a long way! (Note that the pictures show the balancing of the v2 prototype [↓]; in the current version, the balance remained almost exactly the same.)

Connecting to the Spinning Board

With the question of spinning solved, the next problem was to figure out how to implement the "electric bridge" between the stationary and rotating worlds. This turned out to be by far the most challenging part of the project, and the ultimately satisfactory solution had to wait for a few decades.

Enter the Slip Ring

The most common way to handle such cases is the application of slip rings. Unable to find a part suitable for the NTPClock, I had to build my own. The solution I opted for used a ¼ inch headphone jack mounted precisely over the axis of rotation, and contacts pushing against it. The exact implementation went through multiple iterations, and while it worked for periods of time, it had constant issues with reliability. This is not too surprising, considering that parts not designed for this particular purpose are not expected to withstand 500+ k rotations per day, year over year. Despite careful lubrication (that still made a mess on the Spinning Board), the jack and the contacts would wear away and ultimately fail. The problem was only resolved by the contact-free solution of NTPClock v3; but more on that later.

NTPClock v1

In the first prototype, the headphone jack dropped down from a "gantry" above the Spinning Board, and the two power contacts were mounted on the board; the push-button of the single-button interface was right underneath the tip of the jack, and could be activated by gently pushing down on the gantry structure.

While this approached worked well as a proof-of-concept, it had a couple of mechanical issues, which prevented the clock from running reliably for an extended period of time. The headphone jack's alignment was hard to adjust, and the gantry was not rigid enough to maintain consistent contact with the jack.

NTPClock v2

In the next iteration, the mechanical design was turned upside-down, literally. The headphone jack got mounted on the Spinning Board, and the contacts were placed on a small circuit board fastened to a massively reinforced gantry. Because the previous nifty "button trick" no longer applied, the button was brought out to the face plate of the floppy drive, and another contact was added for it to the "slip ring" interface. (Fortunately a stereo headphone jack can still cover these needs.)

Originally the contacts were made out of springy metal strips yanked out of an old edge connector. While these parts were designed to handle friction well, they were too rigid, and started to damage the headphone jack; so the next idea was to replace them with actual springs.

It turned out that the springs being the contacts themselves introduced too much slip ring noise, which made the button utterly unreliable. Because of this, short pieces of wire were attached to the springs, but alas, this time they got the short end of the stick and would gradually wear out over time. Since there was still a small eccentricity in the headphone jack's mounting, more rigid wires would re-introduce the slip ring noise. Bottom line: the design required a radically new solution.

NTPClock v3

That radically new solution came with the advent and wide availability of wireless power transfer technologies; these consist of two inductively coupled coils, where power is transferred via high-frequency alternating magnetic field to maximize efficiency. Each coil is attached to its supporting circuitry, overall forming a wireless DC power transfer system. For the NTPClock, the Taidacent TD-458 kit was chosen, able to provide more than enough juice for the Spinning Board's needs. The transmit and receive modules were mounted on the Canopy Board and Gazebo Board respectively. The Gazebo Board is attached to the Spinning Board, while the Canopy Board is brought right above with two support beams.

The button signal is also transferred wirelessly, but instead of using inductive coupling, the obvious choice was infrared light. Luckily, the floppy drive was able to provide all the necessary parts, in the form of the index-hole-detection LED/photodiode pair. Actually, not exactly... since that LED is very difficult to access, it remained entombed in the belly of the drive, and the Canopy Board received a standard red LED instead (which emits IR in the sensor's desired wavelengths well). By the way, there is also another infrared duo in the drive, detecting the floppy disc's Write Protect cutout; however, they were already spoken for, as described on the NTPClock Hardware page.

Case for the Case

Since no gantry system is screwed to the side of the drive chassis anymore, it was now possible to cover the entire device with a nice acrylic case.

In addition to being decorative, it prevents the spinning parts from colliding with anything from the outside world, and also protects the clock from dust, which used to accumulate on all the parts facing the direction of the spin.