Rolling Ball Clock

I bought this clock some years ago at a "Flohmarkt" (booth sale). Digging through the crud revealed, that is was made by Mattel (order No 3636) under "exclusive license from Arrow Handicraft Corp., Chicago" and is based on U.S.Patent 4.077.198.

The patent (issued March 1978) is available online here.

Assembly and some cleaning was all needed to get it working. Then the tuning started to increase its reliability: Some corners cried for smoothing, the little fence on the top rail was the end, it worked (almost) flawlessly in my office for years on end.

Last spring it suddenly stopped. As it turned out, the tiny motor obviously hadn't survived the continuous abuse by our recently increased mains voltage (230V now, was 220V previously).
I mailed Mattel for a replacement part (ok, I HAD to try ;-) but they never bothered to answer.

So, now I was stuck. Sadly I added "Motor, AC, 220V, 3W, synchronous" to my WANTED list and shelved the clock.

Last autumn fate smiled at me: I found a motor at a surplus trader and couldn't resist the price. But on arrival of the booty I was stuck again: Four wires on the motor, two on my cable... As usual in such situations, my SPLENDID international cavalry - ahem, mailing list came to help and provided the missing spec: The correct phase shifting capacitor was obtained and off I went to the drawing board.

Some gray hairs richer a design was finalized featuring lots of gears (did I mention my fondness of surplus traders ?). I had to drop the idea to hide the mechanism in the little black box which housed the original drive - but on the other hand, why hide running gears at all ?

Two days spent happily in the workshop resulted in this:

Well, now it runs again - and this time I have a spare motor somewhere in my heap...

BTW, this motor - how does it work ?

Some digging in long forgotten basics (alternating current, electromagnetism,...) and a schematic sent by DoN (of said mailing list) led to this PSpice simulation (The used student's version of this simulator can be downloaded FREE at :
Remark: The resistors are necessary to give the ideal inductors some real touch, namely a finite resistance (as measured !).
Obviously the windings of L1 and L2 are fed some current (I(L1) rsp I(L2)) alternatingly. As we all know (don't we ?), sending current through a wire wound round some plain soft iron will convert said iron into a magnet, with polarity depending only on the direction of the current (CW or CCW !). Now look at the animation below: Inductors L1 and L2 are split into several parts of equal size, wired in series but switching sense in each part: If the leftmost part of L1 is wound CW, the next part of L1 is wound CCW and so on. The parts of L2 are built and arranged similarly. As the current through each inductor varies with the mains frequency (50Hz around here, see simulation above), the iron rods through the inductors will change their magnetic polarity in sync. The U-shaped blocks in the animation shall represent permanent magnets. As you see, the permanents are moved left to right merely by attraction of dissimilar poles !
You're still with me ? Just extend the animation to more parts of L1 and L2 and form a ring out of them. Then add some more permanent magnets and glue them to a shaft running in the middle of the inductor ring - done ! The shaft will turn an known and fixed angle (360° / "half the number of parts of L1") for each mains period.

How could we make the motor run backward ? All we need to do is invert the polarity sequence of L2, for example by swapping the leads of the inductor. Another trick with the same result would be moving the lower connection of V1 (V1-) to the other (right hand side) end of C1.

Postscript: Barely had I dropped my tools some wastepaper (they call it advertising) materialized in my mailbox:

Am I disappointed ? No, mine is UNIQUE !

ODTS 2001