Full build details on my site https://www.brettoliver.org.uk/Mystery_Clock/Mystery_Clock.htm In horology, a mystery clock is a timepiece whose working is not easily deducible.

This could be it seems to have no movement at all or the hands do not seem to be connected to any movement.

On this clock time is displayed by 3 white discs that rotate around a blank black face without any obvious connection to a clock movement.

Time is estimated by the position of the discs relative to the four dial fixing screws indicating the quarter hour points.

Credits

This clock is based on The BallClock by Marie and Stefan Kneip

https://github.com/reparator/The-BallClock

Hardware Changes

I have made a few changes to the BallClock.

The most important change is making the clock printable on 220x220 bed printers like my Ender5 pro.

I have done this by cutting the larger 3D parts into quarters. These parts are bonded together with Cyanoacrylate glue, sometimes with straps to add strength.

I have designed a new front bezel and rear enclosure for my version of the clock and these also come in quarters.

I have included all large parts of the original clock in quarters as well, in case you want to build Marie and Stefan Kneip open or closed versions.

Instead of the carbon fiber clock dial I have used a clear 2mm Perspex disc.

The thickness of the disc means that I could not get a ball bearing to stick to the dial.

Instead of ball bearings I have used 3D printed discs with self adhesive magnetic tape inside.

In my version of the clock I have used prebuilt modules where possible and built them onto Vero board or mounted them inside the clock.

I have used a socket mounted SN74HC595N rather than the SMD IC 74HC595D as I am using Vero board.

I have used a RKAT28c MCC Compact Project PCB to house the ATMEGA328P rather than the SMD version https://www.brettoliver.org.uk/Aviator_Clock/Aviator_Clock.htm#modules .

The RTC in the original project uses a DS1307. I have used a module with a DS3231 RTC which has built in temperature compensation.

To drive the 28BYJ-48 stepper motors I have used ULN2003 diver boards on PCB modules. These have been mounted within the clock.

The four switches are mounted on a separate Vero board and protrude through holes in the back of the clock case.

Software changes

Again I have made only minor changes to the software.

On the original clock on initial power up the clock would wait for the White button to be pressed to initialize to the 12 0'clock position.

I found while waiting for the button to be pressed it was drawing over 500mA. I have changed the software so on power up the clock automatically initializes to the 12'oclock position.

The clock now uses max 190mA while the hands are moving and then it drops down to 19mA when the hands reach the 12o'clock position.

I have added a PIR and changed the software so that when the PIR detects no movement for 10 minutes the clock moves the hands to the 12 o'clock position.