In Part 1 of this series, I outlined my concept of a Solstice Clock. Since writing the original article, I have modified a regular quartz clock and have estimated the requirements to drive it.
My original drive circuit was rather complex – excessively so. Having now reconsidered the requirements (for instance, removing the facility to set the time via a serial data connection) and calculated the drive requirements of the clock solenoid, I have found that a minimal implementation of a Solstice Clock could consist of no more than a microcontroller, timing components a couple of BAT54S dual diodes (protection for the 2 port pins connected to the solenoid) and that's about it.
This more minimal system would use less power (runs on a pair of AA cells – the original would have required 4 or 5) and also fit into far less space than the original. The latter consideration could be important if there is little space in the back of the clock for modifications.
Choosing a Chip
Whilst the very minimal version could be implemented with an Atmel ATTiny45 – making for a very small circuit board – I have decided to make my prototype with a little more flexibility which requires more than the 8 pins of the ATTiny45 and have thus selected the ATTiny2313V. The 'V' is significant – this lower-powered version of the device can operate right down to 1.8V with a clock frequency of less than 4MHz or 2.7V with clock frequencies all the way up to 10MHz.
After playing around with a spreadsheet, I checked some crystal values to see what gives the most accurate results. Actually, it wasn't quite that way – I first looked on eBay for cheap job-lots of crystals in the 3Mhz to <5Mhz range and then ran those through the spreadsheet to find the ones most suitable. I now have batches of 4.9152MHz and 3.579545MHz crystals to play with. There is a trade-off between the time resolution that I can obtain and power consumption; the higher the clock frequency, the better the resolution, but the higher the power drain.
More than just the Tropical Year
My original plan, when I was thinking of controlling the clock through a serial interface, was to be able to vary the 'tick' rate to allow for differing year lengths. Whilst I am now using a mean Tropical Year (so a fixed length), I realised that it would take little effort to provide other mean periods that the clock could operate on. The final list is this:
- 1 Tropical Year
- 1 Synodic (Lunar) Month
- 12 Synodic Months
- 13 Synodic Months
These 4 periods would be selectable by two positions of a 4-bit DIP switch. This leaves me with 1 switch to control stop/run and another switch reserved for future use.
I have discovered – somewhat to my annoyance – that most quartz clocks provide a means of setting the hour and minute hands, but not one for setting the position of the second hand. As this hand is representing far more substantial periods of time in my slowed-down clock, I have decided to provide the means of setting this hand through the electronics. The stop/run switch will be set to stop, then a push-button held down until the second hand is at the appropriate position. The hour and minute hands would then be adjusted by the regular twiddler on the back and the stop/run switch set to run when ready.
In the next article of this series are presented the circuit schematic and board layout of the Solstice Clock prototype.