I designed a perpetual calendar mechanism to go into my longcase clock and had it installed and working.  Unfortunately, I couldn't get the clock working so the calendar had no trigger at midnight.  I decided that the clock was far too crowded, so I dumped the calendar section from the clock.  I really want a decent sized and clear to read calendar as the digital clock/calendar I have is almost unreadable, particularly in the evening.

The perpetual calendar has thus bean moved into another project, and this is it.  The whole mechanism has been 3D printed and all the parts are there but they need a new host.  The original calendar mechanism was weight driven with a fly fan to regulate it.  It was triggered from the main clock at midnight from a 2:1 gearing down from the hour shaft and a snail cam.

One of the problems with the perpetual calendar mechanism in the longcase clock was that the cams and gears were on top of each other and some of the workings were obscured.  This can be seen in the photos below.  In the separate calendar unit I plan to spread things out so that everything can be seen.

I particularly like the "Ban-the-Bomb" cam wheel in the right-hand picture.

This photo shows the parts laid out on the table.  To make things work various parts will need redesigning, such as reversing some of the ratchets and changing the levers.  The original had the drums driven from the LHS this will have the drive on the right.

I think It's time for an explanation.  The main feature of the perpetual calendar is that the date is advanced by a number of days at the end of each month depending on the number of days in the month, in addition to the date being advanced by one at midnight every day.  Months have 31, 30, 29 or 28 days depending on the month and whether it's a leap year for February.  Basically the cycle repeats every four years though the century is different for some centuries but I'm not going to concern myself with what happens at the end of the century

To determine how many days advance are required at the end of the month a notched wheel is used with the depth of notch depending on the amount of advance required.  This is shown in the photo below.  The deep notches are February.  At the beginning of the next month the wheel is moved on one notch ready for the end of the month.  This is just part of it - more to follow...

The next problem is that the date wants advancing on a different day for each month with a different number of days.  For instance, for February in a non-leap year, we want to advance the date by 4 days on the 28th, for a leap year a 3 day advance must occur on the 29th etc.  I studied a number of clock mechanisms before I managed to work out how the perpetual calendar part worked - I can't claim credit for the invention, much as I would like to

The diagram below gives a rough idea of how it works.  A lever with probe part is dropped into the slot and then at midnight it is moved out of the slot at the same time turning the date ratchet wheel with a pawl.  The crux of the matter is preventing the pawl on the ratchet wheel from advancing the day by the advance number at each midnight.  This is achieved by means of the "guard" which lifts the pawl away from the ratchet wheel except for the last bit when the pawl pushes the tooth and advances the date by one day.  This can be seen in the photo as the orange pointed piece just above the ratchet wheel.  The orange pawl runs up onto this and back down to turn the wheel one notch.  The yellow pawl rests on a plain cylinder with a peg which will be pushed by the yellow pawl when the wheel is in the right place.

When the date wheel has turned to the point where the days advance is wanted, a second pawl engages with a step and turns the wheel by the required number of days.  This is shown as the orange pawl and green cam in the diagram.

I'm not satisfied with the above explanation and I'll see if I can improve it.  Meanwhile, I'll continue to post my ideas for the separate calendar.  These posts will help with the explanation.

Here is another photo of the parts, rearranged.  I have moved the drum drive back to the right as that seems easiest.  I might even be able to use some of the parts without modification.  Of course, the main drive lever needs rearranging so I have split up the various parts.  The green cam follower with ball bearing roller, runs on and is operated by the green cam.  The lever moves the probe out of the slot and also operates the pawls on the main date wheel (orange).

Advancing the month slotted wheel relies on the red cam for drive and the blue probe lever to determine when, in conjunction with a slot in the date wheel.  The cam allows the probe to move towards the date wheel and either contacts it, in which case nothing happens, or finds the slot and pulls a pawl back on the yellow ratchet wheel attached to the slotted months wheel.  As the cam rotates it lifts the probe out of the slot and pushes the pawl advancing the month wheel.  This will be clearer when I've redesigned and printed the cam followers and main drive lever.

A couple of main lever parts redesigned and printed.  I would like to try to reduce the height if I can.  ATM the cams come further down than I was hoping.  I have a space on the wall where this could go but it would be better if it wasn't as high.

I have an idea for reducing the height.  Another rearrangement of parts.

Here is a more detailed diagram showing the gears that drive the layshaft on which gears drive the drums. 

The cams are on the left and show the main date drive cam in it's fully pushed position and the months wheel drive cam in the fully retracted position.  A ball bearing cam follower on the date drive cam is fitted to a lever (curved) pivoted above it.  Then a straight bar with pivot for the two main date drive cams is supported at the right hand end on another lever pivoted between the date and months wheels.  This latter lever extends below into a curved probe that will drop into the slots when the cam turns to let the bar move leftwards under the action of a spring (not shown).

The other cam drives the months wheel when the date is on day one on the month.  A lever pivoted between cams and date wheel rests on either the date wheel (day other than one) or on the cam if the probe on the other (lower) end finds the slot in the date wheel, on the cam.  On day one of the month the second bar ending in a pawl resting on the months ratchet wheel will be pushed to the right and advance the month wheel.  On other days the bar and pawl is held back by the lever probe resting on the date wheel (no slot).

Another rearrangement that I think will be better.  The above diagram didn't have the lever proportions right - the one below is to scale.  The photo shows roughly the positions of most of the parts.  Some parts need reprinting either reversed or changed in size.

Shorter axle for the drums.  Will now fit in 400mm x 300mm (or 15" x 12") box internal dimensions.

Looking good Gina, is it possible you can add an extra day for the weekend so we can have an extra day off work? ?

Not up to me, of course, but I see what you mean   And an 8 day week would make lots of things simpler   

Now printing new parts.

Gradually printing parts.  The glass plate on my Pilot printer has cracked but I think the printer is still usable.  Here is a photo of where I am ATM with printing the parts.

I've ordered a picture frame for the front :- 16 x 12-inch Picture Photo Frame, Black to match my moon dial and longcase clocks.  The box will consist of pieces of floorboard for the sides with mitred corners and 6mm mahogany plywood for the back - I have these in stock.  I'm hoping I can get away with mounting all the parts on the back panel.

Small changes today.  Redesign and new colour for the cross bar and new colour for the everyday advance pawl.  The same bolt that holds the ball bearing cam follower will also take the support lever pivot. 

I haven't yet decided on the levers and bars to advance the months wheel but I have been thinking about the cam for advancing the date wheel.  I plan to use a stepper motor to drive the cams rather than weight with flyfan governor so the cams can be a bit simpler.  OK, I know I could make things simpler by just driving the drums directly with stepper motors controlled with Arduino and RTC but that's not as much fun   I shall use just one stepper motor that will rotate the cams by one revolution at midnight each day.

Driving the months wheel requires very little force compared with the date wheel which in turn drives 3 of the 4 drums so more of the 360° can be devoted to the date drive.  Previously it was half and half.  The date advance uses up to 40mm - 10mm per day - the 40mm applies to February with 28 days.  The months advance requires only 7mm movement of the advance pawl and only needs enough force to overcome the bearing resistance and detent spring (not made yet).

The sequence of operations is as follows :-

We'll start with date advance bar fully right with the cam follower on the top of the cam.  This would be as in the photo above.  The month advance bar would also be fully right.

1. The date cam lets the date advance bar travel back until the purple probe contacts the slotted month wheel.
2. The cam then pushes the date bar fully to the right again, advancing the date.
3. The date bar then remains in this position for the rest of the cycle.
4. Next the month advance cam follower is moved to the left until either the month probe comes into contact with the date wheel or finds the slot.  Until the slot is in line the cam follower and month advance mechanism stops there but if the slot is in line the month advance moves left pulling the month advance pawl back a notch on the month wheel.
5. Finally the month advance is pushed fully to the right and depending on whether the probe found the slot or not, advances the month wheel one month.

This is just one possible starting point and another would be to start with the date bar to the left with the purple probe resting on the month wheel or in a slot but this would stop the mechanism being manually set up.  Better for the systems to test how much or whether to advance the wheels first.  This would allow the perpetual calendar to be set to the right date manually as the mechanism has no actual date input.

Manual setup would consist of turning the day-of-week drum, then the main date wheel to the day-of-month and finally both the month wheel and the month drum would be set up.  If I get the gears that drive the tens and units of date drums in sync with the date wheel, turning the date wheel should indicate the right numbers.

Here is a linear representation of the motion of the date advance (top graph) and month advance (bottom graph).  The top graph shows the effect of the notch depth representing the number of days in the month.  The bottom either month advance or not.

Calculations for main date drive cam :-

1. Height range required is between 40mm and 50mm.  Try 46mm with max radius = 60mm and min = 14mm.
2. Operation of main date drive is something like between 0° and 270°  (¾ of full cycle)
3. Starting at 10°
4. Dropping to minimum at 40°
5. Rising from minimum to maximum between 40° and 270°.  Range = 270°-40° = 230° -  Motion = 60-14 = 46mm.  2mm per 10°.
6. Stays at maximum for rest of cycle.

Edited to give easier values to work with.

.

I've come to the conclusion that the cams don't need a steady reduction from maximum to minimum and that snail cams with a sudden drop would be alright and save the rotation angle needed for a slow drop.  This means the whole 90° left from the date advance can be used to push the months wheel one step.

Here's a new calendar diagram showing the cams and link bars etc.  The months advance bar will cross in front of the date wheel ending in a pawl to drive the months wheel and a hanger link.  I may make the latter a spring rather than a free rotating lever.