The Antikythera Mechanism 8: That's all for now

I've wrapped up work on the 3D printed Antikythera Mechanism for now. There are a few parts that are incomplete, and I would like to come back to it in future and redesign it. For now, I am out of patience and energy. So let's take a look at what worked and what didn't.

Some things which worked

Overall: the mechanism runs fairly smoothly and without excessive friction. Win!

Frame design: the skeletonized and split frame works well. It is rigid enough and has the right geometry, but without taking an excessive time to print, and it can easily be split into pieces which fit on the print bed.

The splits for the frame and the large gears are also good. Using M3 screws to hold the parts together makes them rigid enough, and adding 2mm dowels as alignment pins also helps. Some of the fixings are a little ugly, especially the brackets to hold the halves of b1 together.

Thin supports for the larger gears where they wobbled on their axes: this eliminated most cases where the mechanism was seizing. I also tried supports for b3/b4, but later found I didn't need them.

Press fit hex fixings for joining gears on the same axis.

Push-on shaft collars for gears where the axle was held only at one end. These worked quite well, I saw them slip off occasionally, and for long term use they should be glued in place. The shaft collars worked particularly well in the cases where I could bury them within the hub of a gear, saving vertical space. Similarly the fixing holes consisting of a hole and a slot worked well. It is possible to get pins into them without excessive force, while having a firm fit which won't work loose.

Most of the axles and pins are standard lengths. In past projects, I've often cut them to length. It's easy with brass and a harder with steel. Standard lengths, by which I mean ones I can easily buy, are better.

The gear sizes are mostly right. I discussed the choice of modules and the exact separation in an earlier post, especially as they deviate from the ones in the HAM.

Some things which didn't

Gear sizes for output: the modules I chose for the output gears meant that they had to sit quite far from the center. It pushed the spirals out while still not leaving much space for the Games, Callippic and Exeligmos dials. With a smaller module they could be moved further in, freeing up space. This is not hard to do but requires reworking the backplate geometry, and I ran out of enthusiasm to do it. Similarly, I never completed the output spirals and a change to the geometry to allow more space for them would make them easier to design.

Some of the shaft collars are not tight enough and slip off. It might work to make them longer where there is enough space. Otherwise, the only solution is to glue them onto the pins.

The planets stages use more vertical space than I would like. I tried thinner gears (doesn't save much, makes the mechanism more unstable), and I looked to see if I could interleave the gears in some way, but didn't find any arrangement that worked.

Finally, as I previous described, the a1 support isn't very stable and could do with a redesign, as part of generally redesigning the mid and back frames.

The Antikythera Mechanism 7: Input and Output

 

The mechanism up to this episode is working, with the only remaining elements being the input and output. The input comes from a gear known as a1, which drives the large b1 gear. a1 is at right angles to b1, but because of the difference in sizes (48 teeth for a1, 223 for b1), a bevel gear doesn't work very well. Instead, I made b1 as a regular flat gear, and then added a flat 48 tooth gear for a1. A set of bevel gears is mounted on top of a1. This assembly was a bit of an afterthought, and so I had to make a bracket which extends the existing back and mid planes. I decided I preferred a simple knob rather than a crank. The result looks like this:

It's a little wobbly but works well enough.

The planets output consists of pointers attached to the nested brass tubes. Most reconstructions use concentric rings. Freeth points out that this is better as it avoids parallax when reading the positions, and also allows the rings to be annotated with synodic events such as opposition and conjunction. By this stage in the project, I was cutting some corners and so went with the simpler scheme. Spencer Connor's version takes a similar approach. The upper output stage also accommodates a bevel gear set driving the moon phase output.

Instead of a 3-D printed dial, I printed the dial onto card. Here it is before mounting on the frame. I again followed Spencer Connor's version by using English month names, with displays for the zodiac and the actual spans of the constellations. The actual Antikythera Mechanism appears to have had a more complex output including an outer calendar ring which could be moved. The diagram was constructed as an SVG using a Python program.

The back outputs consist of the Metonic and Saros outputs and three smaller pointers. The Metonic and Saros should consist of a spiral groove which a pointer moves in, driven by the corresponding output gear. An alternative, with some advantages, is in the Spencer Connor video linked above. The spirals are large and my original plan was to laser cut them into pieces of acrylic or MDF. Unfortunately, my local makerspace (https://web.makespace.org/) has been closed to new members for some time and so I have had to defer this part of the work. I could have followed the model on cults by printing the panel in multiple separate parts. In the end I decided the shelve the back outputs for now. As we will see in the next installment, I've learned a lot by getting this far and may do a redesign for a new version.