I recently joined the Cambridge makespace. It is a very well equipped workshop with a good community of members. Most of the machines require a specific training and last week I did the one for the laser cutter. This is not the first time I have used a laser cutter. I made a few small pieces using the one in the Google LA office back in late 2019 or early 2020. I think it was an Epilog. Makespace has a larger cutter, a LaserScript LS 6090. The Epilog was a little easier to use, for example it has auto focus, and to send a job to it you used a custom printer driver. The LaserScript (usually referred to at the Makespace as HPC) requires loading up a DXF into software called LaserCut, mucking around with it in various ways, and finally downloading a file to the cutter and running it. The software and workflow feel very last century, but it works.
The goal I am working towards is to create a back panel for the functional reproduction of the Antikythera Mechanism which I have described in several previous posts. I stalled out for a while, and recently started working on it again. As well as figuring out the back panel, I've been redesigning some of the mechanism to make it more stable.
The back panel is believed to have looked something like this:
(Attribution: By Tony Freeth - Original publication: The Antikythera Mechanism Research ProjectImmediate The Antikythera Mechanism Research Project, Fair use, https://en.wikipedia.org/w/index.php?curid=36445604).
I wanted to make it transparent to show some of the internal workings. The two large spirals, called the Metonic and Saros outputs) each have a groove in which a pin runs. The pin is mounted on the radial arms and slides along them. When it reaches the end, you have to lift it out of the groove and reset its position.
Cutting this will be a large and difficult project, and so I started with a few test pieces to learn more about what works and what doesn't, and to reinforce the things I learned in the training. Here they are:
At the top is a simple square, with a square line on it, used to get the basic settings right. The material is clear 3mm acrylic. I'll record the power settings here for my own reference. The cut was speed 10, power 100. I later changed the speed to 9. The line was speed 30, power 50. It is very shallow, less than 0.5mm at a guess. The cut only just goes through the material and some of the bottom edges feel a bit rough. Either more power or a slower speed might work better. There vertical edges slope slightly (I think this is called kerf), so that the top edges are about 0.4mm shorter than the bottom ones. In terms of absolute dimensions, the top edges are about 0.6 to 0.8mm smaller than the design (30mm), while the inner line is about 0.2mm smaller than designed (12mm).
The second piece has grooves of width 5mm, 2mm and 1mm. The circles are for M2, M3 and M4 diameters for tapping and clearance, the exact size, and the text and remaining two circles are lines just to see how they looked. One issue I had with this is the material between the grooves warped. I guess it must have softened a bit and sagged. The hole for the exact size test are about 0.4 to 0.5mm larger than they should be, and the other holes are similarly oversized. The holes for tapping took the thread OK, but were a bit loose. I can't say how much this is due to the holes being oversized and how much to my lack of skill in tapping them. I'm used to the dimensions being slightly off from 3D printing and will make adjustments for future work.
My design workflow for this and the final piece was the create a sketch in Fusion and export it as a DXF. In Inkscape, I then ungrouped objects and converted them to paths before saving them as a new DXF. The DXF directly exported from Fusion does not work in LaserCut: some or all of the lines are missing. I also tried using the free ShaperOrigin plug in for Fusion. It saves sketches as SVGs, and these can again be loaded in Inkscape, manipulated and output for the cutter. For reasons I don't recall, I think the second method worked a bit better.
The final piece is a test run for the backplate spirals. It has a smaller radius than I will need (50mm inner radius instead of 75mm or so) and the groove makes only two complete revolutions. The result is very flimsy and pulls itself out of shape. You can see this by looking at the two radial lines at the 3 o'clock position. They should line up. I've yet to try a pin in the groove. I'm not optimistic that it will work well.
This leads to a few ideas for improvements. First, I could leave some tabs in place in the grove, then glue a support piece to the back and cut through the tabs. Other reconstructions definitely use support pieces; for example you can see them in the Chronova Engineering video. I don't know if the slot, currently 1mm, will be wide enough and it would certainly need to be polished for smooth running. Another option is to drop the pin-in-groove idea and use the spiral pointer adopted by Spencer Connor. It's less authentic, and require a 5:1 bevel gear which is rather large, but works nicely.
