I've built robots with cable tendons in the past and always had problems with them (cables can stretch, slip etc..) So it wasn't my first choice, but I think it's the best option here. I'll find out how well it works when the arm is attached.

Tldr: I likeeee

Some of them came out well, some had porosity/inclusions, Some I had to re-do. My sand casting technique is a bit rusty, I've not done this for quite a while.

This is actually the first cycloidal gearbox I've built. The design was largely dictated by the space it had to fit into and the fact it had to be hollow. It turned out fairly well, it does have a small amount of backlash but runs nice and smoothly.

I had no idea someone is building an Antikyhera replica, thanks!

I actually designed it in excel using a formula I found online, then just copied and pasted the values into a gcode file for machining. Only the cycloid discs were CNCed, everything else was done on a manual lathe/mill.

Thanks! Each shoulder unit weighs around 5kg. The motors or joints will have encoders attached for feedback of the position. Yes, I initially wanted to use brushless motors due to their better power/size ratio, unfortunately though they are much higher speed and lower torque than brushed motors meaning they need a bigger gear reduction so the advantage is lost.

Hey Ryan! Good to hear from you, thanks for commenting!

I actually really wanted to use a brushless motor for this, but even the low KV ones have a much higher max. speed and much lower torque at low rpm, meaning I would need more gear reduction, since I want low speed and high torque. Brushed motors are more 80s anyway!

I did measure the backlash and it came out to around 0.5°, though I deliberately cut the discs with some clearance to avoid binding as I was aiming for smooth running rather than low backlash. I'm sure it's possible to do better with DIY equipment, even some of the 3D printed ones I have seen have been pretty good.

@@AndysMachines Thanks :)

The goal hasn't changed since the start, though perhaps it will be time for an overview soon when I start putting all the individual pieces together. Yes it will be an autonomous robot, but it won't be able to 'walk' (refer to ep.1).

@@AndysMachines thanks for replying. I guess I hadn't seen that first video for years and forgot stuff in it.

I actually built that slip roll because I couldn't find a small one that would roll thicker material. Yes, there will be a video on this coming soon!

To keep the top half of the mould in place. Aluminium is denser than sand so sometimes the top box can float on it and the metal will leak out of the sides.

@@AndysMachines I didn’t know that. Is it 100% sand you use or is it mixed with anything?

It's a naturally occurring sand but it's actually a mixture of sand and clay (Bromsgrove greensand).

It was actually a couple of different models I downloaded from Turbosquid for reference.

@@AndysMachines You should really credit the original modelers then. Someone put a lot of work into those models.

I actually bought them, they weren't free, and this was many years ago (more than 10). There are many more better models available now and the ones I originally purchased may no longer be there. There are also many free models available on Thingiverse, but most of them seem to be copies of each other and all seem to have deliberate alterations made, perhaps to get around licensing/copyright laws.

It's a generic chinese 10" bench top lathe, sold under many different names.

I watch his videos, but that's all. Though I think he lives not too far from me.