The main part of this project has been designing a suitable bone capable of being 3D printed. The preface to this project can be found here.

My first bone design was completely from scratch with no prior thought into what it would respond like. I came up with this design solely to get some 3D prints out and see how they came back so I could further iterate to a better design.

Initial bone design

Returned 3d printed bones

A few days later and some super glue, I had my first prototype arm! Obviously given the design with minimal insight, this arm really wasn’t suitable for much at all. The centre core was way too stiff and the distance between centre of the bone and points of applied tension were too close meaning the operating envelope of this arm was minimal along with being exceptionally stiff. The servos used here were standard SG90 and the tendons (string) were connected direct to a standard horizontal servo horn. This also affects the operating envelope.

As I hadn’t controlled the infill settings or anything regarding the slicing, I was dependent on what my 3D printing service provided for me. This was the main reason I decided to get my own 3D printer as figuring those kind of settings would just end up costing a fortune!

A broken bone from poor design and incorrect print settings.

By now however, I hadn’t decided to get my own 3D printer so I went for a few more designs, namely a thinner central part and wider tensioned base. For this print, I went for three different central thicknesses so I could decide which one worked best for me.

Second revision of bone

The second revision was much more useful, I was able to get a much better operating envelope on along with better flex without breakages.

Using another 3d printed bone and servo mount demonstrates much better performance! This was also the point that I decided to use monofilament nylon as the tendon – commonly known as fishing line. The 0.5mm stuff used here was rated up to about 25kg of tension!

Arm performance included added weight and max speeds!

At this point, I was confident that I could then scale up the arm to full TAMI sizes. I wanted to go for a 6 DOF arm with 6 segments giving a total length of ~600mm.

To get to this point, I took one of the bone designs from above and scaled it accordingly. I also added cutouts for the bend sensing components.

Sadly this design also had some oversight still! The central column was still as stiff as my previous design similar to this and I hadn’t accounted for the tearing from the bone tensioning base.

New full size bone design

Another torn base!

It was at this point that I decided that I should 3d print the rest of the bones myself. I went out and bought pretty much the cheapest 3d printer I could find on eBay – the Creality Ender 2 (Amazon link – may expire). And can safely say that I’m pretty damn impressed with it! That along side some TPU filament and I’ve been able to prototype many many more bones to finally get round to what I think is the perfect final bone.

To alleviate the base breakage issue, I’ve added a large fillet to add strength at that point. I could also alter print settings to allow for the best performance from my printer. I ran the filament slightly warmer than nominal temperature to allow for better layer adhesion which is really important in this case. To speed up printing, I placed 4 items on the print bed at once. This caused stringing between the bones but nothing a craft knife and heat gun couldn’t clear up.

(Hopefully) Final bone design

This design includes fillet on the base along side central holes for wire routing and the central channel cutout for bend sensing.

Printing 4 at once on my 3d printer

This was my final bone iteration, I had actually printed enough of a different revision to make some arms but after assembly, I wasn’t particularly happy with their performance.

Multiple bone and arm revisions

It is worth mentioning now how I expect to control multiple arm segments. This can somewhat be seen in the bottom most arm in the above image but I shall explain none the less!

To move a segment without affecting preceeding segments, the point of tension needs to be between the base of the preceeding segment and the base of the proceeding segment.

To ensure tension is only applied to this section, the tendon needs to pass through all bones prior to the preceeding bone without affecting their position. This is achieved by putting the tendon through a piece of pipe, similar to how a bicycle brake works. This method is called a Bowden Cable.

Initially, I made the outer pipe of the tendon out of silicone though after friction and compression issues, decided this wasn’t the best material to use for this task.

I decided after this to go for PTFE tubing instead – the stuff used for 3D printing! This stuff is great for low friction qualities, especially with nylon though has I’ve been having issues with gluing it to the TPU

Until I get the rest of my bone 3D prints sorted, this is the current progress for now!