Work I Completed
This week I have been developing further a new iteration of design for Winston, one which eliminates the issue of friction and accounts for the issue of tension. For friction, I eliminate the problem almost entirely by placing a bike bearing at the joint. This bearing has the axel (the dowel rod) going through it, secured to the inner part, which then attaches to the timing belt directly. The outer part of the bearing is attached to the thigh part of the joint by pressing it into a hole of the same size, being secured by friction. Even before testing with the belt and motors began, the axel appeared to move a lot more smoothly than it has previously. For tension, the issue was finding a balance between having enough tension to move the leg only when required (i.e. not having it be too loose so it can swing and not carry any weight) and having it too tight that it tears components apart. Throughout testing, 4 of the 3D printed joints for the motor to belt had been broken due to too high tension. Part of this was relieved when friction no longer became an issue as the tension requirement diminished slightly. For what remained, however, this was overcome in the most recent leg design which saw an increase in the thickness of the leg. This increased thickness was an amount which still allowed for the current attachment to be installed securely, but when the belt was placed on top, the increased thickness meant that the pulley was unable to angle forward due to the friction, and a new screw attached better inside so that the motor could still easily and reliably turn it. This all worked out very well, as in the most recent tests the leg moves more reliably and precisely to its desired location, and theres enough tension that it doesn’t freely move due to gravity, such that theres hope that it will be able to carry the weight of the body. As of now, the calculations suggest that it cannot currently support that. This however appears to be a result of the smaller motors, and swapping to more powerful motors may overcome this. The calculations to decide this are as follow:
T = flsinθ
Where T is Torque, f is downward force or weight, l is the length of the upper component, and θ is the angle of the leg. To find f, the expected carry weight is multiplied by gravity,
f = 3.5 kg * 9.8 = 34.3 N
The length of the thigh is 15 cm, and we want to find torque when the leg is on a 45˚ angle.
T = 34.3 * 0.15 * sin(45˚) = 3.64 N⋅m
Unfortunately, as the current servo has a max torque of roughly 0.32 N⋅m, whereas the requirement would be 0.91 (total requirement split across 4 legs), which is about 3x larger. This will be tested before modifications are made, especially because the weight used is a rough estimate, hopefully over compensating. Regardless, the leg is at the best it has been and is en route to success.
Reflection
Having applied all the ideas to the newest iteration of Winston’s design has resulted in it working the best it has yet, with smoother, more precise movement and indication that it will be able to support some weight, and if not then it should be somewhat easily modifiable. The physical component for school is due tomorrow, and its a shame that only one leg will be submitted, however, I am happy with the standard to which is has been completed. And, as already mentioned, it shows great promise to be easily adapted into a fully working walking quadruped. There is much more that is needed to be done with this project, but through the 4 main design iterations I have learnt a lot, not only with how to use Fusion360 (although I am still very slow with it), but also how to think about design flaws and creative solutions to best address them. I have more ideas I want to test in future iterations for greater efficiency and smoothness, and am excited to continue and see how it all turns out.