I saw weird spikes in the position data like that before and what the issue was for me was that the SPI bus between the encoder buffer and the arduino(brain) was getting disconnected during vibration. Took me a while to figure that one out. Also for the bellows, this is a blind guess, but you could try adding fillets to the edges(outside) as well as giving it a bit less thickness to the parts you want to bend. Kind of creating a designed "weak point" so they bend correctly.
@harrisonlowАй бұрын
Damn, I can see how that would've taken a while to sort out! Re. bellows: I've considered similar fillets etc. but I'm not sure they'll have much of an effect. I have a feeling it's the internal (concave) bends that are the important ones for making sure the bellows collapse properly and (I think) we want these as pre-folded as possible (which they are now, sans fillet). Also I've found that the bellows print best in vase mode so I'm not sure if variable thicknesses would work here; even if variable thickness vase mode printing is possible (something tells me it isn't 🤔), it would be a pretty small difference between 'thick' and 'thin'.
@hoodioАй бұрын
why do the bellows need to resist torsion? and how much?
@harrisonlowАй бұрын
Good question! The linear actuator legs have the motors aligned colinearly with the leg axis, so when the (fairly large/heavy) motor spins, it generates a good amount of rotational inertia. Without the bellows, this causes the whole leg to twist about its main axis, which pulls/tangles the motor + encoder cables that run from the base to the bottom of the leg. The entire purpose of these bellows is to stop this 'twisting' without impeding on the other two rotational degrees of freedom
@harrisonlowАй бұрын
As for how much torque they need to resist: I have no idea on the torque magnitude, but as long as rotation about the leg's main axis is less than 10-20 degrees, I'm happy 😊