I love the torsion springs; it looks much cleaner and comes a lot closer to the constant-force-spring that we discussed earlier!
I do wonder about the holding torque vs. spring torque tradeoff. There are two limiting factors: on the one side, you have the maximum torque that the motor can exert, minus the spring force decides the maximum acceleration for arm contraction. On the other side, the spring force decides the maximum acceleration for extension. It seems to me that this design may require a little more motor torque than some others. That said, it might still require less than Rostock/Kossel, since it's not holding the print head against gravity.
I still wonder about the planar arms. As long as it doesn't add a lot of complexity, or any exclusion zones (software limits on where the print head can move, based on where mechanical parts run into each other), it should be fine.
I would love to see something like this done with a bend sensor of some variety, so that you could run a PID loop for the length of each arm. The problem that I have seen with using a bend sensor is that there is no way to get a sensor which is sufficiently accurate without exceeding the cost of the rest of the system. (A friend of mine in Engineering Physics suggested using femtosecond laser pulses for positioning, haha!)
I do wonder about the holding torque vs. spring torque tradeoff. There are two limiting factors: on the one side, you have the maximum torque that the motor can exert, minus the spring force decides the maximum acceleration for arm contraction. On the other side, the spring force decides the maximum acceleration for extension. It seems to me that this design may require a little more motor torque than some others. That said, it might still require less than Rostock/Kossel, since it's not holding the print head against gravity.
I still wonder about the planar arms. As long as it doesn't add a lot of complexity, or any exclusion zones (software limits on where the print head can move, based on where mechanical parts run into each other), it should be fine.
I would love to see something like this done with a bend sensor of some variety, so that you could run a PID loop for the length of each arm. The problem that I have seen with using a bend sensor is that there is no way to get a sensor which is sufficiently accurate without exceeding the cost of the rest of the system. (A friend of mine in Engineering Physics suggested using femtosecond laser pulses for positioning, haha!)