Hi Linden! On 07.11.2015 00:55, linden wrote: > Hello All, > > I have progressed a little with the design and modeling of the machine I > asked for help simulating. It has evolved a little over the last few > weeks. have a look at the pdf attached for pictures Another month and I > will be back to the world of fast Internet and I should be able to try > some of the previous suggestions for simulation.
Looks nice :) Since you have 6 actuators, your design will definitely be overconstrained if you only want 5 degrees of freedom. But I have the feeling that you could actually get 6 DOF by having the nominal angle in C about 60° (or even 90°) rotated in respect to the picture you attached. Since you can control the distance between all the joints on the center plate to their slides, this will allow you to rotate the tool plate as well. Even if you don't need rotation around C, I'd suggest to implement this angular offset, as it will most probably make the system more robust against torsional forces. Similar to how hexapods are done. > > The big question is: > > Could linuxCNC something control this? or do I have 2 many variables > with no fixed base position? When you get the kinematics right (which I'd expect to be about midways between the complexity of a delta kins and a hexapod), this should be no problem. > > A few other questions and thoughts I would like to pick your collective > brains on. > > How would you home something like this? As with a hexapod, forward kins are complicated and may have many (practically) invalid joint positions. I'd suggest to have the following homing switches: 1. One for every pole at the top end, which gets activated when the upper slide reaches the top limit. 2. One between every two slides on a pole which is closed when the slides are closer than the nominal position (shown in your PDF) and open otherwise. This must NOT be limit switches, as movement in both directions over the activation point must be possible. Additionally, it will help to have an approximation of the distance between every two slides on a pole. This could be done rather easily by using a linear potentiometer. You could then implement the homing about as follows, using coordinated movements in the XYZABC space (not on a joint basis): 1. When the system gets turned on, all you know is that it must be in some mechanically feasible position. So start by a linear move towards Z+, meaning that all slides run synchronously upwards. Stop as soon as one upper slide reaches the top homing switch. 2. Use the feedback from the pots as an initial guess of the slide distances for the kinematics. Using those, perform a move in the XY plane perpendicularly away from the joint which has it's top slide homed, i.e. moving approximately towards the center position. Stop as soon as the second upper slide hits the top homing switch. Since this movement will also adjust the distance of the slides on a pole, you may also detect one or the other homing switch edge of those. Use this information to correct the joint position guesses. 3. Do the same as in step 2, but this time with a movement perpendicular to the connecting line of the two homed poles. So you should eventually also reach the top homing switch of the third pole, meaning that all upper slides know their positions and the tool is roughly centered. 4. Move all slides down by a few millimeters to get some "working range" for the last step (i.e. coordinated move towards Z-). 5. Perform a movement to X0Y0A0B0C0, i.e. to the approximate nominal position with an aligned tool base. This is possible using the information form the pots. 6. Perform moves from C-30 to C30 (or less, depending on the accuracy of your potentiometer feedbacks), so that all homing switches of the lower slides see at least one edge. As soon as this happens, all joints are homed. All this will require that the tool can rotate around a C axis (which in my terminology is perpendicular to the machine table, NOT aligned to the tool axis). This means that it will be required to have this angular offset of the tool base plate as described above. > I stile haven't simulated the movement digitally yet or built a complete > cardboard and drinking straw model, but i think i have finally settled > on approximate geometry any questions or comments are gladly welcome. > Has any one seen any thing like this before I am sure I am not the > first? Is there any grate big flaw with this logic that I have missed? At least from my feeling, I think that this should work with the mentioned adaptation. I'm not very familiar with complex machine kinematics, however. I find it a cool construction :) No idea if someone else already built that. Good luck! Regards, Philipp
signature.asc
Description: OpenPGP digital signature
------------------------------------------------------------------------------
_______________________________________________ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
