Question about resolution, effector volume and bicep angle

[tapir]

Thanks for this! Sorry if this is not the place to ask these questions but there are so little information about rotary delta projects in web, I thought I might try. Feel free to close the issue if it's not wanted.

1- Looking at the default configuration at https://12343954.github.io/fk/, it comes up with a resolution of 0.463 mm. Is this uniform for all axis? For 3d printer purposes, one would need a resolution of at least 0.1 or 0.05 mm for Z axis and probably even less for XY axis. Could we, let's say, slap on a reduction of 10:1 to the motors and practically have a resolution of 0.046 mm and call it a day? Or do we need to consider something else?

2- The page calculates a cuboid volume for the given configuration but the actual volume is not a cuboid AFAIK. For a traditional (non-rotary) delta it would be triangle like shape with curved sides and spherical void at the top. Would it be a similar shape for rotary-delta? Would it be possible to have this volume plotted in the web interface?

3- I see that limits of the bicep angle (motor angle) is in some cases more than 90 degrees, I can see that this is a theoritical limit but in practice might it have a bed effect as the forearm angle switches polarity?

Best

[12343954]

@tapir asking questions here is ok. and welcome.
this algorithm is not mine. I just put this page https://www.marginallyclever.com/other/samples/fk-ik-test.html here and enhanced some details, and then, implemented it in c++ and open source it.

now answer you questions:

1- I use this algorithm for my delta picking robot, not for 3d printer, so the resolution and accuracy requirements are not so high.
you can use reducer + (stepper/motor) to Improve accuracy.

2- the executable scope should be circular. but it labeled the cuboid, I guess it should be the safe range, not the max range.

3- this algorithm is rough, the result is theoretical value, many situations are not considered. the Endstop(hardware) can prevent the bad accidents.

In fact, after my actual test, this algorithm is inaccurate and needs to be improved.

[tapir]

Thanks for the responses.
Could you point where do you think the inaccuracies are?

Also found this article where it shows the resulting working volume. I'm guessing the cuboid is just a subset of this volume.