Rotational.ElastoBacklash Miscalculation

[JustinKGriffiths@…]

I believe there is a mistake in the Standard Library. Currently I’m using version 2.2.1 but it is also in the latest release, version 2.2.2. In the …Rotational.ElastoBacklash model the torque function has a mistake. The function currently reads as follows,

tau = if b2 > b_min then (if phi_rel > b2 then c*(phi_rel - phi_rel0 - b2)

+ d*w_rel else (if phi_rel < -b2 then c*(phi_rel - phi_rel0 + b2) + d*

w_rel else 0)) else c*(phi_rel - phi_rel0) + d*w_rel;

At first glance this all looks fine but when both the backlash (b2) and phi_rel0 are non-zero then issues arise. Take for instance if b2 > 0 and phi_rel0 > 0 and phi_rel > b2 then (phi_rel – phi_rel0 – b2) < 0 which causes phi_rel to increase instead decrease. The same happens on the negative side when phi_rel < -b2 and phi_rel0 < 0. I believe the function should read,

tau = if b2 > b_min then (if phi_rel – phi_rel0 > b2 then c*(phi_rel - phi_rel0 - b2)

+ d*w_rel else (if phi_rel – phi_rel0 < -b2 then c*(phi_rel - phi_rel0 + b2) + d*

w_rel else 0)) else c*(phi_rel - phi_rel0) + d*w_rel;

This will remove the discontinuities from the equation.

[otter]

The reported bug has been fixed in the maintenance version of the Modelica Standard Library 2.2.2.

In the coming version 3.0, component Backlash is completely newly implemented, to fix non-physical effects (e.g. "pushing" torque).

[otter]

The fix was performed with r1040. In version 3.0, the bug was not present, since already taken care off with the new implementation.