Control-oriented modeling for backlash
Backlash is complex; its model for control design is not.
Kevin C Craig, PhD -- EDN, August 25, 2011
In everyday language, the word “backlash” sounds as undesirable
as its meaning: a strong adverse reaction or a violent
backward movement. In engineering, the situation is no
different. Backlash, the excessive play between machine parts,
as often occurs in gears and flexible couplings, is undesirable
and usually exists with compliance. It gives rise to inaccuracies
in the position and velocity of a machine and to delays
and oscillations.
For any physical system, a hierarchy of models is possible, from the most real, most complex, and more difficult to solve to the less real, less complex, and easier to solve. The model that is most beneficial for control design is the least complex model that still retains sufficient accuracy to capture the gross dynamic behavior of the system. It is critical to strike a delicate balance between dynamic complexity and accuracy in the model.
Figure 1 shows the physical system under investigation with
the accompanying assumptions. In addition, assume that collisions
due to backlash are sufficiently plastic to avoid bouncing.
It is critical that the model capture the fact that the output from
the backlash element causes a torque on—not a displacement
of—the load inertia. This model also captures situations in
which the assumed massless compliant element has damping
(Reference 1). The following demonstration shows the importance
of this model: Imagine that you are compressing with your
hand a massless spring that possesses no internal damping. If you
were to suddenly move your hand away, the spring would stay in
contact because its response is instantaneous; being pure, it has
no mass or damping. If the spring has damping and you repeat
the experiment, however, the spring’s response would not be
instantaneous, and it would start to lose contact.
|
Reference |
|
Talkback






















