|
The most important machine characteristic required to obtain the best finish is smooth motion (which
also extends machine and tool life). The most important factors affecting part accuracy are the precision
of the mechanical components of the machine and the quality of the controller.
There are several
sources of rough machine motion, including limitations in unsophisticated controllers. Interestingly
good controllers can go a long way towards overcoming mechanical problems. Motion, on a two axis
machine, is generally comprised of a sequence of XY positions, and sometimes additional control positions,
e.g. arc centers. The XY position can be the end points of straight lines or arcs, or control points for
bezier curves and splines. The controller generates the machine movement between each point, and controls
the speed. The precise movement path is known as the trajectory - a series of sequential machine XY
positions, each defined for a fixed period, the trajectory time. In order to generate smooth machine
movements for all possible motion profiles, the controller must have a number of characteristics:
a) It must be capable of combining vectors to calculate machine acceleration and deceleration
profiles b) It must be capable of controlling the speed change at every vector intersect
and limit this when required c) To ensure that the speed of machine movement can be
precisely controlled it must be capable of splitting line, arc, bezier and spline movements into straight
line vectors d) These vectors must be short enough to avoid oscillations at vector intersections
e) It must be capable of looking ahead in the program to optimise the speed
f) It must be capable of blending vector intersections
g) It must have "S" curve acceleration and deceleration
h) In order to work with a wide range of machines and drives both the trajectory and servo
loop time must be configurable i) In order to minimise commissioning time and to provide
repeatable performance the controller must use digital servo tuning parameters.
In stepping motor systems, smooth performance requires pulse trains with small frequency jumps.
These cannot be achieved with timers, and software generated pulse trains; special hardware is required,
such as an ASIC. In a servo motion controller the servo loop attempts to make the motors follow
the trajectory. The traditional way to improve accuracy was to tighten the servo loop, this however causes
instabilities. Some modern controllers have servo loops based on axis velocities. Simple velocity based
loops however can decrease manufactured part accuracy especially when changing cutting forces are experienced.
When servo loop feed forward terms are implemented however, part accuracy is excellent.
The following graphs show the impact of velocity feed forward. Graph 1 was taken with no feed forward,
the actual velocity lags the desired velocity. Graph 2 shows the result with feed forward.
Graph 1 above; Graph 2 below
|
|
All mechanical systems have a natural resonant frequency, typically between 5 and 30Hz. If the control
signal is fed to the drive oscillates at or close to the machine's natural frequency, job finish is
impaired. Sophisticated controllers use digital filters to restrict the signal to the drive, overcoming
resonance. Digital controllers can exhibit problems due to quantisation; anti-quantisiation filters
overcome these problems. An integral term which removes steady state errors, can improve part
accuracy. Integral gain, however, has limitations, and if set sufficiently high to remove all steady
state errors during machine motion, instabilities result. An additional "integral limit" term is introduced
into the control loop to prohibit integral windup, this restores stability at the cost of impairing
part accuracy. Sophisticated controllers effectively vary the integral gain depending upon current
conditions, this reduces the potential for destabilising the control loop. Variable integral gain is
effectively achieved by dynamically changing the integral limits which in turn changes the integral time
constant. While the system is stable the integral limit is increased increasing the integral time constant
making the control loop integrate over a longer period. At the earliest sign of control loop instability
the integral limit is decreased, reducing the integral time constant and the impact of integral gain on
the system. The following graphs show the impact of integral gain when the machine moves in an
elliptical shape. Graphs 3, taken with no integral gain, show the trajectory positions as dots and the
machine position as a line. Integral gain was used in graph 4 where the actual machine position follows
the trajectory.
Graphs 3 & 4
New servo control algorithms minimise the computational overhead on the controller;
this maximises machine productivity, the motion smoothness, and the resulting part quality and accuracy.
These algorithms effectively more closely match the control loop to the machine being controlled and also
permit the servo loop to be looser. NEE Controls Ltd has developed PC based multi axis servo
tuning software for analysing and setting servo control loops. This software generates a number of plots
from data recorded by the controller while executing a test part. Once recorded the data is transferred
to the PC via a high speed serial link and plotted. The smoothness of resulting machine movement is
a measure of the quality of the controller, this can be very effectively shown by cutting complex shapes
in clear acrylic on a router.
|
