Mon Oct 6 01:18:51 CEST 2008
Something I forgot about modeling motors is that for AC induction
motors, the torque-speed curve is quite nonlinear.
First, an asynchronous AC motor is basicly a transformer where the
load of the secondary is the rotary dynamic attached to the axis.
This load can be modeled as a torque -> angular velocity transducer.
T -> LOAD -> w_L
The controller that closes this loop maps w_L to T through the the
motor's torque-speed T/w characteristic, which has two inputs: voltage
V and synchronous velocity w_s.
w_L -> M -> T
The nonlinearity between w_L/w_s and T is motor-dependent.
This was helpful:
Torque can be estimated from the motor current+voltage. This allows
to build a controller with simple measurement sensors, but doesn't
allow zero-speed control.
( Explain the flux-linkage estimation. )
With accurate speed/position measurement better control is possible:
"AC motor characteristics require the applied voltage to be
proportionally adjusted whenever the frequency is changed in order
to deliver the rated torque. For example, if a motor is designed to
operate at 460 volts at 60 Hz, the applied voltage must be reduced
to 230 volts when the frequency is reduced to 30 Hz. Thus the ratio
of volts per hertz must be regulated to a constant value (460/60 =
7.67 V/Hz in this case)."
(The idea behind this being that sending DC through a coil is a bad
idea: when the frequency is reduced, the reactive part of the
impedance goes down, which makes the resistive part more significant,
resulting in more heat dissipation.)