[From Bill Powers (930518.2030 MDT)]
Cliff Joslyn (930518) --
Faisal Kadri's model will be a very poor control system if it can
control at all. In the first place, he is confusing the reference
signal with a stimulus input to the system (the Input Blunder).
In the second place, if the reference signal is set to zero the
loop gain will be zero. In the third place, if the sign of the
feedback signal or the reference signal changes, the feedback
will change to positive feedback and the system will run away.
And in the fourth place, he is characterizing the behavior of
this system only for an impulse-stimulus. With an impulse
stimulus (zero duration, infinite amplitude, finite area), the
multiplier will receive a nonzero input only for an infinitesimal
time, after which one input to the multiplier will be zero, so
the output of the multiplier will be zero regardless of the other
input, rendering any feedback completely ineffective. Zero times
anything is zero. All he will get is the impulse-response of the
forward element, A. The slower feedback element H will have no
effect at all.
Note that the actual sensor (where the feedback path departs from
the output) isn't discussed at all, and there is no independent
disturbance applied to the input. Without a disturbance, there is
no way to show that control exists at all. Kadri is offering an
alternative to something he obviously doesn't understand. I give
him an F.
ยทยทยท
-----------------------------------------------------------
Bill P.