# error signals and feedback

[From Bjoern Simonsen (010513;0920 EMT)]

I try to learn more about PCT modeling simulations in PowerSim. I have done
a feedback loop at level 1 and a sketch of five comparators linked together
with the loop at level 1. Over the fifth level comparator I have put in a
constant upper reference (over level 5) with a value 300. All variables are
measured in same units.

I have tried to find answers to some questions that arises when I made the
loops. My achievement was without success.
Therefore I bring my questions to you.

1. I understand it as if higher levels send their error signals to lower
level comparators. Here I have been wondering whether some signals are
linked direct e.g. from level 5 to
level 2, or if e.g. all error signals from level 5 influence comparators at
level 4, and all error signals from the comparators at level 4 influence
comparators at level 3 etc. to level 2. I can imagine a link where errors
from a comparator at level 5 to a comparator at level 4, and from level 4 to
level 3 have the value zero, but where the error from level 3 to level 2
have a certain value different from the values resulted from other than the
zero signals I referred to earlier. This would be an other description of a
direct link from level 5 to level 2.

Am I far away? I need some comments.

2. I understand it as if the feedback signals and disturbances at level 1
affects the input function which gives out a perceptual signal. A branch
carry perceptual signals to input functions at higher levels. Here I have
been wondering whether the error signals at higher levels constitute a
feedback signal to the input function at the same level? Or describes PCT
that we only have feedback signals at level 1?

Am I far away? I need some comments.

Back to my simulating loop.
The disturbance is a range of values changing between 373 to 46:

I find a structure in the comparator values when I run the loop. Here are
the values I can read from the loop the ten first readings:

upper_reference.........:300, 300, 300, 300, 300, 300, 300, 300, 300,
300,...
Input_function_level_5..:321, 10, 299, 1, 306, 6, 301, 4, 303,
6,...
Comperator:level_5......: 21,-290, -1,-299, 6,-294, 1,-296,
3,-294,...
Input_function_level_4..:321, 10, 299, 1, 306, 6, 301, 4, 303,
6,...
Comperator:level_4......:300 300, 300, 300, 300, 300, 300, 300, 300,
300,...
Input_function_level_3..:321, 10, 299, 1, 306, 6, 301, 4, 303,
6,...
Comperator:level_3......: 21,-290, -1,-299, 6,-294, 1,-296,
3,-294,...
Input_function_level_2..:321, 10, 299, 1, 306, 6, 301, 4, 303,
6,...
Comperator:level_2......:300, 300, 300, 300, 300, 300, 300, 300, 300,
300,...
Input_function_level_1..:321, 10, 299, 1, 306, 6, 301, 4, 303,
6,...
Comperator:level_1......: 21,-290, -1,-299, 6,-294, 1,-296,
3,-294,...
Output:function_leve_1..: 21,-290, -1,-299, 6,-294, 1,-296,
3 -294,...
Action_level_1..........:
0,-300,-580,-282,-582,-270,-570,-268,-560,-254,...
Controlled_variable.....:321, -10,-299, 1,-306,
6,-301, -4,-303, -6,...
Disturbance.............:321, 290, 281, 283, 276, 276, 269, 264, 257,
248,...

from the bottom (first reading) there is a disturbance = the controlled
variable. The action is zero as it was before there was a disturbance. The
reference for level 1 is the comparator value at level 2. Because the input
function and perception at level 1 is 321, the comparator at level 1 show
321-300=21. At level 2 the reference (comparator level 3) is 21 and the
perceptual signal is 321. We therefore get an error signal from the
comparator at level 2 to be
321-21=300 For me this is arithmetically OK.
The situation that surprises me is the structure of the comparator values.

The upper reference is 300 because I have put it so. The comparator values
at level 5 are different values under the simulation. Then the comparator
value at level 4 become the same value as the upper reference (300) during
the whole simulation. The same happens at level 2.
My problem is:

_Is it normal for all people who, in a situation, control a certain system
concept (level 11) with a high reference value also control programs (level
9) and categories (level 7) against the same high reference value? (see the
structure in the table over).

Bjoern