Electronic Analog Computing for Feedback Control

YongSik Jeon
[email protected]

Samuel Chun
[email protected]

Computer Science Department

Indiana University
Bloomington, IN 47405

Prof. Jonathan W. Mills

CSCI B644 VLSI Fall 2003

Abstract

This paper describes one of the uses of analog computing in feedback

control systems. Examples of feedback control systems are discussed, and
it shows how the feedback control system improves with different methods
for controlling the input from the output signal, and also shows the
advantage of using analog computing for feedback control systems over
digital computing and an example of an Extended Analog Computer
(EAC) for an analog feedback control system with its benefits.

1. Introduction computers, feedback control systems

have been rapidly developing since the
In recent years, many areas have 1980s and became more applicable in
switched from analog to digital. As we diverse situations.
can see, cell phones, music CDs, DVD
movies, cable TVs, etc. Computing areas One example of analog feedback control
is not an exception thanks to the cheap system is found in our body. Actually,
digital processors, but there are some our bodies have numerous feedback
areas where analog computing is controls. But one of the most obvious
continually being used. One of those one would be the temperature regulator.
areas is feedback control. When we get hot we sweat so the body
cools down and when we get cold we
Feedback control is the mechanism to shiver and burn stored energy so the
maintain an equilibrium state. Without body heats up. Even a small temperature
this there could be no manufacturing, no change in our body can be fatal. So, it is
vehicle, no computers, and no regulated very important that this feedback control
environment. With the help of digital system works properly.
Another example of a simple analog 2. Actual Fields
feedback control system is found in our
bathroom. Whenever we use the toilet Building Automation System
we flush. The flushing mechanism is a
feedback control system. When we flush Most modern buildings use a centralized
the feedback control opens the water control system. It controls temperature
valve to fill up the tank for the next flush and humidity, lights, elevators, fire
and when the water fills up it will shut sprinkler, etc.
the valve off. This feedback control
system was first invented in B.C. 270 by Suspension controller for Automobiles
the Greeks. Their motivation was to
measure time. They had water tanks with Automobile suspension is a shock
water flowing out of them. And they absorber that links the car body to the
needed to keep a constant depth in the axle. It dampens the jarring sustained in
water tanks to make the water flow a moving automobile. Active suspension
constantly. has a controller with a spring and a
damper, which is included in a passive
In this paper, we are not interested in suspension, to control absorbance for a
biological feedback control, nor are we smoother ride.
interested in feedback controls that does
Object tracing control for radar system
not use electricity such as the one in the
toilet. We are only interested in feedback
Aircraft during the 1950s and early
controls that use electronic circuit based
1960s carried analog computers as part
analog computing to decide the input
of their radar equipment. These were
from the output. These systems can be
used to provide targeting information for
found in automotives, satellites, air
guns and missiles. The Heads Up
planes, air conditioners, etc.
Display (HUD) that projected
information onto a piece of glass in front
Why is analog computing more
of the pilot relied upon computer input
preferred for feedback control system
to help the pilot aim his guns or select
than digital computing in some areas and
his weapons.
what are the benefits of using an
Extended Analog Computer (EAC) for Flight control for airplane
the feedback control system? The
following sections will explain how Flight control system for airplanes
analog feedback control works and why controls the attitude and the speed of the
analog computing would continue to be plane. The General Dynamics (now
preferred in some areas of the feedback Lockheed-Martin) F-16, which entered
control systems, and the benefits of EAC service in the late 1970s and has been
feedback control system with an built in large numbers, was the first
applicable example. operational jet fighter to use an analog
flight control system. The pilot steers the
rudder pedals and joystick, but these are
not directly connected to the control
surfaces such as the rudder and ailerons.
Instead, they are connected to a "fly-by-

2
wire" flight control system. Three microphone, pressure and temperature
computers on the aircraft constantly transducers, chemical sensors)
adjust the flight controls to maintain the An actuator affects the plant. (e.g.
aircraft in flight and reply to the hydraulic, pneumatic, electric motors,
commands from the pilot. pumps, heaters, aircraft control)

Flight control for satellite A controller or control processor

processes the sensor signal to drive the
The Apollo Lunar Module also used an actuator. There are analog and digital
analog computer flight control system controllers. (e.g. human operator,
and other U.S. spacecraft such as mechanical controller, electro-
Mercury, Gemini, and Apollo all had mechanical controller, analog electrical
computer flight control systems. controller, digital processor)

Disturbance is a signal from external of

the plant that occurs unpredictably and
3. Feedback Control Systems disturbs the plant from reaching the pre-
specified level. (e.g. wind gusts,
Control systems are designed so that earthquakes, external shaking and
certain designed signals, such as tracking vibration, road surface variations)
errors and actuator inputs, do not exceed
pre-specified levels. Feedback is widely The control law or control algorithm is
used in automatic control systems. Basic the algorithm used by the control
diagram of a feedback control system is processor to derive the actuator signal.
shown in figure 1.

4. Feedback Control

Common setup for feedback control

system is shown in figure 2.

Figure 1. Diagram of feedback control system

The system to be controlled is the plant.

(e.g. CD player, disk drive mechanics,
aircraft, missile, car suspension, engine, Figure 2. Feedback control
industrial process, elevator)
P is the plant.
A sensor measures the quantity to be C is the controller.
controlled. (e.g. radar altimeter, GPS, u is the plant input (actuator signal).
strain gauge, accelerometer, tachometer, y is the plant output (sensor signal).

3
r is the reference or command input t
u (t ) = K p e(t ) + K i e(τ )dτ
(what we'd like y to be). 0

e = r - y is the (tracking) error and can

also be other signals e.g., disturbances K i : integral gain
and noises.
Proportional-Integral controller is used
The goal of the feedback control is to to eliminate steady-state error, but if
make y r, i.e., e small (despite integral gain is mistuned, the system can
variations in P, disturbances). become unstable and the response time
can be slower.
Proportional (P) controller

The equation of the P controller in time

domain:

u (t ) = K p e(t )

K p : proportional gain

Proportional control is the most basic

control that is always used in the
controllers. This is easy to develop, but
cannot remove steady-state error.

Graph 2. Response characteristic of a PI

controller

Proportional-Derivative (PD) controller

The equation of the PD controller in

time domain:

d
u (t ) = K p e(t ) + K d e(t )
dt

K d : derivative gain
Graph 1. Response characteristic of a P PD control increases the stability of the
controller system and makes the response time
faster, but with the presence of noise in
Proportional-Integral (PI) controller the system, it can increase the
differential portion of the equation
The equation of the PI controller in time resulting in negative effect for the input.
domain:

4
 More than 80% of the feedback
controllers are PID controllers in the
actual fields, because its performance is
good and it is easy to tune.

Actual control of the PID controller is

the combination of P, I and D controls.

Graph 3. Response characteristic of a PD

controller

PID (Proportional-Integral-Derivative)

The equation of the PID controller in

time domain:

d t Graph 4. Response characteristic of a PID

u (t ) = K p e(t ) + K d e(t ) + K i e(τ )dτ controller
dt 0

5
Analog controllers

Generally a control task is performed by specifically combined ICs. Simple controls can
be carried out using analog ICs, such as operational amplifier circuitry. An integrated
circuit (IC) is the basic component of "Control Electronics". IC is a small electronic
device made out of a semiconductor material.

R PF 1 d
VOUT (t ) = V IN (t ) + V IN (t ) dt + R D C D V IN (t ) (1)
R PI RI C I dt

Figure 3. An example of an analog circuit for PID control. This analog circuit calculates
PID control response according to Equation 1. Sensor signal VIN is amplified by IC1 (P
mode), integrated by IC2 (I mode), and differentiated by IC3 (D mode), and all three
modes are summed by IC4 to derive control response VOUT.

6
5. Extended Analog Computer (EAC) the conductive sheet, logic gates can be
Feedback System created.

EAC is a very powerful machine, and at This is the mathematical model of the
the same time it is very simple. It is extended analog computer (this model
easily programmable, unlike General has no boundary conditions):
Purpose Analog Computers (GPAC), by
manipulating the Lukasiewicz logic Ii I
arrays (LLA). Not like GPACs, an EAC V p = − j

sources
d i2 sinks
d 2
j
is a Turing-complete model, and this
simple and powerful machine can be
used to directly solve problems like Voltage at any point p is the sum of all
Partial Differential Equations in systems source (Ii) voltages divided by the
and many more. distances to the point minus the sum of
all sink (Ij) voltages divided by the
EAC implementation distances to the point. Until recently,
people believed EACs were not possible
to implement.

EAC feedback control system

Figure 4. Basic diagram of an EAC

Extended analog computers can be

implemented on a conductive sheet with
number of Lukasiewicz logic arrays
(LLA). One or more sources or sinks are
place on the conductive sheet, then
current flows through the conductive
sheet making a gradient pattern. The
current is measured from a chosen
measure point or points. Then the
measured current is fed into the LLAs,
which outputs an analog signal
according to the LLA function.
Figure 5. Example of an EAC feedback control
system for a temperature control
LLAs are somewhat similar to the digital
logic gates, but there is only one input. It
also uses analog signal for its input and Figure 5 is diagram of an EAC feedback
output. There are mainly 27 LLA control system for a cooling system,
functions. With these LLA functions and which can be for a building temperature
controller.

7
There are total of 6 sensor inputs, which converters, and digital signal processors
are from different parts of the floor, to which cost additionally. Even though,
the conductive sheet. The voltage of the the price of digital chips have gone
current from each sensor will reflect the down it is still more expensive. Second,
temperature of that region. This will it consumes less energy. Many feedback
create a gradient manifold of control systems are found in small items,
temperature of the room. such as a MP3 player or a micro robot.
For these kinds of items, power
There are two cooling system connected consumption is a very important issue.
to this EAC by two LLA functions. The Analog feedback control system uses
current from the measure points are fed less power than digital feedback control
into each LLAs. LLAs will decide system. Third, analog system performs
according to its function to turn the better in speed than digital system. Up to
cooler on or not. a certain precision, they can outperform
the digital system.
LLA functions are depicted in Figure 5
in a square box. The x-axis represents There are couple disadvantages of using
the input voltage and the y-axis an analog feedback control system as
represents the output voltage. The LLA well. First, it lacks precision. Analog
function for measure point (S) is feedback control system will outperform
designed to be more sensitive to the digital system, but if the system
temperature change. This can be useful requires higher precision, a digital
if that region is near the window or it is system will be a better solution. Second,
facing south. This region will get heated there is a limitation to apply complicated
faster than other regions of the room due non-linear equations into the systems.
to insulation or the sun. Third, it cannot control a sophisticated
system. It is more efficient to use a
As shown in this example, EAC can be digital system for a sophisticated system
easily used to build a fuzzy cooling such as an electronically controlled
system, which not only saves power but engine that has many sensors and
is efficient and in-expensive. controllers.

We will see many more products with

feedback control which is a kind of
6. Conclusion artificial intelligence. For example, a
chair that adjusts according to the
Many feedback control systems have physical properties of the individual by
switched from analog to digital, because itself. As described in the paper, many
of the power of digital computers. There areas have been digitalized including the
are some advantages and disadvantages feedback control systems, but because of
of using an analog feedback control the advantages of the analog feedback
system. The first advantage is that control system, it will continue to be
analog systems are simple. It also means used, and with EAC feedback control
they are cheap and easy to build. Digital system, simpler and better A.I. system,
system needs A/D converters, D/A like a fuzzy logic, would be possible.

8
References

[1] Carnegie Mellon – PID Tutorial:

http://www.engin.umich.edu/group/ctm/PID/PID.html

[2] Brion Christine, DongInn Kim. Simulating the Recticular Formation via Extended
Analog Computer. Computer Science Department, Indiana University.

[3] C. A. Desoer, M. Vidyasagar. Feedback Systems: Input-Output Properties.

Academic Press, New York, 1975.

[4] R. C. Dorf. Modern Control Systems. Addison-Wesley, 1998.

[5] G. F. Franklin, J. D. Powel. Feedback Control of Dynamic Systems. Addison-

Wesley Publishing Co. 1986.

[6] B. Friedland. Control System Design: An Introduction to State-Space Methods.

McGraw-Hill, 1986.

[7] W. H. Gwon, O. G. Gwon, G. S. Hong, J. H. Lee. Control System Engineering.

Chungmoongak, 1999.

[8] I. M. Horowitz. Synthesis of Feedback Systems. Academic Press, New York,

1963.

[9] Hugh Jack. Dynamic System Modeling and Control. 2003.

[10] F. L. Lewis. Applied Optimal Control and Estimation. Prentice-Hall, 1992.

[11] A. I. Mees. Dynamics of Feedback Systems. Wiley, New York, 1981.

[12] Patricia Mellodge. Feedback Control for a Path Following Robotic Car.
Electrical Engineering, Virginia Polytechnic and State University, 2002.

[13] Jonathan W. Mills. Polymer Processors. Computer Science Department, Indiana

University.

[14] Jonathan W. Mills. Programmable VLSI Extended Analog Computer for

Cyclotron Beam Control. Computer Science Department, Indiana University.

[15] Heather Roinestad, Camilo H. Viecco, Ian Bobbitt, Adam Miller, Stefan
Obereichholz-Bangert, Ryan R. Varic. Study of a Model of an Extended Analog
Computer. Computer Science Department, Indiana University.