EE 476: Laboratory X

Servo Light Tracker


You will build a servo based light tracker which will turn to and follow the brightest light source that it "sees" using a PID algorithm to ensure fast and smooth operation of the servo.  The PID control should allow the light tracker to quickly turn to a bright light source without any overshooting or oscillations. 


The light sensors used in this lab are photo-resistors in can-type packages.  Photo-resistors change their resistance depending on the amount of incident light with typical values being ~.3K ohms under bright light to ~100 K ohms in darkness.  You will need to construct a voltage divider circuit and use the ADC to read the voltage to determine the light intensity seen by the photo-resistor.  An example circuit is shown below, you need to choose a value for R1 that will give you a wide range of voltages at the MCU input pin.  You should use the 5V external AREF for the ADC conversions. 

The photo-resistors should be mounted on top of the servo along with a suitable divider between the two light sensors according to the diagram shown below.  The leads of the photo-resistors should be connected to the above voltage divider circuit using banana clips or other suitable means.  


Additionally, you will need to control the servo using a PWM output from the MCU.  The servo motor expects a 5V pulse ever 20ms with a pulse width of between .5 and 2.5 ms.  The width of the pulse determines the servo's angular position with .5 ms representing 0 degrees adn 2.5 ms representing 180 degrees.  Servo motors create inductive spikes when they are turned on or off, so a protection circuit for the MCU control line and the power rail is necessary.  The first circuit shown below will reduce the inductive spikes on the power rail and make sure that the signals coming from the second opto-isolator circuit are clean.  The pinouts of the TLP371 are shown in the circuit diagram.  Note that the servo must be connected to a separate VCC and GROUND from the MCU. 


PID Control



The term ‘e’ is a function represents the error in the system, or in context of the light tracker problem, the difference between the outputs of the light sensors.  The output is a weighted sum of the error, the integral of the derivative function, and the derivative of the error multiplied by a ‘gain’ value.  The output is a signal sent into the system that ideally should minimize the error as time progresses.  Thus the output is used to correct the position of the servo to ideally obtain a zero error.  The three terms in the above equation are known as the Proportional, Integral, and Derivative terms respectively.  Using just the proportional term to drive the output tends to make the system overshoot as its trying to minimize the error.  In terms of the light tracker, it will make the servo oscillate back and forth as it tries to find a position that will make the inputs to both of the photo sensors equal.  Adding the derivative term will make the system settle faster since its value depends on the slope of the error.  Thus if the error is either increasing or decreasing rapidly, the derivative term will be a large component of the output to drive the system in the direction of zero error.  Thus the derivative term will make the system more stable by applying larger correction if there is a rapid change in the error.  The purpose of the integral term is to reduce the sum of squared errors in the system.  If the error becomes small and constant, the proportional term and derivative term will also be small; however, the integral term is essentially a running average, so it eventually accumulates all of the small errors and apply the appropriate correction to the output.  As a result, the integral term will increase the settling time as it slowly moves the system toward zero error.  Intuitively, the proportional and derivative term will make the light tracker quickly move toward the light source, and the integral term will handle the small corrections so that the light tracker will point exactly at the point where the light sensors report that same output.  Lastly, the ‘gain’ value is the overall gain of the PID control, technically, it can be combined with the and term weights.  Initially set all of the weights to a low value and increment them until the desired operation is reached.  If the tracker is constantly overshooting, turn down the 'gain' term and raise the term. 



Write a program and construct a circuit which will

You will demo all the features above to your TA.

Your written lab report should include:

Copyright Cornell University Dec 2005