Servo Control

EECS 373 Fall 2005

Final Project

Servo Control

Members

Xuetao LI

Introduction

The project is to use two Photo resistors, to detect the angle of the light source, and turn the motor to point to the light source, when we are not using the Photo resistors; we can use 8-bit Nintendo Controller to control the servo to turn right and left.

High Level Design

Provide a high level perspective of your design.

For the Photo resistors detection mode:

I used the Photo resistor and put it in a series circuit and send the input into the ADC and capture the output using external interrupt. Using the data coming from the Photo resistors, I can calculate the direction of the light source.

Below is the high level function block diagram:

For the Nintendo Controller mode:

I use a state machine to generate the necessary signal to get the input from the game controller, and based on the controller input, the motor will turn to left and right.

Note: the design is based on the Futaba S3004servo motor which can only turn from 0 to 100 degrees; In the Demo it’s using a different one.

Below is the high level function block diagram:

Member Task Distribution

Xuetao LI: Hardware design

Software design

Circuit

Interfacing Nintendo Controller

Interfacing Servo motor

Testing and Calibration

Hardware Design

Indicate what high level functions are implemented in hardware.

For the Servo, I use processor timer to implement the pulse every 20ms.

For the Nintendo controller, I use a state machine to generate the latch and pulse, for the timing, I build two macros to generate the starting pulse, and the clock for the state machine.

The Photo resistors is the same list in the project overview slide, Jameco 202391, with resistances range from 5K (light) – 10M (dark),

Two resistors rated at 91KΩ are connected with the photo resistors in series to act as a voltage divider.

The Nintendo controller I am using is the generic 8 bit one, with two B and A button, and has different line routing as the one from Nintendo.

Software Design

The main reason using a software timer rather than a hardware timer is that I need to change the pulse width in a very precise manner, rather than just open and close.

I did anticipate some latency for the software timer, for my case, it is not a big problem for the servo; the servo will turn as long as the frequency is bigger than 50Hz (period 20ms). The period I got is just a little higher than the minimum value, so it is OK.

For the Photo resistor mode, I have Level 4 interrupt for the timer to generate the signal for the servo motor, I also used IRQ7 external interrupt for the interrupt from the ADC, in the IRQ7 interrupt I will save the returned value in a static variable, and also change the channel of the ADC. In the Level 4 interrupt service routine, for every 20ms I will write to the ADC to start the conversion. In the IRQ7 interrupt service routine, I will calculate the pulse width for the servo once I got the output from both channel, I will modify two static variables which store the pulse width of high(1) and low(0).

For Nintendo Controller mode, I use level 4 interrupt to generate signal for the servo just as before, in the interrupt routine I also read in the input from the controller and calculate the pulse width for the servo every 20ms.

All the software are in assembly code, There is only one C function mainc which is just to keep the program running for ever(basically a while(1) loop which nothing inside, same as the ones we have for last several labs).

Here is a high level flow chart of my software:

Photo resistor mode:

For the Nintendo Controller mode:

Results of the Design

I encounter several problems, with the Photo resistors, at the beginning, the resistances seem to be inconsistence, and it turned out to be because it is not soldered, so the wire sometimes will loose and the whole circuit resistance changed. It turned out that I can not sample two signal at the same time, then I have to sample one channel at a time, after I sample both of the channels, then I can do some calculate on it to determined the angle.

In the hardware part, I am using an asynchronized clear 16-bit counter which creates some problems, after Professor Breboh tell me to change to a synchronized one, everything is fine. For the controller mode I have to use hardware to generate two slower clocks, one is for the state machine; the other is for the START signal for the state machine (60Hz).

During the demo day, I found out the project is not working, but the output signal to the servo is correct, it turned out the servo is broken, after we change the servo works as it should be, but the new servo is a little different from the one I am designed around. After that, everything seems to work just as I expected.

Conclusions

It is quite possible to implement my idea with my component, but for the better precision and better sensitivity, I probably need a better ADC(better precision) or possibly two ADC so I can sample the two inputs at the same time rather than separately, and possibly add two more Photo resistors to do vertical detection.

The calibration is based on the situation with light turned on in the lab, so the flash light has to be close to make any differences on the photo resistance. I think I can do the calibration in a dark situation, but in the real world situation we will be in both light and dark situation, so we need a better ADC to extract more precision out, and also have one more Photo resistor to detect the lighting condition of the surrounding and determined the threshold value dynamically.

Media

Photos: