使用超声波和 ATMEGA32 进行距离测量

Question

I'm working on a Distance measurement program using an AVR microcontroller. I use a 16x2 LCD and an ultrasonic sensor along with ATMEGA32A. I wrote a code to display the distance from the Ultrasonic HC-SR04 on the LCD screen, but it gives me false readings, it increases the distance when the object is very near and vice versa. I just want an accurate reading.

Ultrasonic datasheet ATMEGA32A Datasheet

#include <avr/io.h>
#include <avr/interrupt.h>
#include <MrLcd/MrLCDmega32.h>
#define F_CPU 1000000
#include <util/delay.h>
#include <stdlib.h>
#define  Trigger_pin    PD0 /* Trigger pin */


static volatile int pulse = 0;
static volatile int i = 0;

int main(void)
{
    
    Initialise();
    DDRD = 0b11111011;
    _delay_ms(50);

    GICR |= 1<<INT0;
    MCUCR |= 1<<ISC00;
    
    int16_t count_a = 0;
    char show_a[16];

    sei();
    
    while(1)
    {
        PORTD |= (1<<Trigger_pin);
        _delay_us(10);

        PORTD &= ~(1<<Trigger_pin);
        count_a = pulse/58;

        Send_A_String("Distance Sensor");
        GoToMrLCDLocation(1,2);
        Send_A_String("Distance=");
        itoa(count_a,show_a,10);
        Send_A_String(show_a);
        Send_A_String(" ");
        GoToMrLCDLocation(13,2);
        Send_A_String("cm");
        GoToMrLCDLocation(1,1);
    }
}

ISR(INT0_vect)
{
    if(i == 1)
    {
        TCCR1B = 0;
        pulse = TCNT1;
        TCNT1 = 0;
        i = 0;
    }

    if(i==0)
    {
        TCCR1B |= 1<<CS10;
        i = 1;
    }
}

I tried to change the trigger pin definition and define it in the code itself but still no progress.

Update: I changed a bit more in the code but I'm getting hex values when the distance is more than 9, for example, 10 is being displayed as 1e.

This is for initialise function

void Initialise(void)
{        
    DataDir_MrLCDsControl|=1<<LightSwitch|1<<ReadWrite|1<<BipolarMood;  //these information will go towards the LCD
        
    _delay_ms(15);                              // Wait for the LCD to start
    
    Send_A_Command(0x01);   // to clear the screen
    _delay_ms(2);
    Send_A_Command(0x38);   // TO tell LCD about 8 data lines
    _delay_us(50);
    Send_A_Command(0b00001110); //Some cursor command
    _delay_us(50);
}

Answer 1

You are sending pulses at a very rapid rate (determined solely by the display update time), and they are asynchronous to the time/counter reset. You have no idea which pulse triggered the interrupt and it did not start at the same time as the timer.

I would suggest that you reset the counter at the start of the pulse, and capture the counter value on interrupt. When the time has exceeded the maximum range, send a new pulse:

First define some constants:

#define PULSES_PER_CMx100 (F_CPU * 100 / 68600)
#define MAX_RANGE_CM 300
#define MAX_RANGE_COUNT ((MAX_RANGE_CM * PULSES_PER_CMx100) / 100)

Then your measure/display loop might look like:

    pulse = 1 ; // dummy start
    GICR &= ~(1<<INT0) ;  // Disable INT0

    for(;;)
    {
        // Ready for new measurement?...
        if( pulse != 0 )
        {
            // Send pulse and reset timer
            PORTD |= (1<<Trigger_pin) ;
            pulse = 0 ;
            TCNT1 = 0 ;
            _delay_us(10);
            PORTD &= ~(1<<Trigger_pin) ;

            // Wait for echo pulse interrupt...
            GIFR |= 1<<INTF0;  // Clear INT0 pending flag
            GICR |= 1<<INT0 ;  // Enable INT0
        }
        else  // When measurement available...
        {
            int distance_cm = pulse * 100 / PULSES_PER_CMx100 ;

            // display distance
            ...
        }

        // If out of range, timeout, send a new pulse
        if( TCNT1 > MAX_RANGE_COUNT )
        {
            // Force a new pulse to be triggered
            pulse = 1 ;
        }
    }

And the ISR:

ISR(INT0_vect)
{
    pulse = TCNT1;       // Capture time on interrupt
    GICR &= ~(1<<INT0) ; // Disable further interrupts
}

Now bear in mind that that method will take measurements as fast as possible and since you are displaying them for human reading, that is rather unnecessary. You might simply put a delay in the loop - making the pulse timeout unnecessary, or better you could take the mean of multiple measurements to get a more robust measurement, or use a moving average window, with outlier rejection.