AVR Freaks

Disregard, too tired to reply.

I use the 328 a/d. In the init I use ADMUX = 0x40 to use vcc as the ref, and I use DIDR0 = 0x3f; to disable digital inputs on the a/d pins.

This is a very elementary and useless free-running mode ADC example that I had done a couple of years ago and rewrote today.  Wrote it for an ATmega1280 on an original Arduino Mega.  Used some funky retro macros that I found in the tutorials and in forum posts for setting up the LED pins .  I think Danni gets the credit for them.

/*
 * ADCwithapot.c
 *
 * Created: 2/14/2015 04:45:37 PM
 * Author: Larry
 *
 * Potentiometer is wired across +5V VCC and GND
 * with the wiper connected to ADC0.
 * LED1 and LED2 are used as status indicators
 * and are connected to PORTB pins 0 and 1
 * respectively.
 *
 * Code was written for ATmega1280 and comments reflect that
 *
 */

#include <avr/io.h>

struct bits
{
    uint8_t b0:1, b1:1, b2:1, b3:1, b4:1, b5:1, b6:1, b7:1;
} __attribute__((__packed__));

#define SBIT_(port,pin) ((*(volatile struct bits*)&port).b##pin)
#define SBIT(port, pin) SBIT_(port, pin)

enum {LOW, HIGH};

#define PORT PORTB
#define DDR DDRB

#define LED1            SBIT(PORT, PB0)
#define LED1_bm         (1 << PB0)
#define LED1_ACTIVE     HIGH
#define LED1_INACTIVE   !LED1_ACTIVE

#define LED2            SBIT(PORT, PB1)
#define LED2_bm         (1 << PB1)
#define LED2_ACTIVE     HIGH
#define LED2_INACTIVE   !LED2_ACTIVE


int main (void)
{
    // Variable to hold ADC result
    unsigned int adc_data;

    // Initialize LED pins as outputs
    DDR = LED1_bm | LED2_bm;

    // Set up ADMUX - ADC multiplexer selection register
    // Set ADC channel 0 - ADMUX bits 4..0 cleared - default after reset
    // Clear ADLAR bit in AMUX to have right adjusted result -default after reset
    // Voltage Reference Selection - AVCC with external capacitor at AREF pin - ADMUX REFS0 bit 6 set REFS1 bit 7 clear
    ADMUX = (1 << REFS0);

    // Set up ADCSRB - ADC Control and Status Register B
    // Set ADC Auto Trigger Source to Free-Running Mode - ADTS2 ADTS1 ADTS0 bits 2..0 clear - default after reset
    // Set ADC channel 0 - ADMUX bit 5 cleared - default after reset

    // Set up ADCSRA - ADC control and status register A
    // Set ADC prescaler to 128 - 125KHz sample rate @ 16MHz - ADCSRA ADPS2 ADPS1 ADPS0 bits 2..0 set
    // Set ADC ADC Auto Trigger Enable - ADCSRA ADATE bit 5 set
    // Enable ADC - ADCSRA ADEN bit 7 set
    // Start taking continuous samples by writing logical one to ADCSRA ADSC bit 6
    ADCSRA = (1 << ADPS2) | (1 << ADPS1) | (1 << ADPS0) | (1 << ADATE) | (1 << ADEN) | (1 << ADSC);

    // Wait for the first conversion to complete.
    // When the conversion completes the ADC Data Registers are updated
    // and the ADCSRA ADIF bit is set and will remain set until we clear it
    // or a reset occurs.
    while ((ADCSRA & (1 << ADIF)) == 0);
        
    // Since we are in Free-Running Mode new conversions will start immediately after the completion
    // of the current one and will repeat as long as we leave ADSC high.

    while (1)
    {

        // Delay seems to be required - Reason unknown at this time
        asm("NOP");
                
        // Read all 10 bits
        adc_data = ADCW;

        // The compare values are arbitrary and only used to demonstrate that the code works

        if (adc_data == 0)
        {
            LED1 = LED1_INACTIVE;
            LED2 = LED2_INACTIVE;
        }
        else if (adc_data <= 512)
        {
            LED1 = LED1_ACTIVE;
            LED2 = LED2_INACTIVE;
        }
        else if (adc_data == 1023)
        {
            LED1 = LED1_ACTIVE;
            LED2 = LED2_ACTIVE;
        }
        else     // ((adc_data > 512) & (adc_data < 1023))
        {
            LED1 = LED1_INACTIVE;
            LED2 = LED2_ACTIVE;
        }
    }
}

EDIT: Changed code per discussions below.

Lee, I was hoping you would comment.  I knew that part of the code did not make sense, it has bothered me from day one, and even more so when I added the comments yesterday.  So what you are saying is that since it is free-running we don't check for any condition and only loop to read the data registers?  I need more coffee, that is why I am asking.

Actually without the check the code does not work correctly.  What critical piece of the puzzle am I missing here?

EDIT: I added asm("NOP"); in place of the check and now the code works.  I would still like to get your input on all of this. 

Actually without the check the code does not work correctly.  What critical piece of the puzzle am I missing here?

Define "correctly".  Without the check, the first conversion will not have completed, so the first result will be nonsense.  After that, a read of ADC/ADCW will always give the result of the latest conversion, with or without the check.

I guess the check >>does<< serve to burn a couple cycles,  The fact remains that ADIF will always be set so the check is always true.

I'm with Cliff and David -- I've rarely found a use for free-running over the years, and certainly none with multiple channels involved.  The only time I can remember putting it into AVR8 apps is when that is the only ADC channel being used, it is used to monitor raw power, and there is no need to conserve the last bit of supply power.

Lee, definition of correctly is; LED displays change with varying voltages applied to ADC0.  This was not occurring without the delay.  Interesting that a NOP would be enough of a delay to make it work though.

Thanks for the critique, now I can file that away as a very elementary and useless free-running mode ADC example.

Don't you think is worth 1000 words?

Usually the problem with an AVR a/d is there aren't ENOUGH bits, not that are two too many bits. Also, mentioning guns by text are not allowed. By extrapolation, I will assume that mentioning guns pictorially are also not allowed.