Problems measuring time between two pulses

[zanor]

Just for fun I'm trying to make a digital RPM meter for my car. But I'm not able to get exact readings.
I have measured with an oscilloscope that there should be ~0,06802721 sec (14,7Hz) between the pulses when my engine is idling. "TMR_count" in my source code should be 66 then. But I'm getting 61. It is ~0.00512 sec too slow, why?
Is there some delays in timer/interrupt operation I don't know?

I am using a PIC16F84a on a breadboard. Could the "long" way from the crystal to the microchip slow it down? Cables are ~3-4 cm...

Here is my code, in MikroC format.

#include <built_in.h>                                                           // For Lo(); & Hi();

int i = 0;
int TMR_count = 0;  // 1x = 0.001024 sec
int TMR_value = 0;  // 1x = 0.000004 sec
int RPM = 0;
float TIME = 0;

void interrupt() {
  if (INTCON.INTF)
  {
    if (i <= 63)
    {
      TMR_value = TMR0;
      TMR0 = 0;

      TIME = (TMR_count * 0.001024) + (TMR_value * 0.000004);                   // Time between two sparks
      RPM = (1 / TIME) * 60;                                                    // RPM , 2 coiler and wasted spark = 1 spark/revolution

      EEPROM_WRITE(i, TMR_count);
      i++;
      EEPROM_WRITE(i, TMR_value);
      i++;
      
      TMR_count = 0;
    }
    INTCON.INTF = 0;
  }
  if (INTCON.T0IF)
  {
    TMR_count++;
    INTCON.T0IF = 0;
  }
}


void main() {
     PORTA = 0;
     TRISA = 0b00001000;
     PORTB = 0;
     TRISB = 0b00000001;

     PORTB.B3 = 1;
     Delay_MS(1000);
     PORTB.B3 = 0;
     
     INTCON.RBIF = 0;

     INTCON.GIE = 1;                                                            // Enable Interrupts
     INTCON.INTE = 1;                                                           // Enable Interrupt on RB
     INTCON.T0IE = 1;                                                           // Enable timer 0 owerflow interrupt
     
     OPTION_REG = 0x81;                                                         // Pullups DISABLED!!!
     
     while(1);

     }

 

[FvM]

Executing an EEPROM write in your interrupt function results in an exhausted EEPROM within at least a few hours and also huge timing errors.

 

[zanor]

Executing an EEPROM write in your interrupt function results in an exhausted EEPROM within at least a few hours and also huge timing errors.

The EEPROM write is just for debugging, and is only filled once every time the chip is being reset.
I see now in the datasheet that it takes typicaly 4ms for one EEPROM write, that is WAY more than I thought.. Thank you :!:

I'm now going to make some chamges to my program and get back with the result

Added after 2 hours 51 minutes:

Cased closed, it was my EEPROM writes that delayed my timing.... Work fine now ;D

 

[enzo84]

Is there some delays in timer/interrupt operation I don't know?

zanor,
Yes there is a timer between the interrupt is produce and the below line:

TMR_value = TMR0;

This is because the microcontroller always jumps to the same memory location when the interruption occurs to process that kind of interruption is.

Hope, that helps, regards,
Enzo.