if-else statement not working

Question

So I'm facing this problem a lot in my code where the program is not getting into the if else loop properly and just returning the same answer again and again.

void loop() {
double X_value = ((double)(analogRead(X_pin)-487))/540.00;
double Y_value = ((double)(529-analogRead(Y_pin)))/540.00;
double theta_rad = atan((double)Y_value/(double)X_value);
double theta_deg = ((double) theta_rad)*180.00/3.1416;
double temp = (double)abs(tan((double)theta_rad));
if (temp == 0.00) {
temp2 = 0.00;
} else { 
 double temp2 = min(temp,1.00/temp);
}
Serial.print("Theta = ");
Serial.print(theta_deg);
Serial.print("\n");
Serial.print("temp 2 = ");
Serial.print(temp2);
Serial.print("\n");
delay(1000);
}

The answer keeps on coming 0 even when temp is non zero.

Screenshot of result

If anyone knows the reason please help...

• James Waldby - jwpat7: Please edit your question to show actual code used. What's there now has a compile error with temp2 not declared before use. Note, the temp2 declaration shown in the else branch has scope only within the else branch and is not available outside it.
• Edgar Bonet: Note that You may want to use atan2() instead of atan().
• KIIV: And also it's not a good idea to compare floats for equality, as they are not "exact" numbers. For example 0.00 can be anything between -0.0049999.... and 0.0049999..... (roughly)
• Greenonline: The initialisation code is missing. Where is temp2 declared? and X-pin and Y_pin
• Andrew: You use doubles for everything and then use a value of 3.1416 for PI? 1) in arduino doubles are only 32 bit floats not true 64 bit doubles so the constant casting to double is unnecessary. 2) a 32 bit float is accurate to more than 5 figures so that Pi value is the limiting factor in your accuracy.

Answer 1

I'm assuming that temp2 must be defined as a global or you'd be getting compiler errors.
You are re-defining temp2 within the else statement, that redefined value is then only local to that else and is lost as soon as you exit it.

void loop() {
double X_value = ((double)(analogRead(X_pin)-487))/540.00;
double Y_value = ((double)(529-analogRead(Y_pin)))/540.00;
double theta_rad = atan((double)Y_value/(double)X_value);
double theta_deg = ((double) theta_rad)*180.00/3.1416;
double temp = (double)abs(tan((double)theta_rad));
if (temp == 0.00) {
 temp2 = 0.00;
} else { 
/* Bug here. */
/* this temp2 is local to these {} and is lost as soon as we reach the } */
/* at the end of the else. */
/* Remove the word double and it should work */
 double temp2 = min(temp,1.00/temp);
}
Serial.print("Theta = ");
Serial.print(theta_deg);
Serial.print("\n");
Serial.print("temp 2 = ");
Serial.print(temp2);
Serial.print("\n");
delay(1000);
}

Personally I'd replace the entire if ... else with:

double temp = abs(tan(theta_rad)); // removed redundant casts to double
double temp2;
if (temp > 1.0) // 1/temp will be smaller if temp > 1
 temp2 = 1.0/temp;
else
 temp2 = temp;

It'll have exactly the same effect but avoids unnecessary floating point calculations