Thread: nan in output

Hi there,

I am having difficulty getting the right output from this program, it just takes some data and works out the rms. However, the input is supposed to be 16-bit data, hence the forcing to int of the test data which of course should really end up as a float. I don't want decimals at that part of the program.

The program is doing (finally) almost exactly what I am after, but this one bug on some of the output. I think the problem is either:
declaring my variables incorrectly, I need the variable total to get be 64-bit which I thought was long long.

or, there is a problem with the square rooting of some of the data that the program is treating as an imaginary number as such.

Code:

/* Program to compute the RMS value or 1024 Samples 16-bit 2's comp data. 
This program has been designed to run a rolling block of samples so 
that once the initial block of 1024 samples has been aquired the program 
rms is then calculated each time a new sample is added and the oldest sample
is removed from the block*/
#include <stdio.h>
#include <math.h> // Library required for using sqrt function

#define MAX 10 // sample block, in this case 1024
#define DATA 50 // size of test data

//function to calculate rms and print output to screen
float rms(long total)
{ 
     printf("rms = %f\n", sqrt(total / MAX));
     return 0;
}

/* Function to accept input into the program. The input is stored as a 
squared number to simplify the process of swapping the oldest sample 
with the newest sample. */
void input(long samples[]) 
{
     int i = 0;
     int j = 0;
     int value;
     long long total;
     int test[DATA];

     for (j = 0; j < DATA; ++j) { // This loop supplies test data to mimic 
      test[j] = 324 * sin(j); // input from circuit.
      printf("%d\n", test[j]);
     }

     while (i < DATA) {
      total -= samples[i]; // Removes oldest sample from array
      value = test[j];     // Gets new data, in this case from the array "test".
      samples[i] = value * value;
      total += samples[i];
      rms(total);
      ++j;
     }
}

// Function to initialize array for 1024 samples
void init()
{
     int i;
     long samples[MAX];
 
     for (i = 0; i < MAX; ++i) {
      samples[i] = 0;
     }
     input(samples);
}

// Program start
int main(void)
{
     init();
     return 0;
}

Kind regards

P166

Oh yes of course. Well I've fixed this as best I could, but the nan error is still there and I am getting a Segmentation Fault 11 also.

What I am trying to do, si use a rolling array of at the moment 10, that will pass through 50 lines of data. This is to be scaled up to 1024 samples with at least twice the amount for test data. But at present I am using two arrays of 10 and 50. As far as I am aware, a segmentation fault is trying to access unallocated memory. So either I am going outside the bounds of my arrays or the numbers are getting larger than the assignment types I am giving them. Here is the current program.

Code:

/* Program to compute the RMS value or 1024 Samples 16-bit 2's comp data. 
This program has been designed to run a rolling block of samples so 
that once the initial block of 1024 samples has been aquired the program 
rms is then calculated each time a new sample is added and the oldest sample
is removed from the block*/
#include <stdio.h>
#include <math.h> // Library required for using sqrt function

#define MAX 10 // sample block, in this case 1024
#define DATA 50 // size of test data

//function to calculate rms and print output to screen
float rms(long total)
{ 
     printf("rms = %f\n", sqrt(total / MAX));
     return 0;
}

/* Function to accept input into the program. The input is stored as a 
squared number to simplify the process of swapping the oldest sample 
with the newest sample. */
void input(long samples[]) 
{
     int i = 0;
     int j = 0;
     int value;
     long long total;
     int test[DATA];

     for (j = 0; j < DATA; ++j) { // This loop supplies test data to mimic 
      test[j] = 324 * sin(j); // input from circuit.
      printf("%d\n", test[j]);
     }

     while (i < DATA) {
      if (i < MAX) {
           total -= samples[i]; // Removes oldest sample from array
           value = test[j];     // Gets new data, in this case from the array "test".
           samples[i] = value * value;
           total += samples[i];
           rms(total);
           ++j;
           ++i;
      }
      else
           i = 0;
     }
}

// Function to initialize array for 1024 samples
void init()
{
     int i;
     long samples[MAX];
 
     for (i = 0; i < MAX; ++i) {
      samples[i] = 0;
     }
     input(samples);
}

// Program start
int main(void)
{
     init();
     return 0;
}

ok, this is more serious than I thought. This program was running correctly, until I tried to scale it up to roll through the samples. I have just set the variable of j = 10 as the only input to the program, so the result should start at about 3. something, and soon get to 10 and stay at 10, why am I getting 4429.198799?

Right, I have rethought the input function into a slightly more logical structure and it seems to work now. I would still value peoples input as to whether I am missing something.

Am I right in that the previous code was messing with the arrays and accessing them out of their range?

Code:

/* Program to compute the RMS value or 1024 Samples 16-bit 2's comp data. 
This program has been designed to run a rolling block of samples so 
that once the initial block of 1024 samples has been aquired the program 
rms is then calculated each time a new sample is added and the oldest sample
is removed from the block*/
#include <stdio.h>
#include <math.h> // Library required for using sqrt function

#define MAX 10 // sample block, in this case 1024
#define DATA 50 // size of test data

//function to calculate rms and print output to screen
float rms(long total)
{ 
     printf("rms = %f\n", sqrt(total / MAX));
     return 0;
}

/* Function to accept input into the program. The input is stored as a 
squared number to simplify the process of swapping the oldest sample 
with the newest sample. */
void input(long samples[]) 
{
     int i;
     int j;
     int value;
     long total;
     int test[DATA];

/* This for loop is here to represent input data to the program from a circuit.
a sine wave with an amplitude of 340 has been used with the expected result of 
around 240, the same as UK mains voltage. */
     for (j = 0; j < DATA; ++j) {  
      test[j] = 340 * sin(j); 
      printf("%d\n", test[j]);
     }

     j = 0;
     while (j < DATA) {
      for (i = 0; i < MAX; ++i) {
           total -= samples[i]; // Removes oldest sample from array
           value = test[i];     // Gets new data, in this case from the array "test".
           samples[i] = value * value;
           total += samples[i];
           rms(total);
      }
      ++j;
     }
}

// Function to initialize array for 1024 samples
void init()
{
     int i;
     long samples[MAX];
 
     for (i = 0; i < MAX; ++i) {
      samples[i] = 0;
     }
     input(samples);
}

// Program start
int main(void)
{
     init();
     return 0;
}

Kind regards

P166

Yes definitely memory problems, as soon as I put set the amount of samples to 1024, the result comes in at 370725.something. I am expecting 230.something.

What am I doing wrong?

Yes you are absolutely correct there are a few things wrong with that for loop for sine. But as it is doing it's job perfectly, it giving me 16-bit signed values ranging from 0 up to 324 and back down the other way, it is suffice. This is not the problem.

The problem is that if the sample rate is kept below 300, I get a perfect result approximately 221.743 This is the correct result for this calculation.

If I change MAX which is the size of my samples, to 1024, the result is wrong it is 370725.1206. I am curious to know why this is happening. I am sure it is not my sine wave with the decimal values omitted.