topic Calculating Sin/Cos with GFLIB in MCUXpresso IDE

Calculating Sin/Cos with GFLIB

aberger — Thu, 22 Mar 2018 00:53:31 GMT

I am trying to write some simple code making use of GFLIB for an ARM Cortex M4F MCU. I have successfully included the library files, and have some code making use of the GFLIB_Sin_FLT_C function from gflib_FP.h:

In the below code, I am trying to build a sine-wave array (and phase-shifted copies) and store them in current_U, current_V, and current_W.

int main(void) { float current_U[36]; float current_V[36]; float current_W[36]; int i;

for(i=0; i<NUM_PTS; i++) { current_U[i] = GFLIB_Sin_FLT_C(i*10*PI/180.0); // input argument in radians current_V[i] = GFLIB_Sin_FLT_C(i*10*PI/180.0 + 2*PI/3); // phase shift by +120deg current_W[i] = GFLIB_Sin_FLT_C(i*10*PI/180.0 - 2*PI/3); // phase shift by -120deg }

}

Strangely, the current_V calculation gives erroneous results:

Why would this be? Since GFLIB is a pre-compiled library, I can't examine the code that is calculating the sine value. Why would the calculation only fail for one of the function calls?

Re: Calculating Sin/Cos with GFLIB

Carlos_Mendoza — Thu, 22 Mar 2018 16:27:52 GMT

Hi Andy,

Could you tell us which MCU are you using?

Thanks in advance!

Best Regards,
Carlos Mendoza
Technical Support Engineer

Re: Calculating Sin/Cos with GFLIB

aberger — Thu, 22 Mar 2018 16:33:28 GMT

I'm using the MK64FN1M0VMD12 (as a part of the TWR-K64F-120M dev board), and am working in MCUXpresso v10.1.0 build 589.

Re: Calculating Sin/Cos with GFLIB

aberger — Thu, 22 Mar 2018 21:15:19 GMT

Problem solved:

The GFLIB_Sin_FLT_C and Cos functions require the input argument in the range of -pi to +pi. I was inputting values >+pi.

See: https://www.nxp.com/docs/en/user-guide/CM4FGFLIBUG.pdf

Discussion of GFLIB_Sin algorithm on pg 35