CCS/AWR1843BOOST: Some questions about the function convertSNRdBToVar in the MRR demo

Tool/software: Code Composer Studio

Hi,

  In the MRR demo,there is a function convertSNRdBToVar,which is defined as follows:

float convertSNRdBToVar(uint16_t SNRdB,uint16_t bitW, uint16_t n_samples, float resolution)
{
    float fVar, RVar;
    float scaleFac  = (n_samples*resolution);
    float resThresh = 2 * resolution * resolution;//question 2
    float invSNRlin = antilog2(-SNRdB, bitW) * 2; // We assume our estimator is 3dB worse than the CRLB.
    
    /* CRLB for a frequency estimate */
    fVar = (float)invSNRlin * (6.0f/((2.0f*PI_)*(2.0f*PI_))) * recipsp((n_samples*n_samples - 1));
        
    /* Convert to a parameter variance using the scalefactor.*/
    RVar = fVar*scaleFac*scaleFac; //question 1
   
    if (RVar < resThresh)  //question 3
    {
        RVar = resThresh;
    }
    return RVar;
}

1. The variable  scaleFac is a scale factor,why is this variable introduced?

2.The variable resThresh is equal to 2 * resolution * resolution, what does that means?

3.After the variable RVar is obtained, why does RVar  compare with resThresh?

4.Generally,the conditon RVar < resThresh is almost satisfied in all cases, beacuse according to the formula:

RVar = fVar*scaleFac*scaleFac

        =(float)invSNRlin * (6.0f/((2.0f*PI_)*(2.0f*PI_))) * recipsp((n_samples*n_samples - 1))*(n_samples*resolution)*(n_samples*resolution)

        =invSNRlin* (3.0f/((2.0f*PI_)*(2.0f*PI_))) * recipsp((n_samples*n_samples - 1))*n_samples*n_samples*(2 * resolution * resolution)

because recipsp((n_samples*n_samples - 1))*n_samples*n_samples≈1:

        ≈invSNRlin* (3.0f/((2.0f*PI_)*(2.0f*PI_))) *(2 * resolution * resolution)

        =invSNRlin* (3.0f/((2.0f*PI_)*(2.0f*PI_))) *resThresh

because invSNRlin<1:

    =>  RVar <(3.0f/((2.0f*PI_)*(2.0f*PI_))) *resThresh

because (3.0f/((2.0f*PI_)*(2.0f*PI_)))<1:

    =>   RVar< resThresh

It seems like the CRLB is needless.

Thanks,

Regards,

Rata