IIRGen

Generates a low-pass or high-pass IIR filter.

HMPP can generate filters of the Butterworth and Chebyshev1 types. The Butterworth filter has a frequency response that is as flat as possible in the passband, and the Chebyshev1 filter has ripples in the passband.

The function interface is declared as follows:

HmppResult HMPPS_IIRGenLowpass_64f(double rFreq, double ripple, int order, double *pTaps, HmppsIIRFilterType filterType);

HmppResult HMPPS_IIRGenHighpass_64f(double rFreq, double ripple, int order, double *pTaps, HmppsIIRFilterType filterType);

Parameters

Parameter	Description	Value Range	Input/Output
rFreq	Cut-off frequency	(0, 0.5)	Input
ripple	Ripple when the filter type is HmppChebyshev1	(0, INT_MAX]. If the filter type is HmppChebyshev1, the value range is (0, 28].	Input
order	Order of the generated filter	[1, 12]	Input
filterType	Type of the generated filter	HmppButterworth or HmppChebyshev1	Input
pTaps	Taps of the generated filter	Not null	Output

Return Value

Success: HMPP_STS_NO_ERR
Failure: An error code is returned.

Error Codes

Error Code	Description
HMPP_STS_NO_ERR	No error occurs.
HMPP_STS_NULL_PTR_ERR	Any of the specified pointers is null.
HMPP_STS_GEN_ORDER_ERR	The generated order is beyond the range of [1, 12].
HMPP_STS_FILTER_FREQUENCY_ERR	The cut-off frequency is not within the range of (0, 0.5).

The filter type can only be HmppButterworth or HmppChebyshev1. Do not use other values. Otherwise, the result may be incorrect.
The convergence of an IIR filter is sensitive to the quantization error of tap a. Therefore, when using taps of the double type for single-precision computing, you need to confirm that the filter is converged.

Example

The following is the sample code for generating a high-pass filter and a low-pass filter using HMPP.

The high-pass filter is of the Butterworth type, order 9, 1000 Hz sampling frequency, and 400 Hz cut-off frequency. Equivalent code in MATLAB: [b,a] = butter(9,400/500,'high');

The low-pass filter is of the Chebyshev1 type, order 9, 1000 Hz sampling frequency, 400 Hz cut-off frequency, and 0.4 dB passband ripple. Equivalent code in MATLAB: [b,a] = cheby1(9, 0.4, 400/500);

#define ORDER 9
#define TAPS_LEN ((ORDER + 1) * 2)
void IIRGenExample() {
    HmppResult ret;
    double taps[TAPS_LEN];
    ret = HMPPS_IIRGenHighpass_64f(400.0 / 1000.0, 0, ORDER, taps, HmppButterworth);
    printf("IIRGenHighpass ret=%d\n", ret);
    for (int i = 0; i < TAPS_LEN; ++i) {
        printf("%lf ", taps[i]);
    }
    printf("\n");
 
    ret = HMPPS_IIRGenLowpass_64f(400.0 / 1000.0, 0.4, ORDER, taps, HmppChebyshev1);
    printf("IIRGenLowpass ret=%d\n", ret);
    for (int i = 0; i < TAPS_LEN; ++i) {
        printf("%lf ", taps[i]);
    }
    printf("\n");
}

Output:

IIRGenHighpass ret=0
0.000006 -0.000057 0.000229 -0.000534 0.000800 -0.000800 0.000534 -0.000229 0.000057 -0.000006 1.000000 5.386221 13.378550 19.961682 19.623982 13.137028 5.973215 1.775180 0.312381 0.024765 
IIRGenLowpass ret=0
0.080020 0.720182 2.880728 6.721698 10.082547 10.082547 6.721698 2.880728 0.720182 0.080020 1.000000 4.135535 8.510626 10.746179 9.121897 5.230419 1.958336 0.349120 -0.039936 -0.041825

Parent topic: Filtering Algorithms