mbed official / mbed-dsp

CMSIS DSP library

Dependents:   performance_timer Surfboard_ gps2rtty Capstone ... more

Legacy Warning

This is an mbed 2 library. To learn more about mbed OS 5, visit the docs.

cmsis_dsp/FilteringFunctions/arm_biquad_cascade_df1_32x64_q31.c@2:da51fb522205, 2013-05-30 (annotated)

Committer:
emilmont
Date:
Thu May 30 17:10:11 2013 +0100
Revision:
2:da51fb522205
Parent:
1:fdd22bb7aa52
Child:
3:7a284390b0ce
Keep "cmsis-dsp" module in synch with its source

Who changed what in which revision?

UserRevisionLine numberNew contents of line
emilmont 1:fdd22bb7aa52 1 /* ----------------------------------------------------------------------
emilmont 1:fdd22bb7aa52 2 * Copyright (C) 2010 ARM Limited. All rights reserved.
emilmont 1:fdd22bb7aa52 3 *
emilmont 1:fdd22bb7aa52 4 * $Date: 15. February 2012
emilmont 2:da51fb522205 5 * $Revision: V1.1.0
emilmont 1:fdd22bb7aa52 6 *
emilmont 2:da51fb522205 7 * Project: CMSIS DSP Library
emilmont 2:da51fb522205 8 * Title: arm_biquad_cascade_df1_32x64_q31.c
emilmont 1:fdd22bb7aa52 9 *
emilmont 2:da51fb522205 10 * Description: High precision Q31 Biquad cascade filter processing function
emilmont 1:fdd22bb7aa52 11 *
emilmont 1:fdd22bb7aa52 12 * Target Processor: Cortex-M4/Cortex-M3/Cortex-M0
emilmont 1:fdd22bb7aa52 13 *
emilmont 1:fdd22bb7aa52 14 * Version 1.1.0 2012/02/15
emilmont 1:fdd22bb7aa52 15 * Updated with more optimizations, bug fixes and minor API changes.
emilmont 1:fdd22bb7aa52 16 *
emilmont 1:fdd22bb7aa52 17 * Version 1.0.10 2011/7/15
emilmont 1:fdd22bb7aa52 18 * Big Endian support added and Merged M0 and M3/M4 Source code.
emilmont 1:fdd22bb7aa52 19 *
emilmont 1:fdd22bb7aa52 20 * Version 1.0.3 2010/11/29
emilmont 1:fdd22bb7aa52 21 * Re-organized the CMSIS folders and updated documentation.
emilmont 1:fdd22bb7aa52 22 *
emilmont 1:fdd22bb7aa52 23 * Version 1.0.2 2010/11/11
emilmont 1:fdd22bb7aa52 24 * Documentation updated.
emilmont 1:fdd22bb7aa52 25 *
emilmont 1:fdd22bb7aa52 26 * Version 1.0.1 2010/10/05
emilmont 1:fdd22bb7aa52 27 * Production release and review comments incorporated.
emilmont 1:fdd22bb7aa52 28 *
emilmont 1:fdd22bb7aa52 29 * Version 1.0.0 2010/09/20
emilmont 1:fdd22bb7aa52 30 * Production release and review comments incorporated.
emilmont 1:fdd22bb7aa52 31 *
emilmont 1:fdd22bb7aa52 32 * Version 0.0.7 2010/06/10
emilmont 1:fdd22bb7aa52 33 * Misra-C changes done
emilmont 1:fdd22bb7aa52 34 * -------------------------------------------------------------------- */
emilmont 1:fdd22bb7aa52 35
emilmont 1:fdd22bb7aa52 36 #include "arm_math.h"
emilmont 1:fdd22bb7aa52 37
emilmont 1:fdd22bb7aa52 38 /**
emilmont 1:fdd22bb7aa52 39 * @ingroup groupFilters
emilmont 1:fdd22bb7aa52 40 */
emilmont 1:fdd22bb7aa52 41
emilmont 1:fdd22bb7aa52 42 /**
emilmont 1:fdd22bb7aa52 43 * @defgroup BiquadCascadeDF1_32x64 High Precision Q31 Biquad Cascade Filter
emilmont 1:fdd22bb7aa52 44 *
emilmont 1:fdd22bb7aa52 45 * This function implements a high precision Biquad cascade filter which operates on
emilmont 1:fdd22bb7aa52 46 * Q31 data values. The filter coefficients are in 1.31 format and the state variables
emilmont 1:fdd22bb7aa52 47 * are in 1.63 format. The double precision state variables reduce quantization noise
emilmont 1:fdd22bb7aa52 48 * in the filter and provide a cleaner output.
emilmont 1:fdd22bb7aa52 49 * These filters are particularly useful when implementing filters in which the
emilmont 1:fdd22bb7aa52 50 * singularities are close to the unit circle. This is common for low pass or high
emilmont 1:fdd22bb7aa52 51 * pass filters with very low cutoff frequencies.
emilmont 1:fdd22bb7aa52 52 *
emilmont 1:fdd22bb7aa52 53 * The function operates on blocks of input and output data
emilmont 1:fdd22bb7aa52 54 * and each call to the function processes <code>blockSize</code> samples through
emilmont 1:fdd22bb7aa52 55 * the filter. <code>pSrc</code> and <code>pDst</code> points to input and output arrays
emilmont 1:fdd22bb7aa52 56 * containing <code>blockSize</code> Q31 values.
emilmont 1:fdd22bb7aa52 57 *
emilmont 1:fdd22bb7aa52 58 * \par Algorithm
emilmont 1:fdd22bb7aa52 59 * Each Biquad stage implements a second order filter using the difference equation:
emilmont 1:fdd22bb7aa52 60 * <pre>
emilmont 1:fdd22bb7aa52 61 * y[n] = b0 * x[n] + b1 * x[n-1] + b2 * x[n-2] + a1 * y[n-1] + a2 * y[n-2]
emilmont 1:fdd22bb7aa52 62 * </pre>
emilmont 1:fdd22bb7aa52 63 * A Direct Form I algorithm is used with 5 coefficients and 4 state variables per stage.
emilmont 1:fdd22bb7aa52 64 * \image html Biquad.gif "Single Biquad filter stage"
emilmont 1:fdd22bb7aa52 65 * Coefficients <code>b0, b1, and b2 </code> multiply the input signal <code>x[n]</code> and are referred to as the feedforward coefficients.
emilmont 1:fdd22bb7aa52 66 * Coefficients <code>a1</code> and <code>a2</code> multiply the output signal <code>y[n]</code> and are referred to as the feedback coefficients.
emilmont 1:fdd22bb7aa52 67 * Pay careful attention to the sign of the feedback coefficients.
emilmont 1:fdd22bb7aa52 68 * Some design tools use the difference equation
emilmont 1:fdd22bb7aa52 69 * <pre>
emilmont 1:fdd22bb7aa52 70 * y[n] = b0 * x[n] + b1 * x[n-1] + b2 * x[n-2] - a1 * y[n-1] - a2 * y[n-2]
emilmont 1:fdd22bb7aa52 71 * </pre>
emilmont 1:fdd22bb7aa52 72 * In this case the feedback coefficients <code>a1</code> and <code>a2</code> must be negated when used with the CMSIS DSP Library.
emilmont 1:fdd22bb7aa52 73 *
emilmont 1:fdd22bb7aa52 74 * \par
emilmont 1:fdd22bb7aa52 75 * Higher order filters are realized as a cascade of second order sections.
emilmont 1:fdd22bb7aa52 76 * <code>numStages</code> refers to the number of second order stages used.
emilmont 1:fdd22bb7aa52 77 * For example, an 8th order filter would be realized with <code>numStages=4</code> second order stages.
emilmont 1:fdd22bb7aa52 78 * \image html BiquadCascade.gif "8th order filter using a cascade of Biquad stages"
emilmont 1:fdd22bb7aa52 79 * A 9th order filter would be realized with <code>numStages=5</code> second order stages with the coefficients for one of the stages configured as a first order filter (<code>b2=0</code> and <code>a2=0</code>).
emilmont 1:fdd22bb7aa52 80 *
emilmont 1:fdd22bb7aa52 81 * \par
emilmont 1:fdd22bb7aa52 82 * The <code>pState</code> points to state variables array .
emilmont 1:fdd22bb7aa52 83 * Each Biquad stage has 4 state variables <code>x[n-1], x[n-2], y[n-1],</code> and <code>y[n-2]</code> and each state variable in 1.63 format to improve precision.
emilmont 1:fdd22bb7aa52 84 * The state variables are arranged in the array as:
emilmont 1:fdd22bb7aa52 85 * <pre>
emilmont 1:fdd22bb7aa52 86 * {x[n-1], x[n-2], y[n-1], y[n-2]}
emilmont 1:fdd22bb7aa52 87 * </pre>
emilmont 1:fdd22bb7aa52 88 *
emilmont 1:fdd22bb7aa52 89 * \par
emilmont 1:fdd22bb7aa52 90 * The 4 state variables for stage 1 are first, then the 4 state variables for stage 2, and so on.
emilmont 1:fdd22bb7aa52 91 * The state array has a total length of <code>4*numStages</code> values of data in 1.63 format.
emilmont 1:fdd22bb7aa52 92 * The state variables are updated after each block of data is processed; the coefficients are untouched.
emilmont 1:fdd22bb7aa52 93 *
emilmont 1:fdd22bb7aa52 94 * \par Instance Structure
emilmont 1:fdd22bb7aa52 95 * The coefficients and state variables for a filter are stored together in an instance data structure.
emilmont 1:fdd22bb7aa52 96 * A separate instance structure must be defined for each filter.
emilmont 1:fdd22bb7aa52 97 * Coefficient arrays may be shared among several instances while state variable arrays cannot be shared.
emilmont 1:fdd22bb7aa52 98 *
emilmont 1:fdd22bb7aa52 99 * \par Init Function
emilmont 1:fdd22bb7aa52 100 * There is also an associated initialization function which performs the following operations:
emilmont 1:fdd22bb7aa52 101 * - Sets the values of the internal structure fields.
emilmont 1:fdd22bb7aa52 102 * - Zeros out the values in the state buffer.
emilmont 1:fdd22bb7aa52 103 * \par
emilmont 1:fdd22bb7aa52 104 * Use of the initialization function is optional.
emilmont 1:fdd22bb7aa52 105 * However, if the initialization function is used, then the instance structure cannot be placed into a const data section.
emilmont 1:fdd22bb7aa52 106 * To place an instance structure into a const data section, the instance structure must be manually initialized.
emilmont 1:fdd22bb7aa52 107 * Set the values in the state buffer to zeros before static initialization.
emilmont 1:fdd22bb7aa52 108 * For example, to statically initialize the filter instance structure use
emilmont 1:fdd22bb7aa52 109 * <pre>
emilmont 1:fdd22bb7aa52 110 * arm_biquad_cas_df1_32x64_ins_q31 S1 = {numStages, pState, pCoeffs, postShift};
emilmont 1:fdd22bb7aa52 111 * </pre>
emilmont 1:fdd22bb7aa52 112 * where <code>numStages</code> is the number of Biquad stages in the filter; <code>pState</code> is the address of the state buffer;
emilmont 1:fdd22bb7aa52 113 * <code>pCoeffs</code> is the address of the coefficient buffer; <code>postShift</code> shift to be applied which is described in detail below.
emilmont 1:fdd22bb7aa52 114 * \par Fixed-Point Behavior
emilmont 1:fdd22bb7aa52 115 * Care must be taken while using Biquad Cascade 32x64 filter function.
emilmont 1:fdd22bb7aa52 116 * Following issues must be considered:
emilmont 1:fdd22bb7aa52 117 * - Scaling of coefficients
emilmont 1:fdd22bb7aa52 118 * - Filter gain
emilmont 1:fdd22bb7aa52 119 * - Overflow and saturation
emilmont 1:fdd22bb7aa52 120 *
emilmont 1:fdd22bb7aa52 121 * \par
emilmont 1:fdd22bb7aa52 122 * Filter coefficients are represented as fractional values and
emilmont 1:fdd22bb7aa52 123 * restricted to lie in the range <code>[-1 +1)</code>.
emilmont 1:fdd22bb7aa52 124 * The processing function has an additional scaling parameter <code>postShift</code>
emilmont 1:fdd22bb7aa52 125 * which allows the filter coefficients to exceed the range <code>[+1 -1)</code>.
emilmont 1:fdd22bb7aa52 126 * At the output of the filter's accumulator is a shift register which shifts the result by <code>postShift</code> bits.
emilmont 1:fdd22bb7aa52 127 * \image html BiquadPostshift.gif "Fixed-point Biquad with shift by postShift bits after accumulator"
emilmont 1:fdd22bb7aa52 128 * This essentially scales the filter coefficients by <code>2^postShift</code>.
emilmont 1:fdd22bb7aa52 129 * For example, to realize the coefficients
emilmont 1:fdd22bb7aa52 130 * <pre>
emilmont 1:fdd22bb7aa52 131 * {1.5, -0.8, 1.2, 1.6, -0.9}
emilmont 1:fdd22bb7aa52 132 * </pre>
emilmont 1:fdd22bb7aa52 133 * set the Coefficient array to:
emilmont 1:fdd22bb7aa52 134 * <pre>
emilmont 1:fdd22bb7aa52 135 * {0.75, -0.4, 0.6, 0.8, -0.45}
emilmont 1:fdd22bb7aa52 136 * </pre>
emilmont 1:fdd22bb7aa52 137 * and set <code>postShift=1</code>
emilmont 1:fdd22bb7aa52 138 *
emilmont 1:fdd22bb7aa52 139 * \par
emilmont 1:fdd22bb7aa52 140 * The second thing to keep in mind is the gain through the filter.
emilmont 1:fdd22bb7aa52 141 * The frequency response of a Biquad filter is a function of its coefficients.
emilmont 1:fdd22bb7aa52 142 * It is possible for the gain through the filter to exceed 1.0 meaning that the filter increases the amplitude of certain frequencies.
emilmont 1:fdd22bb7aa52 143 * This means that an input signal with amplitude < 1.0 may result in an output > 1.0 and these are saturated or overflowed based on the implementation of the filter.
emilmont 1:fdd22bb7aa52 144 * To avoid this behavior the filter needs to be scaled down such that its peak gain < 1.0 or the input signal must be scaled down so that the combination of input and filter are never overflowed.
emilmont 1:fdd22bb7aa52 145 *
emilmont 1:fdd22bb7aa52 146 * \par
emilmont 1:fdd22bb7aa52 147 * The third item to consider is the overflow and saturation behavior of the fixed-point Q31 version.
emilmont 1:fdd22bb7aa52 148 * This is described in the function specific documentation below.
emilmont 1:fdd22bb7aa52 149 */
emilmont 1:fdd22bb7aa52 150
emilmont 1:fdd22bb7aa52 151 /**
emilmont 1:fdd22bb7aa52 152 * @addtogroup BiquadCascadeDF1_32x64
emilmont 1:fdd22bb7aa52 153 * @{
emilmont 1:fdd22bb7aa52 154 */
emilmont 1:fdd22bb7aa52 155
emilmont 1:fdd22bb7aa52 156 /**
emilmont 1:fdd22bb7aa52 157 * @details
emilmont 1:fdd22bb7aa52 158
emilmont 1:fdd22bb7aa52 159 * @param[in] *S points to an instance of the high precision Q31 Biquad cascade filter.
emilmont 1:fdd22bb7aa52 160 * @param[in] *pSrc points to the block of input data.
emilmont 1:fdd22bb7aa52 161 * @param[out] *pDst points to the block of output data.
emilmont 1:fdd22bb7aa52 162 * @param[in] blockSize number of samples to process.
emilmont 1:fdd22bb7aa52 163 * @return none.
emilmont 1:fdd22bb7aa52 164 *
emilmont 1:fdd22bb7aa52 165 * \par
emilmont 1:fdd22bb7aa52 166 * The function is implemented using an internal 64-bit accumulator.
emilmont 1:fdd22bb7aa52 167 * The accumulator has a 2.62 format and maintains full precision of the intermediate multiplication results but provides only a single guard bit.
emilmont 1:fdd22bb7aa52 168 * Thus, if the accumulator result overflows it wraps around rather than clip.
emilmont 1:fdd22bb7aa52 169 * In order to avoid overflows completely the input signal must be scaled down by 2 bits and lie in the range [-0.25 +0.25).
emilmont 1:fdd22bb7aa52 170 * After all 5 multiply-accumulates are performed, the 2.62 accumulator is shifted by <code>postShift</code> bits and the result truncated to
emilmont 1:fdd22bb7aa52 171 * 1.31 format by discarding the low 32 bits.
emilmont 1:fdd22bb7aa52 172 *
emilmont 1:fdd22bb7aa52 173 * \par
emilmont 1:fdd22bb7aa52 174 * Two related functions are provided in the CMSIS DSP library.
emilmont 1:fdd22bb7aa52 175 * <code>arm_biquad_cascade_df1_q31()</code> implements a Biquad cascade with 32-bit coefficients and state variables with a Q63 accumulator.
emilmont 1:fdd22bb7aa52 176 * <code>arm_biquad_cascade_df1_fast_q31()</code> implements a Biquad cascade with 32-bit coefficients and state variables with a Q31 accumulator.
emilmont 1:fdd22bb7aa52 177 */
emilmont 1:fdd22bb7aa52 178
emilmont 1:fdd22bb7aa52 179 void arm_biquad_cas_df1_32x64_q31(
emilmont 1:fdd22bb7aa52 180 const arm_biquad_cas_df1_32x64_ins_q31 * S,
emilmont 1:fdd22bb7aa52 181 q31_t * pSrc,
emilmont 1:fdd22bb7aa52 182 q31_t * pDst,
emilmont 1:fdd22bb7aa52 183 uint32_t blockSize)
emilmont 1:fdd22bb7aa52 184 {
emilmont 1:fdd22bb7aa52 185 q31_t *pIn = pSrc; /* input pointer initialization */
emilmont 1:fdd22bb7aa52 186 q31_t *pOut = pDst; /* output pointer initialization */
emilmont 1:fdd22bb7aa52 187 q63_t *pState = S->pState; /* state pointer initialization */
emilmont 1:fdd22bb7aa52 188 q31_t *pCoeffs = S->pCoeffs; /* coeff pointer initialization */
emilmont 1:fdd22bb7aa52 189 q63_t acc; /* accumulator */
emilmont 1:fdd22bb7aa52 190 q31_t Xn1, Xn2; /* Input Filter state variables */
emilmont 1:fdd22bb7aa52 191 q63_t Yn1, Yn2; /* Output Filter state variables */
emilmont 1:fdd22bb7aa52 192 q31_t b0, b1, b2, a1, a2; /* Filter coefficients */
emilmont 1:fdd22bb7aa52 193 q31_t Xn; /* temporary input */
emilmont 1:fdd22bb7aa52 194 int32_t shift = (int32_t) S->postShift + 1; /* Shift to be applied to the output */
emilmont 1:fdd22bb7aa52 195 uint32_t sample, stage = S->numStages; /* loop counters */
emilmont 1:fdd22bb7aa52 196 q31_t acc_l, acc_h; /* temporary output */
emilmont 1:fdd22bb7aa52 197 uint32_t uShift = ((uint32_t) S->postShift + 1u);
emilmont 1:fdd22bb7aa52 198 uint32_t lShift = 32u - uShift; /* Shift to be applied to the output */
emilmont 1:fdd22bb7aa52 199
emilmont 1:fdd22bb7aa52 200
emilmont 1:fdd22bb7aa52 201 #ifndef ARM_MATH_CM0
emilmont 1:fdd22bb7aa52 202
emilmont 1:fdd22bb7aa52 203 /* Run the below code for Cortex-M4 and Cortex-M3 */
emilmont 1:fdd22bb7aa52 204
emilmont 1:fdd22bb7aa52 205 do
emilmont 1:fdd22bb7aa52 206 {
emilmont 1:fdd22bb7aa52 207 /* Reading the coefficients */
emilmont 1:fdd22bb7aa52 208 b0 = *pCoeffs++;
emilmont 1:fdd22bb7aa52 209 b1 = *pCoeffs++;
emilmont 1:fdd22bb7aa52 210 b2 = *pCoeffs++;
emilmont 1:fdd22bb7aa52 211 a1 = *pCoeffs++;
emilmont 1:fdd22bb7aa52 212 a2 = *pCoeffs++;
emilmont 1:fdd22bb7aa52 213
emilmont 1:fdd22bb7aa52 214 /* Reading the state values */
emilmont 1:fdd22bb7aa52 215 Xn1 = (q31_t) (pState[0]);
emilmont 1:fdd22bb7aa52 216 Xn2 = (q31_t) (pState[1]);
emilmont 1:fdd22bb7aa52 217 Yn1 = pState[2];
emilmont 1:fdd22bb7aa52 218 Yn2 = pState[3];
emilmont 1:fdd22bb7aa52 219
emilmont 1:fdd22bb7aa52 220 /* Apply loop unrolling and compute 4 output values simultaneously. */
emilmont 1:fdd22bb7aa52 221 /* The variable acc hold output value that is being computed and
emilmont 1:fdd22bb7aa52 222 * stored in the destination buffer
emilmont 1:fdd22bb7aa52 223 * acc = b0 * x[n] + b1 * x[n-1] + b2 * x[n-2] + a1 * y[n-1] + a2 * y[n-2]
emilmont 1:fdd22bb7aa52 224 */
emilmont 1:fdd22bb7aa52 225
emilmont 1:fdd22bb7aa52 226 sample = blockSize >> 2u;
emilmont 1:fdd22bb7aa52 227
emilmont 1:fdd22bb7aa52 228 /* First part of the processing with loop unrolling. Compute 4 outputs at a time.
emilmont 1:fdd22bb7aa52 229 ** a second loop below computes the remaining 1 to 3 samples. */
emilmont 1:fdd22bb7aa52 230 while(sample > 0u)
emilmont 1:fdd22bb7aa52 231 {
emilmont 1:fdd22bb7aa52 232 /* Read the input */
emilmont 1:fdd22bb7aa52 233 Xn = *pIn++;
emilmont 1:fdd22bb7aa52 234
emilmont 1:fdd22bb7aa52 235 /* acc = b0 * x[n] + b1 * x[n-1] + b2 * x[n-2] + a1 * y[n-1] + a2 * y[n-2] */
emilmont 1:fdd22bb7aa52 236
emilmont 1:fdd22bb7aa52 237 /* acc = b0 * x[n] */
emilmont 1:fdd22bb7aa52 238 acc = (q63_t) Xn *b0;
emilmont 1:fdd22bb7aa52 239
emilmont 1:fdd22bb7aa52 240 /* acc += b1 * x[n-1] */
emilmont 1:fdd22bb7aa52 241 acc += (q63_t) Xn1 *b1;
emilmont 1:fdd22bb7aa52 242
emilmont 1:fdd22bb7aa52 243 /* acc += b[2] * x[n-2] */
emilmont 1:fdd22bb7aa52 244 acc += (q63_t) Xn2 *b2;
emilmont 1:fdd22bb7aa52 245
emilmont 1:fdd22bb7aa52 246 /* acc += a1 * y[n-1] */
emilmont 1:fdd22bb7aa52 247 acc += mult32x64(Yn1, a1);
emilmont 1:fdd22bb7aa52 248
emilmont 1:fdd22bb7aa52 249 /* acc += a2 * y[n-2] */
emilmont 1:fdd22bb7aa52 250 acc += mult32x64(Yn2, a2);
emilmont 1:fdd22bb7aa52 251
emilmont 1:fdd22bb7aa52 252 /* The result is converted to 1.63 , Yn2 variable is reused */
emilmont 1:fdd22bb7aa52 253 Yn2 = acc << shift;
emilmont 1:fdd22bb7aa52 254
emilmont 1:fdd22bb7aa52 255 /* Calc lower part of acc */
emilmont 1:fdd22bb7aa52 256 acc_l = acc & 0xffffffff;
emilmont 1:fdd22bb7aa52 257
emilmont 1:fdd22bb7aa52 258 /* Calc upper part of acc */
emilmont 1:fdd22bb7aa52 259 acc_h = (acc >> 32) & 0xffffffff;
emilmont 1:fdd22bb7aa52 260
emilmont 1:fdd22bb7aa52 261 /* Apply shift for lower part of acc and upper part of acc */
emilmont 1:fdd22bb7aa52 262 acc_h = (uint32_t) acc_l >> lShift | acc_h << uShift;
emilmont 1:fdd22bb7aa52 263
emilmont 1:fdd22bb7aa52 264 /* Store the output in the destination buffer in 1.31 format. */
emilmont 1:fdd22bb7aa52 265 *pOut = acc_h;
emilmont 1:fdd22bb7aa52 266
emilmont 1:fdd22bb7aa52 267 /* Read the second input into Xn2, to reuse the value */
emilmont 1:fdd22bb7aa52 268 Xn2 = *pIn++;
emilmont 1:fdd22bb7aa52 269
emilmont 1:fdd22bb7aa52 270 /* acc = b0 * x[n] + b1 * x[n-1] + b2 * x[n-2] + a1 * y[n-1] + a2 * y[n-2] */
emilmont 1:fdd22bb7aa52 271
emilmont 1:fdd22bb7aa52 272 /* acc += b1 * x[n-1] */
emilmont 1:fdd22bb7aa52 273 acc = (q63_t) Xn *b1;
emilmont 1:fdd22bb7aa52 274
emilmont 1:fdd22bb7aa52 275 /* acc = b0 * x[n] */
emilmont 1:fdd22bb7aa52 276 acc += (q63_t) Xn2 *b0;
emilmont 1:fdd22bb7aa52 277
emilmont 1:fdd22bb7aa52 278 /* acc += b[2] * x[n-2] */
emilmont 1:fdd22bb7aa52 279 acc += (q63_t) Xn1 *b2;
emilmont 1:fdd22bb7aa52 280
emilmont 1:fdd22bb7aa52 281 /* acc += a1 * y[n-1] */
emilmont 1:fdd22bb7aa52 282 acc += mult32x64(Yn2, a1);
emilmont 1:fdd22bb7aa52 283
emilmont 1:fdd22bb7aa52 284 /* acc += a2 * y[n-2] */
emilmont 1:fdd22bb7aa52 285 acc += mult32x64(Yn1, a2);
emilmont 1:fdd22bb7aa52 286
emilmont 1:fdd22bb7aa52 287 /* The result is converted to 1.63, Yn1 variable is reused */
emilmont 1:fdd22bb7aa52 288 Yn1 = acc << shift;
emilmont 1:fdd22bb7aa52 289
emilmont 1:fdd22bb7aa52 290 /* Calc lower part of acc */
emilmont 1:fdd22bb7aa52 291 acc_l = acc & 0xffffffff;
emilmont 1:fdd22bb7aa52 292
emilmont 1:fdd22bb7aa52 293 /* Calc upper part of acc */
emilmont 1:fdd22bb7aa52 294 acc_h = (acc >> 32) & 0xffffffff;
emilmont 1:fdd22bb7aa52 295
emilmont 1:fdd22bb7aa52 296 /* Apply shift for lower part of acc and upper part of acc */
emilmont 1:fdd22bb7aa52 297 acc_h = (uint32_t) acc_l >> lShift | acc_h << uShift;
emilmont 1:fdd22bb7aa52 298
emilmont 1:fdd22bb7aa52 299 /* Read the third input into Xn1, to reuse the value */
emilmont 1:fdd22bb7aa52 300 Xn1 = *pIn++;
emilmont 1:fdd22bb7aa52 301
emilmont 1:fdd22bb7aa52 302 /* The result is converted to 1.31 */
emilmont 1:fdd22bb7aa52 303 /* Store the output in the destination buffer. */
emilmont 1:fdd22bb7aa52 304 *(pOut + 1u) = acc_h;
emilmont 1:fdd22bb7aa52 305
emilmont 1:fdd22bb7aa52 306 /* acc = b0 * x[n] + b1 * x[n-1] + b2 * x[n-2] + a1 * y[n-1] + a2 * y[n-2] */
emilmont 1:fdd22bb7aa52 307
emilmont 1:fdd22bb7aa52 308 /* acc = b0 * x[n] */
emilmont 1:fdd22bb7aa52 309 acc = (q63_t) Xn1 *b0;
emilmont 1:fdd22bb7aa52 310
emilmont 1:fdd22bb7aa52 311 /* acc += b1 * x[n-1] */
emilmont 1:fdd22bb7aa52 312 acc += (q63_t) Xn2 *b1;
emilmont 1:fdd22bb7aa52 313
emilmont 1:fdd22bb7aa52 314 /* acc += b[2] * x[n-2] */
emilmont 1:fdd22bb7aa52 315 acc += (q63_t) Xn *b2;
emilmont 1:fdd22bb7aa52 316
emilmont 1:fdd22bb7aa52 317 /* acc += a1 * y[n-1] */
emilmont 1:fdd22bb7aa52 318 acc += mult32x64(Yn1, a1);
emilmont 1:fdd22bb7aa52 319
emilmont 1:fdd22bb7aa52 320 /* acc += a2 * y[n-2] */
emilmont 1:fdd22bb7aa52 321 acc += mult32x64(Yn2, a2);
emilmont 1:fdd22bb7aa52 322
emilmont 1:fdd22bb7aa52 323 /* The result is converted to 1.63, Yn2 variable is reused */
emilmont 1:fdd22bb7aa52 324 Yn2 = acc << shift;
emilmont 1:fdd22bb7aa52 325
emilmont 1:fdd22bb7aa52 326 /* Calc lower part of acc */
emilmont 1:fdd22bb7aa52 327 acc_l = acc & 0xffffffff;
emilmont 1:fdd22bb7aa52 328
emilmont 1:fdd22bb7aa52 329 /* Calc upper part of acc */
emilmont 1:fdd22bb7aa52 330 acc_h = (acc >> 32) & 0xffffffff;
emilmont 1:fdd22bb7aa52 331
emilmont 1:fdd22bb7aa52 332 /* Apply shift for lower part of acc and upper part of acc */
emilmont 1:fdd22bb7aa52 333 acc_h = (uint32_t) acc_l >> lShift | acc_h << uShift;
emilmont 1:fdd22bb7aa52 334
emilmont 1:fdd22bb7aa52 335 /* Store the output in the destination buffer in 1.31 format. */
emilmont 1:fdd22bb7aa52 336 *(pOut + 2u) = acc_h;
emilmont 1:fdd22bb7aa52 337
emilmont 1:fdd22bb7aa52 338 /* Read the fourth input into Xn, to reuse the value */
emilmont 1:fdd22bb7aa52 339 Xn = *pIn++;
emilmont 1:fdd22bb7aa52 340
emilmont 1:fdd22bb7aa52 341 /* acc = b0 * x[n] + b1 * x[n-1] + b2 * x[n-2] + a1 * y[n-1] + a2 * y[n-2] */
emilmont 1:fdd22bb7aa52 342 /* acc = b0 * x[n] */
emilmont 1:fdd22bb7aa52 343 acc = (q63_t) Xn *b0;
emilmont 1:fdd22bb7aa52 344
emilmont 1:fdd22bb7aa52 345 /* acc += b1 * x[n-1] */
emilmont 1:fdd22bb7aa52 346 acc += (q63_t) Xn1 *b1;
emilmont 1:fdd22bb7aa52 347
emilmont 1:fdd22bb7aa52 348 /* acc += b[2] * x[n-2] */
emilmont 1:fdd22bb7aa52 349 acc += (q63_t) Xn2 *b2;
emilmont 1:fdd22bb7aa52 350
emilmont 1:fdd22bb7aa52 351 /* acc += a1 * y[n-1] */
emilmont 1:fdd22bb7aa52 352 acc += mult32x64(Yn2, a1);
emilmont 1:fdd22bb7aa52 353
emilmont 1:fdd22bb7aa52 354 /* acc += a2 * y[n-2] */
emilmont 1:fdd22bb7aa52 355 acc += mult32x64(Yn1, a2);
emilmont 1:fdd22bb7aa52 356
emilmont 1:fdd22bb7aa52 357 /* The result is converted to 1.63, Yn1 variable is reused */
emilmont 1:fdd22bb7aa52 358 Yn1 = acc << shift;
emilmont 1:fdd22bb7aa52 359
emilmont 1:fdd22bb7aa52 360 /* Calc lower part of acc */
emilmont 1:fdd22bb7aa52 361 acc_l = acc & 0xffffffff;
emilmont 1:fdd22bb7aa52 362
emilmont 1:fdd22bb7aa52 363 /* Calc upper part of acc */
emilmont 1:fdd22bb7aa52 364 acc_h = (acc >> 32) & 0xffffffff;
emilmont 1:fdd22bb7aa52 365
emilmont 1:fdd22bb7aa52 366 /* Apply shift for lower part of acc and upper part of acc */
emilmont 1:fdd22bb7aa52 367 acc_h = (uint32_t) acc_l >> lShift | acc_h << uShift;
emilmont 1:fdd22bb7aa52 368
emilmont 1:fdd22bb7aa52 369 /* Store the output in the destination buffer in 1.31 format. */
emilmont 1:fdd22bb7aa52 370 *(pOut + 3u) = acc_h;
emilmont 1:fdd22bb7aa52 371
emilmont 1:fdd22bb7aa52 372 /* Every time after the output is computed state should be updated. */
emilmont 1:fdd22bb7aa52 373 /* The states should be updated as: */
emilmont 1:fdd22bb7aa52 374 /* Xn2 = Xn1 */
emilmont 1:fdd22bb7aa52 375 /* Xn1 = Xn */
emilmont 1:fdd22bb7aa52 376 /* Yn2 = Yn1 */
emilmont 1:fdd22bb7aa52 377 /* Yn1 = acc */
emilmont 1:fdd22bb7aa52 378 Xn2 = Xn1;
emilmont 1:fdd22bb7aa52 379 Xn1 = Xn;
emilmont 1:fdd22bb7aa52 380
emilmont 1:fdd22bb7aa52 381 /* update output pointer */
emilmont 1:fdd22bb7aa52 382 pOut += 4u;
emilmont 1:fdd22bb7aa52 383
emilmont 1:fdd22bb7aa52 384 /* decrement the loop counter */
emilmont 1:fdd22bb7aa52 385 sample--;
emilmont 1:fdd22bb7aa52 386 }
emilmont 1:fdd22bb7aa52 387
emilmont 1:fdd22bb7aa52 388 /* If the blockSize is not a multiple of 4, compute any remaining output samples here.
emilmont 1:fdd22bb7aa52 389 ** No loop unrolling is used. */
emilmont 1:fdd22bb7aa52 390 sample = (blockSize & 0x3u);
emilmont 1:fdd22bb7aa52 391
emilmont 1:fdd22bb7aa52 392 while(sample > 0u)
emilmont 1:fdd22bb7aa52 393 {
emilmont 1:fdd22bb7aa52 394 /* Read the input */
emilmont 1:fdd22bb7aa52 395 Xn = *pIn++;
emilmont 1:fdd22bb7aa52 396
emilmont 1:fdd22bb7aa52 397 /* acc = b0 * x[n] + b1 * x[n-1] + b2 * x[n-2] + a1 * y[n-1] + a2 * y[n-2] */
emilmont 1:fdd22bb7aa52 398
emilmont 1:fdd22bb7aa52 399 /* acc = b0 * x[n] */
emilmont 1:fdd22bb7aa52 400 acc = (q63_t) Xn *b0;
emilmont 1:fdd22bb7aa52 401 /* acc += b1 * x[n-1] */
emilmont 1:fdd22bb7aa52 402 acc += (q63_t) Xn1 *b1;
emilmont 1:fdd22bb7aa52 403 /* acc += b[2] * x[n-2] */
emilmont 1:fdd22bb7aa52 404 acc += (q63_t) Xn2 *b2;
emilmont 1:fdd22bb7aa52 405 /* acc += a1 * y[n-1] */
emilmont 1:fdd22bb7aa52 406 acc += mult32x64(Yn1, a1);
emilmont 1:fdd22bb7aa52 407 /* acc += a2 * y[n-2] */
emilmont 1:fdd22bb7aa52 408 acc += mult32x64(Yn2, a2);
emilmont 1:fdd22bb7aa52 409
emilmont 1:fdd22bb7aa52 410 /* Every time after the output is computed state should be updated. */
emilmont 1:fdd22bb7aa52 411 /* The states should be updated as: */
emilmont 1:fdd22bb7aa52 412 /* Xn2 = Xn1 */
emilmont 1:fdd22bb7aa52 413 /* Xn1 = Xn */
emilmont 1:fdd22bb7aa52 414 /* Yn2 = Yn1 */
emilmont 1:fdd22bb7aa52 415 /* Yn1 = acc */
emilmont 1:fdd22bb7aa52 416 Xn2 = Xn1;
emilmont 1:fdd22bb7aa52 417 Xn1 = Xn;
emilmont 1:fdd22bb7aa52 418 Yn2 = Yn1;
emilmont 1:fdd22bb7aa52 419 /* The result is converted to 1.63, Yn1 variable is reused */
emilmont 1:fdd22bb7aa52 420 Yn1 = acc << shift;
emilmont 1:fdd22bb7aa52 421
emilmont 1:fdd22bb7aa52 422 /* Calc lower part of acc */
emilmont 1:fdd22bb7aa52 423 acc_l = acc & 0xffffffff;
emilmont 1:fdd22bb7aa52 424
emilmont 1:fdd22bb7aa52 425 /* Calc upper part of acc */
emilmont 1:fdd22bb7aa52 426 acc_h = (acc >> 32) & 0xffffffff;
emilmont 1:fdd22bb7aa52 427
emilmont 1:fdd22bb7aa52 428 /* Apply shift for lower part of acc and upper part of acc */
emilmont 1:fdd22bb7aa52 429 acc_h = (uint32_t) acc_l >> lShift | acc_h << uShift;
emilmont 1:fdd22bb7aa52 430
emilmont 1:fdd22bb7aa52 431 /* Store the output in the destination buffer in 1.31 format. */
emilmont 1:fdd22bb7aa52 432 *pOut++ = acc_h;
emilmont 1:fdd22bb7aa52 433 //Yn1 = acc << shift;
emilmont 1:fdd22bb7aa52 434
emilmont 1:fdd22bb7aa52 435 /* Store the output in the destination buffer in 1.31 format. */
emilmont 1:fdd22bb7aa52 436 // *pOut++ = (q31_t) (acc >> (32 - shift));
emilmont 1:fdd22bb7aa52 437
emilmont 1:fdd22bb7aa52 438 /* decrement the loop counter */
emilmont 1:fdd22bb7aa52 439 sample--;
emilmont 1:fdd22bb7aa52 440 }
emilmont 1:fdd22bb7aa52 441
emilmont 1:fdd22bb7aa52 442 /* The first stage output is given as input to the second stage. */
emilmont 1:fdd22bb7aa52 443 pIn = pDst;
emilmont 1:fdd22bb7aa52 444
emilmont 1:fdd22bb7aa52 445 /* Reset to destination buffer working pointer */
emilmont 1:fdd22bb7aa52 446 pOut = pDst;
emilmont 1:fdd22bb7aa52 447
emilmont 1:fdd22bb7aa52 448 /* Store the updated state variables back into the pState array */
emilmont 1:fdd22bb7aa52 449 /* Store the updated state variables back into the pState array */
emilmont 1:fdd22bb7aa52 450 *pState++ = (q63_t) Xn1;
emilmont 1:fdd22bb7aa52 451 *pState++ = (q63_t) Xn2;
emilmont 1:fdd22bb7aa52 452 *pState++ = Yn1;
emilmont 1:fdd22bb7aa52 453 *pState++ = Yn2;
emilmont 1:fdd22bb7aa52 454
emilmont 1:fdd22bb7aa52 455 } while(--stage);
emilmont 1:fdd22bb7aa52 456
emilmont 1:fdd22bb7aa52 457 #else
emilmont 1:fdd22bb7aa52 458
emilmont 1:fdd22bb7aa52 459 /* Run the below code for Cortex-M0 */
emilmont 1:fdd22bb7aa52 460
emilmont 1:fdd22bb7aa52 461 do
emilmont 1:fdd22bb7aa52 462 {
emilmont 1:fdd22bb7aa52 463 /* Reading the coefficients */
emilmont 1:fdd22bb7aa52 464 b0 = *pCoeffs++;
emilmont 1:fdd22bb7aa52 465 b1 = *pCoeffs++;
emilmont 1:fdd22bb7aa52 466 b2 = *pCoeffs++;
emilmont 1:fdd22bb7aa52 467 a1 = *pCoeffs++;
emilmont 1:fdd22bb7aa52 468 a2 = *pCoeffs++;
emilmont 1:fdd22bb7aa52 469
emilmont 1:fdd22bb7aa52 470 /* Reading the state values */
emilmont 1:fdd22bb7aa52 471 Xn1 = pState[0];
emilmont 1:fdd22bb7aa52 472 Xn2 = pState[1];
emilmont 1:fdd22bb7aa52 473 Yn1 = pState[2];
emilmont 1:fdd22bb7aa52 474 Yn2 = pState[3];
emilmont 1:fdd22bb7aa52 475
emilmont 1:fdd22bb7aa52 476 /* The variable acc hold output value that is being computed and
emilmont 1:fdd22bb7aa52 477 * stored in the destination buffer
emilmont 1:fdd22bb7aa52 478 * acc = b0 * x[n] + b1 * x[n-1] + b2 * x[n-2] + a1 * y[n-1] + a2 * y[n-2]
emilmont 1:fdd22bb7aa52 479 */
emilmont 1:fdd22bb7aa52 480
emilmont 1:fdd22bb7aa52 481 sample = blockSize;
emilmont 1:fdd22bb7aa52 482
emilmont 1:fdd22bb7aa52 483 while(sample > 0u)
emilmont 1:fdd22bb7aa52 484 {
emilmont 1:fdd22bb7aa52 485 /* Read the input */
emilmont 1:fdd22bb7aa52 486 Xn = *pIn++;
emilmont 1:fdd22bb7aa52 487
emilmont 1:fdd22bb7aa52 488 /* acc = b0 * x[n] + b1 * x[n-1] + b2 * x[n-2] + a1 * y[n-1] + a2 * y[n-2] */
emilmont 1:fdd22bb7aa52 489 /* acc = b0 * x[n] */
emilmont 1:fdd22bb7aa52 490 acc = (q63_t) Xn *b0;
emilmont 1:fdd22bb7aa52 491 /* acc += b1 * x[n-1] */
emilmont 1:fdd22bb7aa52 492 acc += (q63_t) Xn1 *b1;
emilmont 1:fdd22bb7aa52 493 /* acc += b[2] * x[n-2] */
emilmont 1:fdd22bb7aa52 494 acc += (q63_t) Xn2 *b2;
emilmont 1:fdd22bb7aa52 495 /* acc += a1 * y[n-1] */
emilmont 1:fdd22bb7aa52 496 acc += mult32x64(Yn1, a1);
emilmont 1:fdd22bb7aa52 497 /* acc += a2 * y[n-2] */
emilmont 1:fdd22bb7aa52 498 acc += mult32x64(Yn2, a2);
emilmont 1:fdd22bb7aa52 499
emilmont 1:fdd22bb7aa52 500 /* Every time after the output is computed state should be updated. */
emilmont 1:fdd22bb7aa52 501 /* The states should be updated as: */
emilmont 1:fdd22bb7aa52 502 /* Xn2 = Xn1 */
emilmont 1:fdd22bb7aa52 503 /* Xn1 = Xn */
emilmont 1:fdd22bb7aa52 504 /* Yn2 = Yn1 */
emilmont 1:fdd22bb7aa52 505 /* Yn1 = acc */
emilmont 1:fdd22bb7aa52 506 Xn2 = Xn1;
emilmont 1:fdd22bb7aa52 507 Xn1 = Xn;
emilmont 1:fdd22bb7aa52 508 Yn2 = Yn1;
emilmont 1:fdd22bb7aa52 509
emilmont 1:fdd22bb7aa52 510 /* The result is converted to 1.63, Yn1 variable is reused */
emilmont 1:fdd22bb7aa52 511 Yn1 = acc << shift;
emilmont 1:fdd22bb7aa52 512
emilmont 1:fdd22bb7aa52 513 /* Calc lower part of acc */
emilmont 1:fdd22bb7aa52 514 acc_l = acc & 0xffffffff;
emilmont 1:fdd22bb7aa52 515
emilmont 1:fdd22bb7aa52 516 /* Calc upper part of acc */
emilmont 1:fdd22bb7aa52 517 acc_h = (acc >> 32) & 0xffffffff;
emilmont 1:fdd22bb7aa52 518
emilmont 1:fdd22bb7aa52 519 /* Apply shift for lower part of acc and upper part of acc */
emilmont 1:fdd22bb7aa52 520 acc_h = (uint32_t) acc_l >> lShift | acc_h << uShift;
emilmont 1:fdd22bb7aa52 521
emilmont 1:fdd22bb7aa52 522 /* Store the output in the destination buffer in 1.31 format. */
emilmont 1:fdd22bb7aa52 523 *pOut++ = acc_h;
emilmont 1:fdd22bb7aa52 524
emilmont 1:fdd22bb7aa52 525 //Yn1 = acc << shift;
emilmont 1:fdd22bb7aa52 526
emilmont 1:fdd22bb7aa52 527 /* Store the output in the destination buffer in 1.31 format. */
emilmont 1:fdd22bb7aa52 528 //*pOut++ = (q31_t) (acc >> (32 - shift));
emilmont 1:fdd22bb7aa52 529
emilmont 1:fdd22bb7aa52 530 /* decrement the loop counter */
emilmont 1:fdd22bb7aa52 531 sample--;
emilmont 1:fdd22bb7aa52 532 }
emilmont 1:fdd22bb7aa52 533
emilmont 1:fdd22bb7aa52 534 /* The first stage output is given as input to the second stage. */
emilmont 1:fdd22bb7aa52 535 pIn = pDst;
emilmont 1:fdd22bb7aa52 536
emilmont 1:fdd22bb7aa52 537 /* Reset to destination buffer working pointer */
emilmont 1:fdd22bb7aa52 538 pOut = pDst;
emilmont 1:fdd22bb7aa52 539
emilmont 1:fdd22bb7aa52 540 /* Store the updated state variables back into the pState array */
emilmont 1:fdd22bb7aa52 541 *pState++ = (q63_t) Xn1;
emilmont 1:fdd22bb7aa52 542 *pState++ = (q63_t) Xn2;
emilmont 1:fdd22bb7aa52 543 *pState++ = Yn1;
emilmont 1:fdd22bb7aa52 544 *pState++ = Yn2;
emilmont 1:fdd22bb7aa52 545
emilmont 1:fdd22bb7aa52 546 } while(--stage);
emilmont 1:fdd22bb7aa52 547
emilmont 1:fdd22bb7aa52 548 #endif /* #ifndef ARM_MATH_CM0 */
emilmont 1:fdd22bb7aa52 549 }
emilmont 1:fdd22bb7aa52 550
emilmont 1:fdd22bb7aa52 551 /**
emilmont 1:fdd22bb7aa52 552 * @} end of BiquadCascadeDF1_32x64 group
emilmont 1:fdd22bb7aa52 553 */