arm_biquad_cascade_df1_q31.c

00001 /* ----------------------------------------------------------------------  
00002 * Copyright (C) 2010 ARM Limited. All rights reserved.  
00003 *  
00004 * $Date:        29. November 2010  
00005 * $Revision:    V1.0.3  
00006 *  
00007 * Project:      CMSIS DSP Library  
00008 * Title:        arm_biquad_cascade_df1_q31.c  
00009 *  
00010 * Description:  Processing function for the  
00011 *               Q31 Biquad cascade filter  
00012 *  
00013 * Target Processor: Cortex-M4/Cortex-M3
00014 *  
00015 * Version 1.0.3 2010/11/29 
00016 *    Re-organized the CMSIS folders and updated documentation.  
00017 *   
00018 * Version 1.0.2 2010/11/11  
00019 *    Documentation updated.   
00020 *  
00021 * Version 1.0.1 2010/10/05   
00022 *    Production release and review comments incorporated.  
00023 *  
00024 * Version 1.0.0 2010/09/20   
00025 *    Production release and review comments incorporated.  
00026 *  
00027 * Version 0.0.5  2010/04/26   
00028 *    incorporated review comments and updated with latest CMSIS layer  
00029 *  
00030 * Version 0.0.3  2010/03/10   
00031 *    Initial version  
00032 * -------------------------------------------------------------------- */ 
00033  
00034 #include "arm_math.h" 
00035  
00036 /**  
00037  * @ingroup groupFilters  
00038  */ 
00039  
00040 /**  
00041  * @addtogroup BiquadCascadeDF1  
00042  * @{  
00043  */ 
00044  
00045 /**  
00046  * @brief Processing function for the Q31 Biquad cascade filter.  
00047  * @param[in]  *S         points to an instance of the Q31 Biquad cascade structure.  
00048  * @param[in]  *pSrc      points to the block of input data.  
00049  * @param[out] *pDst      points to the block of output data.  
00050  * @param[in]  blockSize  number of samples to process per call.  
00051  * @return none.  
00052  *  
00053  * <b>Scaling and Overflow Behavior:</b>  
00054  * \par  
00055  * The function is implemented using an internal 64-bit accumulator.  
00056  * The accumulator has a 2.62 format and maintains full precision of the intermediate multiplication results but provides only a single guard bit.  
00057  * Thus, if the accumulator result overflows it wraps around rather than clip.  
00058  * 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).  
00059  * After all 5 multiply-accumulates are performed, the 2.62 accumulator is shifted by <code>postShift</code> bits and the result truncated to  
00060  * 1.31 format by discarding the low 32 bits.  
00061  *  
00062  * \par  
00063  * Refer to the function <code>arm_biquad_cascade_df1_fast_q31()</code> for a faster but less precise implementation of this filter.  
00064  */ 
00065  
00066 void arm_biquad_cascade_df1_q31( 
00067   const arm_biquad_casd_df1_inst_q31 * S, 
00068   q31_t * pSrc, 
00069   q31_t * pDst, 
00070   uint32_t blockSize) 
00071 { 
00072   q31_t *pIn = pSrc;                             /*  input pointer initialization  */ 
00073   q31_t *pOut = pDst;                            /*  output pointer initialization */ 
00074   q31_t *pState = S->pState;                     /*  pState pointer initialization */ 
00075   q31_t *pCoeffs = S->pCoeffs;                   /*  coeff pointer initialization  */ 
00076   q63_t acc;                                     /*  accumulator                   */ 
00077   q31_t Xn1, Xn2, Yn1, Yn2;                      /*  Filter state variables        */ 
00078   q31_t b0, b1, b2, a1, a2;                      /*  Filter coefficients           */ 
00079   q31_t Xn;                                      /*  temporary input               */ 
00080   uint32_t shift = 32u - ((uint32_t) S->postShift + 1u);        /*  Shift to be applied to the output */ 
00081   uint32_t sample, stage = S->numStages;         /*  loop counters                     */ 
00082  
00083  
00084   do 
00085   { 
00086     /* Reading the coefficients */ 
00087     b0 = *pCoeffs++; 
00088     b1 = *pCoeffs++; 
00089     b2 = *pCoeffs++; 
00090     a1 = *pCoeffs++; 
00091     a2 = *pCoeffs++; 
00092  
00093     /* Reading the state values */ 
00094     Xn1 = pState[0]; 
00095     Xn2 = pState[1]; 
00096     Yn1 = pState[2]; 
00097     Yn2 = pState[3]; 
00098  
00099     /* Apply loop unrolling and compute 4 output values simultaneously. */ 
00100     /*      The variable acc hold output values that are being computed:  
00101      *  
00102      *    acc =  b0 * x[n] + b1 * x[n-1] + b2 * x[n-2] + a1 * y[n-1] + a2 * y[n-2]  
00103      */ 
00104  
00105     sample = blockSize >> 2u; 
00106  
00107     /* First part of the processing with loop unrolling.  Compute 4 outputs at a time.  
00108      ** a second loop below computes the remaining 1 to 3 samples. */ 
00109     while(sample > 0u) 
00110     { 
00111       /* Read the input */ 
00112       Xn = *pIn++; 
00113  
00114       /* acc =  b0 * x[n] + b1 * x[n-1] + b2 * x[n-2] + a1 * y[n-1] + a2 * y[n-2] */ 
00115  
00116       /* acc =  b0 * x[n] */ 
00117       acc = (q63_t) b0 *Xn; 
00118       /* acc +=  b1 * x[n-1] */ 
00119       acc += (q63_t) b1 *Xn1; 
00120       /* acc +=  b[2] * x[n-2] */ 
00121       acc += (q63_t) b2 *Xn2; 
00122       /* acc +=  a1 * y[n-1] */ 
00123       acc += (q63_t) a1 *Yn1; 
00124       /* acc +=  a2 * y[n-2] */ 
00125       acc += (q63_t) a2 *Yn2; 
00126  
00127       /* The result is converted to 1.31 , Yn2 variable is reused */ 
00128       Yn2 = (q31_t) (acc >> shift); 
00129  
00130       /* Store the output in the destination buffer. */ 
00131       *pOut++ = Yn2; 
00132  
00133       /* Read the second input */ 
00134       Xn2 = *pIn++; 
00135  
00136       /* acc =  b0 * x[n] + b1 * x[n-1] + b2 * x[n-2] + a1 * y[n-1] + a2 * y[n-2] */ 
00137  
00138       /* acc =  b0 * x[n] */ 
00139       acc = (q63_t) b0 *Xn2; 
00140       /* acc +=  b1 * x[n-1] */ 
00141       acc += (q63_t) b1 *Xn; 
00142       /* acc +=  b[2] * x[n-2] */ 
00143       acc += (q63_t) b2 *Xn1; 
00144       /* acc +=  a1 * y[n-1] */ 
00145       acc += (q63_t) a1 *Yn2; 
00146       /* acc +=  a2 * y[n-2] */ 
00147       acc += (q63_t) a2 *Yn1; 
00148  
00149  
00150       /* The result is converted to 1.31, Yn1 variable is reused  */ 
00151       Yn1 = (q31_t) (acc >> shift); 
00152  
00153       /* Store the output in the destination buffer. */ 
00154       *pOut++ = Yn1; 
00155  
00156       /* Read the third input  */ 
00157       Xn1 = *pIn++; 
00158  
00159       /* acc =  b0 * x[n] + b1 * x[n-1] + b2 * x[n-2] + a1 * y[n-1] + a2 * y[n-2] */ 
00160  
00161       /* acc =  b0 * x[n] */ 
00162       acc = (q63_t) b0 *Xn1; 
00163       /* acc +=  b1 * x[n-1] */ 
00164       acc += (q63_t) b1 *Xn2; 
00165       /* acc +=  b[2] * x[n-2] */ 
00166       acc += (q63_t) b2 *Xn; 
00167       /* acc +=  a1 * y[n-1] */ 
00168       acc += (q63_t) a1 *Yn1; 
00169       /* acc +=  a2 * y[n-2] */ 
00170       acc += (q63_t) a2 *Yn2; 
00171  
00172       /* The result is converted to 1.31, Yn2 variable is reused  */ 
00173       Yn2 = (q31_t) (acc >> shift); 
00174  
00175       /* Store the output in the destination buffer. */ 
00176       *pOut++ = Yn2; 
00177  
00178       /* Read the forth input */ 
00179       Xn = *pIn++; 
00180  
00181       /* acc =  b0 * x[n] + b1 * x[n-1] + b2 * x[n-2] + a1 * y[n-1] + a2 * y[n-2] */ 
00182  
00183       /* acc =  b0 * x[n] */ 
00184       acc = (q63_t) b0 *Xn; 
00185       /* acc +=  b1 * x[n-1] */ 
00186       acc += (q63_t) b1 *Xn1; 
00187       /* acc +=  b[2] * x[n-2] */ 
00188       acc += (q63_t) b2 *Xn2; 
00189       /* acc +=  a1 * y[n-1] */ 
00190       acc += (q63_t) a1 *Yn2; 
00191       /* acc +=  a2 * y[n-2] */ 
00192       acc += (q63_t) a2 *Yn1; 
00193  
00194       /* The result is converted to 1.31, Yn1 variable is reused  */ 
00195       Yn1 = (q31_t) (acc >> shift); 
00196  
00197       /* Every time after the output is computed state should be updated. */ 
00198       /* The states should be updated as:  */ 
00199       /* Xn2 = Xn1    */ 
00200       /* Xn1 = Xn     */ 
00201       /* Yn2 = Yn1    */ 
00202       /* Yn1 = acc    */ 
00203       Xn2 = Xn1; 
00204       Xn1 = Xn; 
00205  
00206       /* Store the output in the destination buffer. */ 
00207       *pOut++ = Yn1; 
00208  
00209       /* decrement the loop counter */ 
00210       sample--; 
00211     } 
00212  
00213     /* If the blockSize is not a multiple of 4, compute any remaining output samples here.  
00214      ** No loop unrolling is used. */ 
00215     sample = (blockSize & 0x3u); 
00216  
00217     while(sample > 0u) 
00218     { 
00219       /* Read the input */ 
00220       Xn = *pIn++; 
00221  
00222       /* acc =  b0 * x[n] + b1 * x[n-1] + b2 * x[n-2] + a1 * y[n-1] + a2 * y[n-2] */ 
00223  
00224       /* acc =  b0 * x[n] */ 
00225       acc = (q63_t) b0 *Xn; 
00226       /* acc +=  b1 * x[n-1] */ 
00227       acc += (q63_t) b1 *Xn1; 
00228       /* acc +=  b[2] * x[n-2] */ 
00229       acc += (q63_t) b2 *Xn2; 
00230       /* acc +=  a1 * y[n-1] */ 
00231       acc += (q63_t) a1 *Yn1; 
00232       /* acc +=  a2 * y[n-2] */ 
00233       acc += (q63_t) a2 *Yn2; 
00234  
00235       /* The result is converted to 1.31  */ 
00236       acc = acc >> shift; 
00237  
00238       /* Every time after the output is computed state should be updated. */ 
00239       /* The states should be updated as:  */ 
00240       /* Xn2 = Xn1    */ 
00241       /* Xn1 = Xn     */ 
00242       /* Yn2 = Yn1    */ 
00243       /* Yn1 = acc    */ 
00244       Xn2 = Xn1; 
00245       Xn1 = Xn; 
00246       Yn2 = Yn1; 
00247       Yn1 = (q31_t) acc; 
00248  
00249       /* Store the output in the destination buffer. */ 
00250       *pOut++ = (q31_t) acc; 
00251  
00252       /* decrement the loop counter */ 
00253       sample--; 
00254     } 
00255  
00256     /*  The first stage goes from the input buffer to the output buffer. */ 
00257     /*  Subsequent stages occur in-place in the output buffer */ 
00258     pIn = pDst; 
00259  
00260     /* Reset to destination pointer */ 
00261     pOut = pDst; 
00262  
00263     /*  Store the updated state variables back into the pState array */ 
00264     *pState++ = Xn1; 
00265     *pState++ = Xn2; 
00266     *pState++ = Yn1; 
00267     *pState++ = Yn2; 
00268  
00269   } while(--stage); 
00270 } 
00271  
00272 /**  
00273   * @} end of BiquadCascadeDF1 group  
00274   */