A bundled STM32F10x Std Periph and CMSIS library
Vous ne pouvez pas sélectionner plus de 25 sujets Les noms de sujets doivent commencer par une lettre ou un nombre, peuvent contenir des tirets ('-') et peuvent comporter jusqu'à 35 caractères.
 
 
 
 
 

406 lignes
12 KiB

  1. /* ----------------------------------------------------------------------
  2. * Copyright (C) 2010-2014 ARM Limited. All rights reserved.
  3. *
  4. * $Date: 12. March 2014
  5. * $Revision: V1.4.4
  6. *
  7. * Project: CMSIS DSP Library
  8. * Title: arm_biquad_cascade_df1_q31.c
  9. *
  10. * Description: Processing function for the
  11. * Q31 Biquad cascade filter
  12. *
  13. * Target Processor: Cortex-M4/Cortex-M3/Cortex-M0
  14. *
  15. * Redistribution and use in source and binary forms, with or without
  16. * modification, are permitted provided that the following conditions
  17. * are met:
  18. * - Redistributions of source code must retain the above copyright
  19. * notice, this list of conditions and the following disclaimer.
  20. * - Redistributions in binary form must reproduce the above copyright
  21. * notice, this list of conditions and the following disclaimer in
  22. * the documentation and/or other materials provided with the
  23. * distribution.
  24. * - Neither the name of ARM LIMITED nor the names of its contributors
  25. * may be used to endorse or promote products derived from this
  26. * software without specific prior written permission.
  27. *
  28. * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
  29. * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
  30. * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
  31. * FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE
  32. * COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
  33. * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,
  34. * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
  35. * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
  36. * CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
  37. * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN
  38. * ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
  39. * POSSIBILITY OF SUCH DAMAGE.
  40. * -------------------------------------------------------------------- */
  41. #include "arm_math.h"
  42. /**
  43. * @ingroup groupFilters
  44. */
  45. /**
  46. * @addtogroup BiquadCascadeDF1
  47. * @{
  48. */
  49. /**
  50. * @brief Processing function for the Q31 Biquad cascade filter.
  51. * @param[in] *S points to an instance of the Q31 Biquad cascade structure.
  52. * @param[in] *pSrc points to the block of input data.
  53. * @param[out] *pDst points to the block of output data.
  54. * @param[in] blockSize number of samples to process per call.
  55. * @return none.
  56. *
  57. * <b>Scaling and Overflow Behavior:</b>
  58. * \par
  59. * The function is implemented using an internal 64-bit accumulator.
  60. * The accumulator has a 2.62 format and maintains full precision of the intermediate multiplication results but provides only a single guard bit.
  61. * Thus, if the accumulator result overflows it wraps around rather than clip.
  62. * 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).
  63. * After all 5 multiply-accumulates are performed, the 2.62 accumulator is shifted by <code>postShift</code> bits and the result truncated to
  64. * 1.31 format by discarding the low 32 bits.
  65. *
  66. * \par
  67. * Refer to the function <code>arm_biquad_cascade_df1_fast_q31()</code> for a faster but less precise implementation of this filter for Cortex-M3 and Cortex-M4.
  68. */
  69. void arm_biquad_cascade_df1_q31(
  70. const arm_biquad_casd_df1_inst_q31 * S,
  71. q31_t * pSrc,
  72. q31_t * pDst,
  73. uint32_t blockSize)
  74. {
  75. q63_t acc; /* accumulator */
  76. uint32_t uShift = ((uint32_t) S->postShift + 1u);
  77. uint32_t lShift = 32u - uShift; /* Shift to be applied to the output */
  78. q31_t *pIn = pSrc; /* input pointer initialization */
  79. q31_t *pOut = pDst; /* output pointer initialization */
  80. q31_t *pState = S->pState; /* pState pointer initialization */
  81. q31_t *pCoeffs = S->pCoeffs; /* coeff pointer initialization */
  82. q31_t Xn1, Xn2, Yn1, Yn2; /* Filter state variables */
  83. q31_t b0, b1, b2, a1, a2; /* Filter coefficients */
  84. q31_t Xn; /* temporary input */
  85. uint32_t sample, stage = S->numStages; /* loop counters */
  86. #ifndef ARM_MATH_CM0_FAMILY_FAMILY
  87. q31_t acc_l, acc_h; /* temporary output variables */
  88. /* Run the below code for Cortex-M4 and Cortex-M3 */
  89. do
  90. {
  91. /* Reading the coefficients */
  92. b0 = *pCoeffs++;
  93. b1 = *pCoeffs++;
  94. b2 = *pCoeffs++;
  95. a1 = *pCoeffs++;
  96. a2 = *pCoeffs++;
  97. /* Reading the state values */
  98. Xn1 = pState[0];
  99. Xn2 = pState[1];
  100. Yn1 = pState[2];
  101. Yn2 = pState[3];
  102. /* Apply loop unrolling and compute 4 output values simultaneously. */
  103. /* The variable acc hold output values that are being computed:
  104. *
  105. * acc = b0 * x[n] + b1 * x[n-1] + b2 * x[n-2] + a1 * y[n-1] + a2 * y[n-2]
  106. */
  107. sample = blockSize >> 2u;
  108. /* First part of the processing with loop unrolling. Compute 4 outputs at a time.
  109. ** a second loop below computes the remaining 1 to 3 samples. */
  110. while(sample > 0u)
  111. {
  112. /* Read the input */
  113. Xn = *pIn++;
  114. /* acc = b0 * x[n] + b1 * x[n-1] + b2 * x[n-2] + a1 * y[n-1] + a2 * y[n-2] */
  115. /* acc = b0 * x[n] */
  116. acc = (q63_t) b0 *Xn;
  117. /* acc += b1 * x[n-1] */
  118. acc += (q63_t) b1 *Xn1;
  119. /* acc += b[2] * x[n-2] */
  120. acc += (q63_t) b2 *Xn2;
  121. /* acc += a1 * y[n-1] */
  122. acc += (q63_t) a1 *Yn1;
  123. /* acc += a2 * y[n-2] */
  124. acc += (q63_t) a2 *Yn2;
  125. /* The result is converted to 1.31 , Yn2 variable is reused */
  126. /* Calc lower part of acc */
  127. acc_l = acc & 0xffffffff;
  128. /* Calc upper part of acc */
  129. acc_h = (acc >> 32) & 0xffffffff;
  130. /* Apply shift for lower part of acc and upper part of acc */
  131. Yn2 = (uint32_t) acc_l >> lShift | acc_h << uShift;
  132. /* Store the output in the destination buffer. */
  133. *pOut++ = Yn2;
  134. /* Read the second input */
  135. Xn2 = *pIn++;
  136. /* acc = b0 * x[n] + b1 * x[n-1] + b2 * x[n-2] + a1 * y[n-1] + a2 * y[n-2] */
  137. /* acc = b0 * x[n] */
  138. acc = (q63_t) b0 *Xn2;
  139. /* acc += b1 * x[n-1] */
  140. acc += (q63_t) b1 *Xn;
  141. /* acc += b[2] * x[n-2] */
  142. acc += (q63_t) b2 *Xn1;
  143. /* acc += a1 * y[n-1] */
  144. acc += (q63_t) a1 *Yn2;
  145. /* acc += a2 * y[n-2] */
  146. acc += (q63_t) a2 *Yn1;
  147. /* The result is converted to 1.31, Yn1 variable is reused */
  148. /* Calc lower part of acc */
  149. acc_l = acc & 0xffffffff;
  150. /* Calc upper part of acc */
  151. acc_h = (acc >> 32) & 0xffffffff;
  152. /* Apply shift for lower part of acc and upper part of acc */
  153. Yn1 = (uint32_t) acc_l >> lShift | acc_h << uShift;
  154. /* Store the output in the destination buffer. */
  155. *pOut++ = Yn1;
  156. /* Read the third input */
  157. Xn1 = *pIn++;
  158. /* acc = b0 * x[n] + b1 * x[n-1] + b2 * x[n-2] + a1 * y[n-1] + a2 * y[n-2] */
  159. /* acc = b0 * x[n] */
  160. acc = (q63_t) b0 *Xn1;
  161. /* acc += b1 * x[n-1] */
  162. acc += (q63_t) b1 *Xn2;
  163. /* acc += b[2] * x[n-2] */
  164. acc += (q63_t) b2 *Xn;
  165. /* acc += a1 * y[n-1] */
  166. acc += (q63_t) a1 *Yn1;
  167. /* acc += a2 * y[n-2] */
  168. acc += (q63_t) a2 *Yn2;
  169. /* The result is converted to 1.31, Yn2 variable is reused */
  170. /* Calc lower part of acc */
  171. acc_l = acc & 0xffffffff;
  172. /* Calc upper part of acc */
  173. acc_h = (acc >> 32) & 0xffffffff;
  174. /* Apply shift for lower part of acc and upper part of acc */
  175. Yn2 = (uint32_t) acc_l >> lShift | acc_h << uShift;
  176. /* Store the output in the destination buffer. */
  177. *pOut++ = Yn2;
  178. /* Read the forth input */
  179. Xn = *pIn++;
  180. /* acc = b0 * x[n] + b1 * x[n-1] + b2 * x[n-2] + a1 * y[n-1] + a2 * y[n-2] */
  181. /* acc = b0 * x[n] */
  182. acc = (q63_t) b0 *Xn;
  183. /* acc += b1 * x[n-1] */
  184. acc += (q63_t) b1 *Xn1;
  185. /* acc += b[2] * x[n-2] */
  186. acc += (q63_t) b2 *Xn2;
  187. /* acc += a1 * y[n-1] */
  188. acc += (q63_t) a1 *Yn2;
  189. /* acc += a2 * y[n-2] */
  190. acc += (q63_t) a2 *Yn1;
  191. /* The result is converted to 1.31, Yn1 variable is reused */
  192. /* Calc lower part of acc */
  193. acc_l = acc & 0xffffffff;
  194. /* Calc upper part of acc */
  195. acc_h = (acc >> 32) & 0xffffffff;
  196. /* Apply shift for lower part of acc and upper part of acc */
  197. Yn1 = (uint32_t) acc_l >> lShift | acc_h << uShift;
  198. /* Every time after the output is computed state should be updated. */
  199. /* The states should be updated as: */
  200. /* Xn2 = Xn1 */
  201. /* Xn1 = Xn */
  202. /* Yn2 = Yn1 */
  203. /* Yn1 = acc */
  204. Xn2 = Xn1;
  205. Xn1 = Xn;
  206. /* Store the output in the destination buffer. */
  207. *pOut++ = Yn1;
  208. /* decrement the loop counter */
  209. sample--;
  210. }
  211. /* If the blockSize is not a multiple of 4, compute any remaining output samples here.
  212. ** No loop unrolling is used. */
  213. sample = (blockSize & 0x3u);
  214. while(sample > 0u)
  215. {
  216. /* Read the input */
  217. Xn = *pIn++;
  218. /* acc = b0 * x[n] + b1 * x[n-1] + b2 * x[n-2] + a1 * y[n-1] + a2 * y[n-2] */
  219. /* acc = b0 * x[n] */
  220. acc = (q63_t) b0 *Xn;
  221. /* acc += b1 * x[n-1] */
  222. acc += (q63_t) b1 *Xn1;
  223. /* acc += b[2] * x[n-2] */
  224. acc += (q63_t) b2 *Xn2;
  225. /* acc += a1 * y[n-1] */
  226. acc += (q63_t) a1 *Yn1;
  227. /* acc += a2 * y[n-2] */
  228. acc += (q63_t) a2 *Yn2;
  229. /* The result is converted to 1.31 */
  230. acc = acc >> lShift;
  231. /* Every time after the output is computed state should be updated. */
  232. /* The states should be updated as: */
  233. /* Xn2 = Xn1 */
  234. /* Xn1 = Xn */
  235. /* Yn2 = Yn1 */
  236. /* Yn1 = acc */
  237. Xn2 = Xn1;
  238. Xn1 = Xn;
  239. Yn2 = Yn1;
  240. Yn1 = (q31_t) acc;
  241. /* Store the output in the destination buffer. */
  242. *pOut++ = (q31_t) acc;
  243. /* decrement the loop counter */
  244. sample--;
  245. }
  246. /* The first stage goes from the input buffer to the output buffer. */
  247. /* Subsequent stages occur in-place in the output buffer */
  248. pIn = pDst;
  249. /* Reset to destination pointer */
  250. pOut = pDst;
  251. /* Store the updated state variables back into the pState array */
  252. *pState++ = Xn1;
  253. *pState++ = Xn2;
  254. *pState++ = Yn1;
  255. *pState++ = Yn2;
  256. } while(--stage);
  257. #else
  258. /* Run the below code for Cortex-M0 */
  259. do
  260. {
  261. /* Reading the coefficients */
  262. b0 = *pCoeffs++;
  263. b1 = *pCoeffs++;
  264. b2 = *pCoeffs++;
  265. a1 = *pCoeffs++;
  266. a2 = *pCoeffs++;
  267. /* Reading the state values */
  268. Xn1 = pState[0];
  269. Xn2 = pState[1];
  270. Yn1 = pState[2];
  271. Yn2 = pState[3];
  272. /* The variables acc holds the output value that is computed:
  273. * acc = b0 * x[n] + b1 * x[n-1] + b2 * x[n-2] + a1 * y[n-1] + a2 * y[n-2]
  274. */
  275. sample = blockSize;
  276. while(sample > 0u)
  277. {
  278. /* Read the input */
  279. Xn = *pIn++;
  280. /* acc = b0 * x[n] + b1 * x[n-1] + b2 * x[n-2] + a1 * y[n-1] + a2 * y[n-2] */
  281. /* acc = b0 * x[n] */
  282. acc = (q63_t) b0 *Xn;
  283. /* acc += b1 * x[n-1] */
  284. acc += (q63_t) b1 *Xn1;
  285. /* acc += b[2] * x[n-2] */
  286. acc += (q63_t) b2 *Xn2;
  287. /* acc += a1 * y[n-1] */
  288. acc += (q63_t) a1 *Yn1;
  289. /* acc += a2 * y[n-2] */
  290. acc += (q63_t) a2 *Yn2;
  291. /* The result is converted to 1.31 */
  292. acc = acc >> lShift;
  293. /* Every time after the output is computed state should be updated. */
  294. /* The states should be updated as: */
  295. /* Xn2 = Xn1 */
  296. /* Xn1 = Xn */
  297. /* Yn2 = Yn1 */
  298. /* Yn1 = acc */
  299. Xn2 = Xn1;
  300. Xn1 = Xn;
  301. Yn2 = Yn1;
  302. Yn1 = (q31_t) acc;
  303. /* Store the output in the destination buffer. */
  304. *pOut++ = (q31_t) acc;
  305. /* decrement the loop counter */
  306. sample--;
  307. }
  308. /* The first stage goes from the input buffer to the output buffer. */
  309. /* Subsequent stages occur in-place in the output buffer */
  310. pIn = pDst;
  311. /* Reset to destination pointer */
  312. pOut = pDst;
  313. /* Store the updated state variables back into the pState array */
  314. *pState++ = Xn1;
  315. *pState++ = Xn2;
  316. *pState++ = Yn1;
  317. *pState++ = Yn2;
  318. } while(--stage);
  319. #endif /* #ifndef ARM_MATH_CM0_FAMILY_FAMILY */
  320. }
  321. /**
  322. * @} end of BiquadCascadeDF1 group
  323. */