1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050 2051 2052 2053 2054 2055 2056 2057 2058 2059 2060 2061 2062 2063 2064 2065 2066 2067 2068 2069 2070 2071 2072 2073 2074 2075 2076 2077 2078 2079 2080 2081 2082 2083 2084 2085 2086 2087 2088 2089 2090 2091 2092 2093 2094 2095 2096 2097 2098 2099 2100 2101 2102 2103 2104 2105 2106 2107 2108 2109 2110 2111 2112 2113 2114 2115 2116 2117 2118 2119 2120 2121 2122 2123 2124 2125 2126 2127 2128 2129 2130 2131 2132 2133 2134 2135 2136 2137 2138 2139 2140 2141 2142 2143 2144 2145 2146 2147 2148 2149 2150 2151 2152 2153 2154 2155 2156 2157 2158 2159 2160 2161 2162 2163 2164 2165 2166 2167 2168 2169 2170 2171 2172 2173 2174 2175 2176 2177 2178 2179 2180 2181 2182 2183 2184 2185 2186 2187 2188 2189 2190 2191 2192 2193 2194 2195 2196 2197 2198 2199 2200 2201 2202 2203 2204 2205 2206 2207 2208 2209 2210 2211 2212 2213 2214 2215 2216 2217 2218 2219 2220 2221 2222 2223 2224 2225 2226 2227 2228 2229 2230 2231 2232 2233 2234 2235 2236 2237 2238 2239 2240 2241 2242 2243 2244 2245 2246 2247 2248 2249 2250 2251 2252 2253 2254 2255 2256 2257 2258 2259 2260 2261 2262 2263 2264 2265 2266 2267 2268 2269 2270 2271 2272 2273 2274 2275 2276 2277 2278 2279 2280 2281 2282 2283 2284 2285 2286 2287 2288 2289 2290 2291 2292 2293 2294 2295 2296 2297 2298 2299 2300 2301 2302 2303 2304 2305 2306 2307 2308 2309 2310 2311 2312 2313 2314 2315 2316 2317 2318 2319 2320 2321 2322 2323 2324 2325 2326 2327 2328 2329 2330 2331 2332 2333 2334 2335 2336 2337 2338 2339 2340 2341 2342 2343 2344 2345 2346 2347 2348 2349 2350 2351 2352 2353 2354 2355 2356 2357 2358 2359 2360 2361 2362 2363 2364 2365 2366 2367 2368 2369 2370 2371 2372 2373 2374 2375 2376 2377 2378 2379 2380 2381 2382 2383 2384 2385 2386 2387 2388 2389 2390 2391 2392 2393 2394 2395 2396 2397 2398 2399 2400 2401 2402 2403 2404 2405 2406 2407 2408 2409 2410 2411 2412 2413 2414 2415 2416 2417 2418 2419 2420 2421 2422 2423 2424 2425 2426 2427 2428 2429 2430 2431 2432 2433 2434 2435 2436 2437 2438 2439 2440 2441 2442 2443 2444 2445 2446 2447 2448 2449 2450 2451 2452 2453 2454 2455 2456 2457 2458 2459 2460 2461 2462 2463 2464 2465 2466 2467 2468 2469 2470 2471 2472 2473 2474 2475 2476 2477 2478 2479 2480 2481 2482 2483 2484 2485 2486 2487 2488 2489 2490 2491 2492 2493 2494 2495 2496 2497 2498 2499 2500 2501 2502 2503 2504 2505 2506 2507 2508 2509 2510 2511 2512 2513 2514 2515 2516 2517 2518 2519 2520 2521 2522 2523 2524 2525 2526 2527 2528 2529 2530 2531 2532 2533 2534 2535 2536 2537 2538 2539 2540 2541 2542 2543 2544 2545 2546 2547 2548 2549 2550 2551 2552 2553 2554 2555 2556 2557 2558 2559 2560 2561 2562 2563 2564 2565 2566 2567 2568 2569 2570 2571 2572 2573 2574 2575 2576 2577 2578 2579 2580 2581 2582 2583 2584 2585 2586 2587 2588 2589 2590 2591 2592 2593 2594 2595 2596 2597 2598 2599 2600 2601 2602 2603 2604 2605 2606 2607 2608 2609 2610 2611 2612 2613 2614 2615 2616 2617 2618 2619 2620 2621 2622 2623 2624 2625 2626 2627 2628 2629 2630 2631 2632 2633 2634 2635 2636 2637 2638 2639 2640 2641 2642 2643 2644 2645 2646 2647 2648 2649 2650 2651 2652 2653 2654 2655 2656 2657 2658 2659 2660 2661 2662 2663 2664 2665 2666 2667 2668 2669 2670 2671 2672 2673 2674 2675 2676 2677 2678 2679 2680 2681 2682 2683 2684 2685 2686 2687 2688 2689 2690 2691 2692 2693 2694 2695 2696 2697 2698 2699 2700 2701 2702 2703 2704 2705 2706 2707 2708 2709 2710 2711 2712 2713 2714 2715 2716 2717 2718 2719 2720 2721 2722 2723 2724 2725 2726 2727 2728 2729 2730 2731 2732 2733 2734 2735 2736 2737 2738 2739 2740 2741 2742 2743 2744 2745 2746 2747 2748 2749 2750 2751 2752 2753 2754 2755 2756 2757 2758 2759 2760 2761 2762 2763 2764 2765 2766 2767 2768 2769 2770 2771 2772 2773 2774 2775 2776 2777 2778 2779 2780 2781 2782 2783 2784 2785 2786 2787 2788 2789 2790 2791 2792 2793 2794 2795 2796 2797 2798 2799 2800 2801 2802 2803 2804 2805 2806 2807 2808 2809 2810 2811 2812 2813 2814 2815 2816 2817 2818 2819 2820 2821 2822 2823 2824 2825 2826 2827 2828 2829 2830 2831 2832 2833 2834 2835 2836 2837 2838 2839 2840 2841 2842 2843 2844 2845 2846 2847 2848 2849 2850 2851 2852 2853 2854 2855 2856 2857 2858 2859 2860 2861 2862 2863 2864 2865 2866 2867 2868 2869 2870 2871 2872 2873 2874 2875 2876 2877 2878 2879 2880 2881 2882 2883 2884 2885 2886 2887 2888 2889 2890 2891 2892 2893 2894 2895 2896 2897 2898 2899 2900 2901 2902 2903 2904 2905 2906 2907 2908 2909 2910 2911 2912 2913 2914 2915 2916 2917 2918 2919 2920 2921 2922 2923 2924 2925 2926 2927 2928 2929 2930 2931 2932 2933 2934 2935 2936 2937 2938 2939 2940 2941 2942 2943 2944 2945 2946 2947 2948 2949 2950 2951 2952 2953 2954 2955 2956 2957 2958 2959 2960 2961 2962 2963 2964 2965 2966 2967 2968 2969 2970 2971 2972 2973 2974 2975 2976 2977 2978 2979 2980 2981 2982 2983 2984 2985 2986 2987 2988 2989 2990 2991 2992 2993 2994 2995 2996 2997 2998 2999 3000 3001 3002 3003 3004 3005 3006 3007 3008 3009 3010
|
//
// Copyright 2014 Ettus Research
// Copyright 2018 Ettus Research, a National Instruments Company
//
// SPDX-License-Identifier: GPL-3.0-or-later
//
#define _USE_MATH_DEFINES
#include "ad9361_device.h"
#include "ad9361_client.h"
#include "ad9361_filter_taps.h"
#include "ad9361_gain_tables.h"
#include "ad9361_synth_lut.h"
#include <uhd/exception.hpp>
#include <uhd/utils/log.hpp>
#include <stdint.h>
#include <boost/format.hpp>
#include <boost/scoped_array.hpp>
#include <chrono>
#include <cmath>
#include <thread>
////////////////////////////////////////////////////////////
// the following macros evaluate to a compile time constant
// macros By Tom Torfs - donated to the public domain
/* turn a numeric literal into a hex constant
(avoids problems with leading zeroes)
8-bit constants max value 0x11111111, always fits in unsigned long
*/
#define HEX__(n) 0x##n##LU
/* 8-bit conversion function */
#define B8__(x) \
((x & 0x0000000FLU) ? 1 : 0) + ((x & 0x000000F0LU) ? 2 : 0) \
+ ((x & 0x00000F00LU) ? 4 : 0) + ((x & 0x0000F000LU) ? 8 : 0) \
+ ((x & 0x000F0000LU) ? 16 : 0) + ((x & 0x00F00000LU) ? 32 : 0) \
+ ((x & 0x0F000000LU) ? 64 : 0) + ((x & 0xF0000000LU) ? 128 : 0)
/* for upto 8-bit binary constants */
#define B8(d) ((unsigned char)B8__(HEX__(d)))
////////////////////////////////////////////////////////////
namespace uhd { namespace usrp {
/* This is a simple comparison for very large double-precision floating
* point numbers. It is used to prevent re-tunes for frequencies that are
* the same but not 'exactly' because of data precision issues. */
// TODO: see if we can avoid the need for this function
int freq_is_nearly_equal(double a, double b)
{
return std::max(a, b) - std::min(a, b) < 1;
}
/***********************************************************************
* Filter functions
**********************************************************************/
/* This function takes in the calculated maximum number of FIR taps, and
* returns a number of taps that makes AD9361 happy. */
int get_num_taps(int max_num_taps)
{
int num_taps = 0;
int num_taps_list[] = {16, 32, 48, 64, 80, 96, 112, 128};
int i;
for (i = 1; i < 8; i++) {
if (max_num_taps >= num_taps_list[i]) {
continue;
} else {
num_taps = num_taps_list[i - 1];
break;
}
}
if (num_taps == 0) {
num_taps = 128;
}
return num_taps;
}
const double ad9361_device_t::AD9361_MAX_GAIN = 89.75;
const double ad9361_device_t::AD9361_MIN_CLOCK_RATE = 220e3;
const double ad9361_device_t::AD9361_MAX_CLOCK_RATE = 61.44e6;
const double ad9361_device_t::AD9361_CAL_VALID_WINDOW = 100e6;
// Max bandwdith is due to filter rolloff in analog filter stage
const double ad9361_device_t::AD9361_MIN_BW = 200e3;
const double ad9361_device_t::AD9361_MAX_BW = 56e6;
/* Startup RF frequencies */
const double ad9361_device_t::DEFAULT_RX_FREQ = 800e6;
const double ad9361_device_t::DEFAULT_TX_FREQ = 850e6;
/* Program either the RX or TX FIR filter.
*
* The process is the same for both filters, but the function must be told
* how many taps are in the filter, and given a vector of the taps
* themselves. */
void ad9361_device_t::_program_fir_filter(
direction_t direction, chain_t chain, int num_taps, uint16_t* coeffs)
{
uint16_t base;
/* RX and TX filters use largely identical sets of programming registers.
Select the appropriate bank of registers here. */
if (direction == RX) {
base = 0x0f0;
} else {
base = 0x060;
}
/* Encode number of filter taps for programming register */
uint8_t reg_numtaps = (((num_taps / 16) - 1) & 0x07) << 5;
uint8_t reg_chain = 0;
switch (chain) {
case CHAIN_1:
reg_chain = 0x01 << 3;
break;
case CHAIN_2:
reg_chain = 0x02 << 3;
break;
default:
reg_chain = 0x03 << 3;
}
/* Turn on the filter clock. */
_io_iface->poke8(base + 5, reg_numtaps | reg_chain | 0x02);
std::this_thread::sleep_for(std::chrono::milliseconds(1));
/* Zero the unused taps just in case they have stale data */
int addr;
for (addr = num_taps; addr < 128; addr++) {
_io_iface->poke8(base + 0, addr);
_io_iface->poke8(base + 1, 0x0);
_io_iface->poke8(base + 2, 0x0);
_io_iface->poke8(base + 5, reg_numtaps | reg_chain | (1 << 1) | (1 << 2));
_io_iface->poke8(base + 4, 0x00);
_io_iface->poke8(base + 4, 0x00);
}
/* Iterate through indirect programming of filter coeffs using ADI recomended
* procedure */
for (addr = 0; addr < num_taps; addr++) {
_io_iface->poke8(base + 0, addr);
_io_iface->poke8(base + 1, (coeffs[addr]) & 0xff);
_io_iface->poke8(base + 2, (coeffs[addr] >> 8) & 0xff);
_io_iface->poke8(base + 5, reg_numtaps | reg_chain | (1 << 1) | (1 << 2));
_io_iface->poke8(base + 4, 0x00);
_io_iface->poke8(base + 4, 0x00);
}
/* UG-671 states (page 25) (paraphrased and clarified):
" After the table has been programmed, write to register BASE+5 with the write bit D2
cleared and D1 high. Then, write to register BASE+5 again with D1 clear, thus
ensuring that the write bit resets internally before the clock stops. Wait 4 sample
clock periods after setting D2 high while that data writes into the table"
*/
_io_iface->poke8(base + 5, reg_numtaps | reg_chain | (1 << 1));
if (direction == RX) {
_io_iface->poke8(base + 5, reg_numtaps | reg_chain);
/* Rx Gain, set to prevent digital overflow/saturation in filters
0:+6dB, 1:0dB, 2:-6dB, 3:-12dB
page 35 of UG-671 */
_io_iface->poke8(
base + 6, 0x02); /* Also turn on -6dB Rx gain here, to stop filter overfow.*/
} else {
/* Tx Gain. bit[0]. set to prevent digital overflow/saturation in filters
0: 0dB, 1:-6dB
page 25 of UG-671 */
_io_iface->poke8(base + 5, reg_numtaps | reg_chain);
}
}
/* Program the RX FIR Filter. */
void ad9361_device_t::_setup_rx_fir(size_t num_taps, int32_t decimation)
{
if (not(decimation == 1 or decimation == 2 or decimation == 4)) {
throw uhd::runtime_error("[ad9361_device_t] Invalid Rx FIR decimation.");
}
boost::scoped_array<uint16_t> coeffs(new uint16_t[num_taps]);
for (size_t i = 0; i < num_taps; i++) {
switch (num_taps) {
case 128:
coeffs[i] =
uint16_t((decimation == 4) ? fir_128_x4_coeffs[i] : hb127_coeffs[i]);
break;
case 96:
coeffs[i] =
uint16_t((decimation == 4) ? fir_96_x4_coeffs[i] : hb95_coeffs[i]);
break;
case 64:
coeffs[i] =
uint16_t((decimation == 4) ? fir_64_x4_coeffs[i] : hb63_coeffs[i]);
break;
case 48:
coeffs[i] =
uint16_t((decimation == 4) ? fir_48_x4_coeffs[i] : hb47_coeffs[i]);
break;
default:
throw uhd::runtime_error(
"[ad9361_device_t] Unsupported number of Rx FIR taps.");
}
}
_program_fir_filter(RX, CHAIN_BOTH, num_taps, coeffs.get());
}
/* Program the TX FIR Filter. */
void ad9361_device_t::_setup_tx_fir(size_t num_taps, int32_t interpolation)
{
if (not(interpolation == 1 or interpolation == 2 or interpolation == 4)) {
throw uhd::runtime_error("[ad9361_device_t] Invalid Tx FIR interpolation.");
}
if (interpolation == 1 and num_taps > 64) {
throw uhd::runtime_error(
"[ad9361_device_t] Too many Tx FIR taps for interpolation value.");
}
boost::scoped_array<uint16_t> coeffs(new uint16_t[num_taps]);
for (size_t i = 0; i < num_taps; i++) {
switch (num_taps) {
case 128:
coeffs[i] = uint16_t(
(interpolation == 4) ? fir_128_x4_coeffs[i] : hb127_coeffs[i]);
break;
case 96:
coeffs[i] =
uint16_t((interpolation == 4) ? fir_96_x4_coeffs[i] : hb95_coeffs[i]);
break;
case 64:
coeffs[i] =
uint16_t((interpolation == 4) ? fir_64_x4_coeffs[i] : hb63_coeffs[i]);
break;
case 48:
coeffs[i] =
uint16_t((interpolation == 4) ? fir_48_x4_coeffs[i] : hb47_coeffs[i]);
break;
default:
throw uhd::runtime_error(
"[ad9361_device_t] Unsupported number of Tx FIR taps.");
}
}
_program_fir_filter(TX, CHAIN_BOTH, num_taps, coeffs.get());
}
/***********************************************************************
* Calibration functions
***********************************************************************/
/* Calibrate and lock the BBPLL.
*
* This function should be called anytime the BBPLL is tuned. */
void ad9361_device_t::_calibrate_lock_bbpll()
{
_io_iface->poke8(0x03F, 0x05); // Start the BBPLL calibration
_io_iface->poke8(0x03F, 0x01); // Clear the 'start' bit
/* Increase BBPLL KV and phase margin. */
_io_iface->poke8(0x04c, 0x86);
_io_iface->poke8(0x04d, 0x01);
_io_iface->poke8(0x04d, 0x05);
/* Wait for BBPLL lock. */
size_t count = 0;
while (!(_io_iface->peek8(0x05e) & 0x80)) {
if (count > 1000) {
throw uhd::runtime_error("[ad9361_device_t] BBPLL not locked");
break;
}
count++;
std::this_thread::sleep_for(std::chrono::milliseconds(2));
}
}
/* Calibrate the synthesizer charge pumps.
*
* Technically, this calibration only needs to be done once, at device
* initialization. */
void ad9361_device_t::_calibrate_synth_charge_pumps()
{
/* If this function ever gets called, and the ENSM isn't already in the
* ALERT state, then something has gone horribly wrong. */
if ((_io_iface->peek8(0x017) & 0x0F) != 5) {
throw uhd::runtime_error("[ad9361_device_t] AD9361 not in ALERT during cal");
}
/* Calibrate the RX synthesizer charge pump. */
size_t count = 0;
_io_iface->poke8(0x23d, 0x04);
while (!(_io_iface->peek8(0x244) & 0x80)) {
if (count > 5) {
throw uhd::runtime_error("[ad9361_device_t] RX charge pump cal failure");
break;
}
count++;
std::this_thread::sleep_for(std::chrono::milliseconds(1));
}
_io_iface->poke8(0x23d, 0x00);
/* Calibrate the TX synthesizer charge pump. */
count = 0;
_io_iface->poke8(0x27d, 0x04);
while (!(_io_iface->peek8(0x284) & 0x80)) {
if (count > 5) {
throw uhd::runtime_error("[ad9361_device_t] TX charge pump cal failure");
break;
}
count++;
std::this_thread::sleep_for(std::chrono::milliseconds(1));
}
_io_iface->poke8(0x27d, 0x00);
}
/* Calibrate the analog BB RX filter.
*
* Note that the filter calibration depends heavily on the baseband
* bandwidth, so this must be re-done after any change to the RX sample
* rate.
* UG570 Page 33 states that this filter should be calibrated to 1.4 * bbbw*/
double ad9361_device_t::_calibrate_baseband_rx_analog_filter(double req_rfbw)
{
double bbbw = req_rfbw / 2.0;
if (bbbw > _baseband_bw / 2.0) {
UHD_LOGGER_DEBUG("AD936X") << "baseband bandwidth too large for current sample "
"rate. Setting bandwidth to: "
<< _baseband_bw;
bbbw = _baseband_bw / 2.0;
}
/* Baseband BW must be between 28e6 and 0.143e6.
* Max filter BW is 39.2 MHz. 39.2 / 1.4 = 28
* Min filter BW is 200kHz. 200 / 1.4 = 143 */
if (bbbw > 28e6) {
bbbw = 28e6;
} else if (bbbw < 0.143e6) {
bbbw = 0.143e6;
}
double rxtune_clk = ((1.4 * bbbw * 2 * M_PI) / M_LN2);
_rx_bbf_tunediv =
std::min<uint16_t>(511, uint16_t(std::ceil(_bbpll_freq / rxtune_clk)));
_regs.bbftune_config = (_regs.bbftune_config & 0xFE)
| ((_rx_bbf_tunediv >> 8) & 0x0001);
double bbbw_mhz = bbbw / 1e6;
double temp = ((bbbw_mhz - std::floor(bbbw_mhz)) * 1000) / 7.8125;
uint8_t bbbw_khz = std::min<uint8_t>(127, uint8_t(std::floor(temp + 0.5)));
/* Set corner frequencies and dividers. */
_io_iface->poke8(0x1fb, (uint8_t)(bbbw_mhz));
_io_iface->poke8(0x1fc, bbbw_khz);
_io_iface->poke8(0x1f8, (_rx_bbf_tunediv & 0x00FF));
_io_iface->poke8(0x1f9, _regs.bbftune_config);
/* RX Mix Voltage settings - only change with apps engineer help. */
_io_iface->poke8(0x1d5, 0x3f);
_io_iface->poke8(0x1c0, 0x03);
/* Enable RX1 & RX2 filter tuners. */
_io_iface->poke8(0x1e2, 0x02);
_io_iface->poke8(0x1e3, 0x02);
/* Run the calibration! */
size_t count = 0;
_io_iface->poke8(0x016, 0x80);
while (_io_iface->peek8(0x016) & 0x80) {
if (count > 100) {
throw uhd::runtime_error("[ad9361_device_t] RX baseband filter cal FAILURE");
break;
}
count++;
std::this_thread::sleep_for(std::chrono::milliseconds(1));
}
/* Disable RX1 & RX2 filter tuners. */
_io_iface->poke8(0x1e2, 0x03);
_io_iface->poke8(0x1e3, 0x03);
return bbbw;
}
/* Calibrate the analog BB TX filter.
*
* Note that the filter calibration depends heavily on the baseband
* bandwidth, so this must be re-done after any change to the TX sample
* rate.
* UG570 Page 32 states that this filter should be calibrated to 1.6 * bbbw*/
double ad9361_device_t::_calibrate_baseband_tx_analog_filter(double req_rfbw)
{
double bbbw = req_rfbw / 2.0;
if (bbbw > _baseband_bw / 2.0) {
UHD_LOGGER_DEBUG("AD936X") << "baseband bandwidth too large for current sample "
"rate. Setting bandwidth to: "
<< _baseband_bw;
bbbw = _baseband_bw / 2.0;
}
/* Baseband BW must be between 20e6 and 0.391e6.
* Max filter BW is 32 MHz. 32 / 1.6 = 20
* Min filter BW is 625 kHz. 625 / 1.6 = 391 */
if (bbbw > 20e6) {
bbbw = 20e6;
} else if (bbbw < 0.391e6) {
bbbw = 0.391e6;
}
double txtune_clk = ((1.6 * bbbw * 2 * M_PI) / M_LN2);
uint16_t txbbfdiv =
std::min<uint16_t>(511, uint16_t(std::ceil(_bbpll_freq / txtune_clk)));
_regs.bbftune_mode = (_regs.bbftune_mode & 0xFE) | ((txbbfdiv >> 8) & 0x0001);
/* Program the divider values. */
_io_iface->poke8(0x0d6, (txbbfdiv & 0x00FF));
_io_iface->poke8(0x0d7, _regs.bbftune_mode);
/* Enable the filter tuner. */
_io_iface->poke8(0x0ca, 0x22);
/* Calibrate! */
size_t count = 0;
_io_iface->poke8(0x016, 0x40);
while (_io_iface->peek8(0x016) & 0x40) {
if (count > 100) {
throw uhd::runtime_error("[ad9361_device_t] TX baseband filter cal FAILURE");
break;
}
count++;
std::this_thread::sleep_for(std::chrono::milliseconds(1));
}
/* Disable the filter tuner. */
_io_iface->poke8(0x0ca, 0x26);
return bbbw;
}
/* Calibrate the secondary TX filter.
*
* This filter also depends on the TX sample rate, so if a rate change is
* made, the previous calibration will no longer be valid.
* UG570 Page 32 states that this filter should be calibrated to 5 * bbbw*/
double ad9361_device_t::_calibrate_secondary_tx_filter(double req_rfbw)
{
double bbbw = req_rfbw / 2.0;
if (bbbw > _baseband_bw / 2.0) {
UHD_LOGGER_DEBUG("AD936X") << "baseband bandwidth too large for current sample "
"rate. Setting bandwidth to: "
<< _baseband_bw;
bbbw = _baseband_bw / 2.0;
}
/* Baseband BW must be between 20e6 and 0.54e6.
* Max filter BW is 100 MHz. 100 / 5 = 20
* Min filter BW is 2.7 MHz. 2.7 / 5 = 0.54 */
if (bbbw > 20e6) {
bbbw = 20e6;
} else if (bbbw < 0.54e6) {
bbbw = 0.54e6;
}
double bbbw_mhz = bbbw / 1e6;
/* Start with a resistor value of 100 Ohms. */
int res = 100;
/* Calculate target corner frequency. */
double corner_freq = 5 * bbbw_mhz * 2 * M_PI;
/* Iterate through RC values to determine correct combination. */
int cap = 0;
int i;
for (i = 0; i <= 3; i++) {
cap =
static_cast<int>(std::floor(0.5 + ((1 / ((corner_freq * res) * 1e6)) * 1e12)))
- 12;
if (cap <= 63) {
break;
}
res = res * 2;
}
if (cap > 63) {
cap = 63;
}
uint8_t reg0d0, reg0d1, reg0d2;
/* Translate baseband bandwidths to register settings. */
if ((bbbw_mhz * 2) <= 9) {
reg0d0 = 0x59;
} else if (((bbbw_mhz * 2) > 9) && ((bbbw_mhz * 2) <= 24)) {
reg0d0 = 0x56;
} else if ((bbbw_mhz * 2) > 24) {
reg0d0 = 0x57;
} else {
reg0d0 = 0x00;
throw uhd::runtime_error(
"[ad9361_device_t] Cal2ndTxFil: INVALID_CODE_PATH bad bbbw_mhz");
}
/* Translate resistor values to register settings. */
if (res == 100) {
reg0d1 = 0x0c;
} else if (res == 200) {
reg0d1 = 0x04;
} else if (res == 400) {
reg0d1 = 0x03;
} else if (res == 800) {
reg0d1 = 0x01;
} else {
reg0d1 = 0x0c;
}
reg0d2 = cap;
/* Program the above-calculated values. Sweet. */
_io_iface->poke8(0x0d2, reg0d2);
_io_iface->poke8(0x0d1, reg0d1);
_io_iface->poke8(0x0d0, reg0d0);
return bbbw;
}
/* Calibrate the RX TIAs.
*
* Note that the values in the TIA register, after calibration, vary with
* the RX gain settings.
* We do not really program the BW here. Most settings are taken form the BB LPF registers
* UG570 page 33 states that this filter should be calibrated to 2.5 * bbbw */
double ad9361_device_t::_calibrate_rx_TIAs(double req_rfbw)
{
uint8_t reg1eb = _io_iface->peek8(0x1eb) & 0x3F;
uint8_t reg1ec = _io_iface->peek8(0x1ec) & 0x7F;
uint8_t reg1e6 = _io_iface->peek8(0x1e6) & 0x07;
uint8_t reg1db = 0x00;
uint8_t reg1dc = 0x00;
uint8_t reg1dd = 0x00;
uint8_t reg1de = 0x00;
uint8_t reg1df = 0x00;
double bbbw = req_rfbw / 2.0;
if (bbbw > _baseband_bw / 2.0) {
UHD_LOGGER_DEBUG("AD936X") << "baseband bandwidth too large for current sample "
"rate. Setting bandwidth to: "
<< _baseband_bw;
bbbw = _baseband_bw / 2.0;
}
/* Baseband BW must be between 28e6 and 0.4e6.
* Max filter BW is 70 MHz. 70 / 2.5 = 28
* Min filter BW is 1 MHz. 1 / 2.5 = 0.4*/
if (bbbw > 28e6) {
bbbw = 28e6;
} else if (bbbw < 0.40e6) {
bbbw = 0.40e6;
}
double ceil_bbbw_mhz = std::ceil(bbbw / 1e6);
/* Do some crazy resistor and capacitor math. */
int Cbbf = (reg1eb * 160) + (reg1ec * 10) + 140;
int R2346 = 18300 * (reg1e6 & 0x07);
double CTIA_fF = (Cbbf * R2346 * 0.56) / 3500;
/* Translate baseband BW to register settings. */
if (ceil_bbbw_mhz <= 3) {
reg1db = 0xe0;
} else if ((ceil_bbbw_mhz > 3) && (ceil_bbbw_mhz <= 10)) {
reg1db = 0x60;
} else if (ceil_bbbw_mhz > 10) {
reg1db = 0x20;
} else {
throw uhd::runtime_error(
"[ad9361_device_t] CalRxTias: INVALID_CODE_PATH bad bbbw_mhz");
}
if (CTIA_fF > 2920) {
reg1dc = 0x40;
reg1de = 0x40;
uint8_t temp = (uint8_t)std::min<uint8_t>(
127, uint8_t(std::floor(0.5 + ((CTIA_fF - 400.0) / 320.0))));
reg1dd = temp;
reg1df = temp;
} else {
uint8_t temp = uint8_t(std::floor(0.5 + ((CTIA_fF - 400.0) / 40.0)) + 0x40);
reg1dc = temp;
reg1de = temp;
reg1dd = 0;
reg1df = 0;
}
/* w00t. Settings calculated. Program them and roll out. */
_io_iface->poke8(0x1db, reg1db);
_io_iface->poke8(0x1dd, reg1dd);
_io_iface->poke8(0x1df, reg1df);
_io_iface->poke8(0x1dc, reg1dc);
_io_iface->poke8(0x1de, reg1de);
return bbbw;
}
/* Setup the AD9361 ADC.
*
* There are 40 registers that control the ADC's operation, most of the
* values of which must be derived mathematically, dependent on the current
* setting of the BBPLL. Note that the order of calculation is critical, as
* some of the 40 registers depend on the values in others. */
void ad9361_device_t::_setup_adc()
{
double bbbw_mhz =
(((_bbpll_freq / 1e6) / _rx_bbf_tunediv) * M_LN2) / (1.4 * 2 * M_PI);
/* For calibration, baseband BW is half the complex BW, and must be
* between 28e6 and 0.2e6. */
if (bbbw_mhz > 28) {
bbbw_mhz = 28;
} else if (bbbw_mhz < 0.20) {
bbbw_mhz = 0.20;
}
uint8_t rxbbf_c3_msb = _io_iface->peek8(0x1eb) & 0x3F;
uint8_t rxbbf_c3_lsb = _io_iface->peek8(0x1ec) & 0x7F;
uint8_t rxbbf_r2346 = _io_iface->peek8(0x1e6) & 0x07;
double fsadc = _adcclock_freq / 1e6;
/* Sort out the RC time constant for our baseband bandwidth... */
double rc_timeconst = 0.0;
if (bbbw_mhz < 18) {
rc_timeconst =
(1
/ ((1.4 * 2 * M_PI) * (18300 * rxbbf_r2346)
* ((160e-15 * rxbbf_c3_msb) + (10e-15 * rxbbf_c3_lsb) + 140e-15)
* (bbbw_mhz * 1e6)));
} else {
rc_timeconst =
(1
/ ((1.4 * 2 * M_PI) * (18300 * rxbbf_r2346)
* ((160e-15 * rxbbf_c3_msb) + (10e-15 * rxbbf_c3_lsb) + 140e-15)
* (bbbw_mhz * 1e6) * (1 + (0.01 * (bbbw_mhz - 18)))));
}
double scale_res = sqrt(1 / rc_timeconst);
double scale_cap = sqrt(1 / rc_timeconst);
double scale_snr = (_adcclock_freq < 80e6) ? 1.0 : 1.584893192;
double maxsnr = 640 / 160;
/* Calculate the values for all 40 settings registers.
*
* DO NOT TOUCH THIS UNLESS YOU KNOW EXACTLY WHAT YOU ARE DOING. kthx.*/
uint8_t data[40];
data[0] = 0;
data[1] = 0;
data[2] = 0;
data[3] = 0x24;
data[4] = 0x24;
data[5] = 0;
data[6] = 0;
data[7] = std::min<uint8_t>(124,
uint8_t(std::floor(-0.5
+ (80.0 * scale_snr * scale_res
* std::min<double>(1.0, sqrt(maxsnr * fsadc / 640.0))))));
double data007 = data[7];
data[8] = std::min<uint8_t>(255,
uint8_t(std::floor(0.5
+ ((20.0 * (640.0 / fsadc) * ((data007 / 80.0))
/ (scale_res * scale_cap))))));
data[10] = std::min<uint8_t>(127,
uint8_t(std::floor(
-0.5
+ (77.0 * scale_res * std::min<double>(1.0, sqrt(maxsnr * fsadc / 640.0))))));
double data010 = data[10];
data[9] = std::min<uint8_t>(127, uint8_t(std::floor(0.8 * data010)));
data[11] = std::min<uint8_t>(255,
uint8_t(std::floor(
0.5
+ (20.0 * (640.0 / fsadc) * ((data010 / 77.0) / (scale_res * scale_cap))))));
data[12] = std::min<uint8_t>(127,
uint8_t(std::floor(
-0.5
+ (80.0 * scale_res * std::min<double>(1.0, sqrt(maxsnr * fsadc / 640.0))))));
double data012 = data[12];
data[13] = std::min<uint8_t>(255,
uint8_t(std::floor(
-1.5
+ (20.0 * (640.0 / fsadc) * ((data012 / 80.0) / (scale_res * scale_cap))))));
data[14] = 21 * uint8_t(std::floor(0.1 * 640.0 / fsadc));
data[15] = std::min<uint8_t>(127, uint8_t(1.025 * data007));
double data015 = data[15];
data[16] = std::min<uint8_t>(127,
uint8_t(std::floor(
(data015
* (0.98 + (0.02 * std::max<double>(1.0, (640.0 / fsadc) / maxsnr)))))));
data[17] = data[15];
data[18] = std::min<uint8_t>(127, uint8_t(0.975 * (data010)));
double data018 = data[18];
data[19] = std::min<uint8_t>(127,
uint8_t(std::floor(
(data018
* (0.98 + (0.02 * std::max<double>(1.0, (640.0 / fsadc) / maxsnr)))))));
data[20] = data[18];
data[21] = std::min<uint8_t>(127, uint8_t(0.975 * data012));
double data021 = data[21];
data[22] = std::min<uint8_t>(127,
uint8_t(std::floor(
(data021
* (0.98 + (0.02 * std::max<double>(1.0, (640.0 / fsadc) / maxsnr)))))));
data[23] = data[21];
data[24] = 0x2e;
data[25] =
uint8_t(std::floor(128.0 + std::min<double>(63.0, 63.0 * (fsadc / 640.0))));
data[26] = uint8_t(std::floor(std::min<double>(
63.0, 63.0 * (fsadc / 640.0) * (0.92 + (0.08 * (640.0 / fsadc))))));
data[27] = uint8_t(std::floor(std::min<double>(63.0, 32.0 * sqrt(fsadc / 640.0))));
data[28] =
uint8_t(std::floor(128.0 + std::min<double>(63.0, 63.0 * (fsadc / 640.0))));
data[29] = uint8_t(std::floor(std::min<double>(
63.0, 63.0 * (fsadc / 640.0) * (0.92 + (0.08 * (640.0 / fsadc))))));
data[30] = uint8_t(std::floor(std::min<double>(63.0, 32.0 * sqrt(fsadc / 640.0))));
data[31] =
uint8_t(std::floor(128.0 + std::min<double>(63.0, 63.0 * (fsadc / 640.0))));
data[32] = uint8_t(std::floor(std::min<double>(
63.0, 63.0 * (fsadc / 640.0) * (0.92 + (0.08 * (640.0 / fsadc))))));
data[33] = uint8_t(std::floor(std::min<double>(63.0, 63.0 * sqrt(fsadc / 640.0))));
data[34] = std::min<uint8_t>(127, uint8_t(std::floor(64.0 * sqrt(fsadc / 640.0))));
data[35] = 0x40;
data[36] = 0x40;
data[37] = 0x2c;
data[38] = 0x00;
data[39] = 0x00;
/* Program the registers! */
for (size_t i = 0; i < 40; i++) {
_io_iface->poke8(0x200 + i, data[i]);
}
}
/* Calibrate the baseband DC offset.
* Disables tracking
*/
void ad9361_device_t::_calibrate_baseband_dc_offset()
{
_io_iface->poke8(0x18b, 0x83); // Reset RF DC tracking flag
_io_iface->poke8(0x193, 0x3f); // Calibration settings
_io_iface->poke8(0x190, 0x0f); // Set tracking coefficient
// write_ad9361_reg(device, 0x190, /*0x0f*//*0xDF*/0x80*1 | 0x40*1 | (16+8/*+4*/)); //
// Set tracking coefficient: don't *4 counter, do decim /4, increased gain shift
_io_iface->poke8(0x194, 0x01); // More calibration settings
/* Start that calibration, baby. */
size_t count = 0;
_io_iface->poke8(0x016, 0x01);
while (_io_iface->peek8(0x016) & 0x01) {
if (count > 100) {
throw uhd::runtime_error(
"[ad9361_device_t] Baseband DC Offset Calibration Failure");
break;
}
count++;
std::this_thread::sleep_for(std::chrono::milliseconds(5));
}
}
/* Calibrate the RF DC offset.
* Disables tracking
*/
void ad9361_device_t::_calibrate_rf_dc_offset()
{
/* Some settings are frequency-dependent. */
if (_rx_freq < 4e9) {
_io_iface->poke8(0x186, 0x32); // RF DC Offset count
_io_iface->poke8(0x187, 0x24);
_io_iface->poke8(0x188, 0x05);
} else {
_io_iface->poke8(0x186, 0x28); // RF DC Offset count
_io_iface->poke8(0x187, 0x34);
_io_iface->poke8(0x188, 0x06);
}
_io_iface->poke8(0x185, 0x20); // RF DC Offset wait count
_io_iface->poke8(0x18b, 0x83); // Disable tracking
_io_iface->poke8(0x189, 0x30);
/* Run the calibration! */
size_t count = 0;
_io_iface->poke8(0x016, 0x02);
while (_io_iface->peek8(0x016) & 0x02) {
if (count > 200) {
throw uhd::runtime_error(
"[ad9361_device_t] RF DC Offset Calibration Failure");
break;
}
count++;
std::this_thread::sleep_for(std::chrono::milliseconds(50));
}
_io_iface->poke8(0x18b, 0x8d); // Enable RF DC tracking
}
void ad9361_device_t::_configure_bb_dc_tracking()
{
if (_use_dc_offset_tracking)
_io_iface->poke8(0x18b, 0xad); // Enable BB tracking
else
_io_iface->poke8(0x18b, 0x8d); // Disable BB tracking
}
void ad9361_device_t::_configure_rx_iq_tracking()
{
if (_use_iq_balance_tracking)
_io_iface->poke8(0x169, 0xcf); // Enable Rx IQ tracking
else
_io_iface->poke8(0x169, 0xc0); // Disable Rx IQ tracking
}
/* Single shot Rx quadrature calibration
*
* Procedure documented in "AD9361 Calibration Guide". Prior to calibration,
* state should be set to ALERT, FDD, and Dual Synth Mode. Rx quadrature
* tracking will be disabled, so run before or instead of enabling Rx
* quadrature tracking.
*/
void ad9361_device_t::_calibrate_rx_quadrature()
{
/* Configure RX Quadrature calibration settings. */
_io_iface->poke8(0x168, 0x03); // Set tone level for cal
_io_iface->poke8(0x16e, 0x25); // RX Gain index to use for cal
_io_iface->poke8(0x16a, 0x75); // Set Kexp phase
_io_iface->poke8(0x16b, 0x95); // Set Kexp amplitude
_io_iface->poke8(0x057, 0x33); // Power down Tx mixer
_io_iface->poke8(0x169, 0xc0); // Disable tracking and free run mode
/* Place Tx LO within passband of Rx spectrum */
double current_tx_freq = _tx_freq;
_tune_helper(TX, _rx_freq + _rx_bb_lp_bw / 2.0);
size_t count = 0;
_io_iface->poke8(0x016, 0x20);
while (_io_iface->peek8(0x016) & 0x20) {
if (count > 1000) {
throw uhd::runtime_error(
"[ad9361_device_t] Rx Quadrature Calibration Failure");
break;
}
count++;
std::this_thread::sleep_for(std::chrono::milliseconds(5));
}
_io_iface->poke8(0x057, 0x30); // Re-enable Tx mixers
_tune_helper(TX, current_tx_freq);
}
/* TX quadrature calibration routine.
*
* The TX quadrature needs to be done twice, once for each TX chain, with
* only one register change in between. Thus, this function enacts the
* calibrations, and it is called from calibrate_tx_quadrature. */
void ad9361_device_t::_tx_quadrature_cal_routine()
{
/* This is a weird process, but here is how it works:
* 1) Read the calibrated NCO frequency bits out of 0A3.
* 2) Write the two bits to the RX NCO freq part of 0A0.
* 3) Re-read 0A3 to get bits [5:0] because maybe they changed?
* 4) Update only the TX NCO freq bits in 0A3.
* 5) Profit (I hope). */
uint8_t reg0a3 = _io_iface->peek8(0x0a3);
uint8_t nco_freq = (reg0a3 & 0xC0);
_io_iface->poke8(0x0a0, 0x15 | (nco_freq >> 1));
reg0a3 = _io_iface->peek8(0x0a3);
_io_iface->poke8(0x0a3, (reg0a3 & 0x3F) | nco_freq);
/* It is possible to reach a configuration that won't operate correctly,
* where the two test tones used for quadrature calibration are outside
* of the RX BBF, and therefore don't make it to the ADC. We will check
* for that scenario here. */
double max_cal_freq =
(((_baseband_bw * _tfir_factor) * ((nco_freq >> 6) + 1)) / 32) * 2;
double bbbw = _baseband_bw / 2.0; // bbbw represents the one-sided BW
if (bbbw > 28e6) {
bbbw = 28e6;
} else if (bbbw < 0.20e6) {
bbbw = 0.20e6;
}
if (max_cal_freq > bbbw)
throw uhd::runtime_error("[ad9361_device_t] max_cal_freq > bbbw");
_io_iface->poke8(0x0a1, 0x7B); // Set tracking coefficient
_io_iface->poke8(0x0a9, 0xff); // Cal count
_io_iface->poke8(0x0a2, 0x7f); // Cal Kexp
_io_iface->poke8(0x0a5, 0x01); // Cal magnitude threshold VVVV
_io_iface->poke8(0x0a6, 0x01);
/* The gain table index used for calibration must be adjusted for the
* mid-table to get a TIA index = 1 and LPF index = 0. */
if (_rx_freq < 1300e6) {
_io_iface->poke8(0x0aa, 0x22); // Cal gain table index
} else {
_io_iface->poke8(0x0aa, 0x25); // Cal gain table index
}
_io_iface->poke8(0x0a4, 0xf0); // Cal setting conut
_io_iface->poke8(0x0ae, 0x00); // Cal LPF gain index (split mode)
/* Now, calibrate the TX quadrature! */
size_t count = 0;
_io_iface->poke8(0x016, 0x10);
while (_io_iface->peek8(0x016) & 0x10) {
if (count > 100) {
throw uhd::runtime_error(
"[ad9361_device_t] TX Quadrature Calibration Failure");
break;
}
count++;
std::this_thread::sleep_for(std::chrono::milliseconds(10));
}
}
/* Run the TX quadrature calibration.
*/
void ad9361_device_t::_calibrate_tx_quadrature()
{
/* Make sure we are, in fact, in the ALERT state. If not, something is
* terribly wrong in the driver execution flow. */
if ((_io_iface->peek8(0x017) & 0x0F) != 5) {
throw uhd::runtime_error(
"[ad9361_device_t] TX Quad Cal started, but not in ALERT");
}
/* Turn off free-running and continuous calibrations. Note that this
* will get turned back on at the end of the RX calibration routine. */
_io_iface->poke8(0x169, 0xc0);
/* This calibration must be done in a certain order, and for both TX_A
* and TX_B, separately. Store the original setting so that we can
* restore it later. */
uint8_t orig_reg_inputsel = _regs.inputsel;
/***********************************************************************
* TX1/2-A Calibration
**********************************************************************/
_regs.inputsel = _regs.inputsel & 0xBF;
_io_iface->poke8(0x004, _regs.inputsel);
_tx_quadrature_cal_routine();
/***********************************************************************
* TX1/2-B Calibration
**********************************************************************/
_regs.inputsel = _regs.inputsel | 0x40;
_io_iface->poke8(0x004, _regs.inputsel);
_tx_quadrature_cal_routine();
/***********************************************************************
* fin
**********************************************************************/
_regs.inputsel = orig_reg_inputsel;
_io_iface->poke8(0x004, orig_reg_inputsel);
}
/***********************************************************************
* Other Misc Setup Functions
***********************************************************************/
/* Program the mixer gain table.
*
* Note that this table is fixed for all frequency settings. */
void ad9361_device_t::_program_mixer_gm_subtable()
{
// clang-format off
uint8_t gain[] = { 0x78, 0x74, 0x70, 0x6C, 0x68, 0x64, 0x60, 0x5C, 0x58,
0x54, 0x50, 0x4C, 0x48, 0x30, 0x18, 0x00 };
uint8_t gm[] = { 0x00, 0x0D, 0x15, 0x1B, 0x21, 0x25, 0x29, 0x2C, 0x2F, 0x31,
0x33, 0x34, 0x35, 0x3A, 0x3D, 0x3E };
// clang-format on
/* Start the clock. */
_io_iface->poke8(0x13f, 0x02);
/* Program the GM Sub-table. */
int i;
for (i = 15; i >= 0; i--) {
_io_iface->poke8(0x138, i);
_io_iface->poke8(0x139, gain[(15 - i)]);
_io_iface->poke8(0x13A, 0x00);
_io_iface->poke8(0x13B, gm[(15 - i)]);
_io_iface->poke8(0x13F, 0x06);
_io_iface->poke8(0x13C, 0x00);
_io_iface->poke8(0x13C, 0x00);
}
/* Clear write bit and stop clock. */
_io_iface->poke8(0x13f, 0x02);
_io_iface->poke8(0x13C, 0x00);
_io_iface->poke8(0x13C, 0x00);
_io_iface->poke8(0x13f, 0x00);
}
/* Program the gain table.
*
* There are three different gain tables for different frequency ranges! */
void ad9361_device_t::_program_gain_table()
{
/* Figure out which gain table we should be using for our current
* frequency band. */
uint8_t(*gain_table)[3] = NULL;
uint8_t new_gain_table;
if (_rx_freq < 1300e6) {
gain_table = gain_table_sub_1300mhz;
new_gain_table = 1;
} else if (_rx_freq < 4e9) {
gain_table = gain_table_1300mhz_to_4000mhz;
new_gain_table = 2;
} else if (_rx_freq <= 6e9) {
gain_table = gain_table_4000mhz_to_6000mhz;
new_gain_table = 3;
} else {
new_gain_table = 1;
throw uhd::runtime_error("[ad9361_device_t] Wrong _rx_freq value");
}
/* Only re-program the gain table if there has been a band change. */
if (_curr_gain_table == new_gain_table) {
return;
} else {
_curr_gain_table = new_gain_table;
}
/* Okay, we have to program a new gain table. Sucks, brah. Start the
* gain table clock. */
_io_iface->poke8(0x137, 0x1A);
/* IT'S PROGRAMMING TIME. */
uint8_t index = 0;
for (; index < 77; index++) {
_io_iface->poke8(0x130, index);
_io_iface->poke8(0x131, gain_table[index][0]);
_io_iface->poke8(0x132, gain_table[index][1]);
_io_iface->poke8(0x133, gain_table[index][2]);
_io_iface->poke8(0x137, 0x1E);
_io_iface->poke8(0x134, 0x00);
_io_iface->poke8(0x134, 0x00);
}
/* Everything above the 77th index is zero. */
for (; index < 91; index++) {
_io_iface->poke8(0x130, index);
_io_iface->poke8(0x131, 0x00);
_io_iface->poke8(0x132, 0x00);
_io_iface->poke8(0x133, 0x00);
_io_iface->poke8(0x137, 0x1E);
_io_iface->poke8(0x134, 0x00);
_io_iface->poke8(0x134, 0x00);
}
/* Clear the write bit and stop the gain clock. */
_io_iface->poke8(0x137, 0x1A);
_io_iface->poke8(0x134, 0x00);
_io_iface->poke8(0x134, 0x00);
_io_iface->poke8(0x137, 0x00);
}
/* Setup gain control registers.
*
* This really only needs to be done once, at initialization.
* If AGC is used the mode select bits (Reg 0x0FA) must be written manually */
void ad9361_device_t::_setup_gain_control(bool agc)
{
/* The AGC mode configuration should be good for all cases.
* However, non AGC configuration still used for backward compatibility. */
if (agc) {
/*mode select bits must be written before hand!*/
_io_iface->poke8(0x0FB, 0x08); // Table, Digital Gain, Man Gain Ctrl
_io_iface->poke8(0x0FC, 0x23); // Incr Step Size, ADC Overrange Size
_io_iface->poke8(0x0FD, 0x4C); // Max Full/LMT Gain Table Index
_io_iface->poke8(0x0FE, 0x44); // Decr Step Size, Peak Overload Time
_io_iface->poke8(0x100, 0x6F); // Max Digital Gain
_io_iface->poke8(0x101, 0x0A); // Max Digital Gain
_io_iface->poke8(0x103, 0x08); // Max Digital Gain
_io_iface->poke8(0x104, 0x2F); // ADC Small Overload Threshold
_io_iface->poke8(0x105, 0x3A); // ADC Large Overload Threshold
_io_iface->poke8(0x106, 0x22); // Max Digital Gain
_io_iface->poke8(0x107, 0x2B); // Large LMT Overload Threshold
_io_iface->poke8(0x108, 0x31);
_io_iface->poke8(0x111, 0x0A);
_io_iface->poke8(0x11A, 0x1C);
_io_iface->poke8(0x120, 0x0C);
_io_iface->poke8(0x121, 0x44);
_io_iface->poke8(0x122, 0x44);
_io_iface->poke8(0x123, 0x11);
_io_iface->poke8(0x124, 0xF5);
_io_iface->poke8(0x125, 0x3B);
_io_iface->poke8(0x128, 0x03);
_io_iface->poke8(0x129, 0x56);
_io_iface->poke8(0x12A, 0x22);
} else {
_io_iface->poke8(0x0FA, 0xE0); // Gain Control Mode Select
_io_iface->poke8(0x0FB, 0x08); // Table, Digital Gain, Man Gain Ctrl
_io_iface->poke8(0x0FC, 0x23); // Incr Step Size, ADC Overrange Size
_io_iface->poke8(0x0FD, 0x4C); // Max Full/LMT Gain Table Index
_io_iface->poke8(0x0FE, 0x44); // Decr Step Size, Peak Overload Time
_io_iface->poke8(0x100, 0x6F); // Max Digital Gain
_io_iface->poke8(0x104, 0x2F); // ADC Small Overload Threshold
_io_iface->poke8(0x105, 0x3A); // ADC Large Overload Threshold
_io_iface->poke8(0x107, 0x31); // Large LMT Overload Threshold
_io_iface->poke8(0x108, 0x39); // Small LMT Overload Threshold
_io_iface->poke8(0x109, 0x23); // Rx1 Full/LMT Gain Index
_io_iface->poke8(0x10A, 0x58); // Rx1 LPF Gain Index
_io_iface->poke8(0x10B, 0x00); // Rx1 Digital Gain Index
_io_iface->poke8(0x10C, 0x23); // Rx2 Full/LMT Gain Index
_io_iface->poke8(0x10D, 0x18); // Rx2 LPF Gain Index
_io_iface->poke8(0x10E, 0x00); // Rx2 Digital Gain Index
_io_iface->poke8(0x114, 0x30); // Low Power Threshold
_io_iface->poke8(0x11A, 0x27); // Initial LMT Gain Limit
_io_iface->poke8(0x081, 0x00); // Tx Symbol Gain Control
}
}
/* Setup the RX or TX synthesizers.
*
* This setup depends on a fixed look-up table, which is stored in an
* included header file. The table is indexed based on the passed VCO rate.
*/
void ad9361_device_t::_setup_synth(direction_t direction, double vcorate)
{
/* The vcorates in the vco_index array represent lower boundaries for
* rates. Once we find a match, we use that index to look-up the rest of
* the register values in the LUT. */
int vcoindex = 0;
for (size_t i = 0; i < 53; i++) {
vcoindex = i;
if (vcorate > vco_index[i]) {
break;
}
}
if (vcoindex > 53)
throw uhd::runtime_error("[ad9361_device_t] vcoindex > 53");
/* Parse the values out of the LUT based on our calculated index... */
uint8_t vco_output_level = synth_cal_lut[vcoindex][0];
uint8_t vco_varactor = synth_cal_lut[vcoindex][1];
uint8_t vco_bias_ref = synth_cal_lut[vcoindex][2];
uint8_t vco_bias_tcf = synth_cal_lut[vcoindex][3];
uint8_t vco_cal_offset = synth_cal_lut[vcoindex][4];
uint8_t vco_varactor_ref = synth_cal_lut[vcoindex][5];
uint8_t charge_pump_curr = synth_cal_lut[vcoindex][6];
uint8_t loop_filter_c2 = synth_cal_lut[vcoindex][7];
uint8_t loop_filter_c1 = synth_cal_lut[vcoindex][8];
uint8_t loop_filter_r1 = synth_cal_lut[vcoindex][9];
uint8_t loop_filter_c3 = synth_cal_lut[vcoindex][10];
uint8_t loop_filter_r3 = synth_cal_lut[vcoindex][11];
/* ... annnd program! */
if (direction == RX) {
_io_iface->poke8(0x23a, 0x40 | vco_output_level);
_io_iface->poke8(0x239, 0xC0 | vco_varactor);
_io_iface->poke8(0x242, vco_bias_ref | (vco_bias_tcf << 3));
_io_iface->poke8(0x238, (vco_cal_offset << 3));
_io_iface->poke8(0x245, 0x00);
_io_iface->poke8(0x251, vco_varactor_ref);
_io_iface->poke8(0x250, 0x70);
_io_iface->poke8(0x23b, 0x80 | charge_pump_curr);
_io_iface->poke8(0x23e, loop_filter_c1 | (loop_filter_c2 << 4));
_io_iface->poke8(0x23f, loop_filter_c3 | (loop_filter_r1 << 4));
_io_iface->poke8(0x240, loop_filter_r3);
} else if (direction == TX) {
_io_iface->poke8(0x27a, 0x40 | vco_output_level);
_io_iface->poke8(0x279, 0xC0 | vco_varactor);
_io_iface->poke8(0x282, vco_bias_ref | (vco_bias_tcf << 3));
_io_iface->poke8(0x278, (vco_cal_offset << 3));
_io_iface->poke8(0x285, 0x00);
_io_iface->poke8(0x291, vco_varactor_ref);
_io_iface->poke8(0x290, 0x70);
_io_iface->poke8(0x27b, 0x80 | charge_pump_curr);
_io_iface->poke8(0x27e, loop_filter_c1 | (loop_filter_c2 << 4));
_io_iface->poke8(0x27f, loop_filter_c3 | (loop_filter_r1 << 4));
_io_iface->poke8(0x280, loop_filter_r3);
} else {
throw uhd::runtime_error("[ad9361_device_t] [_setup_synth] INVALID_CODE_PATH");
}
}
/* Tune the baseband VCO.
*
* This clock signal is what gets fed to the ADCs and DACs. This function is
* not exported outside of this file, and is invoked based on the rate
* fed to the public set_clock_rate function. */
double ad9361_device_t::_tune_bbvco(const double rate)
{
UHD_LOG_TRACE("AD936X", "[ad9361_device_t::_tune_bbvco] rate=" << rate);
/* Let's not re-tune to the same frequency over and over... */
if (freq_is_nearly_equal(rate, _req_coreclk)) {
return _adcclock_freq;
}
_req_coreclk = rate;
const double fref = 40e6;
const int modulus = 2088960;
const double vcomax = 1430e6;
const double vcomin = 672e6;
double vcorate;
int vcodiv;
/* Iterate over VCO dividers until appropriate divider is found. */
int i = 1;
for (; i <= 6; i++) {
vcodiv = 1 << i;
vcorate = rate * vcodiv;
if (vcorate >= vcomin && vcorate <= vcomax)
break;
}
if (i == 7)
throw uhd::runtime_error("[ad9361_device_t] _tune_bbvco: wrong vcorate");
UHD_LOG_TRACE("AD936X",
boost::format("[ad9361_device_t::_tune_bbvco] vcodiv=%d vcorate=%.10f") % vcodiv
% vcorate);
/* Fo = Fref * (Nint + Nfrac / mod) */
int nint = static_cast<int>(vcorate / fref);
UHD_LOG_TRACE("AD936X", "[ad9361_device_t::_tune_bbvco] (nint)=" << (vcorate / fref));
int nfrac = static_cast<int>(
std::lround(((vcorate / fref) - (double)nint) * (double)modulus));
UHD_LOG_TRACE("AD936X",
"[ad9361_device_t::_tune_bbvco] (nfrac)=" << ((vcorate / fref) - (double)nint)
* (double)modulus);
UHD_LOG_TRACE("AD936X",
boost::format("[ad9361_device_t::_tune_bbvco] nint=%d nfrac=%d") % nint % nfrac);
const double actual_vcorate =
fref * ((double)nint + ((double)nfrac / (double)modulus));
UHD_LOG_TRACE("AD936X",
boost::format("[ad9361_device_t::_tune_bbvco] actual vcorate=%.10f")
% actual_vcorate);
/* Scale CP current according to VCO rate */
const double icp_baseline = 150e-6;
const double freq_baseline = 1280e6;
double icp = icp_baseline * (actual_vcorate / freq_baseline);
int icp_reg = static_cast<int>(icp / 25e-6) - 1;
_io_iface->poke8(0x045, 0x00); // REFCLK / 1 to BBPLL
_io_iface->poke8(0x046, icp_reg & 0x3F); // CP current
_io_iface->poke8(0x048, 0xe8); // BBPLL loop filters
_io_iface->poke8(0x049, 0x5b); // BBPLL loop filters
_io_iface->poke8(0x04a, 0x35); // BBPLL loop filters
_io_iface->poke8(0x04b, 0xe0);
_io_iface->poke8(0x04e, 0x10); // Max accuracy
_io_iface->poke8(0x043, nfrac & 0xFF); // Nfrac[7:0]
_io_iface->poke8(0x042, (nfrac >> 8) & 0xFF); // Nfrac[15:8]
_io_iface->poke8(0x041, (nfrac >> 16) & 0xFF); // Nfrac[23:16]
_io_iface->poke8(0x044, nint); // Nint
_calibrate_lock_bbpll();
_regs.bbpll = (_regs.bbpll & 0xF8) | i;
_bbpll_freq = actual_vcorate;
_adcclock_freq = (actual_vcorate / vcodiv);
return _adcclock_freq;
}
/* This function re-programs all of the gains in the system.
*
* Because the gain values match to different gain indices based on the
* current operating band, this function can be called to update all gain
* settings to the appropriate index after a re-tune. */
void ad9361_device_t::_reprogram_gains()
{
set_gain(RX, CHAIN_1, _rx1_gain);
set_gain(RX, CHAIN_2, _rx2_gain);
set_gain(TX, CHAIN_1, _tx1_gain);
set_gain(TX, CHAIN_2, _tx2_gain);
}
/* This is the internal tune function, not available for a host call.
*
* Calculate the VCO settings for the requested frquency, and then either
* tune the RX or TX VCO. */
double ad9361_device_t::_tune_helper(direction_t direction, const double value)
{
/* The RFPLL runs from 6 GHz - 12 GHz */
const double fref = 80e6;
const int modulus = 8388593;
const double vcomax = 12e9;
const double vcomin = 6e9;
double vcorate;
int vcodiv;
/* Iterate over VCO dividers until appropriate divider is found. */
int i;
for (i = 0; i <= 6; i++) {
vcodiv = 2 << i;
vcorate = value * vcodiv;
if (vcorate >= vcomin && vcorate <= vcomax)
break;
}
if (i == 7)
throw uhd::runtime_error("[ad9361_device_t] RFVCO can't find valid VCO rate!");
int nint = static_cast<int>(vcorate / fref);
int nfrac = static_cast<int>(((vcorate / fref) - nint) * modulus);
double actual_vcorate = fref * (nint + (double)(nfrac) / modulus);
double actual_lo = actual_vcorate / vcodiv;
if (direction == RX) {
_req_rx_freq = value;
/* Set band-specific settings. */
if (value < _client_params->get_band_edge(AD9361_RX_BAND0)) {
_regs.inputsel = (_regs.inputsel & 0xC0) | 0x30; // Port C, balanced
} else if ((value >= _client_params->get_band_edge(AD9361_RX_BAND0))
&& (value < _client_params->get_band_edge(AD9361_RX_BAND1))) {
_regs.inputsel = (_regs.inputsel & 0xC0) | 0x0C; // Port B, balanced
} else if ((value >= _client_params->get_band_edge(AD9361_RX_BAND1))
&& (value <= 6e9)) {
_regs.inputsel = (_regs.inputsel & 0xC0) | 0x03; // Port A, balanced
} else {
throw uhd::runtime_error(
"[ad9361_device_t] [_tune_helper] INVALID_CODE_PATH");
}
_io_iface->poke8(0x004, _regs.inputsel);
/* Store vcodiv setting. */
_regs.vcodivs = (_regs.vcodivs & 0xF0) | (i & 0x0F);
/* Setup the synthesizer. */
_setup_synth(RX, actual_vcorate);
/* Tune!!!! */
_io_iface->poke8(0x233, nfrac & 0xFF);
_io_iface->poke8(0x234, (nfrac >> 8) & 0xFF);
_io_iface->poke8(0x235, (nfrac >> 16) & 0xFF);
_io_iface->poke8(0x232, (nint >> 8) & 0xFF);
_io_iface->poke8(0x231, nint & 0xFF);
_io_iface->poke8(0x005, _regs.vcodivs);
/* Lock the PLL! */
std::this_thread::sleep_for(std::chrono::milliseconds(2));
if ((_io_iface->peek8(0x247) & 0x02) == 0) {
throw uhd::runtime_error("[ad9361_device_t] RX PLL NOT LOCKED");
}
_rx_freq = actual_lo;
return actual_lo;
} else {
_req_tx_freq = value;
/* Set band-specific settings. */
if (value < _client_params->get_band_edge(AD9361_TX_BAND0)) {
_regs.inputsel = _regs.inputsel | 0x40;
} else if ((value >= _client_params->get_band_edge(AD9361_TX_BAND0))
&& (value <= 6e9)) {
_regs.inputsel = _regs.inputsel & 0xBF;
} else {
throw uhd::runtime_error(
"[ad9361_device_t] [_tune_helper] INVALID_CODE_PATH");
}
_io_iface->poke8(0x004, _regs.inputsel);
/* Store vcodiv setting. */
_regs.vcodivs = (_regs.vcodivs & 0x0F) | ((i & 0x0F) << 4);
/* Setup the synthesizer. */
_setup_synth(TX, actual_vcorate);
/* Tune it, homey. */
_io_iface->poke8(0x273, nfrac & 0xFF);
_io_iface->poke8(0x274, (nfrac >> 8) & 0xFF);
_io_iface->poke8(0x275, (nfrac >> 16) & 0xFF);
_io_iface->poke8(0x272, (nint >> 8) & 0xFF);
_io_iface->poke8(0x271, nint & 0xFF);
_io_iface->poke8(0x005, _regs.vcodivs);
/* Lock the PLL! */
std::this_thread::sleep_for(std::chrono::milliseconds(2));
if ((_io_iface->peek8(0x287) & 0x02) == 0) {
throw uhd::runtime_error("[ad9361_device_t] TX PLL NOT LOCKED");
}
_tx_freq = actual_lo;
return actual_lo;
}
}
/* Configure the various clock / sample rates in the RX and TX chains.
*
* Functionally, this function configures AD9361's RX and TX rates. For
* a requested TX & RX rate, it sets the interpolation & decimation filters,
* and tunes the VCO that feeds the ADCs and DACs.
*/
double ad9361_device_t::_setup_rates(const double rate)
{
/* If we make it into this function, then we are tuning to a new rate.
* Store the new rate. */
_req_clock_rate = rate;
UHD_LOG_TRACE("AD936X", "[ad9361_device_t::_setup_rates] rate=" << rate);
/* Set the decimation and interpolation values in the RX and TX chains.
* This also switches filters in / out. Note that all transmitters and
* receivers have to be turned on for the calibration portion of
* bring-up, and then they will be switched out to reflect the actual
* user-requested antenna selections. */
int divfactor = 0;
_tfir_factor = 0;
_rfir_factor = 0;
if (rate < 0.33e6) {
// RX1 + RX2 enabled, 3, 2, 2, 4
_regs.rxfilt = B8(11101111);
// TX1 + TX2 enabled, 3, 2, 2, 4
_regs.txfilt = B8(11101111);
divfactor = 48;
_tfir_factor = 4;
_rfir_factor = 4;
} else if (rate < 0.66e6) {
// RX1 + RX2 enabled, 2, 2, 2, 4
_regs.rxfilt = B8(11011111);
// TX1 + TX2 enabled, 2, 2, 2, 4
_regs.txfilt = B8(11011111);
divfactor = 32;
_tfir_factor = 4;
_rfir_factor = 4;
} else if (rate <= 20e6) {
// RX1 + RX2 enabled, 2, 2, 2, 2
_regs.rxfilt = B8(11011110);
// TX1 + TX2 enabled, 2, 2, 2, 2
_regs.txfilt = B8(11011110);
divfactor = 16;
_tfir_factor = 2;
_rfir_factor = 2;
} else if ((rate > 20e6) && (rate < 23e6)) {
// RX1 + RX2 enabled, 3, 2, 2, 2
_regs.rxfilt = B8(11101110);
// TX1 + TX2 enabled, 3, 1, 2, 2
_regs.txfilt = B8(11100110);
divfactor = 24;
_tfir_factor = 2;
_rfir_factor = 2;
} else if ((rate >= 23e6) && (rate < 41e6)) {
// RX1 + RX2 enabled, 2, 2, 2, 2
_regs.rxfilt = B8(11011110);
// TX1 + TX2 enabled, 1, 2, 2, 2
_regs.txfilt = B8(11001110);
divfactor = 16;
_tfir_factor = 2;
_rfir_factor = 2;
} else if ((rate >= 41e6) && (rate <= 58e6)) {
// RX1 + RX2 enabled, 3, 1, 2, 2
_regs.rxfilt = B8(11100110);
// TX1 + TX2 enabled, 3, 1, 1, 2
_regs.txfilt = B8(11100010);
divfactor = 12;
_tfir_factor = 2;
_rfir_factor = 2;
} else if ((rate > 58e6) && (rate <= 61.44e6)) {
// RX1 + RX2 enabled, 2, 1, 2, 2
_regs.rxfilt = B8(11001110);
// TX1 + TX2 enabled, 2, 1, 1, 2
_regs.txfilt = B8(11010010);
divfactor = 8;
_tfir_factor = 2;
_rfir_factor = 2;
} else {
// should never get in here
throw uhd::runtime_error("[ad9361_device_t] [_setup_rates] INVALID_CODE_PATH");
}
UHD_LOG_TRACE("AD936X", "[ad9361_device_t::_setup_rates] divfactor=" << divfactor);
/* Tune the BBPLL to get the ADC and DAC clocks. */
const double adcclk = _tune_bbvco(rate * divfactor);
double dacclk = adcclk;
// We keep the DAC clock <= 336 MHz, either the ADC clock or 1/2 the ADC
// clock.
// Note that current recommendations from ADI are 640 MHz for the ADC and
// 320 MHz for the DAC. We are thus overclocking the AD9361, since we are
// using these no longer valid recommendations, but they seem to work and we
// consider the risk of changing these higher than the benefit. We will add
// a log statement, though, so we can track back suboptimal RF performance
// to overclocked settings.
if (adcclk > 336e6) {
/* Make the DAC clock = ADC/2 */
_regs.bbpll = _regs.bbpll | 0x08;
dacclk = adcclk / 2.0;
} else {
_regs.bbpll = _regs.bbpll & 0xF7;
}
if (dacclk > 320e6) {
UHD_LOG_DEBUG("AD936X",
"[ad9361_device_t::_setup_rates] DAC rate is overclocked! dacclk="
<< (dacclk / 1e6) << " MHz.");
}
if (dacclk > 640e6) {
UHD_LOG_DEBUG("AD936X",
"[ad9361_device_t::_setup_rates] ADC rate is overclocked! adcclk="
<< (adcclk / 1e6) << " MHz.");
}
/* Set the dividers / interpolators in AD9361. */
_io_iface->poke8(0x002, _regs.txfilt);
_io_iface->poke8(0x003, _regs.rxfilt);
_io_iface->poke8(0x004, _regs.inputsel);
_io_iface->poke8(0x00A, _regs.bbpll);
UHD_LOG_TRACE("AD936X", "[ad9361_device_t::_setup_rates] adcclk=" << adcclk);
_baseband_bw = (adcclk / divfactor);
/*
The Tx & Rx FIR calculate 16 taps per clock cycle. This limits the number of
available taps to the ratio of DAC_CLK/ADC_CLK to the input data rate multiplied
by 16. For example, if the input data rate is 25 MHz and DAC_CLK is 100 MHz, then the
ratio of DAC_CLK to the input data rate is 100/25 or 4. In this scenario, the total
number of taps available is 64.
Also, whilst the Rx FIR filter always has memory available for 128 taps, the Tx FIR
Filter can only support a maximum length of 64 taps in 1x interpolation mode, and 128
taps in 2x & 4x modes.
*/
const size_t max_tx_taps =
std::min<size_t>(std::min<size_t>((16 * (int)((dacclk / rate) + 0.5)), 128),
(_tfir_factor == 1) ? 64 : 128);
const size_t max_rx_taps =
std::min<size_t>((16 * (size_t)((adcclk / rate) + 0.5)), 128);
const size_t num_tx_taps = get_num_taps(max_tx_taps);
const size_t num_rx_taps = get_num_taps(max_rx_taps);
_setup_tx_fir(num_tx_taps, _tfir_factor);
_setup_rx_fir(num_rx_taps, _rfir_factor);
return _baseband_bw;
}
/***********************************************************************
* Publicly exported functions to host calls
**********************************************************************/
void ad9361_device_t::initialize()
{
std::lock_guard<std::recursive_mutex> lock(_mutex);
/* Initialize shadow registers. */
_regs.vcodivs = 0x00;
_regs.inputsel = 0x30;
_regs.rxfilt = 0x00;
_regs.txfilt = 0x00;
_regs.bbpll = 0x02;
_regs.bbftune_config = 0x1e;
_regs.bbftune_mode = 0x1e;
/* Initialize private VRQ fields. */
_rx_freq = DEFAULT_RX_FREQ;
_tx_freq = DEFAULT_TX_FREQ;
_req_rx_freq = 0.0;
_req_tx_freq = 0.0;
_baseband_bw = 0.0;
_req_clock_rate = 0.0;
_req_coreclk = 0.0;
_bbpll_freq = 0.0;
_adcclock_freq = 0.0;
_rx_bbf_tunediv = 0;
_curr_gain_table = 0;
_rx1_gain = 0;
_rx2_gain = 0;
_tx1_gain = 0;
_tx2_gain = 0;
_use_dc_offset_tracking = true;
_use_iq_balance_tracking = true;
_rx1_agc_mode = GAIN_MODE_SLOW_AGC;
_rx2_agc_mode = GAIN_MODE_SLOW_AGC;
_rx1_agc_enable = false;
_rx2_agc_enable = false;
_rx_analog_bw = 0;
_tx_analog_bw = 0;
_rx_tia_lp_bw = 0;
_tx_sec_lp_bw = 0;
_rx_bb_lp_bw = 0;
_tx_bb_lp_bw = 0;
/* Reset the device. */
_io_iface->poke8(0x000, 0x01);
_io_iface->poke8(0x000, 0x00);
std::this_thread::sleep_for(std::chrono::milliseconds(20));
/* Check device ID to make sure iface works */
uint32_t device_id = (_io_iface->peek8(0x037) & 0xf8);
if (device_id != 0x8) {
throw uhd::runtime_error(
str(boost::format("[ad9361_device_t::initialize] Device ID readback failure. "
"Expected: 0x8, Received: 0x%x")
% device_id));
}
/* There is not a WAT big enough for this. */
_io_iface->poke8(0x3df, 0x01);
_io_iface->poke8(0x2a6, 0x0e); // Enable master bias
_io_iface->poke8(0x2a8, 0x0e); // Set bandgap trim
/* Set RFPLL ref clock scale to REFCLK * 2 */
_io_iface->poke8(0x2ab, 0x07);
_io_iface->poke8(0x2ac, 0xff);
/* Enable clocks. */
switch (_client_params->get_clocking_mode()) {
case clocking_mode_t::AD9361_XTAL_N_CLK_PATH: {
_io_iface->poke8(0x009, 0x17);
} break;
case clocking_mode_t::AD9361_XTAL_P_CLK_PATH: {
_io_iface->poke8(0x009, 0x07);
_io_iface->poke8(0x292, 0x08);
_io_iface->poke8(0x293, 0x80);
_io_iface->poke8(0x294, 0x00);
_io_iface->poke8(0x295, 0x14);
} break;
default:
throw uhd::runtime_error("[ad9361_device_t] NOT IMPLEMENTED");
}
std::this_thread::sleep_for(std::chrono::milliseconds(20));
/* Tune the BBPLL, write TX and RX FIRS. */
_setup_rates(50e6);
/* Setup data ports (FDD dual port DDR):
* FDD dual port DDR CMOS no swap.
* Force TX on one port, RX on the other. */
switch (_client_params->get_digital_interface_mode()) {
case AD9361_DDR_FDD_LVCMOS: {
_io_iface->poke8(
0x010, 0xc8); // Swap I&Q on Tx, Swap I&Q on Rx, Toggle frame sync mode
_io_iface->poke8(0x011, 0x00);
_io_iface->poke8(0x012, 0x02);
} break;
case AD9361_DDR_FDD_LVDS: {
_io_iface->poke8(0x010, 0xcc); // Swap I&Q on Tx, Swap I&Q on Rx, Toggle frame
// sync mode, 2R2T timing.
_io_iface->poke8(0x011, 0x00);
_io_iface->poke8(0x012, 0x10);
// LVDS Specific
_io_iface->poke8(0x03C, 0x23);
_io_iface->poke8(0x03D, 0xFF);
_io_iface->poke8(0x03E, 0x0F);
} break;
default:
throw uhd::runtime_error("[ad9361_device_t] NOT IMPLEMENTED");
}
/* Data delay for TX and RX data clocks */
digital_interface_delays_t timing = _client_params->get_digital_interface_timing();
uint8_t rx_delays = ((timing.rx_clk_delay & 0xF) << 4) | (timing.rx_data_delay & 0xF);
uint8_t tx_delays = ((timing.tx_clk_delay & 0xF) << 4) | (timing.tx_data_delay & 0xF);
_io_iface->poke8(0x006, rx_delays);
_io_iface->poke8(0x007, tx_delays);
/* Setup AuxDAC */
_io_iface->poke8(0x018, 0x00); // AuxDAC1 Word[9:2]
_io_iface->poke8(0x019, 0x00); // AuxDAC2 Word[9:2]
_io_iface->poke8(0x01A, 0x00); // AuxDAC1 Config and Word[1:0]
_io_iface->poke8(0x01B, 0x00); // AuxDAC2 Config and Word[1:0]
_io_iface->poke8(0x023, 0xFF); // AuxDAC Manaul/Auto Control
_io_iface->poke8(0x026, 0x00); // AuxDAC Manual Select Bit/GPO Manual Select
_io_iface->poke8(0x030, 0x00); // AuxDAC1 Rx Delay
_io_iface->poke8(0x031, 0x00); // AuxDAC1 Tx Delay
_io_iface->poke8(0x032, 0x00); // AuxDAC2 Rx Delay
_io_iface->poke8(0x033, 0x00); // AuxDAC2 Tx Delay
/* LNA bypass polarity inversion
* According to the register map, we should invert the bypass path to
* match LNA phase. Extensive testing, however, shows otherwise and that
* to align bypass and LNA phases, the bypass inversion switch should be
* turned off.
*/
_io_iface->poke8(0x022, 0x0A);
/* Setup AuxADC */
_io_iface->poke8(0x00B, 0x00); // Temp Sensor Setup (Offset)
_io_iface->poke8(0x00C, 0x00); // Temp Sensor Setup (Temp Window)
_io_iface->poke8(0x00D, 0x00); // Temp Sensor Setup (Manual Measure)
_io_iface->poke8(0x00F, 0x04); // Temp Sensor Setup (Decimation)
_io_iface->poke8(0x01C, 0x10); // AuxADC Setup (Clock Div)
_io_iface->poke8(0x01D, 0x01); // AuxADC Setup (Decimation/Enable)
/* Setup control outputs. */
_io_iface->poke8(0x035, 0x01);
_io_iface->poke8(0x036, 0xFF);
/* Setup GPO */
_io_iface->poke8(0x03a, 0x27); // set delay register
_io_iface->poke8(0x020, 0x00); // GPO Auto Enable Setup in RX and TX
_io_iface->poke8(0x027, 0x03); // GPO Manual and GPO auto value in ALERT
_io_iface->poke8(0x028, 0x00); // GPO_0 RX Delay
_io_iface->poke8(0x029, 0x00); // GPO_1 RX Delay
_io_iface->poke8(0x02A, 0x00); // GPO_2 RX Delay
_io_iface->poke8(0x02B, 0x00); // GPO_3 RX Delay
_io_iface->poke8(0x02C, 0x00); // GPO_0 TX Delay
_io_iface->poke8(0x02D, 0x00); // GPO_1 TX Delay
_io_iface->poke8(0x02E, 0x00); // GPO_2 TX Delay
_io_iface->poke8(0x02F, 0x00); // GPO_3 TX Delay
_io_iface->poke8(0x261, 0x00); // RX LO power
_io_iface->poke8(0x2a1, 0x00); // TX LO power
_io_iface->poke8(0x248, 0x0b); // en RX VCO LDO
_io_iface->poke8(0x288, 0x0b); // en TX VCO LDO
_io_iface->poke8(0x246, 0x02); // pd RX cal Tcf
_io_iface->poke8(0x286, 0x02); // pd TX cal Tcf
_io_iface->poke8(0x249, 0x8e); // rx vco cal length
_io_iface->poke8(0x289, 0x8e); // rx vco cal length
_io_iface->poke8(0x23b, 0x80); // set RX MSB?, FIXME 0x89 magic cp
_io_iface->poke8(0x27b, 0x80); // "" TX //FIXME 0x88 see above
_io_iface->poke8(0x243, 0x0d); // set rx prescaler bias
_io_iface->poke8(0x283, 0x0d); // "" TX
_io_iface->poke8(0x23d, 0x00); // Clear half VCO cal clock setting
_io_iface->poke8(0x27d, 0x00); // Clear half VCO cal clock setting
/* The order of the following process is EXTREMELY important. If the
* below functions are modified at all, device initialization and
* calibration might be broken in the process! */
_io_iface->poke8(0x015, 0x04); // dual synth mode, synth en ctrl en
_io_iface->poke8(0x014, 0x05); // use SPI for TXNRX ctrl, to ALERT, TX on
_io_iface->poke8(0x013, 0x01); // enable ENSM
std::this_thread::sleep_for(std::chrono::milliseconds(1));
_calibrate_synth_charge_pumps();
_tune_helper(RX, _rx_freq);
_tune_helper(TX, _tx_freq);
_program_mixer_gm_subtable();
_program_gain_table();
_setup_gain_control(false);
set_bw_filter(RX, _baseband_bw);
set_bw_filter(TX, _baseband_bw);
_setup_adc();
_calibrate_baseband_dc_offset();
_calibrate_rf_dc_offset();
_calibrate_rx_quadrature();
/*
* Rx BB DC and IQ tracking are both disabled by calibration at this
* point. Only issue commands if tracking needs to be turned on.
*/
if (_use_dc_offset_tracking)
_configure_bb_dc_tracking();
if (_use_iq_balance_tracking)
_configure_rx_iq_tracking();
_last_rx_cal_freq = _rx_freq;
_last_tx_cal_freq = _tx_freq;
// cals done, set PPORT config
switch (_client_params->get_digital_interface_mode()) {
case AD9361_DDR_FDD_LVCMOS: {
_io_iface->poke8(0x012, 0x02);
} break;
case AD9361_DDR_FDD_LVDS: {
_io_iface->poke8(0x012, 0x10);
} break;
default:
throw uhd::runtime_error("[ad9361_device_t] NOT IMPLEMENTED");
}
_io_iface->poke8(0x013, 0x01); // Set ENSM FDD bit
_io_iface->poke8(0x015, 0x04); // dual synth mode, synth en ctrl en
/* Default TX attentuation to 10dB on both TX1 and TX2 */
_io_iface->poke8(0x073, 0x00);
_io_iface->poke8(0x074, 0x00);
_io_iface->poke8(0x075, 0x00);
_io_iface->poke8(0x076, 0x00);
/* Setup RSSI Measurements */
_io_iface->poke8(0x150, 0x0E); // RSSI Measurement Duration 0, 1
_io_iface->poke8(0x151, 0x00); // RSSI Measurement Duration 2, 3
_io_iface->poke8(0x152, 0xFF); // RSSI Weighted Multiplier 0
_io_iface->poke8(0x153, 0x00); // RSSI Weighted Multiplier 1
_io_iface->poke8(0x154, 0x00); // RSSI Weighted Multiplier 2
_io_iface->poke8(0x155, 0x00); // RSSI Weighted Multiplier 3
_io_iface->poke8(0x156, 0x00); // RSSI Delay
_io_iface->poke8(0x157, 0x00); // RSSI Wait
_io_iface->poke8(0x158, 0x0D); // RSSI Mode Select
_io_iface->poke8(0x15C, 0x67); // Power Measurement Duration
/* Turn on the default RX & TX chains. */
set_active_chains(true, false, false, false);
/* Set TXers & RXers on (only works in FDD mode) */
_io_iface->poke8(0x014, 0x21);
/* We won't be using immediate-Tx-attenuation, so we de-assert "Mask Clr
* Atten Update". */
_io_iface->poke8(0x077, 0x00);
}
void ad9361_device_t::set_io_iface(ad9361_io::sptr io_iface)
{
_io_iface = io_iface;
}
/* This function sets the RX / TX rate between AD9361 and the FPGA, and
* thus determines the interpolation / decimation required in the FPGA to
* achieve the user's requested rate.
*
* This is the only clock setting function that is exposed to the outside. */
double ad9361_device_t::set_clock_rate(const double req_rate)
{
std::lock_guard<std::recursive_mutex> lock(_mutex);
if (req_rate > 61.44e6) {
throw uhd::runtime_error(
"[ad9361_device_t] Requested master clock rate outside range");
}
UHD_LOG_TRACE("AD936X", "[ad9361_device_t::set_clock_rate] req_rate=" << req_rate);
/* UHD has a habit of requesting the same rate like four times when it
* starts up. This prevents that, and any bugs in user code that request
* the same rate over and over. */
if (freq_is_nearly_equal(req_rate, _req_clock_rate)) {
// We return _baseband_bw, because that's closest to the
// actual value we're currently running.
return _baseband_bw;
}
/* We must be in the SLEEP / WAIT state to do this. If we aren't already
* there, transition the ENSM to State 0. */
uint8_t current_state = _io_iface->peek8(0x017) & 0x0F;
switch (current_state) {
case 0x05:
/* We are in the ALERT state. */
_io_iface->poke8(0x014, 0x21);
std::this_thread::sleep_for(std::chrono::milliseconds(5));
_io_iface->poke8(0x014, 0x00);
break;
case 0x0A:
/* We are in the FDD state. */
_io_iface->poke8(0x014, 0x00);
break;
default:
throw uhd::runtime_error(
"[ad9361_device_t] [set_clock_rate:1] AD9361 in unknown state");
break;
};
/* Store the current chain / antenna selections so that we can restore
* them at the end of this routine; all chains will be enabled from
* within setup_rates for calibration purposes. */
uint8_t orig_tx_chains = _regs.txfilt & 0xC0;
uint8_t orig_rx_chains = _regs.rxfilt & 0xC0;
/* Call into the clock configuration / settings function. This is where
* all the hard work gets done. */
double UHD_UNUSED(rate) = _setup_rates(req_rate);
UHD_LOG_TRACE("AD936X", "[ad9361_device_t::set_clock_rate] rate=" << rate);
/* Transition to the ALERT state and calibrate everything. */
_io_iface->poke8(0x015, 0x04); // dual synth mode, synth en ctrl en
_io_iface->poke8(0x014, 0x05); // use SPI for TXNRX ctrl, to ALERT, TX on
_io_iface->poke8(0x013, 0x01); // enable ENSM
std::this_thread::sleep_for(std::chrono::milliseconds(1));
_calibrate_synth_charge_pumps();
_tune_helper(RX, _rx_freq);
_tune_helper(TX, _tx_freq);
_program_mixer_gm_subtable();
_program_gain_table();
_setup_gain_control(false);
_reprogram_gains();
set_bw_filter(RX, _baseband_bw);
set_bw_filter(TX, _baseband_bw);
_setup_adc();
_calibrate_baseband_dc_offset();
_calibrate_rf_dc_offset();
_calibrate_rx_quadrature();
/*
* Rx BB DC and IQ tracking are both disabled by calibration at this
* point. Only issue commands if tracking needs to be turned on.
*/
if (_use_dc_offset_tracking)
_configure_bb_dc_tracking();
if (_use_iq_balance_tracking)
_configure_rx_iq_tracking();
_last_rx_cal_freq = _rx_freq;
_last_tx_cal_freq = _tx_freq;
// cals done, set PPORT config
switch (_client_params->get_digital_interface_mode()) {
case AD9361_DDR_FDD_LVCMOS: {
_io_iface->poke8(0x012, 0x02);
} break;
case AD9361_DDR_FDD_LVDS: {
_io_iface->poke8(0x012, 0x10);
} break;
default:
throw uhd::runtime_error("[ad9361_device_t] NOT IMPLEMENTED");
}
_io_iface->poke8(0x013, 0x01); // Set ENSM FDD bit
_io_iface->poke8(0x015, 0x04); // dual synth mode, synth en ctrl en
/* End the function in the same state as the entry state. */
switch (current_state) {
case 0x05:
/* We are already in ALERT. */
break;
case 0x0A:
/* Transition back to FDD, and restore the original antenna
* / chain selections. */
_regs.txfilt = (_regs.txfilt & 0x3F) | orig_tx_chains;
_regs.rxfilt = (_regs.rxfilt & 0x3F) | orig_rx_chains;
_io_iface->poke8(0x002, _regs.txfilt);
_io_iface->poke8(0x003, _regs.rxfilt);
_io_iface->poke8(0x014, 0x21);
break;
default:
throw uhd::runtime_error(
"[ad9361_device_t] [set_clock_rate:2] AD9361 in unknown state");
break;
};
return get_clock_rate();
}
/* This function returns the RX / TX rate between AD9361 and the FPGA.
*/
double ad9361_device_t::get_clock_rate() const
{
return _baseband_bw;
}
/* Set which of the four TX / RX chains provided by AD9361 are active.
*
* AD9361 provides two sets of chains, Side A and Side B. Each side
* provides one TX antenna, and one RX antenna. The B200 maintains the USRP
* standard of providing one antenna connection that is both TX & RX, and
* one that is RX-only - for each chain. Thus, the possible antenna and
* chain selections are:
*
* B200 Antenna AD9361 Side AD9361 Chain
* -------------------------------------------------------------------
* TX / RX1 Side A TX1 (when switched to TX)
* TX / RX1 Side A RX1 (when switched to RX)
* RX1 Side A RX1
*
* TX / RX2 Side B TX2 (when switched to TX)
* TX / RX2 Side B RX2 (when switched to RX)
* RX2 Side B RX2
*/
void ad9361_device_t::set_active_chains(bool tx1, bool tx2, bool rx1, bool rx2)
{
std::lock_guard<std::recursive_mutex> lock(_mutex);
/* Clear out the current active chain settings. */
_regs.txfilt = _regs.txfilt & 0x3F;
_regs.rxfilt = _regs.rxfilt & 0x3F;
/* Turn on the different chains based on the passed parameters. */
if (tx1) {
_regs.txfilt = _regs.txfilt | 0x40;
}
if (tx2) {
_regs.txfilt = _regs.txfilt | 0x80;
}
if (rx1) {
_regs.rxfilt = _regs.rxfilt | 0x40;
}
if (rx2) {
_regs.rxfilt = _regs.rxfilt | 0x80;
}
/* Check for FDD state */
uint8_t set_back_to_fdd = 0;
uint8_t ensm_state = _io_iface->peek8(0x017) & 0x0F;
if (ensm_state == 0xA) // FDD
{
/* Put into ALERT state (via the FDD flush state). */
_io_iface->poke8(0x014, 0x01);
set_back_to_fdd = 1;
}
/* Wait for FDD flush state to complete (if necessary) */
while (ensm_state == 0xA || ensm_state == 0xB)
ensm_state = _io_iface->peek8(0x017) & 0x0F;
/* Turn on / off the chains. */
_io_iface->poke8(0x002, _regs.txfilt);
_io_iface->poke8(0x003, _regs.rxfilt);
/*
* Last unconditional Tx calibration point. Any later Tx calibration will
* require user intervention (currently triggered by tuning difference that
* is > 100 MHz). Late calibration provides better performance.
*/
if (tx1 | tx2)
_calibrate_tx_quadrature();
/* Put back into FDD state if necessary */
if (set_back_to_fdd)
_io_iface->poke8(0x014, 0x21);
}
/* Setup Timing mode depending on active channels.
*
* LVDS interface can have two timing modes - 1R1T and 2R2T
*/
void ad9361_device_t::set_timing_mode(const ad9361_device_t::timing_mode_t timing_mode)
{
switch (_client_params->get_digital_interface_mode()) {
case AD9361_DDR_FDD_LVCMOS: {
switch (timing_mode) {
case TIMING_MODE_1R1T: {
_io_iface->poke8(0x010,
0xc8); // Swap I&Q on Tx, Swap I&Q on Rx, Toggle frame sync mode
break;
}
case TIMING_MODE_2R2T: {
throw uhd::runtime_error("[ad9361_device_t] [set_timing_mode] 2R2T "
"timing mode not supported for CMOS");
break;
}
default:
UHD_THROW_INVALID_CODE_PATH();
}
break;
}
case AD9361_DDR_FDD_LVDS: {
switch (timing_mode) {
case TIMING_MODE_1R1T: {
_io_iface->poke8(0x010, 0xc8); // Swap I&Q on Tx, Swap I&Q on Rx,
// Toggle frame sync mode, 1R1T timing.
break;
}
case TIMING_MODE_2R2T: {
_io_iface->poke8(0x010, 0xcc); // Swap I&Q on Tx, Swap I&Q on Rx,
// Toggle frame sync mode, 2R2T timing.
break;
}
default:
UHD_THROW_INVALID_CODE_PATH();
}
break;
}
default:
throw uhd::runtime_error("[ad9361_device_t] NOT IMPLEMENTED");
}
}
/* Tune the RX or TX frequency.
*
* This is the publicly-accessible tune function. It makes sure the tune
* isn't a redundant request, and if not, passes it on to the class's
* internal tune function.
*
* After tuning, it runs any appropriate calibrations. */
double ad9361_device_t::tune(direction_t direction, const double value)
{
std::lock_guard<std::recursive_mutex> lock(_mutex);
double last_cal_freq;
if (direction == RX) {
if (freq_is_nearly_equal(value, _req_rx_freq)) {
return _rx_freq;
}
last_cal_freq = _last_rx_cal_freq;
} else if (direction == TX) {
if (freq_is_nearly_equal(value, _req_tx_freq)) {
return _tx_freq;
}
last_cal_freq = _last_tx_cal_freq;
} else {
throw uhd::runtime_error("[ad9361_device_t] [tune] INVALID_CODE_PATH");
}
/* If we aren't already in the ALERT state, we will need to return to
* the FDD state after tuning. */
int not_in_alert = 0;
if ((_io_iface->peek8(0x017) & 0x0F) != 5) {
/* Force the device into the ALERT state. */
not_in_alert = 1;
_io_iface->poke8(0x014, 0x01);
}
/* Tune the RF VCO! */
double tune_freq = _tune_helper(direction, value);
/* Run any necessary calibrations / setups */
if (direction == RX) {
_program_gain_table();
}
/* Update the gain settings. */
_reprogram_gains();
/*
* Only run the following calibrations if we are more than 100MHz away
* from the previous Tx or Rx calibration point. Leave out single shot
* Rx quadrature unless Rx quad-cal is disabled.
*/
if (std::abs(last_cal_freq - tune_freq) > AD9361_CAL_VALID_WINDOW) {
/* Run the calibration algorithms. */
if (direction == RX) {
_calibrate_rf_dc_offset();
if (!_use_iq_balance_tracking)
_calibrate_rx_quadrature();
if (_use_dc_offset_tracking)
_configure_bb_dc_tracking();
_last_rx_cal_freq = tune_freq;
} else {
_calibrate_tx_quadrature();
_last_tx_cal_freq = tune_freq;
}
/* Rx IQ tracking can be disabled on Rx or Tx re-calibration */
if (_use_iq_balance_tracking)
_configure_rx_iq_tracking();
}
/* If we were in the FDD state, return it now. */
if (not_in_alert) {
_io_iface->poke8(0x014, 0x21);
}
return get_freq(direction);
}
/* Get the current RX or TX frequency. */
double ad9361_device_t::get_freq(direction_t direction)
{
std::lock_guard<std::recursive_mutex> lock(_mutex);
if (direction == RX)
return _rx_freq;
else
return _tx_freq;
}
/* Set the gain of RX1, RX2, TX1, or TX2.
*
* Note that the 'value' passed to this function is the gain index
* for RX. Also note that the RX chains are done in terms of gain, and
* the TX chains are done in terms of attenuation. */
double ad9361_device_t::set_gain(direction_t direction, chain_t chain, const double value)
{
std::lock_guard<std::recursive_mutex> lock(_mutex);
if (direction == RX) {
int gain_index = static_cast<int>(value);
/* Clip the gain values to the proper min/max gain values. */
if (gain_index > 76)
gain_index = 76;
if (gain_index < 0)
gain_index = 0;
if (chain == CHAIN_1) {
_rx1_gain = value;
_io_iface->poke8(0x109, gain_index);
} else {
_rx2_gain = value;
_io_iface->poke8(0x10c, gain_index);
}
return gain_index;
} else {
/* Each gain step is -0.25dB. Calculate the attenuation necessary
* for the requested gain, convert it into gain steps, then write
* the attenuation word. Max gain (so zero attenuation) is 89.75.
* Ugly values will be written to the attenuation registers if
* "value" is out of bounds, so range checking must be performed
* outside this function.
*/
const double atten = AD9361_MAX_GAIN - value;
const uint32_t attenreg = uint32_t(atten * 4);
if (chain == CHAIN_1) {
_tx1_gain = value;
_io_iface->poke8(0x073, attenreg & 0xFF);
_io_iface->poke8(0x074, (attenreg >> 8) & 0x01);
} else {
_tx2_gain = value;
_io_iface->poke8(0x075, attenreg & 0xFF);
_io_iface->poke8(0x076, (attenreg >> 8) & 0x01);
}
/* After we've set both registers for the new gain, we apply it (assert
* "Immediately Update TPC Atten"). This avoids the attenuation from
* intermediately becoming very small (i.e., high output power) during
* the set-gain cycle.
*/
_io_iface->poke8(0x07c, 0x40);
return AD9361_MAX_GAIN - ((double)(attenreg) / 4);
}
}
void ad9361_device_t::output_test_tone() // On RF side!
{
std::lock_guard<std::recursive_mutex> lock(_mutex);
/* Output a 480 kHz tone at 800 MHz */
_io_iface->poke8(0x3F4, 0x0B);
_io_iface->poke8(0x3FC, 0xFF);
_io_iface->poke8(0x3FD, 0xFF);
_io_iface->poke8(0x3FE, 0x3F);
}
void ad9361_device_t::digital_test_tone(bool enb) // Digital output
{
std::lock_guard<std::recursive_mutex> lock(_mutex);
_io_iface->poke8(0x3F4, 0x02 | (enb ? 0x01 : 0x00));
}
void ad9361_device_t::data_port_loopback(const bool loopback_enabled)
{
std::lock_guard<std::recursive_mutex> lock(_mutex);
_io_iface->poke8(0x3F5, (loopback_enabled ? 0x01 : 0x00));
}
/* Read back the internal RSSI measurement data. The result is in dB
* but not in absolute units. If absolute units are required
* a bench calibration should be done.
* -0.25dB / bit 9bit resolution.*/
double ad9361_device_t::get_rssi(chain_t chain)
{
uint32_t reg_rssi = 0;
uint8_t lsb_bit_pos = 0;
if (chain == CHAIN_1) {
reg_rssi = 0x1A7;
lsb_bit_pos = 0;
} else {
reg_rssi = 0x1A9;
lsb_bit_pos = 1;
}
uint8_t msbs = _io_iface->peek8(reg_rssi);
uint8_t lsb = ((_io_iface->peek8(0x1AB)) >> lsb_bit_pos) & 0x01;
uint16_t val = ((msbs << 1) | lsb);
double rssi =
(-0.25f * ((double)val)); //-0.25dB/lsb (See Gain Control Users Guide p. 25)
return rssi;
}
/*
* Returns the reading of the internal temperature sensor.
* One point calibration of the sensor was done according to datasheet
* leading to the given default constant correction factor.
*/
double ad9361_device_t::_get_temperature(const double cal_offset, const double timeout)
{
// set 0x01D[0] to 1 to disable AuxADC GPIO reading
uint8_t tmp = 0;
tmp = _io_iface->peek8(0x01D);
_io_iface->poke8(0x01D, (tmp | 0x01));
_io_iface->poke8(0x00B, 0); // set offset to 0
_io_iface->poke8(0x00C, 0x01); // start reading, clears bit 0x00C[1]
auto end_time = std::chrono::steady_clock::now()
+ std::chrono::milliseconds(int64_t(timeout * 1000));
// wait for valid data (toggle of bit 1 in 0x00C)
while (((_io_iface->peek8(0x00C) >> 1) & 0x01) == 0) {
std::this_thread::sleep_for(std::chrono::microseconds(100));
if (std::chrono::steady_clock::now() > end_time) {
throw uhd::runtime_error(
"[ad9361_device_t] timeout while reading temperature");
}
}
_io_iface->poke8(0x00C, 0x00); // clear read flag
uint8_t temp = _io_iface->peek8(0x00E); // read temperature.
double tmp_temp = temp / 1.140f; // according to ADI driver
tmp_temp = tmp_temp + cal_offset; // Constant offset acquired by one point
// calibration.
return tmp_temp;
}
double ad9361_device_t::get_average_temperature(
const double cal_offset, const size_t num_samples)
{
double d_temp = 0;
for (size_t i = 0; i < num_samples; i++) {
double tmp_temp = _get_temperature(cal_offset);
d_temp += (tmp_temp / num_samples);
}
return d_temp;
}
/*
* Enable/Disable DC offset tracking
*
* Only disable BB tracking while leaving static RF and BB DC calibrations enabled.
* According to correspondance from ADI, turning off Rx BB DC tracking clears the
* correction words so we don't need to be concerned with leaving the calibration
* in a bad state upon disabling. Testing also confirms this behavior.
*
* Note that Rx IQ tracking does not show similar state clearing behavior when
* disabled.
*/
void ad9361_device_t::set_dc_offset_auto(direction_t direction, const bool on)
{
if (direction == RX) {
_use_dc_offset_tracking = on;
_configure_bb_dc_tracking();
} else {
throw uhd::runtime_error(
"[ad9361_device_t] [set_dc_offset_auto] Tx DC tracking not supported");
}
}
/*
* Enable/Disable IQ balance tracking
*
* Run static Rx quadrature calibration after disabling quadrature tracking.
* This avoids the situation where a user might disable tracking when the loop
* is in a confused state (e.g. at or near saturation). Otherwise, the
* calibration setting could be forced to and left in a bad state.
*/
void ad9361_device_t::set_iq_balance_auto(direction_t direction, const bool on)
{
if (direction == RX) {
_use_iq_balance_tracking = on;
_configure_rx_iq_tracking();
if (!on) {
_io_iface->poke8(0x014, 0x05); // ALERT mode
_calibrate_rx_quadrature();
_io_iface->poke8(0x014, 0x21); // FDD mode
}
} else {
throw uhd::runtime_error(
"[ad9361_device_t] [set_iq_balance_auto] Tx IQ tracking not supported");
}
}
/* Sets the RX gain mode to be used.
* If a transition from an AGC to an non AGC mode occurs (or vice versa)
* the gain configuration will be reloaded. */
void ad9361_device_t::_setup_agc(chain_t chain, gain_mode_t gain_mode)
{
uint8_t gain_mode_reg = 0;
uint8_t gain_mode_prev = 0;
uint8_t gain_mode_bits_pos = 0;
gain_mode_reg = _io_iface->peek8(0x0FA);
gain_mode_prev = (gain_mode_reg & 0x0F);
if (chain == CHAIN_1) {
gain_mode_bits_pos = 0;
} else if (chain == CHAIN_2) {
gain_mode_bits_pos = 2;
} else {
throw uhd::runtime_error("[ad9361_device_t] Wrong value for chain");
}
gain_mode_reg = (gain_mode_reg & (~(0x03 << gain_mode_bits_pos))); // clear mode bits
switch (gain_mode) {
case GAIN_MODE_MANUAL:
// leave bits cleared
break;
case GAIN_MODE_SLOW_AGC:
gain_mode_reg = (gain_mode_reg | (0x02 << gain_mode_bits_pos));
break;
case GAIN_MODE_FAST_AGC:
gain_mode_reg = (gain_mode_reg | (0x01 << gain_mode_bits_pos));
break;
default:
throw uhd::runtime_error("[ad9361_device_t] Gain mode does not exist");
}
_io_iface->poke8(0x0FA, gain_mode_reg);
uint8_t gain_mode_status = _io_iface->peek8(0x0FA);
gain_mode_status = (gain_mode_status & 0x0F);
/*Check if gain mode configuration needs to be reprogrammed*/
if (((gain_mode_prev == 0) && (gain_mode_status != 0))
|| ((gain_mode_prev != 0) && (gain_mode_status == 0))) {
if (gain_mode_status == 0) {
/*load manual mode config*/
_setup_gain_control(false);
} else {
/*load agc mode config*/
_setup_gain_control(true);
}
}
}
void ad9361_device_t::set_agc(chain_t chain, bool enable)
{
if (chain == CHAIN_1) {
_rx1_agc_enable = enable;
if (enable) {
_setup_agc(chain, _rx1_agc_mode);
} else {
_setup_agc(chain, GAIN_MODE_MANUAL);
}
} else if (chain == CHAIN_2) {
_rx2_agc_enable = enable;
if (enable) {
_setup_agc(chain, _rx2_agc_mode);
} else {
_setup_agc(chain, GAIN_MODE_MANUAL);
}
} else {
throw uhd::runtime_error("[ad9361_device_t] Wrong value for chain");
}
}
void ad9361_device_t::set_agc_mode(chain_t chain, gain_mode_t gain_mode)
{
if (chain == CHAIN_1) {
_rx1_agc_mode = gain_mode;
if (_rx1_agc_enable) {
_setup_agc(chain, _rx1_agc_mode);
}
} else if (chain == CHAIN_2) {
_rx2_agc_mode = gain_mode;
if (_rx2_agc_enable) {
_setup_agc(chain, _rx2_agc_mode);
}
} else {
throw uhd::runtime_error("[ad9361_device_t] Wrong value for chain");
}
}
std::vector<std::string> ad9361_device_t::get_filter_names(direction_t direction)
{
auto& filters = (direction == RX) ? _rx_filters : _tx_filters;
std::vector<std::string> ret;
ret.reserve(filters.size());
for (auto& filter : filters) {
ret.push_back(filter.first);
}
return ret;
}
filter_info_base::sptr ad9361_device_t::get_filter(
direction_t direction, chain_t chain, const std::string& name)
{
auto& filters = (direction == RX) ? _rx_filters : _tx_filters;
if (!filters.count(name)) {
throw uhd::runtime_error(
"ad9361_device_t::get_filter this filter does not exist: " + name);
}
// Check entry 0 in the tuple (the getter) exists before calling it
if (!std::get<0>(filters[name])) {
throw uhd::runtime_error(
"ad9361_device_t::get_filter this filter can not be read: " + name);
}
return std::get<0>(filters[name])(chain);
}
void ad9361_device_t::set_filter(direction_t direction,
chain_t chain,
const std::string& name,
filter_info_base::sptr filter)
{
auto& filters = (direction == RX) ? _rx_filters : _tx_filters;
if (!filters.count(name)) {
throw uhd::runtime_error(
"ad9361_device_t::set_filter this filter does not exist: " + name);
}
// Check entry 1 in the tuple (the setter) exists before calling it
if (!std::get<1>(filters[name])) {
throw uhd::runtime_error(
"ad9361_device_t::set_filter this filter can not be written: " + name);
}
std::get<1>(filters[name])(chain, filter);
}
double ad9361_device_t::set_bw_filter(direction_t direction, const double rf_bw)
{
// both low pass filters are programmed to the same bw. However, their cutoffs will
// differ. Together they should create the requested bb bw. Select rf_bw if it is
// between AD9361_MIN_BW & AD9361_MAX_BW.
const double clipped_bw = std::min(std::max(rf_bw, AD9361_MIN_BW), AD9361_MAX_BW);
if (direction == RX) {
_rx_bb_lp_bw = _calibrate_baseband_rx_analog_filter(clipped_bw); // returns bb bw
_rx_tia_lp_bw = _calibrate_rx_TIAs(clipped_bw);
_rx_analog_bw = clipped_bw;
} else {
_tx_bb_lp_bw = _calibrate_baseband_tx_analog_filter(clipped_bw); // returns bb bw
_tx_sec_lp_bw = _calibrate_secondary_tx_filter(clipped_bw);
_tx_analog_bw = clipped_bw;
}
return (clipped_bw);
}
void ad9361_device_t::_set_fir_taps(
direction_t direction, chain_t chain, const std::vector<int16_t>& taps)
{
size_t num_taps = taps.size();
size_t num_taps_avail = _get_num_fir_taps(direction);
if (num_taps == num_taps_avail) {
boost::scoped_array<uint16_t> coeffs(new uint16_t[num_taps_avail]);
for (size_t i = 0; i < num_taps_avail; i++) {
coeffs[i] = uint16_t(taps[i]);
}
_program_fir_filter(direction, chain, num_taps_avail, coeffs.get());
} else if (num_taps < num_taps_avail) {
throw uhd::runtime_error(
"ad9361_device_t::_set_fir_taps not enough coefficients.");
} else {
throw uhd::runtime_error("ad9361_device_t::_set_fir_taps too many coefficients.");
}
}
size_t ad9361_device_t::_get_num_fir_taps(direction_t direction)
{
uint8_t num = 0;
if (direction == RX)
num = _io_iface->peek8(0x0F5);
else
num = _io_iface->peek8(0x065);
num = ((num >> 5) & 0x07);
return ((num + 1) * 16);
}
size_t ad9361_device_t::_get_fir_dec_int(direction_t direction)
{
uint8_t dec_int = 0;
if (direction == RX)
dec_int = _io_iface->peek8(0x003);
else
dec_int = _io_iface->peek8(0x002);
/*
* 0 = dec/int by 1 and bypass filter
* 1 = dec/int by 1
* 2 = dec/int by 2
* 3 = dec/int by 4 */
dec_int = (dec_int & 0x03);
if (dec_int == 3) {
return 4;
}
return dec_int;
}
std::vector<int16_t> ad9361_device_t::_get_fir_taps(direction_t direction, chain_t chain)
{
int base;
size_t num_taps = _get_num_fir_taps(direction);
uint8_t config;
uint8_t reg_numtaps = (((num_taps / 16) - 1) & 0x07) << 5;
config = reg_numtaps | 0x02; // start the programming clock
if (chain == CHAIN_1) {
config = config | (1 << 3);
} else if (chain == CHAIN_2) {
config = config | (1 << 4);
} else {
throw uhd::runtime_error(
"[ad9361_device_t] Can not read both chains synchronously");
}
if (direction == RX) {
base = 0xF0;
} else {
base = 0x60;
}
_io_iface->poke8(base + 5, config);
std::vector<int16_t> taps;
uint8_t lower_val;
uint8_t higher_val;
uint16_t coeff;
for (size_t i = 0; i < num_taps; i++) {
_io_iface->poke8(base, 0x00 + i);
lower_val = _io_iface->peek8(base + 3);
higher_val = _io_iface->peek8(base + 4);
coeff = ((higher_val << 8) | lower_val);
taps.push_back(int16_t(coeff));
}
config = (config & (~(1 << 1))); // disable filter clock
_io_iface->poke8(base + 5, config);
return taps;
}
/*
* Returns either RX TIA LPF or TX Secondary LPF
* depending on the direction.
* See UG570 for details on used scaling factors. */
filter_info_base::sptr ad9361_device_t::_get_filter_lp_tia_sec(direction_t direction)
{
double cutoff = 0;
if (direction == RX) {
cutoff = 2.5 * _rx_tia_lp_bw;
} else {
cutoff = 5 * _tx_sec_lp_bw;
}
filter_info_base::sptr lp(new analog_filter_lp(
filter_info_base::ANALOG_LOW_PASS, false, 0, "single-pole", cutoff, 20));
return lp;
}
/*
* Returns RX/TX BB LPF.
* See UG570 for details on used scaling factors. */
filter_info_base::sptr ad9361_device_t::_get_filter_lp_bb(direction_t direction)
{
double cutoff = 0;
if (direction == RX) {
cutoff = 1.4 * _rx_bb_lp_bw;
} else {
cutoff = 1.6 * _tx_bb_lp_bw;
}
filter_info_base::sptr bb_lp(new analog_filter_lp(filter_info_base::ANALOG_LOW_PASS,
false,
1,
"third-order Butterworth",
cutoff,
60));
return bb_lp;
}
/*
* For RX direction the DEC3 is returned.
* For TX direction the INT3 is returned. */
filter_info_base::sptr ad9361_device_t::_get_filter_dec_int_3(direction_t direction)
{
// clang-format off
uint8_t enable = 0;
double rate = _adcclock_freq;
double full_scale;
size_t dec = 0;
size_t interpol = 0;
filter_info_base::filter_type type = filter_info_base::DIGITAL_I16;
std::string name;
int16_t taps_array_rx[] = {55, 83, 0, -393, -580, 0, 1914, 4041, 5120, 4041, 1914, 0, -580, -393, 0, 83, 55};
int16_t taps_array_tx[] = {36, -19, 0, -156, -12, 0, 479, 233, 0, -1215, -993, 0, 3569, 6277, 8192, 6277, 3569, 0, -993, -1215, 0, 223, 479, 0, -12, -156, 0, -19, 36};
std::vector<int16_t> taps;
// clang-format on
filter_info_base::sptr ret;
if (direction == RX) {
full_scale = 16384;
dec = 3;
interpol = 1;
enable = _io_iface->peek8(0x003);
enable = ((enable >> 4) & 0x03);
taps.assign(
taps_array_rx, taps_array_rx + sizeof(taps_array_rx) / sizeof(int16_t));
} else {
full_scale = 8192;
dec = 1;
interpol = 3;
uint8_t use_dac_clk_div = _io_iface->peek8(0x00A);
use_dac_clk_div = ((use_dac_clk_div >> 3) & 0x01);
if (use_dac_clk_div == 1) {
rate = rate / 2;
}
enable = _io_iface->peek8(0x002);
enable = ((enable >> 4) & 0x03);
if (enable == 2) // 0 => int. by 1, 1 => int. by 2 (HB3), 2 => int. by 3
{
rate /= 3;
}
taps.assign(
taps_array_tx, taps_array_tx + sizeof(taps_array_tx) / sizeof(int16_t));
}
ret = filter_info_base::sptr(new digital_filter_base<int16_t>(type,
(enable != 2) ? true : false,
2,
rate,
interpol,
dec,
full_scale,
taps.size(),
taps));
return ret;
}
filter_info_base::sptr ad9361_device_t::_get_filter_hb_3(direction_t direction)
{
uint8_t enable = 0;
double rate = _adcclock_freq;
double full_scale = 0;
size_t dec = 1;
size_t interpol = 1;
filter_info_base::filter_type type = filter_info_base::DIGITAL_I16;
int16_t taps_array_rx[] = {1, 4, 6, 4, 1};
int16_t taps_array_tx[] = {1, 2, 1};
std::vector<int16_t> taps;
if (direction == RX) {
full_scale = 16;
dec = 2;
enable = _io_iface->peek8(0x003);
enable = ((enable >> 4) & 0x03);
taps.assign(
taps_array_rx, taps_array_rx + sizeof(taps_array_rx) / sizeof(int16_t));
} else {
full_scale = 2;
interpol = 2;
uint8_t use_dac_clk_div = _io_iface->peek8(0x00A);
use_dac_clk_div = ((use_dac_clk_div >> 3) & 0x01);
if (use_dac_clk_div == 1) {
rate = rate / 2;
}
enable = _io_iface->peek8(0x002);
enable = ((enable >> 4) & 0x03);
if (enable == 1) {
rate /= 2;
}
taps.assign(
taps_array_tx, taps_array_tx + sizeof(taps_array_tx) / sizeof(int16_t));
}
filter_info_base::sptr hb =
filter_info_base::sptr(new digital_filter_base<int16_t>(type,
(enable != 1) ? true : false,
2,
rate,
interpol,
dec,
full_scale,
taps.size(),
taps));
return hb;
}
filter_info_base::sptr ad9361_device_t::_get_filter_hb_2(direction_t direction)
{
uint8_t enable = 0;
double rate = _adcclock_freq;
double full_scale = 0;
size_t dec = 1;
size_t interpol = 1;
filter_info_base::filter_type type = filter_info_base::DIGITAL_I16;
int16_t taps_array[] = {-9, 0, 73, 128, 73, 0, -9};
std::vector<int16_t> taps(
taps_array, taps_array + sizeof(taps_array) / sizeof(int16_t));
digital_filter_base<int16_t>::sptr hb_3 =
std::dynamic_pointer_cast<digital_filter_base<int16_t>>(
_get_filter_hb_3(direction));
digital_filter_base<int16_t>::sptr dec_int_3 =
std::dynamic_pointer_cast<digital_filter_base<int16_t>>(
_get_filter_dec_int_3(direction));
if (direction == RX) {
full_scale = 256;
dec = 2;
enable = _io_iface->peek8(0x003);
} else {
full_scale = 128;
interpol = 2;
enable = _io_iface->peek8(0x002);
}
enable = ((enable >> 3) & 0x01);
if (!(hb_3->is_bypassed())) {
if (direction == RX) {
rate = hb_3->get_output_rate();
} else if (direction == TX) {
rate = hb_3->get_input_rate();
if (enable) {
rate /= 2;
}
}
} else { // else dec3/int3 or none of them is used.
if (direction == RX) {
rate = dec_int_3->get_output_rate();
} else if (direction == TX) {
rate = dec_int_3->get_input_rate();
if (enable) {
rate /= 2;
}
}
}
filter_info_base::sptr hb(new digital_filter_base<int16_t>(type,
(enable == 0) ? true : false,
3,
rate,
interpol,
dec,
full_scale,
taps.size(),
taps));
return hb;
}
filter_info_base::sptr ad9361_device_t::_get_filter_hb_1(direction_t direction)
{
uint8_t enable = 0;
double rate = 0;
double full_scale = 0;
size_t dec = 1;
size_t interpol = 1;
filter_info_base::filter_type type = filter_info_base::DIGITAL_I16;
std::vector<int16_t> taps;
int16_t taps_rx_array[] = {
-8, 0, 42, 0, -147, 0, 619, 1013, 619, 0, -147, 0, 42, 0, -8};
int16_t taps_tx_array[] = {
-53, 0, 313, 0, -1155, 0, 4989, 8192, 4989, 0, -1155, 0, 313, 0, -53};
digital_filter_base<int16_t>::sptr hb_2 =
std::dynamic_pointer_cast<digital_filter_base<int16_t>>(
_get_filter_hb_2(direction));
if (direction == RX) {
full_scale = 2048;
dec = 2;
enable = _io_iface->peek8(0x003);
enable = ((enable >> 2) & 0x01);
rate = hb_2->get_output_rate();
taps.assign(
taps_rx_array, taps_rx_array + sizeof(taps_rx_array) / sizeof(int16_t));
} else if (direction == TX) {
full_scale = 8192;
interpol = 2;
enable = _io_iface->peek8(0x002);
enable = ((enable >> 2) & 0x01);
rate = hb_2->get_input_rate();
if (enable) {
rate /= 2;
}
taps.assign(
taps_tx_array, taps_tx_array + sizeof(taps_tx_array) / sizeof(int16_t));
}
filter_info_base::sptr hb(new digital_filter_base<int16_t>(type,
(enable == 0) ? true : false,
4,
rate,
interpol,
dec,
full_scale,
taps.size(),
taps));
return hb;
}
filter_info_base::sptr ad9361_device_t::_get_filter_fir(
direction_t direction, chain_t chain)
{
double rate = 0;
size_t dec = 1;
size_t interpol = 1;
size_t max_num_taps = 128;
uint8_t enable = 1;
digital_filter_base<int16_t>::sptr hb_1 =
std::dynamic_pointer_cast<digital_filter_base<int16_t>>(
_get_filter_hb_1(direction));
if (direction == RX) {
dec = _get_fir_dec_int(direction);
if (dec == 0) {
enable = 0;
dec = 1;
}
interpol = 1;
rate = hb_1->get_output_rate();
} else if (direction == TX) {
interpol = _get_fir_dec_int(direction);
if (interpol == 0) {
enable = 0;
interpol = 1;
}
dec = 1;
rate = hb_1->get_input_rate();
if (enable) {
rate /= interpol;
}
}
max_num_taps = _get_num_fir_taps(direction);
filter_info_base::sptr fir(
std::make_shared<digital_filter_fir<int16_t>>(filter_info_base::DIGITAL_FIR_I16,
(enable == 0) ? true : false,
5,
rate,
interpol,
dec,
32767,
max_num_taps,
_get_fir_taps(direction, chain)));
return fir;
}
void ad9361_device_t::_set_filter_fir(
direction_t direction, chain_t channel, filter_info_base::sptr filter)
{
digital_filter_fir<int16_t>::sptr fir =
std::dynamic_pointer_cast<digital_filter_fir<int16_t>>(filter);
// only write taps. Ignore everything else for now
_set_fir_taps(direction, channel, fir->get_taps());
}
/*
* If BW of one of the analog filters gets overwritten manually,
* _tx_analog_bw and _rx_analog_bw are not valid any more!
* For useful data in those variables set_bw_filter method should be used
*/
void ad9361_device_t::_set_filter_lp_bb(
direction_t direction, filter_info_base::sptr filter)
{
analog_filter_lp::sptr lpf = std::dynamic_pointer_cast<analog_filter_lp>(filter);
double bw = lpf->get_cutoff();
if (direction == RX) {
// remember: this function takes rf bw as its input and calibrated to 1.4 x the
// given value
_rx_bb_lp_bw = _calibrate_baseband_rx_analog_filter(2 * bw / 1.4); // returns bb
// bw
} else {
// remember: this function takes rf bw as its input and calibrates to 1.6 x the
// given value
_tx_bb_lp_bw = _calibrate_baseband_tx_analog_filter(2 * bw / 1.6);
}
}
void ad9361_device_t::_set_filter_lp_tia_sec(
direction_t direction, filter_info_base::sptr filter)
{
analog_filter_lp::sptr lpf = std::dynamic_pointer_cast<analog_filter_lp>(filter);
double bw = lpf->get_cutoff();
if (direction == RX) {
// remember: this function takes rf bw as its input and calibrated to 2.5 x the
// given value
_rx_tia_lp_bw = _calibrate_rx_TIAs(2 * bw / 2.5); // returns bb bw
} else {
// remember: this function takes rf bw as its input and calibrates to 5 x the
// given value
_tx_sec_lp_bw = _calibrate_secondary_tx_filter(2 * bw / 5);
}
}
}} // namespace uhd::usrp
|