***************************************************************** * ARRL Radio Designer 1.0 File EXRF9509.TXT V1.0 (31 Jul 1995) * * Exploring RF -- September 1995 *QST*, pp 99-101 * * compiled by David Newkirk, WJ1Z (dnewkirk@arrl.org) * * * * Constrained Values for a More-Optimizable BIP * * and * * Testing Active-TWO Noise Modeling * * in ARRL Radio Designer 1.0 and 1.5 * ***************************************************************** * * For greatest ease in switching back and forth between gradient * and random optimization of a given circuit file, it's arguably * better to use *constrained* values--values entered as R=?5KOH * 10KOH 50KOH? rather than just unconstrained values like R=?10KOH? * --because the range of values chooseable by ARD's optimizer differs * between gradient and random optimization. (See page 19-3 of * The ARRL Radio Designer Manual for details.) This can make quite * a difference when you optimize a BIP or FET for congruence with * a databank-data-characterized TWO for the reasons inferrable * from the coverage on that page of the manual, and as explained * in a bit more detail in September 1995 Exploring RF. To that end, * here's an optimizable BIP template reworked to include constrained * values for all of its parameters. BLK BIP 1 2 0 A=?.85 .9 .999? RE=?0.5 1 15? + F=?0.1GHZ 1GHZ 10GHZ? T=?0.005NS 0.01NS 0.1NS? + CE=?0.01PF 0.1PF 5PF? CI=?0.005PF 1PF 10PF? + RC=?100 1000 10000? RO=?0.001 0.1 5? + CO=?0.001PF 1PF 10PF? RB1=?0.005 1 10? + RC1=?1E-7 1E-6 1E-5? RB2=?0.001 0.1 5? + CBE=?0.005PF 5PF 10PF? CBC=?0.001PF 0.1PF 5PF? + CCE=?1E-5PF 1E-4PF 1E-3PF? + LB=?1E-7NH 1E-6NH 1E-5NH? LC=?1E-5NH 1E-4NH 1E-3NH? + LE=?0.001NH 0.1NH 1NH? OPTBIP:2POR 1 2 0 END * ARRL Radio Designer README.TXT V1.03 put it like this: * * > ARD 1.0 inaccurately models the noise of black-box * > elements even when appropriate, valid noise-parameter data * > is present in the DATA block. This bug causes meaningless, * > "noisy" noise-related report traces (and their tabular * > equivalents) in ARD 1.0 simulations. * * and you can use this circuit file to see this bug in action. * Analyzing this file and plotting traces for MS21 and NF will * reveal unrealistic values and numerical noise in portions of * both traces. * * If you have access to any of ARRL Radio Designer's superset * programs--Super-Compact, Microwave Harmonica or SCOPE--you can * see how this circuit file's active-TWO noise modeling works * in the absence of the active-TWO-noise-modeling bug. * * So much for confirming the bad news. The good news is that * an ARRL Radio Designer maintenance release, ARD 1.5, will be * available as of September 1. It will be free if you bought your * copy of ARRL Radio Designer 1.0 on or after August 1, 1995, * and \$10 if you bought your copy of ARRL Radio Designer 1.0 on * or before July 31, 1995. How do you get it? **After August * 15**, contact ARRL Publication Sales at ARRL HQ (voice * 203-594-0200 and fax 594-0303, or e-mail ltardette@arrl.org) * for details. * * And yes, this circuit file doesn't contain September 1995 * Exploring RF's "Table 2 netlist" because Table 2 didn't * contain a complete one. Table *3* did, however, and that's * what's in this circuit file. * BLK TWO 1 2 0 AVANTBIP NOISETST:2POR 1 2 END FREQ STEP 0.1GHZ 4.0GHZ 0.1GHZ END DATA AVANTBIP:S 0.1GHZ .74 -40 25.80 156 .017 81 .93 -14 0.5GHZ .61 -126 12.46 108 .035 44 .57 -31 1.0GHZ .57 -161 6.80 87 .040 38 .46 -33 1.5GHZ .57 -180 4.67 75 .052 47 .43 -34 2.0GHZ .58 166 3.55 64 .060 54 .41 -38 2.5GHZ .59 160 2.92 59 .068 58 .40 -39 3.0GHZ .61 150 2.42 50 .080 63 .39 -46 3.5GHZ .62 142 2.09 41 .091 61 .41 -54 4.0GHZ .62 134 1.84 32 .106 59 .42 -62 NOI RN 0.1GHZ 1.3 .12 5 .17 0.5GHZ 1.3 .10 25 .17 1.0GHZ 1.4 .06 50 .16 2.0GHZ 1.7 .25 172 .16 4.0GHZ 3.0 .48 -131 .24 END