QST Exploring RF—1995 This file contains the six Exploring RF columns from the 1995 issues of QST. Users of ARRL Radio Designer software will find these columns to be useful application notes that explore ways of using ARRL Radio Designer to solve thorny circuit-analysis problems of the sort encountered in RF design. If you are using Adobe Acrobat version 2.1, with its WEBLINK.API plug-in, and if you have a World Wide Web browser installed and configured to work with Acrobat, you can click on the links in these articles to download referenced files from the ARRL Web server or to generate an email message to a referenced email address. Enjoy. And let these be only your starting point for using ARRL Radio Designer to help you design and analyze the circuits in your next RF project! Conducted By David Newkirk, WJ1Z Exploring RF Senior Assistant Technical Editor e-mail: [email protected] Now that ARRL Radio Designer is on the street, we’re hoping The T Network in Schematic and Netlist Form that ham interest in exploring radio-frequency (RF) technology Figure 1 shows Andy’s T net’s schematic with a few additions through computer-based modeling will take off in a way that hasn’t and changes. Here’s the corresponding ARD netlist block: been possible before. This column is part of taking that show on the BLK road—a place for suggesting ARRL Radio Designer applications CAP 1 2 C=?20PF 82.1733PF 1000PF? Q=1000 ; C/IN\ we hope will be interesting to many hams. IND 2 0 L=?0.1UH 30.2528UH 35UH? Q1=200 F=2.52MHZ;L But this column won’t stop there. Many useful electronic and CAP 2 3 C=240PF Q=1000 ; C/OUT\ RF circuit design programs have been with us for years in various RES 3 0 R=10 ; this is the load forms, and new ones keep appearing. For starters, I’m thinking of ONEPORT:1POR 1 END the various commercial products based on the University of Cali- fornia at Berkeley’s SPICE (Simulation Program with Integrated Question marks bracket the component values we want to op- Circuit Emphasis)—at least one of which, MicroSim Corp’s timize. Note that instead of just specifying a single value, we’ve PSpice, is available free in a limited-function evaluation version, indicated ranges for CIN and L. (In keeping with Andy Griffith’s and other of which, MICRO-CAP, is available jointly from suggested tuning technique, we start out by setting COUT to a fixed Addison-Wesley Publishing and Benjamin/Cummings Publishing value and leaving it alone.) In ARD-speak, this constrains their in a low-cost student edition. Just sketching the abilities of the values to the ranges shown in Figure 1. Note also that the optimizer many SPICE derivatives would take several pages. A diskful of won’t touch the Q and F values associated with CIN and L. This is antenna, transmission-line and propagation software now ships appropriate, considering that we can’t adjust them at will in a real with every new ARRL Antenna Book sold, and software is also antenna tuner. available with The ARRL UHF/Microwave Experimenter’s We’re interested in seeing how the network plays at 160 meters, Manual—what can those programs do? And what about the so we’ll set up a frequency block that combines good resolution freeware and shareware that’s almost everywhere? There’s a lot to across the band with enough wide-range exponential steps for be excited about if you’re interested in using computer-aided de- graphing the network’s harmonic reduction: sign (CAD) and computer-aided engineering (CAE) in Amateur FREQ Radio, and QST will share that excitement through Exploring RF. STEP 1.8MHZ 2.0MHZ 1KHZ ESTP 0.5MHZ 30MHZ 300 Optimizing Circuit Performance with ARRL Radio Designer END One of the ARRL Radio Designer features we’re most excited Finally, we add an optimization (OPT) block that tells ARD about is the program’s ability to optimize circuits—to make cir- which circuit to optimize—a netlist may contain more than one, cuits work better, or to modify circuit performance according to after all—at what frequency or frequencies, in what way, and to specified goals. I suggested in October QST’s ARRL Radio what degree: Designer article1 that one use for this feature might be to tweak a OPT 2-meter receiver preamp for a better noise figure, but that’s a pretty ONEPORT F=1.9MHZ far-out example for many of us. Then I went on to describe an RZ11=50 antenna-tuner optimization that’s a very nice example of optimiz- IZ11=0 ing a circuit almost, but not exactly, like the one actually shown.2 TERM=1E-5 Elsewhere in this issue, Andy Griffith’s “Getting the Most Out END of Your T-Network Antenna Tuner” includes some eye-opening This block tells ARRL Radio Designer to optimize the perfor- findings on just how lossy a T network can be when transforming mance of the circuit block ONEPORT so that, at 1.9 MHz, the real some impedances to 50 W . ARRL Radio Designer, manipulated by (resistive) part of its impedance is 50 W and the imaginary (reac- yours truly, backed up Andy’s findings and even provided a few tive) part of its impedance is j0 W ; that is, 50+j0 W . The graphics. Here’s how I did it—and how you can use ARRL Radio TERM=1E-5 line tells ARD to terminate optimization—that is, to Designer’s optimizer to simulate adjustable T networks, too. consider the optimization goals reached—when the optimizer’s error function reaches or drops below 1· 10–5, or 0.00001. The OPT and END statements merely identify the block and delimit it so ARRL Radio Designer knows where it starts and stops. Table 1 shows the corresponding ARD netlist, all set to go. This column isn’t the place for excruciating detail about exactly how ARD calculates the optimizer’s error function. (The ARRL Radio Designer Manual explains how optimization goals factor into the error function, and how you can specify their weight—the degree to which they contribute to the number.) To get started with optimization, all we need to know is that the greater the error func- tion, the farther we are from a successful optimization. Running It Figure 1—Many of us use a capacitor-inductor-capacitor T network Once we’ve got our optimizable file ready to go, we click on similar to this one as an antenna tuner. With an SWR or reflected power meter acting as an indicator, we adjust its C and L values to Optim (or press Shift+F10). ARRL Radio Designer analyzes the transform the antenna’s impedance (here, 10 W , resistive) to 50 W , circuit first, and then pops its Optimization and Optimization Data resistive (50+j0 W when written in rectangular notation) at our dialogs (Figure 2).3 frequencies of interest. (I specify L’s Q at 2.52 MHz to tie in with When the optimizer succeeds in pushing its error function to a lab Q-meter conventions, as pages 5-7 and 5-8 of The ARRL number equal or less than the OPT block’s TERM value—or to 0 if Radio Designer Manual explain.) The circled numbers indicate you didn’t specify one—it halts, and the message “Optimization nodes for ARRL Radio Designer analysis, which treats the tuner as a one-port network imaginatively named ONEPORT. Completed” appears in the Optimization dialog.4 Click the Done 80 From January 1995 QST © 1995 ARRL Figure 2—Its pre-optimization analysis complete, ARRL Radio Figure 3—The 160-meter tuner’s frequency response for 10+j0 to Designer pops its Optimization and Optimization Data dialogs. All 50+j0 W transformations based on various C settings. I used OUT we need to do now is enter a number in the Iterations edit control netlist blocks like that shown in Table 3 to do these analyses. If (I always enter 1000), clear the Display box (ARD’s optimizer runs harmonic rejection is what you’re after, the capacitor-inductor- much faster when it doesn’t stop to do display calculations capacitor T network isn’t the way to get it. between iterations), and click Optimize. Leaving the optimizer set to its Random default is best for now; we’ll have to discuss random versus gradient optimization some other time. Table 2 The T Network Reconfigured as a Two-Port for Frequency-Response Simulation (Values for a 240-pF Table 1 C Shown) Simulating an Adjustable T-Network Antenna Tuner with OUT BLK ARRL Radio Designer CAP 1 2 C=122.811PF Q=1000 BLK IND 2 0 L=19.3847UH Q=200 F=2.52MHZ CAP 1 2 C=?20PF 82.1733PF 1000PF? Q=1000 ; C/IN\ CAP 2 3 C=240PF Q=1000 IND 2 0 L=?0.1UH 30.2528UH 35UH? Q1=200 F=2.52MHZ;L TRF 3 4 0 0 R1=10 R2=50 CAP 2 3 C=240PF Q=1000 ; C/OUT\ OUT240PF:2POR 1 4 RES 3 0 R=10 ; this is the tuner’s load END ONEPORT:1POR 1 This block replaces ONEPORT’s RES line with an ideal transformer END (TRF) connected between nodes 3 and 4, with node 0 common for FREQ both windings. The phrase R1=10 R2=50 defines the transformer STEP 1.8MHZ 2.0MHZ 1KHZ as presenting 10 W to the network when its secondary is terminated ESTP 0.5MHZ 30MHZ 300 by 50 W —which it will be when we plot OUT240PF’s response as the END S-parameter MS using ARD’s default 50+j0 W terminations.Note: OPT 21 Use this block in addition to, not in place of, the ONEPORT block in ONEPORT Table 1. F=1.9MHZ RZ11=50 IZ11=0 TERM=1E-5 Table 3 END Simulating a T-Network Tuner Feeding a Reactive Antenna at 7.2 MHz BLK button. “Optimization complete,” responds ARRL Radio Designer. CAP 1 2 C=?20PF 100PF 240PF? Q=1000 ; C/IN\ “Do you want to update circuit?” Click Yes (if you don’t want to IND 2 0 L=?0.1UH 30.2528UH 35UH? Q1=200 F=2.52MHZ;L keep and use the updated values, click No), and ARD updates each CAP 2 3 C=240PF Q=1000 ; C/OUT\ optimizable component’s initial nominal value with the new value ONE 3 0 ANTENNA found by the optimizer. A cheerful “WARNING - REANALYSIS ONEPORT:1POR 1 IS REQUIRED AFTER OPTIMIZATION” message appears in the END screen’s lower left corner. When you see this, just click the Ana FREQ button or press F10. STEP 7.1MHZ 7.3MHZ 1KHZ If we want to determine the tuner’s network’s frequency re- ESTP 1MHZ 30MHZ 300 END sponse, we need to reconfigure it as a two-port network to see how OPT its amplitude response varies with frequency. Table 2 shows how ONEPORT to do that, and Figure 3 graphs the analysis results based on this F=7.2MHZ approach. RZ11=50 IZ11=0 Specifying Reactive Antenna Loads TERM=1E-5 So far, we’ve used optimization only in transforming purely END resistive loads to 50 W resistive—50+j0 W . In the ONEPORT block DATA for our T-net circuit file, we specified the load with the netlist line ANTENNA: Z RI RES 3 0 R=10—a 10+j0-W load. Often, however, the antenna- 60 +60 system load we want to transform to 50+j0 W will be reactive. Its END January 1995 81 j (imaginary) part will be a nonzero positive or negative number. ware for RF Circuit Simulation and Analysis,” QST, Oct 1994, pp 21-26. How do we specify that? 2The “oops” is that I misspecified one of the network’s capacitors in the netlist. ARRL Radio Designer found a match with C , shown in the If we happen to know what series or parallel combination of article’s Figure 9 as connected to common on the left sidPe of L , actually S resistance and capacitance (or inductance) will exhibit that load at connected to common from the junction of L and C . So the network S S our frequency of interest, we can connect them between nodes 3 I actually optimized was: and 0 of ONEPORT and optimize away. If we don’t, ARRL Radio Designer’s ONE element, a programmable one-port “black box,” can serve. ONE is a neat solution because we can specify its characteristics in terms of S, Y or Z parameters. Using ONE is a bit more involved than just coding in a few components, however, because we need to specify it and its impedance data in the netlist. That means our netlist will have to include a DATA block. Table 3 puts it all together. The netlist line ONE 3 0 ANTENNA connects a ONE element between nodes 3 and 0 and labels it and not exactly the tuner described by Rohde. Having reoptimized the circuit after finding and fixing that error, I can now report that the good ANTENNA. The label is important because ARRL Radio Designer news is that Rohde’s network can indeed transform a 1.6–j2246 W load needs it to link this particular ONE—a more complex netlist might to 50+j0 W at 1.83 MHz if modified as shown above, but that the bad contain several different ONEs, after all—with its data in the DATA news is that it can’t transform that load to 50+j0 W at 1.83 MHz as it was originally specified by Rohde. Live and learn. The corrected circuit file block: is now part of the information file ARD.TXT, available from the ARRL DATA BBS (203-666-0578) and the ARRL Technical Information Service’s ANTENNA: Z RI Internet-accessible information server ([email protected]). 3If you’ve set ARD’s Auto Reporter to pop Report Editor or Quick Reporter 60 +60 after every analysis, the one you’ve selected will pop before the Opti- END mization dialogs appear, and you may even have to close it or move it This DATA block tells ARD to provide impedance (Z) data in to work with the Optimization dialogs. Frankly, Auto Reporter irks me— I can’t stand that big Linear Reports dialog popping when all I want to real-imaginary (RI) form for the element labeled ANTENNA—in do is load predefined reports, which pressing F7 after analysis handily this case, an inductively reactive load of 60+j60 W at all frequen- facilitates—so I keep Auto Reporter set to None. cies.5 The modeling’s yours to do, but here’s a hint: The optimizer 4Well, okay—ARD’s optimizer also stops if it completes the specified won’t reach its goal with C fixed at 240 pF. You’ll have to try number of iterations before finding a solution. The thing to do then— OUT depending on whether you think a solution is possible—is to specify a lower value, or make COUT optimizable. more iterations. If the error function hasn’t been dropping through many iterations, a successful solution may not be possible within the value Free Electronic Goodies limits set by the netlist. 5If we don’t specify a frequency for the ONE data, ARRL Radio Designer You can get expanded versions of the Table 1 and Table 3 assumes that we mean “at all frequencies”—and this is important. Just netlists, as well as the one I used to generate Figure 3, in the file as the purely resistive tuner load we’ve specified so far is unrealistic EXRF9501.TXT, available from the ARRL HQ BBS (203-666- because no antenna exhibits a constant, pure resistance across even a small frequency range, a reactive load that doesn’t vary with frequency 0578) and Internet info server ([email protected]). As always, the is also unrealistic. ARRL Radio Designer lets us specify frequencies for latest ARRL Radio Designer news is also available from those black-box-element data, so if we happened to know our antenna’s actual sources in the file ARD.TXT. impedance across the 1.8 to 2.0-MHz range, we could exactly specify it in the DATA block with multiple frequency-specific lines. Note, though, Notes that it’s perfectly okay to use a frequency-nonspecific impedance value 1David Newkirk, WJ1Z, “Introducing ARRL Radio Designer: New Soft- in adjusting/optimizing a tuner at any one frequency. ers, antennas and accessories to consider for your home listening New Books station, and on collecting QSL cards and other mementos and souve- nirs from SW broadcasters. Amateurs who operate HF probably al- ready know most of the necessary theory regarding HF antennas, reception and equipment, although you may find a few pointers about the slightly different requirements of setting up an effective SW SHORTWAVE RADIO LISTENING FOR BEGINNERS monitoring station. and It’s also gratifying to see her mention the growing use of ham THE SHORTWAVE LISTENER’S Q & A BOOK radio in schools and through the US space program, such as shuttle By Anita Louise McCormick, KA8KGI communications. The educational aspects of Amateur Radio are important, especially in a book for listeners, because stumbling Published by TAB Books, Division of McGraw-Hill Inc, Blue Ridge across ham communication while tuning for other radio services is Summit, PA 17294-0850; 91/4· 71/4 inches, black-and-white photos often a young person’s first exposure to Amateur Radio. and drawings. Shortwave Radio Listening for Beginners, First edi- The final chapter covers radio waves, propagation and other such tion 1993, 176 pp, $10.95; The Shortwave Listener’s Q & A Book, information that may help a listener better understand how and why First edition 1994, 144 pp, retail price $12.95. he can hear different signals at different times, and help determine Reviewed By Brian Battles, WS1O the best antenna(s) and listening times to hear the stations of interest. QST Features Editor Three adequate appendices list SW stations, radio listening clubs With the publication of these two books, it appears that Anita and radio sources. There’s a handy glossary and reasonably compre- McCormick is establishing herself as America’s “First Lady of hensive index. Shortwave Listening.” She certainly logs an enormous amount of Although it’s complete as a standalone work, The Shortwave shortwave radio listening (SWLing) time from her home in Hunting- Listener’s Q & A Book is a handy companion volume. It depends on ton, West Virginia, and she’s enthusiastic and knowledgeable in whether you prefer a straight text or question-and-answer format. It her work to bring the hobby into the home of anyone who reads her covers essentially the same topics, but organized in a style that might books. be more appealing to some readers. Shortwave Radio Listening for Beginners is a clear, nontechnical Both books are illustrated with many black-and-white photo- introduction to SWLing for anyone who has yet to venture into the graphs and helpful diagrams, and the writing style is friendly and airwaves outside the amateur bands. Anita briefly covers the begin- nontechnical. So if you know a new ham or nonham who has yet to nings of radio and a little bit of broadcasting history, and then goes be introduced to listening to shortwave radio and scanners, one or on to discuss domestic AM broadcast DXing, VHF/UHF scanning, both of these books could make a fine gift…or as a loan from your ham radio and more. She lists popular frequencies and modes for personal radio library. At just a little more than 10 bucks each, I rate worldwide reception of broadcasters, utility, relay, government, these books a good value for someone starting to get curious about pirate and clandestine stations. There are tips on the kinds of receiv- the signals waiting all across the busy RF spectrum. 82 Exploring RF Conducted By David Newkirk, WJ1Z Senior Assistant Technical Editor e-mail: [email protected] Modeling a Direct-Conversion Receiver’s Audio Response and Gain with ARRL Radio Designer You can just about go bonkers in getting started with a power- version or transducer action, it can make itself pretty useful in ful tool like ARRL Radio Designer1 if doing so means doing too helping us understand, design, and modify the linear subsystems many all-new things at once. That’s one of the reasons why ARD in a direct-conversion radio. Roger Hayward’s “Ugly Weekender” ships with seven example circuits:2 receiver (June 1992 QST4)—a sound, friendly direct-conversion design for 7 MHz—is a good example for study because most of •a double-tuned-circuit filter for 13.975 to 14.475 MHz its active devices are easily modelable discrete transistors. (I per- •a gang-tuned double-tuned-circuit filter for 80 meters sonally like the Ugly Weekender because it’s easy to build, uses •a simple JFET MF/HF preamp no hard-to-get parts, and sounds like a real radio. Connect it and •a BJT post-mixer amplifier with feedback its companion 1.5-W transmitter to a low half-wave dipole just •an op-amp audio filter after lunch some Saturday or Sunday, and you’ll have to work •a six-crystal ladder filter for 3.579545 MHz hard not to contact someone on QRP CW.) •a 9-MHz diplexer The entire Ugly Weekender Receiver (UWR) schematic The examples are there to help you learn the ARRL Radio covered a full QST page. Our ARDized UWR schematic (Figure Designer way of doing things, of course, but they are also in- 1) takes less space because we need to model only four of the tended to demonstrate that ARD can do a good job of telling you original’s 11 transistors—just those that amplify and band-limit things you already know. (Maybe you don’t check each year’s the UWR’s audio. Table 1 shows Figure 1 in ARRL Radio new White Pages to see if your phone number is correct, but I use Designer netlist form. that simple test to determine whether or not there’s hope for the rest of the book!) Once you’re satisfied that a new tool can do the Netlist Highlights old jobs you know, you’re more likely to trust it to do new jobs Structure your old tools can’t do. This month’s modeling example—simulating the audio chan- How we structure an ARRL Radio Designer netlist for a par- nel of a heterodyne direct-conversion receiver3—is another ticular circuit depends somewhat on what we want to know about potential ARD trust-builder. Many of us have built heterodyne the circuit we’re modeling. We know, for instance, that a direct- direct-conversion receivers, so we know what they do: They het- conversion receiver containing little or no RF amplification—the erodyne RF signals to AF, amplify the resulting audio, and shove UWR uses none—must be capable of something like 80 to 100 dB it at us through electroacoustic transducers (headphones or speak- of audio gain. We also know (by carefully reading citation trails ers). (Actually, this is also all a superheterodyne receiver ulti- through several articles’ worth of footnotes) that the UWR’s first mately does. The difference is that a superhet also does one or audio preamp was designed to terminate the receiver’s double- more RF-to-RF conversions before completing that final RF-to- balanced diode mixer in something reasonably close to 50 W , AF conversion.) resistive, at least in the audio range. Since we’re out to see whether Although ARRL Radio Designer can’t model frequency con- ARRL Radio Designer can correctly tell us things we already know, what we already know is what we want to find out. The Table 1 netlist represents the Ugly Weekender Receiver in 1Notes appear on page 78. three circuit blocks—MIXER, Q8-9-10 and Q12—so we can Figure 1—We need concern ourselves with only four of the Ugly Weekender Receiver’s 11 transistors in modeling its basic perfor- mance with ARRL Radio Designer. I added the circled node numbers as I coded the circuit into an ARD netlist (Table 1). 76 From March 1995 QST © 1995 ARRL Table 1 Simulating the Ugly Weekender Receiver’s Audio Amplifier with ARRL Radio Designer BLK ; this is MIXER (Figure 1A) RES 1 0 R=150 RES 1 2 R=36 RES 2 0 R=150 MIXER:2POR 1 2 END BLK ; this is Q8-9-10 (Figure 1B) CAP 10 0 C=0.1UF ; C30 CAP 10 11 C=10UF ; C27 BIP 13 12 11 A=0.99 RE=(26/0.636); Q8 RES 12 0 R=10KOH ; R30 CAP 13 0 C=10UF ; C28 Figure 2—Wes Hayward’s NPNBIAS.EXE, one of over 20 utilities RES 13 0 R=22KOH ; R33 on the software disk shipped with ARRL’s reissue of his RES 13 0 R=100KOH ; R32 Introduction to Radio Frequency Design book, provided me with CAP 12 0 C=0.1UF ; C31 the collector currents coded into the BIP netlist lines in Table 1. BIP 12 14 15 A=0.99 RE=(26/0.5); Q9 RES 15 0 R=10KOH ; R36 CAP 15 0 C=10UF ; C32 RES 14 18 R=4.7KOH ; R35 RES 18 0 R=1KOH ; R34 CAP 18 0 C=10UF ; C33 CAP 14 0 C=0.1UF ; C34 BIP 14 18 16 A=0.99 RE=(26/0.886); Q10 RES 16 0 R=10KOH ; R37 CAP 16 17 C=10UF ; C38 Q8-9-10:2POR 10 17 END BLK ; this is Q12 (Figure 1C) RES 40 0 R=5KOH ; R38 (GAIN control at max) CAP 40 41 C=10UF ; C39 RES 41 42 R=1KOH ; R39 RES 42 0 R=10KOH ; R40 RES 42 44 R=10KOH ; R41 CAP 44 0 C=0.1UF ; C40 RES 44 43 R=47KOH ; R42 BIP 42 43 0 A=0.99 RE=(26/1.919); Q12 RES 43 0 R=3.3KOH ; R43 CAP 43 45 C=10UF ; C41 CAP 45 0 C=0.01UF ; C42 Q12:2POR 40 45 END Figure 3—ARRL Radio Designer can evaluate voltage gain in several ways. This graph merely expresses in decibels the ratio BLK ; this block chains MIXER, Q8-9-10 and Q12 of the UWR’s output voltage (at the output terminals of the Q12 MIXER 10 20 block, terminated in 2000 + j0 W , the circuit’s anticipated Q8-9-10 20 30 headphone load) to input voltage (at the input terminals of the Q12 30 40 MIXER block, with MIXER’s input terminated in 50 + j0 W )—just SYSVGAIN:2POR 10 40 the sort of measurement we’re implying when we say that a END direct conversion receiver must have “80 to 100 dB” of audio FREQ gain. Our model predicts that the Ugly Weekender Receiver’s ESTP 20HZ 20KHZ 500 maximum audio gain—without its optional LM386 audio power END amp IC—is just under 94 dB at its passband peak. FREQ ESTP 20HZ 20KHZ 500 zero in on the radio’s subsystems. To model how these blocks END work together as one big system, we chain them end to end (using a new set of arbitrary node numbers that happen not to duplicate Transistors those of the constituent blocks) in a fourth netlist block: The Table 1 netlist models the Ugly Weekender’s four 2N3904 bipolar junction transistors (BJTs) with ARD’s BIP element. Be- BLK cause our modeling goals are simple—we’re pretty much after MIXER 10 20 just gain and impedance, and at audio frequencies to boot—we Q8-9-10 20 30 can get away without specifying any more than the transistors’ Q12 30 40 alphas and emitter resistances, which happen to be the only two SYSVGAIN:2POR 10 40 BIP parameters ARD absolutely can’t live without. END An alpha of 0.99—corresponding to a beta of 100—is a safe ARRL Radio Designer will use this block to calculate the Ugly assumption for a garden-variety small-signal transistor like the Weekender’s overall audio voltage gain, so we’ll name it 2N3904. For emitter resistance (RE), we insert a formula (26 [(cid:176) C, SYSVGAIN. that is—room temperature] ‚ collector current in milliamperes) Finally, we tell ARD to calculate these four blocks’ perfor- that comes from the well-established transistor model developed mance at 500 exponentially stepped frequencies from 20 Hz to by Ebers and Moll. Yanked out of Table 1 and grouped, the UWR’s 20 kHz: four 3904s look like this: March 1995 77 Figure 4—The Ugly Weekender’s first audio preamp, Q8, is Figure 5—Finally, in five acts, a “what if?” detective vignette from biased to terminate the radio’s mixer with a resistive load the tattered casebook of Inspector ARD: (1) If Q8’s input imped- reasonably close to 50 W across the span of common CW ance is supposed to be flat and resistive across our audio range receiving pitches. ARRL Radio Designer can report complex of interest, how come its magnitude (Figure 4’s MZ curve) takes 11 impedances in magnitude and R + jX (that is, real-imaginary) off so noticeably below 1 kHz? (2) Hey, the imaginary part of Q8’s form, so this graph shows Q8’s input impedance in terms of input impedance—Figure 4’s IZ curve—pretty much tracks the magnitude (MZ , triangular marker, 0 to 160-W scale), resistive reciprocal of that MZ rise over1 1the same range! (3) Hmm, that component (RZ11, square marker, 0 to 160-W scale) and imagi- IZ curve shows only11 negative values. By convention, we ascribe nary (reactive) c1o1mponent (IZ , diamond marker, 0 to –160-W ne1g1ative numbers to capacitive reactance, so the culprit must be 11 scale. a capacitor between the circuit’s input terminal and Q8’s emitter. (4) That would probably be C27, shown as 10 m F in Figure 1. Come to think of it, 10 m F does seem a bit small for a part that’s supposed to act as a low-Z series element down to a few hundred hertz. So what would happen if we made C27 100-m F instead? (5) Answer: a Q8 input impedance that’s almost purely resistive BIP 13 12 11 A=0.99 RE=(26/0.636); Q8 from 100 Hz to 10 kHz. Case closed! BIP 12 14 15 A=0.99 RE=(26/0.5); Q9 BIP 14 18 16 A=0.99 RE=(26/0.886); Q10 BIP 42 43 0 A=0.99 RE=(26/1.919); Q12 We could hand-calculate the transistors’ collector currents, of 3I used the adjective heterodyne here to give us an imperfect start in course—a drag—or snip our own Ugly Weekenders’ 2N3904 differentiating among the various types of direct-conversion receiv- collector leads and measure the currents (highly recommendable ers, which have actually always included “crystal radios” and regen- to keep your modeling on the rails, if you can take the time). Short erative receivers (oscillating or not). Any receiver that converts RF directly to audio is a direct-conversion receiver, our tradition of using on time, I instead used NPNBIAS.EXE (Figure 2), one of the 20+ the term more narrowly notwithstanding. utility programs included on the software disk shipped with 4Roger Hayward, KA7EXM, “The ‘Ugly Weekender’ II: Adding a Junk- ARRL’s hot-off-the-press reissue of Wes Hayward’s Introduc- Box Receiver,” QST, Jun 1992, pp 27-30. tion to Radio Frequency Design book, to give me ballpark collec- tor-current numbers. Modeling Results New Products Figures 3, 4, and 5 tell the rest of the story. Our Ugly Week- ender Receiver model predicts realistic audio gain—just below 94 dB, max—and a useful degree of AF bandwidth limiting. And PC PakRatt 2.0 AIMED AT WINDOWS POWER USERS Q8, the UWR’s post-mixer preamp, does indeed exhibit a reason- (cid:224) PC PakRatt for Windows 2.0 from Advanced Electronic Applica- ably resistive input impedance that’s reasonably close to 50 W tions is a full-featured Windows-based control program that’s aimed across the audio range of interest. at power users of AEA’s entire line of multimode data controllers Now you’ll have to excuse me—this modeling has got me so (PK-900, DSP-1232, DSP-2232, PK-88 and the PK-232MBX). revved up about the Ugly Weekender that I think I’ll go contact Version 2 also supports the PK-96 1200/9600-bit/s packet controller somebody with the real thing! and the PK-12 1200-bit/s packet controller. All PakRatt features are just a mouse-click away: Run two mul- Free Electronic Goodies timode controllers at the same time; try HF and VHF packet, AMTOR, You can get expanded versions of the Table 1 netlist in the file Baudot RTTY, Morse code, ASCII, NAVTEX, PACTOR (with ANSI EXRF9503.TXT, available from the ARRL HQ BBS (203-666- graphics); or let your multimode box figure out what that mysterious 0578) and Internet info server ([email protected]). As always, the signal is (signal analysis mode). latest ARRL Radio Designer news is also available from those Other features: simplified mail-drops; file transfers; split-screen sources in the file ARD.TXT. operation; macros; built-in QSO logging; compatibility with AEA’s Log Windows 2.0 software; and more. Requirements: At least one of Notes the above-mentioned AEA multimode data controllers and an IBM- 1David Newkirk, WJ1Z, “Introducing ARRL Radio Designer: New compatible computer running Microsoft Windows. Software for RF Circuit Simulation and Analysis,” QST, Oct 1994, pp 21-26. Price: $129; available from AEA or your local Amateur Radio 2Seven circuits, but eight examples. ARRL Radio Designer’s Example dealer. Upgrades for users of version 1.0 are available from AEA. 2 is just Example 1—the 20-meter filter—with at least one error inten- Contact Advanced Electronic Applications at PO Box C2160, tionally added to each of its netlist lines to give you experience in debugging a buggy ARD circuit file. Lynnwood, WA 98036; tel 206-774-5554, fax 206-775-2340. 78 Conducted By David Newkirk, WJ1Z Exploring RF Senior Assistant Technical Editor e-mail: [email protected] Math in a Box, Transistor Modeling, and a New Meeting Place Someone told me that each equation I included in the book in and get whatever value out of it we can. would halve the sales. I therefore resolved not to have any This is a terrific time to make peace with math because this is equations at all. In the end, however, I did put in one the time of the personal computer. We use computers to do so famous equation, E = mc2. I hope that this will not scare many things that it’s easy to forget that all they ultimately do is off half of my potential readers. math. Whatever a computer does—e-mail, processing words, bal- —Steven J. Hawking, in acknowledgments to his book ancing a checkbook, stress-analyzing an aircraft wing, wafting A Brief History of Time you on a cybertrip down a virtual Amazon—it does by crunching numbers. If you can represent a thing or the action of a thing The TV you watch, or the radio transceiver which with you numerically—that is, if you can model it mathematically—you talk to the world, is a math machine—math put to work in the form can simulate it and its action with a computer. of a kinetic sculpture built from copper, carbon, plastic, silicon Since radio-electronic circuitry just does math, a computer and steel. You feed radio (or audio) and ac-line energy—both of can readily simulate, predict and analyze the action of radio-elec- which can be readily described numerically—in. The TV or radio tronic circuitry.1 What makes a computer a vastly more powerful performs mathematical operations on those inputs, and a desired, tool for doing this than a slide rule or a simple calculator is that quantifiable result comes out: a sound, a moving picture, a dec- it can contain math in addition to just solving it. You can use a laration of rook to queen’s bishop three from a chess opponent in computer to determine, say, a coil’s reactance (ac resistance) even Spain. if you know absolutely nothing about how to do it with math. All Because radio-electronic circuitry “just does math,” math can you do is run a program that contains the right equation(s) and lets predict and analyze the action of radio-electronic circuitry. Until you interact with them in a suitably friendly way. You tell the recently, enjoying the fruits of heavier-duty radio math pretty computer your coil’s inductance and the frequency at which you much led to two choices: become fluent in enough math to do your need to know its reactance, and the computer tells you the answer. own designing, or be content with someone else’s designs. The This is why we’re so excited about putting ARRL Radio first choice leads smack into the field of radio engineering, which Designer2 into the hands of radio hobbyists. Quite a few hams isn’t for everyone. The second choice amounts to little more than would like to tinker with the nuts and bolts of radio-electronic just consuming radio products engineered by someone else— circuitry in ways that their modest math fluency just doesn’t al- another entirely acceptable approach to a radio hobby. Crossing low. They know many of the issues, concepts and quantities in- from designer to user is relatively easy—just relax and float down- volved in RF and electronic design, but aren’t comfortable with stream, preferably without tools—but moving from user toward manipulating them numerically. For them—as well as for math- designer generally involves a growing struggle with math. fluent hams who want to spend less time doing busywork—ARRL Math is actually a language—invariably a second language— Radio Designer is a natural. that’s greatly more abstract than any mother tongue. The snag is I’m not saying that ARRL Radio Designer just hands over the that, by tradition, the math education we receive in our formative keys to the radio-engineering castle for $150 plus shipping. What years rarely reflects this fact. Our introduction to numbers, count- the program does do is stuff a large array of muscular radio- ing, addition, subtraction, multiplication and division proceeds electronics math—math you can use regardless of how well you smoothly (one might even say naturally) enough. But for many of can speak it—into an affordable package. The more you already us—probably most, if Steven Hawking’s aside about equations in happen to know about radio and electronics design, the faster and A Brief History of Time is a trustworthy indicator—a point even- farther you’ll be able to go with ARD. Backing up your ARRL tually comes beyond which we just don’t “get” how some string Radio Designer explorations with reading in solid textbooks like of alien, inorganic gibberish (say, y = mx+b) can possibly mean ARRL’s reissue of Hayward’s Introduction to Radio Frequency what its English supporting text says it means (“the definition of Design can brighten your light even faster. a line”). Remedial instruction fails, we may get another dose of More on Transistor Modeling in ARRL Radio Designer, that slow-acting poison—”anyone can do this if they really try” Part 1 (struggling Morse code learners, does this ring a bell?)—real life or an acceptable substitute beckons, and we bolt for the door. If modeling a circuit required us to exactly simulate the perfor- However sordid our educational pasts, many of us survive mance of the real thing, circuit simulators wouldn’t be the major Amateur Radio license exams, slog our way through to a license, industrial and educational tools they are. Simulating even some- and find this hobby a blast. Some of us—by no means all of us— thing as simple as a one-transistor 2-meter preamp would be so discover in ourselves that fiddling with what’s inside a radio box complex and painful that it wouldn’t be worth doing. Luckily, can be quite a kick. Just building radios from scratch or kits may much as the NTSC TV system can simulate moving pictures by satisfy this discovery. To others, a little reading here and a bit tossing 525-line stills at us at the rate of 30 per second, simulated of tentative calculator poking there—successfully scaling a circuit behavior can be considerably grainier than a circuit’s real- 223-MHz antenna to 440 MHz, say—may suggest that the time life performance and still return news we can use. for Peace Talks with Math may have come. For example, in simulating a direct-conversion receiver’s audio Suddenly, math’s prickly language seems to be talking about channel in March 1995 Exploring RF, we used ARRL Radio things that count: deepening a TVI filter’s stopband loss until the neighbors resume smiling, or flattening out an amplifier’s phase 1It gets better: Since radio-electronic circuitry just does math, and a response so it doesn’t cream 9600-baud data. Now math’s alien computer just does math, a computer—properly configured and pro- tongue refers to ideas we can wrap our minds around—problems grammed, of course—can also do the work of radio-electronic cir- that get past our math antibodies because they’re not just made up cuitry. We call this digital signal processing, and it’s an increasingly big deal. like math-textbook examples. It’s still tough, because it’s just not 2Available from ARRL Publications Sales for $150 plus shipping. See our thing and probably never will be. But now we’re going to dig the ad elsewhere in this issue. 90 From May 1995 QST © 1995 ARRL want—by its actual part number, such as 2N5179, 2N3819 or 1N914—from a menu, and the program duly adds the correspond- ing code and device parameters to your netlist. ARRL Radio Designer lets you do something a little different. Instead of device libraries, it comes with device databanks— manufacturer-supplied ASCII files that convey, in S-parameter (scattering parameter) form, the actual measured performance of RF devices operating under specific conditions. You can use this data directly or indirectly. The direct method involves plugging the databank data directly into ARD’s two-port [TWO] black-box element and using that TWO instead of a BIP or FET. The indirect method makes use of databank data and ARD’s optimization en- gine to adjust the values of a BIP or FET’s many parameters until its performance matches the measured performance of an actual device as reflected in databank data.4 S-Parameter Basics S parameters reflect a standardized way of characterizing how a device behaves in response to signal energy applied to its ports— its signal inputs and outputs—usually with all of its ports termi- nated in identical, standard impedances (commonly, 50 W resis- tive). A transistor, for instance, is a two-port device. By Figure 1—ARRL Radio Designer’s bipolar junction transistor convention, the ports are labeled with numbers, Port 1 being the (BIP) element allows you specify a BJT’s internal structure in great detail. You can safely accept ARD’s defaults for many of input and Port 2 being the output. these values when modeling at audio and low radio frequencies, Signal energy applied to one port of a two-port device comes but supplying true-to-life detail for accurate modeling at higher out two places: at the same port (the device reflects some of the frequencies is a challenge in simulating with ARD. Most of the energy back to the generator) and at the other port. How much device characteristics listed in transistor databooks can’t be signal comes out relative to the applied signal tells us the device’s directly plugged into this model! gain (which can be negative; that is, a loss); how much signal reflects back out tells us something about the impedance match between that port’s impedance and our signal generator. Deter- Designer’s BIP (bipolar junction transistor) element to supply only mining the phase of the output or reflection signals relative to the the roughest possible sketches of the circuit’s four 2N3904 transis- phase of the applied signals tells us even more about the device tors.3 But if you’re already been thumbing through The ARRL Radio under test. Designer Manual, you know that the BIP element can handle much Figure 2 shows this idea graphically. An S parameter is a volt- more detail. Take a deep breath and look at Figure 1. age ratio (commonly, but not always, expressed in decibels) Precise active-device modeling lies at the heart of any circuit annotated with two subscript numbers that indicate the ports in- simulator worth using, so a radio engineer might wade right into volved. For instance, a device’s forward transmission gain, S21 the complexity of ARD’s BIP model with a bleat of pure joy. But many of us will wonder how we’re ever going to model anything 4Although I’ve been mentioning only S parameters, ARD can also handle much more complex than a p attenuator if Figure 1 is any indica- TWO data in Y or Z-parameter form. tion of what we’ll need to succeed. A quick glance at the nearest transistor data manual seems to seal our doom: Of all the data listed for a particular type, almost none of it—if any—consists of numbers than we can plug directly into an ARD netlist! Some circuit simulators—some of the SPICE variants come to Table 2 mind—approach this problem by providing device libraries con- Two-Port S-Parameter Data Equivalent to a 2N3553 taining pre-coded device parameters. You select the part you Transistor (Operating at V =28 and I =30 mA) from the CE C Motorola File 2N3553A.S2P 3We—uh, that is, I—also supplied a first-rate howler when I wrote in that 0.045 0.76 –165 10.39 96 0.03 38 0.23 -87 column that the 26 in the emitter resistance formula (R =26‚ I ) con- veys “(cid:176) C, that is—room temperature.” Since that formEula isC just a 0.046 0.75 –166 10.09 95 0.03 46 0.23 -87 variation on Ohm’s Law, the 26 actually means 26 millivolts, the room- 0.047 0.77 –166 9.84 94 0.03 48 0.22 –88 temperature value of V , which is the thermal equivalent of voltage in 0.048 0.75 –166 9.78 94 0.02 44 0.22 –87 T the transistor’s semiconductor material. 0.049 0.77 –166 9.67 94 0.03 45 0.22 –89 0.05 0.76 –167 9.39 94 0.03 51 0.21 –88 0.051 0.78 –168 9.23 93 0.03 51 0.21 –89 0.052 0.76 –168 9.04 93 0.03 49 0.21 –88 0.053 0.77 –167 8.92 92 0.03 52 0.2 –87 Table 1 0.054 0.77 –168 8.7 92 0.03 44 0.21 –87 Two-Port S-Parameter Data Equivalent to an MRF586 0.055 0.76 –168 8.66 92 0.03 49 0.21 –85 Transistor (Operating at V =10 and I =90 mA) from the CE C Like the .FLP-format data shown in Table 1, each of these lines ARRL Radio Designer Databank File MOTOROLA.FLP conveys frequency and the S parameters MS , PS , MS , PS , 11 11 21 21 100MHZ .36 -123 15.68 107 .05 57 .44 –77 MS12, PS12, MS22 and PS22. The difference between this data format 300MHZ .33 180 5.78 83 .1 61 .32 –117 (that used by EESOF’s Touchstone circuit simulator, and one of 500MHZ .34 154 3.44 70 .15 59 .39 –122 several S-parameter data formats now standard in the RF CAD 1000MHZ .31 118 1.84 43 .25 51 .49 –133 community) and that of Table 1 is that an .S2P file conveys frequency in gigahertz but does not include frequency units. Because ARD Each dataset line conveys frequency and the S parameters MS11 assumes that you’re talking hertz if you specify frequencies without (magnitude of S11), PS11 (phase of S11), MS21 (magnitude of S21), PS21 units, you must add the suffix GHZ, with no leading space—0.045GHZ (phase of S21), MS12 (magnitude of S12), PS12 (phase of S12), MS22 instead of 0.045 GHZ—to each frequency spec in any .S2P data you (magnitude of S22) and PS22 (phase of S22). use in an ARD netlist. May 1995 91 ARRLCAD: A Meeting Place in Cyberspace If you can send and receive Internet e-mail, you can now share your questions, answers, ideas and views about ARRL Radio De- signer and other Amateur-Radio-related circuit and antenna de- sign and simulation tools with other users. Thanks to the generous provision to ARRL of facilities and volunteer support by TAPR, Tucson Amateur Packet Radio Corp, I’m pleased to announce the startup of ARRL’s computer-aided-design e-mail reflector, ARRLCAD. You may already know of e-mail reflectors as mailing lists. A mailing list automatically sends copies of any messages sent in by Figure 2—S parameters corresponding to input reflection, its subscribers back out to all of its subscribers. It’s somewhat like forward transmission gain, reverse transmission gain and output reflection can quite closely characterize a small-signal linear a UseNet newsgroup, the difference being that you don’t need device—in this case, a two-port device. The text describes one access to UseNet newsgroups to participate. All you need is the way you can use ARD’s TWO element to insert manufacturer- ability to send and receive Internet e-mail. supplied S-parameter data directly into your netlists. To subscribe to ARRLCAD, send an e-mail message to [email protected] with text that reads (“S sub two one”), is the ratio of the voltage at Port 2 to the voltage subscribe arrlcad FirstName LastName applied to Port 1—a value that must be expressed as a vector to convey the two signals’ relative phase. To discuss S parameters in which FirstName and LastName mean exactly that. (My sub- more readily and to communicate them in tabular form, we split scription message read subscribe arrlcad david newkirk, each of the four basic ones into separate magnitude and phase for example.) The reflector software will confirm your subscrip- components: MS (magnitude of input reflection) and PS tion with a welcoming information message. Subscribing to 11 11 (phase of input reflection); MS (magnitude of forward trans- ARRLCAD costs nothing at all. 21 mission gain) and PS (phase of forward transmission gain); 21 ARRL Radio Designer Front-Line Support MS (magnitude of reverse transmission gain) and PS (phase of 12 12 reverse transmission gain); and MS (magnitude of output re- ARRL HQ’s Technical Information Service provides front- 22 flection) and PS (phase of output reflection). line support for ARRL Radio Designer. Just ask for the Technical 22 One big deal about S parameters is that they can cram a lot of Information Service at 203-666-1541 (fax 203-665-7531), or send very realistic information about a device’s behavior into just a email to [email protected]. few standardized numerical values. If we know the S parameters Free Electronic Goodies of a given transistor operating under known conditions, we have a very useful picture of how it looks to the outside world—a This month’s circuit example is so straightforward that it picture we can paste directly into ARRL Radio Designer. This doesn’t warrant the creation of an EXRF9505.TXT file, but as a leads to another big deal about S parameters: RF-device manufac- consolation prize I’ve added to the heap the file ARD9410.TXT, turers commonly make their products’ S parameters freely avail- an ASCII-text version of the October 1994 QST article that intro- able in forms that ARD can easily digest! Tables 1 and 2 show two duced ARRL Radio Designer. To find out how to get it and the very similar S-parameter formats that ARD can handle. files ARD.TXT, EXRF9501.TXT and EXRF9503.TXT from the ARRL Technical Information Service’s automated info server Plugging S-Parameters into ARRL Radio Designer ([email protected]), send the server an e-mail message that reads just Using Table 1’s data to synthesize a transistor in an ARD netlist INDEX is so straightforward that I won’t even show you a complete netlist. HELP Instead of connecting a BIP to nodes 3, 4 and 5, connect a TWO to and you’ll receive instructions by return e-mail. ARD.TXT, those nodes in the same sense and give it a distinctive label: ARD9410.TXT, EXRF9501.TXT and EXRF9503.TXT are also TWO 3 4 5 MRF586 available from the ARRL BBS (203-666-0578) and via FTP from the ARRL subdirectory at oak.oakland.edu. At the end of your file—after your FREQ block’s END state- ment—add the TWO’s S-parameter data in a DATA block: DATA MRF586: S 100MHZ .36 -123 15.68 107 .05 57 .44 -77 New Products 300MHZ .33 180 5.78 83 .1 61 .32 -117 500MHZ .34 154 3.44 70 .15 59 .39 -122 1000MHZ .31 118 1.84 43 .25 51 .49 -133 HAND-MADE KEYS/PADDLES END (cid:224) If a hand-made key that won’t slide around and won’t be damaged The data’s header tags it as S-parameter information intended for by a fall to the floor is in your future, consider the line of keys use with an element labeled MRF586. Additional header informa- and paddles produced by Karl Reiber, N6NXM, of Bakersfield, tion is necessary for Y or Z-parameter data, and for S-parameter California. data derived at impedances other than 50 W . (See pages 15-50 and Karl’s keys, made one at a time by special order, come in vari- 15-51 of The ARRL Radio Designer Manual for detail on the TWO ous sizes and configurations and can be made from steel, brass and element, and pages 17-7 and 17.8 of the Manual for how to specify aluminum. They feature highly polished and tested needle bearings data for TWO.) for smooth operation. Each piece is guaranteed for three years. If This example is simplistic in that its DATA block doesn’t in- he can’t repair your key, N6NXM will build you a new one. clude any numbers ARD can use to predict how a transistor-fla- Contact Titan Communications (ask for Jerry, KD6EAG) at vored TWO element may generate noise. We’ll overcome this limi- 805-399-1321, or the designer, Karl Reiber, N6NXM, 446 tation in the next edition of Exploring RF. Sycamore Dr, Bakersfield, CA 93308; 805-393-5855. 92