LPCAD30 Software for Designing LPDAs: A Review

L. B. Cebik, W4RNL

or many years, the DOS-based program LPCAD has been used by designers of log periodic dipole arrays (LPDAs). Written by veteran antenna engineer and designer, Roger Cox, WB0DGF, the program has been available at Roger's web site or in the collection of programs included on CD-ROM with The ARRL Antenna Book. I have relied upon versions 2.7 and 2.8 for initial LPDA designs in my own work.

Roger has just released version 3.0 of the venerable program. It is still DOS based, with a Windows version still in the planning stages. However, the program includes a number of improvements and updates that offer the individual LPDA designer many enhancements. The program now segments elements in saved .NEC files in better accord with general NEC modeling requirements. As well, the NEC file output also includes summary specifications as part of the CM or comment structure. The program also calculates the feeder impedance based on user specification of a desired feedline input impedance. In the NEC file, we also find full frequency sweep specifications and requests for both azimuth (phi) and elevation (theta) patterns. Perhaps the most notable change to the calculations is the Tau-tapering of element diameters after the user specifies the diameter of the rear element. The calculations now operate with a constant Length/Diameter (L/D) ratio in accord with the original theory of LPDAs.

The program includes several screens of instructions and explanations that every user should read carefully. Among those pages is a complete log of program modifications with time. For example, version 3.0 no longer produces an .LPA files, but will read older ones. LPA and LPD files are internal to LPCAD and save essential design and calculation information. LPD is an enlarged version of LPA, making the latter superfluous. The instruction set also reveals some of the program maximums. LPCAD will handle up to 60 elements with a maximum frequency of 10,000 MHz. The minimum feeder (phase-line) impedance is 75 Ohms. (To actually use 75-Ohm feeders, one must specify square feeder lines, since round lines interpenetrate in the neighborhood of 80 Ohms.) The instruction set also shows the equations for Tau and Sigma, the key design elements of an LPDA. It also presents the equation for calculating the optimum value of Sigma for any selected value of Tau. In accord with its engineering handbook origins, the program uses for calculations an upper frequency limit that is 1.3 times the upper frequency specified by the user to ensure that there are enough active element forward of the most active element. Hence, calculating the required number of LPDA elements requires an equation based on user inputs, and the instruction set shows how the program does this job.

Among the data in the instruction screens is a table of expected gain and front-to-back values for various ranges of Tau. Tau limits run from 0.80 to 0.96. The lower the value of Tau, the lower the gain potential for an LPDA. For the acceptable values of Tau, standard handbook graphs show a gain range from 5.3 to 10.0 dBi in free space. However, for any selected value of Tau, do not expect gain at the high end of the scale for that value unless you also use the optimum value of Sigma. For a Tau of 0.9, optimum Sigma is 0.17. That value results in a very long and impractical LPDA design. Sigma values between 0.03 and 0.06 are far more common in both amateur and commercial arrays. The result is not only lower gain, but often the need for compensatory modifications to the initial design, a major subject in my 2-volume series of LPDA Notes. The front-to-back ratio for LPDAs is very much tied to their forward-gain potential. However, do not take a single reading within an LPDA's operating range. Instead, sample the entire passband. Gain, front-to-back ratio, and feedpoint impedance all describe undulating curves that only become nearly flat when we use the highest values of Tau in combination with optimized values of Sigma. However, few people can handle an LPDA for 14-30 MHz with a boomlength of over 1200'.

Perhaps the best way to provide a feel for how to use the program is to run through many of the program screens. Fig. 1 shows the main menu. Although you have options to recall existing designs, let's begin by designing an LPDA from scratch. Hence, out first stop is option D. We shall design in inches, although the program offers the option of designing in meters.

The design screen has two forms, depending on whether you design to values of Tau and Sigma or design by specifying the desired number of elements and the boomlength. Fig. 2 shows the screen after selecting a Tau of 0.92 and a Sigma of 0.04. Common to both design avenues is the user input of the lowest used frequency and the highest used frequency. As well, the user specifies the diameter of the rear element. (The program calculates the diameter of all other elements.) The design screen summarizes the initial inputs and calculations for Tau, Sigma, boomlength, and the number of required elements. If we do not like the results of the initial selections--either from the calculated information to come or from modeling exercises from the saved design--we can always recall the file and use the edit function to change something.

Many designers begin with a set of constraints, such as a maximum boomlength and a fixed number of elements. LPCAD allows initial design using these factors, as shown in Fig. 3. Since I used the output of the first screen as the input for the alternative screen, we should expect very similar results, allowing for rounding in the calculations.

I purposely selected a narrow passband (10 meters) for our sample in order to hold down the number of elements in the demonstration design (called "review1"). Otherwise, I could not fit what comes next on a single screen. Once we have accepted the design parameters, the next main menu option is C to calculate all the physical dimensions of the design. Fig. 4 shows this screen through all of the physical element data.

The program splits the screen into element spacing and element length sections. Element spacing show both the distance between elements and the total boom length up to any given element. Sigma gives the initial spacing, but successive spacing adjustments are based on Tau. The lower part of the figure shows the length of each element and its diameter based on the user-specified value for the rearmost element. Both the length and the diameter progressions rest on the selected value of Tau (or the value calculated by initial setting the boomlength and the number of elements. Since the calculations shown result from inputting a boomlength and the number of elements, we find that the adjustment or Tau value is between 0.922 and 0.923 simply by dividing one value by an adjacent value.

Fig. 5 shows an extension of the calculated data screen. The limitations of DOS screens result in scrolling of the previously calculated data, although you have the option of printing out the entire set of results from the main-menu P option. In this section of the design work, we can specify a desired characteristic impedance of the phase line (called the feeder on the screen). The program remembers the most recently used value and offers it. In this case, I refused the offer. Instead I register my preference for a 50-Ohm feedpoint impedance. The program calculated the phase-line impedance that would come closest to this value, namely, the minimum value of 75 Ohms. The calculated feedpoint impedance is 53 Ohms.

The only practical way to achieve a 75-Ohm feedline is with square (or L-stock) phase-lines. Since transferring the design to any form of MININEC requires physical phase lines, I also calculated them, using a 0.75" dimension. Fig. 6 shows the resulting calculation. Although there are more complex methods of calculating the exact characteristic impedance of flat-face transmission lines, the standard approximation used by the program provides results that are accurate for all practical line Zo values within normal manufacturing or construction limits. Few amateurs can produce a line spacing that is exactly and consistently 1.064", whether using independent lines or a dual-boom form of construction.

If we return to the main menu, we can review our design work using the E option. Since that data summarizes what we have seen on the screens so far, let's bypass that step and save the design. Since I tend to use NEC for most of my LPDA test models, let's select S for save from the main menu and N for a NEC file from the save menu. Fig. 7 shows the results of our saving work. I gave the file a name and added only 2 comments. The program will add summary data in other comment lines automatically. The rest of my saving work involved specifying the lowest frequency in the test sweep, the increment between sweep steps, and the total number of steps. The program offers you the number of steps needed to sweep the program from the starting frequency to the highest frequency entered on the basic design screen. For this exercise, that number is the desired one. However, in many LPDA designs, you may specify a top frequency that is considerably higher than the upper end of the passband. Remember that standard calculations use a factor of 1.3 times the specified upper frequency for the calculations. For smoother upper-end responses with lower Sigma values, you may need to specify an upper frequency that is about 1.25 to 1.3 times the actual top operating frequency. The result is a set of calculations for a frequency that reaches 1.6 to 1.7 times the topmost operating frequency. In such cases, you will have to determine the number of sweep steps needed to cover the operating passband.

One limitation of DOS is the difficulty of combining data and graphics on the same screen. Before we look at the NEC file that we just produced, let's pause at the main menu and check the P or print/view option. We may view either the dimensions or the layout of the antenna design, but not both at the same time. Fig. 8 shows the layout with limited data underneath. The layout is quite useful in revealing the general proportions of the array, from which we might make some initial construction decisions.

Although we might close LPCAD 3.0 at this point, our work does not end. So let's leave the program open and return to Windows. Let's open an antenna-modeling program suitable for the model that we just saved. Since the model is in NEC, I opened GNEC and then opened the saved model, review1.nec. The model appears in Fig. 8. The first item to note is the automatic comment lines that summarize the design data in addition to the comment lines that I had added at the program prompts.

The model uses inches in the geometry section, since that was the unit of measure used in the design. The Tau-based progression of element radii also appear. Here, the designer must make some choices, including whether he or she wishes to use the calculated element-diameter values or to replace them with actual planned values. One might independently determine the effective diameters of stepped-diameter elements actually planned for the array. Or one might use a uniform diameter throughout the design. However, at this stage, I would not recommend replacing the uniform-diameter elements with the multiple entries required to implement stepped diameter elements, since we have not yet learned anything about the design's performance. We do not know if the work involved will be worthwhile.

The LPCAD saved model uses lossless or perfect wire for the elements. We may add a conductivity entry for the model either now or later. The FR entry shows the specified sweep, and the two RP entries show the requested E-plane and H-plane patterns for the free-space environment. Advanced versions of EZNEC allow the user to input files in .NEC format. However, since EZNEC is basically a single-frequency program (with a special sweep facility), it will ignore the sweep request and use only the initial frequency. As well, it will ignore the second RP pattern request. Within those limits, EZNEC will accept the design with no faults. For NEC-based programs using the standard command sets, the saved LPCAD model file will work with either NEC-2 or NEC-4.

Now let's learn about the probable operating characteristics of the LPDA that we just designed. If we refer to Fig. 2, we can recall that we set the lower and upper frequency limits to the band edges of the 10-meter amateur band. As well, we used modest values of both Tau and Sigma. Hence, we may not encounter modeling results that approach the levels calculated by standard LPDA design methods. The first clue appears in Fig. 10.

The table of 50-Ohm SWR values shows us that the design does not achieve a 2:1 50-Ohm SWR across the passband. As we move upward in frequency, the resistive component of the feedpoint impedance goes down into the 25-Ohm range. One alternative is to return to LPCAD and to request a higher feedpoint impedance value, hoping that the newly calculated phase-line Zo will achieve a better SWR curve. Alternatively, we can modify the NEC file by raising the impedance of each TL entry. The -75 entries indicate the places to make changes, with the - sign needed to ensure that each phase-line section reverses itself between elements. We might wish to save the file under a slightly revised name. If we examine the SWR table a second time, we find that the SWR, the resistance, and the reactance all undergo--even within the small operating passband--the undulations that I noted as endemic to LPDA antennas.

The gain curve for the initial model appears in Fig. 11. It shows another limitation of using the passband limits as the LPDA design limits. The gain peaks at mid-band and tails off severely at both band edges. Nowhere does it reach the handbook values for a Tau of 0.9, that is a minimum gain of 6.6 dBi. So rather than just playing with the phase-line impedance, we may wish to do a complete redesign of the array, perhaps setting 27.5 and 36 MHz has the design limits. We shall end up with a longer and more heavily populated LPDA, but the performance is likely to improve.

The front-to-back curve in Fig. 12 begins well enough at above 20 dB. However, by the upper end of the band, the value has dropped to a mediocre 14 dB. Again, the average value is well below the standard calculations (about 20 dB) due to the combination of narrow design frequency limits and low values of Tau and Sigma. The curve also illustrates another feature of LPDAs: the gain and front-to-back curves both undulate, but not synchronously.

I have explored a specific mediocre LPDA design for several reasons. One purpose has been to illustrate that LPDA designs are not automatically good, despite that fact that LPCAD performs its calculations with superb accuracy. The initial model, review1, performs little better than a wide-band 2-element Yagi, despite its 5 elements.

My second reason for using the test model was to illustrate why I left LPCAD open while testing the model in GNEC. Neither program is an end in itself. Instead, the programs become synergistic in a multi-step process of design, testing, redesign, re-testing, etc. The user is the vital link in making the program do useful work in terms of winding up with a truly superior LPDA antenna design. Do not expect LPCAD to design the ultimate wide-band amateur antenna on the first try--or perhaps even on the 10th try. Instead, use LPCAD as an important tool to ease the design work without relieving you of the privilege and the duty of making design decisions.

Used wisely, LPCAD 3.0 is an improved tool for the LPDA designer.

-30-

~ antenneX ~ February 2006 Online Issue #106 ~