NEC-LIST: Problems with PATCHES (SP cards)

I have downloaded from the site http://www.qsl.net/wb6tpu/swindex.html
two different instances of the double precision NEC2 (nec2d) program.

A) nec2d.f, contained in the ZIP file nec2_src_tar.Z; I will label it
nec2d_A, in the following;

B) nec2d.for, contained in the ZIP file nec-pc-x.zip; I will label it
nec2d_B, in the following;

Both of them were compiled without any problem using Digital FORTRAN
compiler 5.0 of Digital Equipment.

These two versions give identical results with geometries containing
only wires, but perform quite differently when the geometry contains
PATCHES (SP cards).

For example, I ran Example 4, as reported in NEC-2 Manual, Part III:
User's Guide, pp. 98-101, (NOTE: similar considerations apply to
Example 6 which also uses patches)

CE EXAMPLE 4. T ANTENNA ON A BOX OVER PERFECT GROUND
SP 0 0 .1 .05 .05 0. 0. .01
SP 0 0 .05 .1 .05 0. 90. .01
GX 0 110
SP 0 0 0. 0. .1 90. 0. .04
GW 1 4 0. 0. .1 0. 0. .3
.001
GW 2 2 0. 0. .3 .15 0. .3
.001
GW 3 2 0. 0. .3 -.15 0. .3
.001
GE 1
GN 1
EX 0 1 1 0 1.
RP 0 10 4 1001 0. 0. 10. 30.
EN

and I obtained quite different results from nec2d_A and nec2d_B. As a
synthetic indicator of these results, I transcribe here the antenna
input impedance :

Nec2d_A: Z=6.09397E+01 -j1.98192E+02 ohm
Nec2d_B: Z=1.80727E+02 +j2.17654E+02 ohm
For the same problem the Manual (p. 99) gives Z=1.80727E+02
+j2.17655E+02.

It is evident that the two results are very different, and that the
second one (Nec2d_B) is practically identical to that of the Manual.

Now, the question is: which one (if any) of the two nec2d program
versions give the correct answer?

To clarify that point, I ran Example 4 at a much lower frequency (in
the above computations, the frequency was 2.998E+02 MHz), where the
input impedance is expected to be essentially capacitive (i.e. real
part very small and imaginary part large and NEGATIVE). I did that
with: FREQUENCY= 1.0000E+00 MHZ and I got these results:

nec2d_A: Z=4.37694E-04 -j1.59564E+05 ohm
nec2d_B: Z=1.38704E-01 +j7.45156E+05 ohm

It is clear that only nec2d_A gives the correct behavior (negative
imaginary part) and nec2d_B is IN ERROR. But nec2d_B, as I have shown
before, is the program version that at 2.998E+02 MHz gives results
equal to those of the Manual. As a logical consequence, I must
conclude that the solution of Example 4 given in the Manual is WRONG.

Does someone can give me a commentary to the above, rather surprising,
conclusion? Did I make some mistake? Is that of patches a yet well
known problem to people working with nec2d? .

Thanks in advance for your answers.

Marco Bini

-----------------------------------------------------------
  Dr. Marco BINI
  CNR-IROE
  National Research Council
  Electromagnetic Wave Research Institute "Nello Carrara"
  Via Panciatichi, 64
  50127 Firenze - Italy
  Ph: +39 055 4235.236; Fax: +39 055 417102;
  e-mail: bini_at_iroe.fi.cnr.it
-----------------------------------------------------------
Received on Fri Mar 24 2000 - 05:59:08 EST