Re: NEC-LIST: Scattered field from a lossy sphere

Hallo Ed,

Regarding your query on scatterring from a lossy sphere and methods to
get such results:

I believe we could do it fairly easily using the SuperNEC code. You
may have picked up from some earlier replies to nec list queries that
we have successfully used a loaded wire grid to represent a lossy
surface. We have presented a paper on the topic, unfortunately at the
Quebec ANTEM conference, which is not that widely attended. We are
currently preparing a paper to a journal on the topic.

Our methodology essentially was quite simple. You load each segment in
a wire grid model with a resistance equal to the per square resistance
of the real life surface (this gives the correct per square wire grid
value as well as long as you consider a square group of grids,
regardless of size of the square).

We validated the results using rigorous measurements of variation in
dipole input impedance with distance away from a conductive sheet in
small steps starting a 0.05 wavelength spacing (between dipole
parallel to sheet) going away until 2 wavelengths. We used the
following sheets (All 0.7 x 1.1 wavelength rectangular):

a) Aluminium (assumed perfectly conducting)
b) Carbon fibre woven composite (0.5 Ohm per square)
c) Teledeltos paper (850 Ohm per square)

Measurements were done at 350MHz in an anechoic chamber. We simulated
exactly the same situation using the method mentioned above on
SuperNEC. We achieved a mean difference of 1.2 Ohm on real parts and
1.5 Ohm on imaginary over a variation of 45 Ohm in real values and 28
Ohm in imaginary values for Teledeltos (1.41/1.4 diff. over 87/54
range for carbon fibre). I must note that the measurements required
considerable care, but that we achieved this correlation using exact
real life dimensions for simulations (dipole length, radius and sheet
dimensions). Normally the temptation is to "twiddle" dipole
length/radius to get agreement - this is in fact perhaps necessary
with thicker dipoles.

The new SuperNEC interface, of which you have a beta release version,
has a "assembly" which only requires the user to input the sphere
radius and it will generate a gridded sphere. You then just have to
load all the segments with the correct value of resistance per square
to get a penetrable surface representation.

I do believe that volumetric lossy solids can also be treated in this
fashion by making 3-d grids (like crystal lattice). If thicker
materials which are still thin relative to wavelength are required
then two grid layers should work fine as well. It also seems likely
that capacitive loading should be capable of modelling dielectric
properties in a similar fashion. The method really deserves further
study, since I believe it will be most convenient if lossy/dielectric
materials can be handled by simple wire MoM programs.

Regards
Andre Fourie

Tel: Intl + 27 11 4030380
Fax: Intl + 27 11 4030381
Website: http://www.poynting.co.za
email: fourie_at_poynting.co.za
Papermail: Dr APC Fourie, PO Box 318, Wits, 2050, South Africa
Received on Wed Aug 26 1998 - 12:11:23 EDT