Excellent summary of what to expect from the world of modeling.
Just to add few notes to illustrate the roadblocks.
There was that big controversy about loaded antennas, coils etc.
please see http://www.k3bu.us/loadingcoils.htm
In the nutshell, EZNEC treated loading coil as a 0 size L inductor, which did
not account for the current distribution along the physical coil and was
"proof" for W8JI and W7EL that the current does not drop along the coil
(quarter wave standing wave radiator case), while my and W9UCW practical
experience and real life tests indicated substantial (~50%) drop.
The major impact on modeling, especially of multi-element loaded arrays, using
"their" assumptions would magnify the effect and errors, giving wrong results.
W7EL eventually updated the EZNEC and allowed physical modeling of the coil by
specifying turns etc.
I am always critically looking at the results of modeling and for unusual
designs I would try to verify that on hardware models, scaled down etc. When I
tried to soft model my Razors (Quad/Yagi arrays) the EZNEC gives different
results than my real life models and actual arrays. Yagi designs are now well
captured, provisions for clamps, taper, mounting hardware can be accommodated
(thanks W2PV).
IMHO Cebik was too much involved with modeling and not much with building real
stuff. He did lot of studies of investigating changes in model parameters,
which is good indication of effect of various changes on the array design. But
careful correlation to real life hardware "models" is always proof of model's
validity and any found deviations should be fed back to "fathers" of modeling
programs to incorporate the nuances to improve the accuracy of modeling.
I found modeling very useful when trying to relate soft models to hard arrays,
is to use the resonant frequencies of elements out of the antenna and then
resonate the hard element before it is inserted in the hard array. This will
account for variations in hardware, like wire, material, clamps, loading
elements etc. Q might change a bit, but element resonance is what is most
important. The final check is to compare the resonant frequency of soft and
hard array models.
Optimization feature is alwasy fun to play with and can point to some
interesting solutions.
Another caution is to modeling in free space. Who runs antennas in free space?
When I was hard modeling my Razors on 2m scale range, I found that antenna
should be designed for the height and ground conditions it will be used at.
Might seem too picky, but when we are searching for that fraction of a dB, it
all ads up.
Effect of sea water is more like 10 - 15 dB in real life vs. some 4 dB
indicated by models.
Step back and take the modeling results with the grain of salt and beware of
"soft antenna gurus".
I am looking forward to design some super antennas for salt marshes of N2EE
www.TeslaRadio.org - plenty of space and quiet location, wanna join the Tesla
Sparks?
Yuri, K3BU.us
> When I first started using modeling software, the temptation was to<BR>>
> accept those results as real and I quickly succumbed. I came up with
> some models that showed excellent gain and put one up.
>
> I quickly discovered that there were some gotcha's in modeling where
> free/inexpensive programs got lost in computer floating point issues.
>
> I discovered that my particular backyard dirt was not the same
> as the
> assumptions in the programs and measurements could vary wildly from
> the model, especially with low-band vertical antennas, that wire in
> and on dirt was still a considerably unsolved problem.
>
> 73, Guy. K2AV
> _______________________________________________
> UR RST IS ... ... ..9 QSB QSB - hw? BK
>
_______________________________________________
UR RST IS ... ... ..9 QSB QSB - hw? BK
|