art wrote:
The most basic of all programs for antennas come from Roy
They do nothing but number crunching like a calculator and
will give you an answer close to what other programs provide
but not the same. The program does NOT help the user in any
way other than give you an answer regarding the performance
of what you provide.

Which, is, of course, the "sine qua non" for a *modeling* program.. it
should take your model and tell you what the performance will be.

It does NOT give you any help as to where
you could benefit in any way. When you move beyond the most
basic of antenna programes you can obtain help fr4om the
programs in that you don't have to specify actual dimensions
which may be useles because you can alow those dimensions
to be variable to allow the computor to guide you in the right
direction
to meet your desires.

This is an "antenna designing" program. Many such programs will use one
or more "antenna modeling" programs as part of their operation,
although, it's not by any means universal.


The cost of these type programs are similar
to eznec but can go up as high as a couple of thousand dollars
tho most amateurs should be satisfied with the cheapest versions

"antenna design" is a fairly wide field, and people literally spend
their life becoming good at one small part of it. There's a lot of
judgement and skill in antenna design, particularly when it comes to
things like mechanical/electrical tradeoffs and manufacturability.
There are tools designed to address one niche or another (e.g. there's
programs that are designed to optimize electrical performance microstrip
patch arrays, there's programs that are designed to optimize Yagi-Udas,
etc.)

Invariably, such tools are (at least originally) designed to be used by
a person who will do the "higher level" trades (Do I use a rectangular
or circular array of patches? What mechanical tolerances can I hold in
manfuacturing? Can I hold a 5 meter by 4 meter array flat enough to
actually work at Ka-band?)

There is some work on integrating all of these, but it's still baby
steps (for instance, taking a Solidworks model and turning it into a
meshed grid for modeling, or trying to integrate electrical, optical,
and mechanical models for large dishes (e.g. IMOS))

Some programs are designed around the yagi only for simplification.

yes.. back before computers got cheap, people worked out clever
analytical models for certain classes of antennas.. arrays of parallel
straight thin elements would be one that's particularly amenable to such
analysis. No surprise that as computers came to be more common, such
models would be first ones to be implemented.
These ofcourse need to be avoided since they are based on the
yagi being unbeatable.

Not at all. it's that people had equations for Yagis (based on
empirical experience that Yagis worked and met at least some of the
requirements), and people tend to want to work with what is familiar.
If for no other reason than you can compare the output of the modeling
code (or the optimization code) with something you've actually built and
see if it matches (aka validation).

So if a choice has to be made then programs
with variable dimension abilities together with a sufficient large
number of pulses are by far superior toi any other computor program.

Well, sure.. if you're going to any sort of Finite Element analysis (of
which the method of moments methods are just one subset), more elements
is better. But there's issues and concerns there, too: computational
resources is one, roundoff and numerical precision is another. Start
looking at models with hundreds of thousands of very tiny pieces, and it
becomes quite the numerical analysis/computer science challenge to
effectively compute it. And there are people working on it. I'm aware
of several efforts to implement some MoM and FDTD codes on large (1000
processor) cluster computers. (say you want to simulate an entire ship,
airplane, or spacecraft)
None of these programs agree with each other because of built in
errors

Error is the wrong word here (although technically correct), because it
is perjorative and implies that there is a fundamental bug, which is
generally not the case. All modeling codes are inaccurate to some
degree, partly because of the limited fidelity of the model input
(surely you don't want to spend the time to put in the actual atomic
composition of the elements) and partly because of a deliberate tradeoff
between speed and uncertainty (most people would rather have an answer
in a few minutes accurate to 1% than an answer next week accurate to 0.01%)

As a side point some programs provide errors because the user
doesn't understand the thinking behind garbage in and garbage out
because there is no oversight with respect to programmers error.

This is true of any modeling code. Better codes DO some reasonableness
checks for nonphysical structures and such. But, just like using a
chainsaw to saw down trees more rapidly than using a handsaw, there's
some assumption that the user has some skill.

On 9 Mar, 09:50, Jim Lux wrote:
art wrote:
The most basic of all programs for antennas come from Roy
They do nothing but number crunching like a calculator and
will give you an answer close to what other programs provide
but not the same. The program does NOT help the user in any
way other than give you an answer regarding the performance
of what you provide.

Which, is, of course, the "sine qua non" for a *modeling* program.. it
should take your model and tell you what the performance will be.

It does NOT give you any help as to where

you could benefit in any way. When you move beyond the most
basic of antenna programes you can obtain help fr4om the
programs in that you don't have to specify actual dimensions
which may be useles because you can alow those dimensions
to be variable to allow the computor to guide you in the right
direction
to meet your desires.

This is an "antenna designing" program. Many such programs will use one
or more "antenna modeling" programs as part of their operation,
although, it's not by any means universal.

The cost of these type programs are similar

to eznec but can go up as high as a couple of thousand dollars
tho most amateurs should be satisfied with the cheapest versions

"antenna design" is a fairly wide field, and people literally spend
their life becoming good at one small part of it. There's a lot of
judgement and skill in antenna design, particularly when it comes to
things like mechanical/electrical tradeoffs and manufacturability.
There are tools designed to address one niche or another (e.g. there's
programs that are designed to optimize electrical performance microstrip
patch arrays, there's programs that are designed to optimize Yagi-Udas,
etc.)

Invariably, such tools are (at least originally) designed to be used by
a person who will do the "higher level" trades (Do I use a rectangular
or circular array of patches? What mechanical tolerances can I hold in
manfuacturing? Can I hold a 5 meter by 4 meter array flat enough to
actually work at Ka-band?)

There is some work on integrating all of these, but it's still baby
steps (for instance, taking a Solidworks model and turning it into a
meshed grid for modeling, or trying to integrate electrical, optical,
and mechanical models for large dishes (e.g. IMOS))

Some programs are designed around the yagi only for simplification.

yes.. back before computers got cheap, people worked out clever
analytical models for certain classes of antennas.. arrays of parallel
straight thin elements would be one that's particularly amenable to such
analysis. No surprise that as computers came to be more common, such
models would be first ones to be implemented.

These ofcourse need to be avoided since they are based on the
yagi being unbeatable.

Not at all. it's that people had equations for Yagis (based on
empirical experience that Yagis worked and met at least some of the
requirements), and people tend to want to work with what is familiar.
If for no other reason than you can compare the output of the modeling
code (or the optimization code) with something you've actually built and
see if it matches (aka validation).

So if a choice has to be made then programs

with variable dimension abilities together with a sufficient large
number of pulses are by far superior toi any other computor program.

Well, sure.. if you're going to any sort of Finite Element analysis (of
which the method of moments methods are just one subset), more elements
is better. But there's issues and concerns there, too: computational
resources is one, roundoff and numerical precision is another. Start
looking at models with hundreds of thousands of very tiny pieces, and it
becomes quite the numerical analysis/computer science challenge to
effectively compute it. And there are people working on it. I'm aware
of several efforts to implement some MoM and FDTD codes on large (1000
processor) cluster computers. (say you want to simulate an entire ship,
airplane, or spacecraft)

None of these programs agree with each other because of built in
errors

Error is the wrong word here (although technically correct), because it
is perjorative and implies that there is a fundamental bug, which is
generally not the case. All modeling codes are inaccurate to some
degree, partly because of the limited fidelity of the model input
(surely you don't want to spend the time to put in the actual atomic
composition of the elements) and partly because of a deliberate tradeoff
between speed and uncertainty (most people would rather have an answer
in a few minutes accurate to 1% than an answer next week accurate to 0.01%)

As a side point some programs provide errors because the user
doesn't understand the thinking behind garbage in and garbage out
because there is no oversight with respect to programmers error.

This is true of any modeling code. Better codes DO some reasonableness
checks for nonphysical structures and such. But, just like using a
chainsaw to saw down trees more rapidly than using a handsaw, there's
some assumption that the user has some skill.

So Jim to sum up all you have said can I say that the production of
gaussian arrays by following NEC parameters is O.K. or correct and we
should leave well alone. Or should we remove the causes of its
creation
since it apparently is agreed that it is impossible? Since you work
with space antennas can you say it is not worth looking into or will
that be the same answer that you give your superviser next year as to
why you did not look into it?
I assure you Jim that the further you get into this the more will be
divulged as to what we have misunderstood in the past.
For a man in your position it is imperitive that you avail yourself of
a gaussian array as I have stated. Examine it as to why the product is
resonant in situ and also elements individually and try to come up as
to why
something that some would suggest was a bug produced such a symetrical
product? One would also question why NEC4 corroberated the performance
of the array and why internal algorithms did not deny it in the face
of the superior yagi .In no way does the production of an array by a
computor program provide cover for any pre stated contention . You of
course have to do what you think is correct in your work but one thing
that really disapoints me is that tho I do not expect to get a
response from the South African company as to how or what their
program provides in these circumstances I am more than a little
dismayed that the owner of 4NEC2 is not cooperating. He asked what AO
provided but now refuses to provide what
his program provides, thus I have a sneaking suspicion that all he did
was convert AO and AOP from DOS to WINDOWS without checking content
such that the contended bug or supposed error has carried thru today
without review.
I suppose we can all raise the flag and say it wasn't my job and close
our eyes to the whole saga and let other countries pursue for them
selves and reap the rewards while America sleeps.
Art XG

art wrote:
I am more than a little
dismayed that the owner of 4NEC2 is not cooperating. He asked what AO
provided but now refuses to provide what
his program provides, thus I have a sneaking suspicion that all he did
was convert AO and AOP from DOS to WINDOWS without checking content
such that the contended bug or supposed error has carried thru today
without review.

Uhh.. I think Arie is doing what he can. Have YOU downloaded 4nec2 and
taken the time to learn how to use it and loaded your model in and run
it? It's your design, and YOU should put the work into verifying it.
Sure, you might have to figure out how to convert your conceptual design
into a format amenable for 4nec2 (or EZNEC or any of the other
modeling/optimizing codes). But this is something that any antenna
designer has to do. The program is just a tool. You have to invest the
effort in effectively using the tool, or deciding whether it's an
appropriate tool to use for answering YOUR questions.

It's not like Arie should do this as a demo, just to convince you to
download a FREE program. If you were a large company that was
contemplating spending $60K on a big modeling code, it might be
reasonable to ask the vendor of the code to run a sample model to see if
you've got a handle on the interfaces, etc. But the vendor's work in
that case fits in the category of "marketing expenses".

And, it's not real clear what the question is that you expect to have
answered by 4nec2 or AO, or whatever.


As far as the source of what Arie has done... I have NO first hand
knowledge other than observing the changes it's gone through over the
years. I'm pretty sure it does NOT use the same optimizers or
optimization method as AO. For that matter, his optimizers have evolved
over the years as new optimization algorithms come along.

And, on a more philosophical note.. it's pretty darn offensive to allege
that Arie's merely copied someone else's work, especially since Brian's
codes haven't ever been published as source code, to my knowledge.
Arie's put a lot of work into this, and is being quite gracious in
giving to us to use for free.