Jim is right. There is almost no difference in a V and a catenary as far
as the antenna is concerned.

It would really wind up being an exercise of "can we really model a
catenary?"

If anyone disagrees, we will do it. (NOTE: I said "we", not just me)

John S wrote:
Jim is right. There is almost no difference in a V and a catenary as far
as the antenna is concerned.

It would really wind up being an exercise of "can we really model a
catenary?"

If anyone disagrees, we will do it. (NOTE: I said "we", not just me)

It depends on how close you want the model to be, but in general all you
do is break the catenary, or any curve you want, into a series of
straight line segments.



--
Jim Pennino

John S wrote in :

Jim is right. There is almost no difference in a V and a catenary as far
as the antenna is concerned.

It would really wind up being an exercise of "can we really model a
catenary?"

If anyone disagrees, we will do it. (NOTE: I said "we", not just me)


No worries, I'm willing to accept that, now and as of last night's talk too.
This is useful to me as it is, it gives a good example of diminishing returns
in some lines of exploration.

On 10/13/2014 12:38 PM, wrote:
John S wrote:
Jim is right. There is almost no difference in a V and a catenary as far
as the antenna is concerned.

It would really wind up being an exercise of "can we really model a
catenary?"

If anyone disagrees, we will do it. (NOTE: I said "we", not just me)

It depends on how close you want the model to be, but in general all you
do is break the catenary, or any curve you want, into a series of
straight line segments.

Yes, of course. And, with the free version of EZNEC, one must be careful
not to exceed the max segments allowed.

John S wrote in :

Yes, of course. And, with the free version of EZNEC, one must be careful
not to exceed the max segments allowed.


I've started reading the manual, I suspect there's little danger of that.
At least with EZNEC+ 4 onwards, not sure about standard version. I suspect
like curves built in segments in Sketchup, or the straight bars in the chains
of the Clifton Suspension Bridge, the amount of fine tuning you'll get in
using more than about 24 segments for a catenary might be an exercise in
diminishing returns, and that even just 3 to 5 might be adequate, if the
deviation from straight is small.

John S wrote:
On 10/13/2014 12:38 PM, wrote:
John S wrote:
Jim is right. There is almost no difference in a V and a catenary as far
as the antenna is concerned.

It would really wind up being an exercise of "can we really model a
catenary?"

If anyone disagrees, we will do it. (NOTE: I said "we", not just me)

It depends on how close you want the model to be, but in general all you
do is break the catenary, or any curve you want, into a series of
straight line segments.

Yes, of course. And, with the free version of EZNEC, one must be careful
not to exceed the max segments allowed.

Not really a problem as it does not take many segments to represent the
ends, which has a slow change, as the center part with a more rapid
change.

If I were going to do it, I would use something like a spreadsheet
to plot the curve then draw straight line segments on the curve and
plug those directly into EZNEC.

The extreme case is modeling a loop as a geometric figure with straight
side.

EZNEC will generate loops with whatever number of sides you want and
thus it is fairly easy to see when increasing the number of sides
gives diminishing returns in the difference between the loops.


--
Jim Pennino

wrote in news
EZNEC will generate loops with whatever number of sides you want and
thus it is fairly easy to see when increasing the number of sides
gives diminishing returns in the difference between the loops.


Precisely so. Let me guess, at around 24 segments for a circle, it gets
pretty good?

wrote in news
EZNEC will generate loops with whatever number of sides you want and
thus it is fairly easy to see when increasing the number of sides
gives diminishing returns in the difference between the loops.


I imagine the simplest way (the way I'd do it if I was coding EZNEC myself
(as if I could!)), I'd allow the user to set the segment count not by direct
number, but by limiting angle to dictate segment length before generating a
new segment automatically, that way ANY curve will be met with an appropriate
array of segemnts, and changing the angle changes count, process time,
accuracy...

On 10/14/2014 12:18 PM, wrote:
John S wrote:
On 10/13/2014 12:38 PM, wrote:
John S wrote:
Jim is right. There is almost no difference in a V and a catenary as far
as the antenna is concerned.

It would really wind up being an exercise of "can we really model a
catenary?"

If anyone disagrees, we will do it. (NOTE: I said "we", not just me)

It depends on how close you want the model to be, but in general all you
do is break the catenary, or any curve you want, into a series of
straight line segments.

Yes, of course. And, with the free version of EZNEC, one must be careful
not to exceed the max segments allowed.

Not really a problem as it does not take many segments to represent the
ends, which has a slow change, as the center part with a more rapid
change.

If I were going to do it, I would use something like a spreadsheet
to plot the curve then draw straight line segments on the curve and
plug those directly into EZNEC.

Interestin that you suggest that. See below. Note that I am not
affiliated with EZNEC in anyway other than as a very satisfied user.

The extreme case is modeling a loop as a geometric figure with straight
side.

EZNEC will generate loops with whatever number of sides you want and
thus it is fairly easy to see when increasing the number of sides
gives diminishing returns in the difference between the loops.


While checking the EZNEC Web site to see if I had the latest version two
days ago, I discovered a dynamite Excel spreadsheet application. It is
called AutoEZ. Just today I learned how to use it to generate a curve of
antenna efficiency vs permeability of the wire.

I also was able to reproduce the list I posted earlier of efficiency vs
antenna length in a matter of seconds. I am flabbergasted with this tool.

It seems to have an optimize tool that I have yet to explore.

There is a free version with limitations.

Lostgallifreyan wrote:
wrote in news
EZNEC will generate loops with whatever number of sides you want and
thus it is fairly easy to see when increasing the number of sides
gives diminishing returns in the difference between the loops.


Precisely so. Let me guess, at around 24 segments for a circle, it gets
pretty good?

A quick run of a 4, 8, and 16 sided loop shows:

gain impedance
3.07 126
3.34 136
3.41 139

So it appears that by 8 sides you are already pretty close.


--
Jim Pennino