is there somewhere I can read up on AGT and adjusting it, i didnt see much in the nec manual

My AGT adjustment procedure:
http://www.digitalhome.ca/forum/showthread.php?p=971475
ERRATA in line 4: To increase AGT, increase the wire radius & vice versa.
In the cases I've seen, there is minimal change to AGT after adjusting
for the mid-band frequency. If there is, you might have to define different
SOURCE wire radii for runs across several frequency bands....bletch....
And always bear in mind NEC runs are only accurate to about 1 dB and
perhaps more when compared to actual metal.....

Additional info re export data to Excel and the various RP cards I use:
http://www.digitalhome.ca/forum/showthread.php?p=981550

Excellent, Holl-Ands, good starting point for me, thanks, ill read up

Keep in mind the NEC engine was developed for the Navy, which only has to deal with many,
but relatively narrow bandwidths. The uhf-tv band is pushing it.
The "Navy" deals with very wide bandwidths: HF Band is 2-30 MHz
and military UHF band is 225-400 MHz. Sometimes a single antenna
covers the entire band....although partial band antennas are also used.
Whip antennas require a variable L/C antenna tuner to adjust the
antenna impedance to the transmitter....but most other antennas do not.

Croatia's Navy used NEC to model HF band (2-10 MHz) antennas,
as they interact with the superstructure....and each other. It is also
important to provide as much isolation as possible between antennas:
http://bib.irb.hr/datoteka/375488.NVISant_e.pdf
Observe the very large number of wire segments.....

NEC models are then compared to actual patterns on (typ. 1/48) scale models:
http://www.swri.org/3pubs/ttoday/summer98/model.htm

One of the NEC developers (SPAWAR SSC, aka NRaD, aka NOSC, aka NELC) uses
the Shipboard Model Range Facility on top of Point Loma, here in San Diego, CA:
http://www.spawar.navy.mil/sandiego/facilities/html/shipboard_antenna_model_range.html
JPL used it in the Mars Lander development in addition to CST & WIPL (NEC sim):
http://trs-new.jpl.nasa.gov/dspace/bitstream/2014/40721/1/06-1936.pdf

Never mind. Got it.

If RC is the reflection coefficient

then:

1+|RC|
SWR = --------
1-|RC|

and:


___________________
/ (Ra - R0)^2 + Xa^2
|RC| = \ / ----------------------
\/ (Ra + R0)^2 + Xa^2

where Ra, R0 are actual and characteristic impedance
and Xa is the actual reactance.


(from the "radio laboratory handbook")

the Shipboard Model Range Facility on top of Point Loma, here in San Diego, CA:
Cool test range. :p

is there somewhere I can read up on AGT and adjusting it, i didnt see much in the nec manual
Having a wire radius print out in front of you is very handy.

A wire radius table would be handy....but most are diameter....
so try to remember to divide by 2:
http://www.powerstream.com/Wire_Size.htm
http://en.wikipedia.org/wiki/American_wire_gauge#Table_of_AWG_wire_sizes

im getting a "geometry card data error"as I go to run the file....

in the gemetry edit, and viewer, all appears normal,


any idea where the error can be, im using symbols, thanks


EDIT: OK solved, missing a RAD data

Segmentation questions:


1. I noticed many files using the EK-extended thin wire kernel, but according to my calcs and necs manual for most files its not needed since most wire radii are less than 5.4 mm when using the segmentation length limits with the the governing upper UHF freq of 689 mhz

2. i noticed when using autosegmentation, it changes the nec file, thus if saved it wipes out the symbols, is the only way to keep the symbols and the updated segmentation to use symbols/variables for the segmentation in the nec file

thanks

Since I don't think it hurts anything, I use EK in ALL my 4nec2 files
so I don't have to worry about whether it is "needed" or not, esp as
I try to see how well a UHF antenna performs in Hi-VHF and vice versa...

Yeah, Geometry Editor is pretty stupid....it's okay as a separate design exercise.....
But then it's back to the Symbols.....let's see...19.359032 would be L2.....
GeoEdit also generates a bunch of round off errors.....

To determine segmentation, I enable Autosegment(21) and do the AGT=1.0
optimization. When finished, I hit F8 key to see the NEC Output File.
Then I copy the segment numbers into my 4nec2 file. No need for GeoEdit.
Next time I run the file, I usually disable Autosegment....but I leave it enabled
if I am running Parameter Sweeps or Optimization, cuz dimensions are changing.

thanks holl_ands- still reading & learning here, and a lot from you!

Optimization questions:

1. in an initial optimization is it better to run it on all available variables (which make sense to try to change) at the same time, or on fewer or only one variable at at a time

2. can we use 100% weighting factor in more than one of the SWR, Gain, R-in fields at the same time on the same run, or is it designed to add up to no more than a max. 100% on all fields combined (SWR=70%, Gain=20%, R-in-10% = tot = 100%, for example)

3. If we set Theta = 90, and select Tot Gain, is this not the same result as setting Theta = 90, and selecting Hor. Gain

4. I noticed for symbols and parameterization in nec files, the variables used often include a formula such as a geometry type for say determining the postion of an angular element (bowtie), as opposed to using a variable for each x,y,z coordinate of the element endpoints (2), is this just to reduce the number of variables in the file, or does it make a difference in the actual way the optimizer arrives at a solution

thank you

when doing the adjusting AGT procedure, such as by adjusting the source wire radius, length, and or segmentation, if we continue until the AGT =1, and such that the radius is so small and warning, say tiny (in the order of tenths of aa mm), does this mean the model is basically "no good" and cannot be improved in its present form?

No...it means the segmentation needs to be changed...usually on the SOURCE wire.
If you used 3 segments, try just 1....and vice versa....

Try a variety of autosegment settings....but don't be afraid to "try something else"....
Manually guesstimate your own segmentation....
Fortunately, I've only run across a few of these problematic situations....

If that doesn't work, try something DIFFERENT, such as a pair of wires feeding
the antenna with the SOURCE (EX) wire simulating an off-set Balun....
Ken Nist (hdtvprimer) has some 4-Bay and 8-Bay examples where he used a
very LONG twinlead feed, although I found it caused other problems....

I use geometric statements, such as the angle of the bowtie whisker, so that
I can quickly CHANGE parameters to find the "best" settings....although I
sometimes use Parameter Sweep at three fixed freqs to get an initial "feel",
eventually I'll do some standard runs with slightly different parameter settings.
Obviously, varying just one parameter is better than a group of parameters.

Other than using the Parameter Sweep capability, I have been unable to figure
out how to SUCCESSFULLY use the various Optimizers...maybe it's just me....
And, as I've mentioned before, the (correct) values displayed in the Pattern
box frequently don't match the values displayed elsewhere.....

No...it means the segmentation needs to be changed...usually on the SOURCE wire.
If you used 3 segments, try just 1....and vice versa....
From what Ive read, the Source wire should have more than 1 segment when using the NEC2 engine, although in some cases Ive found it unavoidable. :( Ken Nist uses the pricey NEC4 engine, which from his models, seems to be OK with a 1 segment Source wire.

Besides what holl-ands listed above, try changing the radius of the rest of the wires of the model when in that situation.

sorry about the spelling

hello there

i understand the source wire and segs. but to change the complete wire size for the whole drawing say ,a larger radiaus . then the whole result would change in the preformace of the antenna ?. some times worse of what your trying to do .. .question if you do that . then if your model becomes worse at 10 gage , how would you model it at 12 gage ? with out the problem of souce wire ? to see the results ? is the model useless ? or do i need to learn how to optimize , or to resize to find the gage it is working at ?

Changing the wire radius of the whole model is a piece of cake with the "NEC Editor (new)". Right click on the "Radius" and choose the "Make same value" option. :)

We've been talking about trying to make adjustments so AGT=1.0 (or 0.99), so the model is accurate.

To begin with, the SOURCE wire is oftentimes a FAKE....it's just a means to apply
voltage to the feedpoint (the EMPTY space "occupied" by the SOURCE wire).
So adjusting the fictious SOURCE wire size is also a FAKE....

On the other hand, making small changes to ALL of the wires in the model might result
in small changes to the SWR, but it has very little effect on AGT and for sure won't "fix"
the problem we were talking about....unrealistically sized SOURCE wires and instability
in the resultant AGT....

Thank you, all excellent, helpful posts here!

a few more questions:


1. what exactly is the "normalization" setting in the expert doing, i see some members use it, others dont, i didnt see much in the documentation re. this

2. when initially drawing/designing an antenna in geo edit, say for the UHF broad band, what is the frequency that should be entered in that windows frequency box, I assume since the auto segment uses this, it should be set at the highest frequency, say 689 Mhz

3. I believe Holl_Ands, you have said that in a manual frequency sweep, the program uses the azimuth with the maximum gain for the lowest (start)frequency in the sweep for the gain azimuth of all subsequent frequencies in the sweep, so does this mean that setting the PHi/Theta manually in the sweep window box will not give me the frequency sweep gains at those set angles

thanks for all the assistance ;)

1. what exactly is the "normalization" setting in the expert doing, i see some members use it, others dont, i didnt see much in the documentation
Good question. The 4nec2 Help file states this:
Experienced NEC users can use the ‘expert’ button to set special far-field calculating options. For additional info about these ‘XNDA’ options, please consult the Nec-2 or Nec-4 user-guide. Default XNDA settings are:

1: Ver/Hor/total field,
0: No normalization
0: Power-gain
2: Compute average-gain




Searching thru the 3 part NEC2 manual, I cant find anything on "normalization". I generally choose 5: tot-gain normalization, (it seems to me to make the most sense) while Ive seen some others choose 0: no normalization.

Edit: found a little more on it here under the RP card : http://www.nec2.org/part_3/cards/rp.html
N - (column 18) causes normalized gain for the specified field points to be printed after the standard gain output. The number of field points for which the normalized gain can be printed is limited by an array dimension in the program. In the demonstration program, the limit is 600 points. If the number of field points exceeds this limit, the remaining points will be omitted from the normalized gain. The gain may be normalized to its maximum or to a value input in field F6. The type of gain that is normalized is determined by the value of N as follows:
N = 0 no normalized gain.
= 1 major axis gain normalized.
= 2 minor axis gain normalized.
= 3 vertical axis gain normalized.
= 4 horizontal axis gain normalized.
= 5 total gain normalized.

it should be set at the highest frequency, say 689 Mhz Channel 51 ends at 698mhz. Dyslexia can be a pain, heh.