Hi

To get started with nec2 I tried to simulate a simple dipole.
The graphical view of the pattern looks okay, however the gain seems
to be to high. To what are the DBs refered in the output file?
Or what do I have to change to get the gain in dBi ?

Here is my input file:
-----------------------------------------
CM Test
CE
GW 1 20 0 0.005 0 0 7.5 0 0.01
GW 2 20 0 -0.005 0 0 -7.5 0 0.01
GW 3 1 0 -0.005 0 0 0.005 0 0.01
GE 0
EX 0 3 1 1 10. 0 0 0 0 0
FR 0 30 0 0 6. 1. 0 0 0 0
RP 0 37 72 1001 0. 0. 5. 5. 1000 1.
EN
-----------------------------------------

Detlef

PS:
If I calculate the gain with the formula:

4 * Pi * r^2 * |E|^2
Gain = -----------------------
P_in * 2 * 120 * Pi

by using the E-Field magnitude values from nec2's output I get also
those very high values. (?)

"Detlef Conradin" wrote in message
...
Hi

To get started with nec2 I tried to simulate a simple dipole.
The graphical view of the pattern looks okay, however the gain seems
to be to high. To what are the DBs refered in the output file?
Or what do I have to change to get the gain in dBi ?

Here is my input file:
-----------------------------------------
CM Test
CE
GW 1 20 0 0.005 0 0 7.5 0 0.01
GW 2 20 0 -0.005 0 0 -7.5 0 0.01
GW 3 1 0 -0.005 0 0 0.005 0 0.01
GE 0
EX 0 3 1 1 10. 0 0 0 0 0
FR 0 30 0 0 6. 1. 0 0 0 0
RP 0 37 72 1001 0. 0. 5. 5. 1000 1.
EN
-----------------------------------------

Detlef

PS:
If I calculate the gain with the formula:

4 * Pi * r^2 * |E|^2
Gain = -----------------------
P_in * 2 * 120 * Pi

by using the E-Field magnitude values from nec2's output I get also
those very high values. (?)


Hi, you have a segment length violation in TAG 3; it is very short, being
only 5mm. Also short when compared to the segment length in TAGs 1 and 2
at 37.5 cm. It is a good idea to maintain equal segment lengths in the
entire model, with minimum lengths not exceeding 0.001 wavelengths on any
frequency. I have modified your code as shown below. There is nothing
wrong with breaking the wire up into three tags, but they should all run the
same way. Your end TAGs are running in opposite directions, which will
cause current discontinuities in plots, and NEC output file data.

There was something weird about your RP card also, but have not yet delved
into what is wrong. I have not checked, but your computation of gain with
the E-field should now be correct.

Cebik's book on NEC modelling, at www.nittany-scientific.com, for $50 is
well worth the money.

Regards,

Frank


CM Test
CE
GW 1 21 0 -7.5 0 0 7.5 0 0.01
GS 0 0 1.000000
GE 0
GN -1
EX 0 1 11 00 10. 0 0 0
FR 0 30 0 0 6. 1. 0 0 0 0
RP 0 181 1 0000 -90 0 1.00000 1.00000 1000 1.
EN

"Detlef Conradin" wrote in message
...
Hi

To get started with nec2 I tried to simulate a simple dipole.
The graphical view of the pattern looks okay, however the gain seems
to be to high. To what are the DBs refered in the output file?
Or what do I have to change to get the gain in dBi ?

Here is my input file:
-----------------------------------------
CM Test
CE
GW 1 20 0 0.005 0 0 7.5 0 0.01
GW 2 20 0 -0.005 0 0 -7.5 0 0.01
GW 3 1 0 -0.005 0 0 0.005 0 0.01
GE 0
EX 0 3 1 1 10. 0 0 0 0 0
FR 0 30 0 0 6. 1. 0 0 0 0
RP 0 37 72 1001 0. 0. 5. 5. 1000 1.
EN
-----------------------------------------

Detlef

PS:
If I calculate the gain with the formula:

4 * Pi * r^2 * |E|^2
Gain = -----------------------
P_in * 2 * 120 * Pi

by using the E-Field magnitude values from nec2's output I get also
those very high values. (?)


PS, also confused by your formula. Which Value of E are you using? Are you
trying to calculate the "Total radiated power" (TRP)? If so you need to
integrate the power density over a spherical region. The calculation is
greatly simplified if you take advantage of any natural symetry in the
radiation pattern -- i.e. integrate from axis end to axis end, then multiply
by 2*PI -- as in elementary calculus, you are using a single integral for a
"Solids of revolution" method, instead of a double (Surface integral).

If you are trying to caluculate the gain in a given direction you need to
take 20*log(E/Ei). Where E is the NEC calculated E field in the desired
direction, and Ei is the computed E field from an ideal isotropic source
with the same input power as the test antenna.

As for your RP card, I notice that you are attempting to vary both "Theta"
and "Phi". I usually keep one fixed and vary the other, but to be honest I
am not sure if what you are doing is incorrect, although I notice that the
"Phi" pattern is only computed for one frequency. The only other difference
is that I normally do not use gain averaging. Also using 1 degree
increments does not seem to slow things down very much. With your RP card I
was also getting some weird pattern discontinuities.

Regards,

Frank

........entire model, with minimum lengths not exceeding 0.001 wavelengths
on any
frequency. I have ....

Oops, of course, I meant: "Not less than 0.001 wavelengths".

Frank

To get started with nec2 I tried to simulate a simple dipole.
The graphical view of the pattern looks okay, however the gain seems
to be to high. To what are the DBs refered in the output file?
[...]
If I calculate the gain with the formula:

4 * Pi * r^2 * |E|^2
Gain = -----------------------
P_in * 2 * 120 * Pi

by using the E-Field magnitude values from nec2's output I get also
those very high values. (?)

PS, also confused by your formula. Which Value of E are you using? Are you
The formula should be correct:
I just use the magnitude value of the E Field. The intensitiy
W_rad is then 0.5 * |E|^2 / Z_w (E and H are in phase and
orthogonal in the far field)

The radiation density U is then r^2 * W_rad (Unit: Watt per solid angle)

The radiation density U_0 of an isotropic radiator with a total radiated
power P_rad is: P_rad / (4 * Pi).

The gain is then U/U_0 (or in dBi: 10*log(U/U_0) ).

trying to calculate the "Total radiated power" (TRP)? If so you need to
integrate the power density over a spherical region. The calculation is
Yes, but nec2 shows the total radiated power in its output.

As for your RP card, I notice that you are attempting to vary both "Theta"
and "Phi". I usually keep one fixed and vary the other, but to be honest I
am not sure if what you are doing is incorrect, although I notice that the
"Phi" pattern is only computed for one frequency. The only other difference
No thats no true. In the output file I can see E_phi as well as E_theta for
every combination of phi and theta (for every frequency).
I use the programm Xnecview to view the pattern and it has problems if you
don't have points over the full sphere.

is that I normally do not use gain averaging. Also using 1 degree
The average gain (over the full sphere) should probably be 1 (or 0 dBi).
I get 0.991.

Thanks for you help! The bad computation was caused by the stupid
segementation I had chosen.

Detlef

4 * Pi * r^2 * |E|^2
Gain = -----------------------
P_in * 2 * 120 * Pi

by using the E-Field magnitude values from nec2's output I get also
those very high values. (?)

PS, also confused by your formula. Which Value of E are you using? Are
you
The formula should be correct:
I just use the magnitude value of the E Field. The intensitiy
W_rad is then 0.5 * |E|^2 / Z_w (E and H are in phase and
orthogonal in the far field)

The radiation density U is then r^2 * W_rad (Unit: Watt per solid angle)

The radiation density U_0 of an isotropic radiator with a total radiated
power P_rad is: P_rad / (4 * Pi).

The gain is then U/U_0 (or in dBi: 10*log(U/U_0) ).

trying to calculate the "Total radiated power" (TRP)? If so you need to
integrate the power density over a spherical region. The calculation is
Yes, but nec2 shows the total radiated power in its output.

As for your RP card, I notice that you are attempting to vary both
"Theta"
and "Phi". I usually keep one fixed and vary the other, but to be honest
I
am not sure if what you are doing is incorrect, although I notice that
the
"Phi" pattern is only computed for one frequency. The only other
difference
No thats no true. In the output file I can see E_phi as well as E_theta
for
every combination of phi and theta (for every frequency).
I use the programm Xnecview to view the pattern and it has problems if you
don't have points over the full sphere.

is that I normally do not use gain averaging. Also using 1 degree
The average gain (over the full sphere) should probably be 1 (or 0 dBi).
I get 0.991.

Thanks for you help! The bad computation was caused by the stupid
segementation I had chosen.

Detlef

Very interesting Detlef. I must admit I did not study the formula too
closely, so will look at it again, to be sure I understand it.

My version of NEC2 has a built in NEC-View utility (NEC-Win Pro), so am not
familiar with Xnecview. I will try your full spherical pattern RP card, and
see if I can run it. When I first cut and pasted your code I was getting
strange FORTRAN code error messages -- something I have never seen before --
and the program was locking up (Windows XP OS). Modifying the RP card cured
the lock up problems. Examination of the segment currents would also
probably have indicated the segmentation problem area. At 1 degree
intervals the NEC output file would be huge, so can see why you chose 5
degree increments.

As for "Total radiated power" (TRP) I assume you mean the "Power budget"
output. I had thought this is only valid for free space computations. With
antennas placed close to the ground the "Power budget" only seems to account
for copper losses. Just a minute, now you have made me start thinking. My
"Network Loss" line in the power budget output always shows zero. Did I
miss something? Do I have to invoke something in the RP card to make
"Network loss" non-zero, and thus arrive at a true TRP. Heck! To think all
this time I have been running Excel numerical integration routines to
determine the TRP.

Regards,

Frank

...........................
If I calculate the gain with the formula:

4 * Pi * r^2 * |E|^2
Gain = -----------------------
P_in * 2 * 120 * Pi

by using the E-Field magnitude values from nec2's output I get also
those very high values. (?)

PS, also confused by your formula. Which Value of E are you using? Are
you
The formula should be correct:
I just use the magnitude value of the E Field. The intensitiy
W_rad is then 0.5 * |E|^2 / Z_w (E and H are in phase and
orthogonal in the far field)

The radiation density U is then r^2 * W_rad (Unit: Watt per solid angle)

The radiation density U_0 of an isotropic radiator with a total radiated
power P_rad is: P_rad / (4 * Pi).

The gain is then U/U_0 (or in dBi: 10*log(U/U_0) ).

Ok, now I understand.


trying to calculate the "Total radiated power" (TRP)? If so you need to
integrate the power density over a spherical region. The calculation is
Yes, but nec2 shows the total radiated power in its output.

Have checked "Power Budget" output, and confirm that using it for TRP is
valid in free space only.


No thats no true. In the output file I can see E_phi as well as E_theta
for
every combination of phi and theta (for every frequency).

I agree, my problem is that NEC-Win Pro cannot use this information for its
graphical utilities.

I use the programm Xnecview to view the pattern and it has problems if you
don't have points over the full sphere.

Ok, I see that Xnecview will only run under Linux/Unix OS. My version of
NEC only runs in Windows.

is that I normally do not use gain averaging. Also using 1 degree
The average gain (over the full sphere) should probably be 1 (or 0 dBi).
I get 0.991.

Regards,

Frank

Frank wrote:

Cebik's book on NEC modelling, at www.nittany-scientific.com, for $50 is
well worth the money.

Or try 4nec2 at http://home.ict.nl/~arivoors/. It has built-in
geometry- and segment-checks which will deliver instant error and
warning reports. Furthermore it's completely free to use
and will run Nec2 (and Nec4, if you have a Nec4 license).

Arie.