开云体育

I only start to know EZNEC well. If you install EZNEC, it files the manual (in English) in the form of a Word file. It is huge because the author explains a lot about good and bad modeling as well as ways to identify bad modeling. It is this notice that I translated into French and completed.

As far as I know NEC motors, other than EZNEC built-in NEC-2D, are unaffordable for an OM (QSJ).
--
F1AMM (Fran?ois)

Be aware that EZNec with the built-in NEC engine doesn't work well on higher frequencies like 50MHz.? The resonant point is off by a couple hundred kilohertz.? The latest LLNL engine gets it right.
Mike W9MDB

On 10/14/22 10:21 AM, F1AMM wrote:

I only start to know EZNEC well. If you install EZNEC, it files the manual (in English) in the form of a Word file. It is huge because the author explains a lot about good and bad modeling as well as ways to identify bad modeling. It is this notice that I translated into French and completed.
As far as I know NEC motors, other than EZNEC built-in NEC-2D, are unaffordable for an OM (QSJ).

NEC2 is free, as is 4nec2, as is MMANA.

NEC4 (or NEC5) require a one time license fee to Lawrence Livermore.

Yes, one of the huge advantages of EZNEC is the modeling assistance in the manual, which is actually generally applicable to all method of moments codes (e.g. NEC).


Antenna modeling is one of those idiosyncratic things that you learn by doing, especially the subtle points and tricks.

For instance, gridding for solids (or boxes) is a whole art in itself.

Attached are a couple of screen shots from 4nec2, for a model built with some custom built software. In one, I'm gridding a 6U cubesat that has 4 2.5m long booms. In the other, I'm gridding the ground plane under an antenna for an array radio telescope.

To keep this remotely NanoVNA related, one test of the gridding is whether the modeled feedpoint Z matches the measured feedpoint Z, and that's a lot easier with a NanoVNA. It's small enough to put inside a mockup satellite, so you don't have to worry about cables.

On 10/14/22 10:28 AM, Michael Black via groups.io wrote:

Be aware that EZNec with the built-in NEC engine doesn't work well on higher frequencies like 50MHz.? The resonant point is off by a couple hundred kilohertz.? The latest LLNL engine gets it right.

Can you provide more information? NEC *defaults* to 300 MHz (1 meter lambda), it's basically just geometry. I've done a *lot* of NEC modeling and never seen errors like that.

I *have* seen errors due to unwise segmentation, and with single precision, and situations where the assumption of uniform current distribution on the surface of a wire is broken. But that's not frequency related.

Mike W9MDB
On Friday, October 14, 2022 at 12:22:13 PM CDT, F1AMM <18471@...> wrote:
I only start to know EZNEC well. If you install EZNEC, it files the manual (in English) in the form of a Word file. It is huge because the author explains a lot about good and bad modeling as well as ways to identify bad modeling. It is this notice that I translated into French and completed.
As far as I know NEC motors, other than EZNEC built-in NEC-2D, are unaffordable for an OM (QSJ).

I am considering learning to use one of these programs to model 2.4 GHz PCB antennas. Are there some that are better at that frequency range?

None!
They do wire type antennas only.

I believe that NEC derivative modeling software only goes into the 300 MHz range.

The two higher-frequency software packages of which I am aware are CST Studio and Comsol Multiphysics. CST offers a partially-functional free version. I don’t know if Comsol offers something similar.

Others will probably have more ideas.

DaveD

On 10/14/22 12:21 PM, Douglas Butler wrote:

I am considering learning to use one of these programs to model 2.4 GHz PCB antennas. Are there some that are better at that frequency range?

In general Method of Moments (MMANA, NEC and programs that use NEC) is not a great way to model surfaces like patches.


If you're making simple patches, then the equations in textbooks like Pozar or Kraus will get you close enough to where you're more likely to have problems from manufacturing and material tolerances.

PUFF (which is ancient) will model patches.

I don't think there's a "student version" of HFSS or other similar codes.

On 10/14/22 1:26 PM, Dave Daniel wrote:

I believe that NEC derivative modeling software only goes into the 300 MHz range.

NEC is not frequency limited - you can model wire antennas at 30 GHz if you like.

The problem is that nobody builds wire antennas at 10 GHz - the losses get high. Then you're doing microstrip patches, and while NEC models patches or gridded equivalents, it doesn't do things like account for a substrate and ground plane.

You need a different code.

Jim, I've not experienced that problem when the model is properly set up.
I've designed and constructed many antennas from HF through L-band and the
1.42 GHz deep space neutral hydrogen emission using 4NEC2 which uses the
basic NEC2 engine. All have worked fine for me when finally implemented.

However, all the NEC engines have trouble handling "ground". Once the feed
impedance of a modeled antenna is determined with the NEC engines (EZNEC or
4NEC2 and some others), I then transition to SimSmith for the matching
solution.

Dave - W?LEV

--
Dave - W?LEV

Thank hou for correcting me.

DaveD

Sonnet USA supplies a far-field radiation pattern program as an option on
all versions except the free SonnetLite. It probably costs an arm and a leg
to get it, though. I've been using SonnetLite for a long time to analyze
and design pc patterns like microstrip printed filters, microstrip matching
networks, etc.

73, Zack W9SZ

OPENEMS.DE

Open EMS is to HFSS as
Linux? is to? Windows

Produced by a community of EE and software geeks.
Be sure to download the tutorials.
Like HFSS, it has quite the vertical learning curve.? ?Kent

Back in 2011, I used something called FEKO Lite to model a PCB bowtie antenna with a tapered matching feedline that had to function both as an antenna and a rudder for an autonomous submarine. Periodically the sub would stick its rudder in the air to get a GPS fix and exchange satellite data. I sure wish I had a VNA for that project! Now I have the nanoVNA but FEKO may be beyond my budget.

On 10/14/22 1:50 PM, W0LEV wrote:

Jim, I've not experienced that problem when the model is properly set up.
I've designed and constructed many antennas from HF through L-band and the
1.42 GHz deep space neutral hydrogen emission using 4NEC2 which uses the
basic NEC2 engine. All have worked fine for me when finally implemented.

my experience likewise.

However, all the NEC engines have trouble handling "ground". Once the feed
impedance of a modeled antenna is determined with the NEC engines (EZNEC or
4NEC2 and some others), I then transition to SimSmith for the matching
solution.

It sort of depends on what "ground" is and how close you are. NEC4 does much better with antennas that are close to or touch the soil.

But NEC2 does pretty good, if you're aware of the limitations. A low dipole that's lambda/10 is modeled pretty much the same as NEC4 and with analytical results and actual test data.

I've done a lot of modeling of wires a short distance off the surface of the Moon. The analytical and NEC and other codes all agree.

Sometime ago, George Hagn and others at SRI compared measurements of actual antennas over actual measured soil properties with models and had good agreement.

The usual challenge is that soil isn't uniform and NEC has no way to model that.

Dave - W?LEV
On Fri, Oct 14, 2022 at 5:56 PM Jim Lux <jimlux@...> wrote:

On 10/14/22 10:28 AM, Michael Black via groups.io wrote:

Be aware that EZNec with the built-in NEC engine doesn't work well on

higher frequencies like 50MHz. The resonant point is off by a couple
hundred kilohertz. The latest LLNL engine gets it right.
Can you provide more information? NEC *defaults* to 300 MHz (1 meter
lambda), it's basically just geometry. I've done a *lot* of NEC
modeling and never seen errors like that.

I *have* seen errors due to unwise segmentation, and with single
precision, and situations where the assumption of uniform current
distribution on the surface of a wire is broken. But that's not
frequency related.

Mike W9MDB



On Friday, October 14, 2022 at 12:22:13 PM CDT, F1AMM <

18471@...> wrote:

I only start to know EZNEC well. If you install EZNEC, it files the

manual (in English) in the form of a Word file. It is huge because the
author explains a lot about good and bad modeling as well as ways to
identify bad modeling. It is this notice that I translated into French and
completed.

As far as I know NEC motors, other than EZNEC built-in NEC-2D, are

unaffordable for an OM (QSJ).