With reference to 300Ohm's post #49 -> http://www.digitalhome.ca/forum/showpost.php?p=1110605&postcount=49
... its best to get more modelers into this discussion , rather than just you and me alone. holl_ands will probably be back, and maybe Snowman53 too. mclapp, since he's very good with EZNEC can provide some insight too.
That would be great. I did mention "Perhaps the "resident modeling wizards" will seriously take a look at my modeling challenge.
I would suggest that you enlarge the scope of the discussion to think about collectively generating a more detailed modeler's guideline that Stampeder could make into a "sticky"
For me, there are really two important questions that snowman53's auto-segmentation analysis raises. I had mentioned them in my previous posts on this subject, but I guess I wasn't that clear. So let me have another go and provide some points that might start the discussion with a wider viewpoint.
Question 1. How to determine an antenna model's "reliability" for posted results of Gain/Swr etc
The late L. B. Cebik (W4RNL) pioneered the use of MoM nec programs in the ARRL as a better way to compare commercial antennas for Radio Hams, and to provide a software tool to better design "Home-Brew" Radio Ham antennas covering a wide range from 80m up to "Moon-Bounce" frequencies.
His recommendation for determining a model's reliability is to run two tests
: not only AGT
at at least one frequency (within the designed bandwidth
), BUT one should also run Convergence Tests
, to determine how increased segmentation affects the results of a specific nec model.
The value of these two tests is simply this: It is applied to a specific antenna model, hence comparisons of models of different antennas that have been found to be "reliable" via these two tests,
provides the best and only assurance that the comparison is valid.
The AGT test should be done at at least one frequency inside
the designed bandwidth and Convergence tests should be done from the minimum recommended density of 10 segments per 1/2 wave up to the max limit of segmentation that meets the segmentation rules.
In 4nec2, Arie has developed a Convergence test that does exactly that. It is the "Conv-test" function in the Calculate->Start optimizer (F12) tab of the main window of 4nec2. I would recommend that you ensure your Auto-segmentation setting is disabled before running the Convergence and AGT tests. If you have never run the Conv-test function in 4nec2, take a close look at it. It is an extremely powerful tool that Arie has built into 4nec2 !
I really believe that running both tests will provide the answer to Snowman53's query "I am not sure one can trust NEC to provide consistent values between different types of antennas which may use different segment sizes."
Question 2. Why is Auto-segmentation used ? Is it ever Disabled ? ...and other questions
Auto-segmenation is a useful computer aid for 4nec2 modelers, especially when the model is of an antenna with complex geometry. Snowman53's analysis shows that setting auto-seg to >13 gives good convergence on the specific SBGH and DBGH models that he considered in his analysis. It does not necessarily imply that all models will be reliable if you use auto-seg at a setting >13 (say 21 or 22). Instead, you should test each specific model using AGT and Covergence tests. That will give a measure of the model's reliability; using auto-seg does not give such assurances.
My modeling challenge was geared to explore what the auto-seg algorithm in 4nec2 actually does for complex antenna geometries, including a mesh screen. I was not familiar with the details of the snowman53 models, hence my suggestion for the GH6.3-narod with Screen. Feel free to explore with any other model.
Personally I don't use auto-seg. From what I understand, it is very useful when first setting up your model segmentation and should be used when running sweep optimizations. I suspect that most other times your should disable auto-segmentation. There might be some other strange effects that others have noticed, using auto-seg. You probably have the most experience of anyone on this forum with mesh-screen models; would you recommend using auto-seg with mesh screens? From Arie's help file, it's interesting to note that you can selectively turn-off auto-segs for specific wire elements in your model via a preceding + <sign>. This would allow the modeler some control over the auto-seg algorithm.
I see more and more new comers to 4nec2 modeling on this forum. I know from my journey that there is lots to learn and even more to continue to learn - for all of us. I really believe a more detailed 4nec2 guideline "sticky" that provides some explanation on these subjects would be very useful and would further grow the reputation for DigitalHome.ca as the world's best modeling forum.
Anyway...time to go... I hope folks find this helpful in better defining my thoughts. I'm way over my allotted time... you know ...Zapperman is calling ...so it's back to "kompozer" and more html stuff.
4NEC2 Auto-segmentation Help Notes
When optimizing and/or sweeping antenna performance, it is likely thatdimensions and/or frequencies are changing, so it is possible that certain NEC requirements, such as the minimum number of segments per half-wave are not met any more. To overcome this problem, Auto-Segmentation is used.
If this feature is enabled (See Settingssettings on Main window), each calculation/optimization/sweeping step, the wire length for each wire is checked against the actual wave-length and the specified number of segments per wire, and the position for voltage/current sources, RLC-loading and transmission lines is updated.
If a certain number of segments is already specified in the input file, at least this number of segments is set with auto-segmentation. This to avoid auto-segmentation decrease the number of segments specified for critical regions. If you do not want to specify a minimum, set the number of segments to one.
If the number of segments is preceded by a “+” sign the number of segments is fixed, meaning, the segments for this wire are not affected by the auto- segmentation process.
When enabling the Auto-Segmentation feature, you are asked to specify the number of segments per half-wave. Although, mostly accuracy improves if more segments are specified, also the computation time is increased. For ‘simple’ structures, without sharp angles, a value from 10 to 20 will be sufficient. If precise impedance values are required or if wires joining at sharp angles are used more segments could be required. If you are not sure what amount to specify, please use increasing values and compare the results. (You could do this using the Sweepereval function). Also abrupt current changes on the structure might indicate an insufficient number of segments. One place more segments aren’t better is if wires of different diameters are connected.
If computer performance is an issue, you could consider not to the use the auto-segmentation feature. If this feature is disabled, the processing especially for large input files with many wires is speed-up.