Цитата(Ooze @ May 17 2012, 02:18)

Добрый день!
Подскажите кто знает.
Я не слежу за обовлениями в ЦСТ после версии 2008 года. Реализовано ли в более поздних версиях возбуждение частично заполненых диелектриком волноводов (заполненых от порта)? Имеется в виду, можно ли узнать критические частоы гибридных волн в такой структуре или, например, коэф. передачи? Возможно, есть другие программы где это можно просчитать?
По крайней мере в версии 2011 такая возможность имеется. Вот текст и карртинки из хэлпа:
Нажмите для просмотра прикрепленного файлаа) Inhomogeneous Waveguide Ports (Special Treatment)
A waveguide is called inhomogeneous whenever more than one different dielectric material is present in the cross-section of the waveguide. As mentioned before this is the case for microstrip lines and coplanar lines as well as for dielectrically loaded waveguides, representing QTEM and no QTEM modes respectively.
The most important fact of inhomogeneous waveguide ports is that the mode pattern is frequency dependent. As an example in the pictures below the basic TE mode of a dielectrically loaded waveguide is presented for three different frequency points. The higher the frequency (from left to right), the more the field is concentrated in the material with the higher dielectric value (colored in light brown).
The frequency domain solvers automatically recalculate the mode patterns for every frequency point so that this frequency dependent behavior does not cause a difficulty for the analysis.
In contrast, the time domain solver uses the same mode pattern for the entire frequency band which may cause port mode mismatches at frequencies other than the mode calculation frequency. The error increases with increasing distance to the mode calculation frequency.
By default, the transient solver computes the mode pattern at the center frequency of the frequency band, but this behavior can be changed by specifying the Mode calculation frequency in the solver specials dialog box on the Waveguide page.
Despite this small mismatch at the ports, broadband simulation results will still be sufficiently accurate in most cases. However, possibilities to achieve very high accuracy requirements or very large bandwidths are discussed in the following for microstrip lines / coplanar lines (QTEM Modes) and dielectrically loaded waveguides (No QTEM Modes).
б) Dielectrically Loaded Waveguides (No QTEM Modes)
The following explanations therefore exclude these types of waveguides and focus on dielectrically loaded waveguides. A typical example is shown in the following picture:
In terms of assigning ports to these waveguides, the procedure is very similar to the Empty waveguide case explained earlier in this section. The main difference here is that the port mode pattern is no longer frequency independent.
The frequency domain solvers automatically recalculate the mode patterns for every frequency point so that this frequency dependency does not constitute a difficulty for the analysis.
In contrast, the time domain solver uses the same mode pattern for the entire frequency band which may cause port mode mismatches at frequencies other than the mode calculation frequency. The error increases with increasing distance to the mode calculation frequency.
If you are interested in only very small frequency ranges, a viable solution may be to set the mode calculation frequency equal to the center frequency of this frequency band. This setting can be made by specifying the Mode calculation frequency in the solver specials dialog box on the Waveguide page.
However, if you are interested in broadband results within larger frequency ranges, the transient solver needs to be advised to make a special (and computationally expensive) treatment for the inhomogeneous ports. The Broadband for inhomogeneous waveguide ports (no QTEM modes) option can also be activated on the Waveguide page of the solver’s specials dialog box:
This option will provide much more accurate broadband results for this type of port. Unfortunately, it is limited to the fundamental (propagating) mode only.
Видимо, для таких расчетов лучше подходит F-solver.