Output CPM

Error Indicator

When a filter is defined the topology matrix defines the wanted structure of the final coupling matrix.

It is, however, not always given that it is actually possible for the synthesizer to implement the wanted filter characteristic as defined by the topology matrix (see Topology Matrix). To help the user and alert him or her about the accuracy of the synthesized coupling matrix, an error indicator has been included, which visually shows the obtained accuracy.

The error indicator is a help to the user - showing the possible quality of the synthesized coupling matrix.
A red bar shows the error level: A large red area means that the error is also high.
The error is calculated by comparing the ideal characteristic as defined by the filter input parameters, to the characteristic as calculated from the synthesized coupling matrix.
- The ideal characteristic is the one which is displayed when the 'Plot Using Filter Data' button is selected
- The characteristic, which the coupling matrix represents, is plotted when the 'Plot Using Coupling Matrix' button' is selected.
The error is therefore the average difference (in dB) between these two characteristics (only S21 is used to calculate the error).
The error is only calculated in the displayed frequency band. Changing the span may therefore also change the synthesis error.
The actual error is read out when the mouse cursor is placed over the error bar.
If a non-zero error results, the user should select the 'Plot Using Coupling Matrix' button to see whether or not this error is a problem.
If the wanted characteristic differs significantly from the synthesized characteristic (one or more transmission zeroes may for example be missing), the topology matrix has to be changed and/or the filter order needs to be increased (see Topology Matrix).
When the coupling matrix is manually edited the error related to these changes is also displayed.
The "dynamic range" of the synthesizer is limited. This means that if S21 gets below -150 dB to -200 dB there may be a difference between the ideal characteristic and the characteristic as calculated from the coupling matrix. This error is caused by numerical inaccuracies, and is therefore not necessarily an error in the synthesized filter. Try to narrow in the frequency span.
The combination of very narrow bandwidths (< 1 %) and large filter orders (N>10) may give rise to numerical inaccuracies on the calculated couplings, which may indicate false synthesis errors. At present the best fix to this problem is either to decrease the filter order or increase the bandwidth.