Input and Load Reflection Coefficients

In the case of Homer Autotuner, two reflection coefficients are provided (in all its various formats):

  • Input reflection coefficient GI

  • Load reflection coefficient GL

In the case of Homer Analyzer alone, the two quantities are identical and we shall speak only of load reflection coefficient GL.

Reflection coefficient must always be defined at a certain cross-sectional plane along the transmission line.

Input Reflection Coefficient

Input reflection coefficient (GI) is the quantity seen at the tuner input. It is identical with or very close to reflection coefficient GM inherently measured by the Analyzer part of Autotuner. For impedance match, GI must be forced as close to zero as possible.

Input reference plane is fixed at the axis of the tuning stub 1 (the one closest to generator, see Fig. 4).

Load Reflection Coefficient

Load reflection coefficient (GL) is the reflection coefficient of the object (e.g., an applicator) connected at the tuner output.

Load reference plane can be arbitrarily shifted by the user (see the Plane Ext item of Homer Menu). For zero shift, load reference plane coincides with the output flange of the Homer, be it the Autotuner (see Fig. 4) or the Analyzer alone.

Transformed Input Reflection Coefficient

Input reflection coefficient (GI) can be, before displaying, subject to a user-defined bilinear transform, accounting for instance for a two-port network connected at the Homer input. For details, see Transforming Input Reflection Coefficient section.

Deembedding

Because the Homer Autotuner includes tuning stubs inserted between the Analyzer part and the actual load, the input reflection coefficient GI generally differs from the load reflection coefficient GL. The effect of the tuning stubs can, however, be accounted for by a mathematical process called deembedding. In this way, although the stubs may be extended, you are able to observe what is going on “behind the tuner”, i.e., to see the reflection coefficient GL of your applicator without need of physically retracting the stubs. This is why in the case of Autotuner the load reflection coefficient is alternatively termed deembedded reflection coefficient.

In HomSoft, you can observe either the input or the load reflection coefficient in each display window and in Results windows. You may, for instance, open two identically formatted display windows, one showing the input reflection coefficient, the other showing the load reflection coefficient. To switch between the two reflection coefficient displays, proceed as follows:

In Display Windows

  • Click into the desired display window to make it active.

  • Click the Trace|Show Load menu item, or

  • Activate either the or the toolbar button.

In Full-Size (Maxi) Results Window

  • Check Show Load checkbox.

In Reduced-Size (Mini) Results Window

  • Right-click the magnitude display panel (top left) to invoke pop-up menu.

  • Check the Show Load menu item in the pop-up menu.

Deembedded Accuracy

If tuning stubs of Autotuner are inserted, they can mask the deembedded (load) reflection coefficient GL in the sense that large variations of GL may cause only small variations of the measured input reflection coefficient GI. This is because the main reflection contributor may be the stubs themselves rather than load (see Fig. 4). The corresponding areas in the Smith chart can easily be 10:1 in size if the stubs are deeply inserted. Deembedding the reflection coefficient (removing the effect of tuner from GI to arrive at GL) means the opposite process: we try to transform those small variations of GI to large variations of GL. Therefore, even a small error uI in determining GI will be “blown up” and can cause large error uL in the determining of GL. If you keep the actual load reflection coefficient constant and move stubs to deeper positions, the displayed deembedded GL should ideally also remain constant. But due to this “blown-up” measurement error, the actually displayed GL will still vary. You can gain a feeling for what you may expect by performing such experiment.

Since autotuning is based on such indirectly obtained GL, tuning of highly mismatched loads is less accurate and may require more than one tuning step. Under very adverse conditions (poor magnetron signal) the tuning stubs may get trapped in the full-extension positions. Invoking All Stubs Home procedure often helps in such cases, but the only proper solution is to improve your applicator or generator.

Fig. 4.  Load (deembedded) reflection coefficient measurement error uL.