
For FullText PDF, please login, if you are a member of IEICE,
or go to Pay Per View on menu list, if you are a nonmember of IEICE.

Variance Distribution of Reflection Coefficients in SixPort Reflectometer
Manabu KINOSHITA Hajime SUZUKI Toshiyuki YAKABE Hatsuo YABE
Publication
IEICE TRANSACTIONS on Electronics
Vol.E77C
No.6
pp.930934 Publication Date: 1994/06/25
Online ISSN:
DOI:
Print ISSN: 09168516 Type of Manuscript: Special Section PAPER (Special Issue on Measurement Techniques for Microwave/Millimeter Wave) Category: Keyword: sixport, reflectometer, random error,
Full Text: PDF(488.4KB)>>
Summary:
This paper discusses the effect of random errors in power meter readings by the sixport reflectometer. By means of sixport techniques, the determination of the reflection coefficient (Γ) of a divice under test is reduced to the problem of finding a common intersection of three circles in the complex plane. Since the intersection usually forms a cluster due to the effect of measurement error, the extraction of a single value from the cluster including the radical center of the three circles is required. Two types of methods are presented for determining Γ. One uses a linear solution for the radical center, and the other is a statistically based nonlinear solution. In order to improve measurement accuracy, the effect of random errors in the sidearm power meter readings and due to the influence of the qpoint locations are investigated for each method. By adding a random variation of 0.5% onto each of the three port power ratios, the variance distributions of Γ over the entire area of the Smith chart are simulated for comparison of these two solutions. The three dimensional variance distribution chart reveals that only the nonlinear solution suffers a variance increase shown as a ridgelike peak along the lines of centers of the three circles. As a result of computer simulations, it is clarified that the reflectometer has the property of measurement accuracy dependence on the value of Γ. A new type of sixport model is suggested, which is unlikely to be affected by random errors in the nonlinear solution.

