gin&pez@arg
1) The pdf document you have provided and archived for us is written for a C++ complier. This document plus the subtitle of this post "combined fortran, maxima and gnuplot code for [errors in "error" models]", implies to me your intention to port your Fortran code into the C++ environment, and provide the C++ version here for the convenience of this group. Is my interpretation of your intent correct on this?
2) Please confirm that the G-mini equation you have derived and posted here:
is a "minimal" one port subset of the full set of two port equations posted here:
3) The Nominal values for the SOL standards in the G-mini equation assumptions in #17 at:
/g/nanovna-users/message/4747are (numerically) defined as perfect. Neither of the above equations provide for standards characterization influences. Are the standards inaccuracies defined and embedded in the uncertainty boundary definitions?
4) The equations defined at
a) The set of Equations (0) through (8) are void of calibration/characterization data, and only require reflection SOL calibration measurements of a single port (S11?). Please confirm that both one and two port calibration is embedded in these calculations, and also confirm that SOL measurements are required on only one port, with the through reference plane defined by the SOL standards used to perform the calibration.
b) Equation (0 ) and its inverse Equation (8) are unclear. I assume the upper case Gamma is the computed reflection coefficient, and the lower case gamma is the measured (DUT) reflection coefficient. Please confirm.
c) In Equation (0), please confirm that S12S21 is an intermediate variable defined in equation (6), and does not imply S12 * S21.
d) In Equation (0), S11 is summed with the computed result of all remaining calculations defined as Equation (0). In other words S11 + (S12S21 * G) not (S11 + S12S21) * G
d) Equations (1) through (4) define the common intermediate denominator D, and intermediate numerators N11, N21N21, and N22 used in Equations (5), (6), and (7) respectively.
e) Equations (5) and(7) are the final calculated reflection coefficient results.
f) Equation (6) is a calculated transmission coefficient. Please explain how to differentiate between forward and reverse transmission results.
g) Uncertainties due to errors in accuracy, and those accounted for through characterization of the calibration standards used (by their absence here) are removed to the DERDEI contour calculations by the relevance of their contributions to the overall uncertainty in the accuracy of the measurements.
h) These equations alone will yield measurement results no better, and no less accurate than existing calibration processes, but are computationally much more efficient.
5) If my interpretation in 5 a) is correct, and only one port is required to be SOL calibrated; have you validated and compared these equations independently for each port in a full two-port system, calibrated with one set of SOL measurements performed at both ports uniquely and independently? This "validation" task cannot be completed on the NanoVNA which has only one reflection port.
6) The forthcoming C++ software (ported Fortran code) will produce the DERDEI (uncertainty) contour calculations given reflection and/or transmission coefficient measurement results and a file of uncertainty boundary estimates as inputs. It is anticipated that the DERDEI uncertainty contour tool will be independent of the calibration algorithm outlined in 4 above, and thus serve the VNA community overall, without compromise, using reflection/transmission coefficient results provided from any source. To that end; I continue to recommend the use of the Touchstone format as the data format of choice for data entry into the DERDEI tool.
73
Gary, N3GO
Gyula Molnar
Rich NE1EE
Oristo
