# Coefficients of fringing capacitance polynomial

Question asked by frequencylover on Jan 4, 2013
Latest reply on Jan 27, 2013 by Dr_joel
Hi,

I'm working on a project that needs the characterization of a SOLT calibration kit.
The kit represents a transition from 3.5mm SMA connector to a microstrip line of length 20mm.
The measurement for frequencies above 3GHz requires the modelization of the open fringing capacitance as a polynomial of order 3.

I perform the following steps:

1- 1-port calibration at the end of the test cable (85033D calibration kit).
2- Connect short standard (20mm 50ohm microstrip line ended by 2-parallel 50ohm resistors) and make port extension until 180 degrees phase shift.
3- Disconnect short standard and connect the open one.
4- Adjust the delay until the phase is negative (Negative phase across the whole band).
5- Extract the formatted RI data and trace the parallel capacitance values as a function of frequency.
6- Use mathematical tool to fit the graph in order to evaluate the coefficients of the capacitance polynomial.

Note that the open phase trace is neither flat nor linear and reaches -28 deg at the end of the band (6 GHz) if we want negative phase for the whole operating band.

My problem is that the values of the coefficients of the capacitance polynomial are big.
(C0=38.33E-15; C1=79676.1E-27; C2=-35687.91E-36; C3=4270.35E-45).
The VNA doesn't accept values more that 10000 or less than -10000.

Actually, I tried to divide the band into two sub-bands, but that same problem results.
Moreover, I tried to minimize the band in order to achieve flat phase (less than 1 deg variation) but also the same trouble arises.