Some effects of error term interpolation on network analyzer uncertainties

Practically, many network analyzer measurements (both large signal and small signal) are performed at frequencies other than those where the calibration was performed in order to save time and effort. Typically, the error terms are simply interpolated in some way to calibrate at the new frequencies. Under what circumstances and how this interpolation is done can significantly affect the uncertainties of the measurement. This paper explores the characteristics of this interpolative error and how its impact can sometimes be assessed. Of particular note is that the hardware dependence of this effect is considerably larger than for many other components of uncertainty.