SC5305A Operating & Programming Manual
Rev 2.1.0
28
Measured IF gain and phase error response is available for both filter paths; the user simply needs to
properly select the path of interest. The measurement is made using a vector network analyzer in the
frequency domain, covered by fifty-one evenly spaced frequency points. The amplitude (gain) error
values are measured with respect to the center frequency and are given in decibels, while the phase
error values are in radians. The phase errors are deviations from linear phase. Each set of calibrated
points consists of a 3x51 floating point array (see Table 7 as an example). There are several ways to
apply the frequency domain calibration:
1.
Determine a fitted polynomial function for the amplitude error (gain) and multiply this function with
the uncorrected amplitude spectrum. Add the two values if dealing in decibels. Additionally,
determine a fitted polynomial function for the phase error and add values derived from this function
with the uncorrected phase. To derive, let
be the measured uncorrected value,
be
the fitted polynomial to the calibrated error values, and
be the corrected measured value.
Also let
denote the principle value of the phase of the above terms, and we can relate all the
terms as
From the above equation, we see that the magnitude terms are multiplied and the phase terms
added. In the discrete sense (digitized) for every frequency value,
, we apply the above equation
to correct for the non-ideality of the IF filter.
2.
The other method finds the magnitude and error points through interpolation methods such as
Spline, then multiplying the error magnitude with the uncorrected magnitude and adding the error
and uncorrected phases. This is similar to method 1, but instead of using a fitted function to obtain
the error values, interpolation is used. Interpolation is generally a slower process. This is the method
implemented in the library function
sc5305a_CalcIfResponseCorrection
.
