User guide

This function, applied to the high level and low level data points, yields new
threshold vs. value combinations.
In the area of low BER (typically below 10
-4
), these new data pairs should fit to two
straight lines, although a couple of assumptions and approximations have been
made.
To determine the gradient and offset of these lines, a linear regression is performed.
This is illustrated in the figure below.
A straight line can be expressed as:
Y = A + BX
where Y is the inverse error function of BER, and X is D, the decision threshold.
The following calculations are performed for the high and low level data:
S XY -
(S X )(S Y )
n
S X
2
-
(S X )
2
n
B =
S Y
n
A = - B
S X
n
where n is the number of respective data points.
The results of the linear regression are displayed in the QBER vs. Threshold graph.
6 Advanced Analysis
284 Agilent J-BERT N4903B High-Performance Serial BERT