Tektronix OM4006D, User Manual

Page: 63 / 205

Operating basics
When the Nyquist fi lter type option is selected a fi lter is inserted such that the
combination of the signal’s impulse response with the fi lter’s impulse response is
a Nyquist function, having zero ISI. In principle, there are many possible Nyquist
functions. The Nyquist function that has been implemented is a raised cosine
function, and the steepness (roll-off factor) of the raised cosine is matched to
the steepness of the signal spectrum. With the Nyquist
fi lter type, the ISI seen
in the eye diagrams should be minimal, but the fi lter may not suppress noise
as well as the matched fi lter.
The matched and Nyquist fi lters are available in the MATLAB workspace in
variables FIR (actual fi lter) and FIRCent (centered version). The fi lter can be used
later as a user-speci fi ed fi lter by assigning FIRCent to UserFilter. For example,
the Nyquist fi lter may be calculated accurately using a long record, and then
recalled later to be applied to short records.
Direct assignment of pattern variables.
When the transmitter is sending a PRBS pattern that is not one of the standard
patterns provided in the drop-down list, you can assign the PRBS polynomial
directly in the MATLAB Engine Command Window in the OUI. The acceptable
PRBS polynomials are of the form X
A
+ X
B
+ ... +1, where A > B. For example,
in the polynomial X
7
+X
5
+1, A = 7 and B = 5. This can be assigned to the Real
tributary of the X-polarization as PattXRe.PRBSGens = [5 7]; shown in the
following
fi gure. Select any standard PRBS in the Analysis Parameters tab. That
value will be overridden by the statement in the Matlab Engine Window. Any
PRBS polynomial can be speci
fi ed in this manner, enabling the use of different
sequences having the same length. For example, [5 9] and [4 6 7 9] are both valid
2
9-1
PRBS sequences.
Direct assignment of pattern variables when not using a PRBS.
When the
transmitter is sending something other than a PRBS, even if it is just a DQPSK
pre-code, the analyzer needs to know what data is being sent in order to calculate
the BER. In this case, it is necessary to load your pattern into MATLAB and
assign it to the pattern variable. You must also select User Pattern for the data
content in the Analysis Parameters tab.
PattXRe.Values = Seq1;
PattXRe.SyncFrameEnd = 100;
PattXlm.Values = Seq2;
OM4000D Series Coherent Lightwave Signal Analyzer 39