HP 86037B, Инструкция по эксплуатации

Страница: 124 / 204

5-4
Maintenance
Troubleshooting Common Problems
If the RF signal power is low
❒
Perform a modulator bias adjustment.
If the data looks incorrect
❒
If the test path contains an optical amplifier, check for possible overdriving of
the receiver input of the CD test set. The input level should be less than –3
dBm. Also assure the measurement wavelength range is appropriate for the
pass band of the amplified system.
If problems occur testing narrowband devices
The five-term Sellmeier fit is generally inappropriate for very narrowband
devices, such as Fabry-Perot filters. The result can be roughness in the fitted
curve and multiple values of lambda zero. This is not a problem when testing
fibers.
If results of different curve fit models do not agree
The three curve fit types differ in their ability to conform to the actual disper-
sion curve of the tested fiber. As a result, a variation on the order of a few hun-
dredths to a few tenths of a nanometer can be observed in the value of lambda
zero. The three-term Sellmeier curve fit tends to yield a lower lambda zero
value than the other two fits.
Although the five-term Sellmeier curve fit has greater ability to fit the mea-
sured data, it is also more easily affected by noise. Therefore, the quadratic fit
will generally give a more repeatable value of lambda zero. For most DSF, the
mean values of lambda zero returned by the five-term and quadratic curve fits
are in good agreement while the three-term fit yields a lower value.
If fitted curve does not follow the actual data
The shape of the curve fitted to the relative group delay data depends upon
the device under test and the type of fit selected. Conditions of high DUT
insertion loss, low DUT dispersion or narrow wavelength range can also affect
the quality of the fit. The system will always try to fit the selected curve to the