Insertion Loss Deviation
Impedance uniformity is an increasingly important parameter
to understand, measure, and quantify for high speed full duplex
transmission systems. The most common way to specify cable roughness
or impedance uniformity has been to measure return loss. Since
return loss is a reflection measurement, the amount of impedance
variation measured becomes restricted at high frequencies to
the first few meters of cabling. There is an interest in looking
at the degree of impedance uniformity over an entire 100 meter
segment in such a way as the high frequency components or roughness
are not masked or attenuated by distance.
One way to accomplish these objectives is to make a through
measurement rather than a reflection measurement. When insertion
loss is measured on links exhibiting structural impedance
variations, a ripple occurs in the insertion loss results
at high frequencies (typically above 75 MHz). This ripple
increases in magnitude as a function of frequency and the
amount of structure in the cable. Insertion loss deviation
is a measure of the worst case difference in magnitude between
the expected insertion loss and the actual measured insertion
loss. Insertion loss deviation is measured by first finding
the insertion loss, and then computing the maximum amplitude
across the specified frequency range between the insertion
loss and the least squares curve that fits the insertion loss
data.
The term "insertion loss" is used instead of attenuation
because attenuation assumes matching impedance between the
system under test and the test device. For insertion loss
measurements the test device is set at 100 ohms and the system
under test may have an input impedance between 85 and 115
ohms.
Experiments show that return loss is not necessarily correlated
to insertion loss deviation.
Results Interpretation
While insertion loss deviation is under study as a Category
6 link test, there are as yet no pass/fail limits set. All
that can be said is the minimum possible insertion loss is
desirable.
As an illustration of insertion loss deviation, two Category
5 cables and one Category 6 cable were tested with a network
analyzer. Attenuation and return loss were measured, then
insertion loss deviation computed. All three results were
plotted on the same graph to 300 MHz.
Category 5 cable C shows a correspondence between an insertion
loss minima at 112 MHz and a return loss maxima. The worst
case insertion loss deviation on cable C was slightly less
than 2 dB. The worst case insertion loss deviation on cable
B was much worse, at 8 dB, yet cable B showed better return
loss performance. This suggests some structure effects are
only evident at higher frequencies. Because return loss is
a reflection measurement, much of these high frequency effects
are not seen if they are more than a few meters from the measurement
port (due to attenuation effects).
|