598
4, APRIL 2001
High-Frequency Characterization of Twisted-Pair Cables
José E. Schutt-Ainé, Senior Member, IEEE
Abstract—A wave propagation method model is proposed for
the extraction of propagation parameters of high-bit rate digital
subscriber twisted-pair cables. The model is derived from the scat-
tering parameters of a transmission line. The frequency depen-
dence and the skin effect are studied. Experimental simulations are
performed to evaluate the accuracy of the extracted parameters.
Good agreement is found in the cases studied.
Index Terms—Loss, scattering parameters, skin effect, twisted
pairs.
I. INTRODUCTION
T HE ABILITY to deliver multimedia services will increas-
ingly depend on the propagation characteristics of the loop
plant. Recent developments supporting high-bit rate digital sub-
scriber line (HDSL) and asynchronous digital subscriber line
(ADSL) were aimed at utilizing the existing infrastructure of
copper twisted-pair wires. Present subscriber lines are providing
digital access at high data rates, such that several major ef-
fects related to propagation degrade the performance of the loop
plant. These effects include the crosstalk and propagation loss.
Cable characteristics have been examined and studied for ef-
fects related to near-end crosstalk (NEXT) and far-end crosstalk
(FEXT) [1]–[3]. Other effects related to propagation loss have
also been investigated. However, no systematic characterization
of the loss and skin effect have been performed from an electro-
magnetic propagation standpoint.
In this study, we present an approach for the characterization
of the propagation properties of twisted-pair cables. This char-
acterization approach aims at determining the electrical param-
eters of twisted pairs via measurements. Scattering parameters
are used to relate measurements to the electrical parameters, and
measurements are performed and validated via simulation.
II. SCATTERING PARAMETERS OF TWISTED-PAIR CABLE
In order to fully characterize the cable properties, a set of net-
work parameters must be chosen. Since the high-frequency be-
havior of the twisted pairs is the focus of the study, we choose
a scattering parameter representation to characterize the cables
[4]–[6]. The parameters of interest are the resistance , induc-
tance , conductance , and capacitance per unit length.
These parameters define the propagation constant and the
characteristic impedance as
(1)
Paper approved by C.-L. Wang, the Editor for Wireless Spread Spectrum of
the IEEE Communications Society. Manuscript received December 17, 1999;
revised July 28, 2000 and September 20, 2000.
The author is with the Department of Electrical and Computer En-
gineering, University of Illinois, Urbana, IL 61801 USA (e-mail:
jose@decwa.ece.uiuc.edu).
Publisher Item Identifier S 0090-6778(01)03142-7.
0090–6778/01$10.00 © 2001 IEEE
IEEE TRANSACTIONS ON COMMUNICATIONS, VOL. 49, NO.
From network analyzer measurements at a frequency , with
, the scattering parameters are obtained. They can be
related to the line parameters using the relations [4]
and
(2)
(3)
is the reflection coefficient, is the propagation function,
and are the measured reflection and transmission scattering
parameters, respectively, is the length of the transmission line
and is the reference impedance of the measurement system.
can be rewritten as
(4)
in which the propagation constant and the attenuation con-
stant are separated. In most practical cases involving cables,
dielectric losses are very small so that the conductance per unit
length, can be ignored; also, when the ratio is very
small, an approximation for the complex propagation constant
gives
(5)
where is the propagation velocity in the cable. The first term
of is associated with the attenuation along the direction of
propagation. The second term is associated with the phase shift
and velocity. They can be extracted using
and (6)
At each frequency, and are obtained from the measured
scattering parameters using (2). Next, (5) and (6) are used to
approximate and . The results can then be combined with
(1) to obtain , , and .
III. MEASUREMENTS AND EXTRACTION
In order to extract the parameters of several types of
twisted-pair cables, measurements were performed in the time
and frequency domains using short and long segments of cables
of each type. The different types of cable that were measured
are listed in Table I.
In each case, a short segment of the cable was first measured
on a time-domain reflectometer (TDR). Since the total length
of a transmission line is small, the loss can be neglected and
the characteristic impedance and the propagation velocity can
IEEE TRANSACTIONS ON COMMUNICATIONS, VOL. 49, NO. 4, APRIL 2001 599

TABLE I
TYPES AND SPECIFICATIONS FOR MEASURED CABLES

Fig. 1. TDR response for CAT-5, 1-m cable. The observed trace yields a
characteristic impedance of approximately 100
. The propagation velocity is
approximately 0.21 m/ns or 0.74 c, where c is the speed of light.

be readily obtained. Such a TDR response is shown in Fig. 1 for

a type 1 category 5 (CAT5) cable. This measurement provides
and .
Next, swept frequency measurements are performed on a long
section (100 m) of the same cable using a network analyzer.
Using (2) and (3) with the measured and , we extract
for all the measured frequency points. Also, it is obvious that
so that can be extracted as
Fig. 2. Plots of resistance and propagation velocity ratio (v =c) versus
frequency obtained from measurement of type 1 cable.
(7)

and since , then

from which

From the network analyzer measurements, the resistance per

unit length and velocity ratio are obtained. Plots of the resistance
per unit length and the relative propagation velocity are
shown in Figs. 2 and 3. As anticipated, an increase in resistance
with frequency is observed due to the skin effect; this increase
varies depending on the cable type. Fluctuations of the charac-
teristics were the result of shape and twisting irregularities of
the cables; thus a curve fitting scheme was applied to the resis-
tance per unit length data and forced to take the form

(8)

where is the resistance per unit length in ohms per meter,

is the frequency in gigahertz, is the skin effect resistance
factor, and is the exponent. Another curve fitting scheme is
applied to the relative velocity as
Fig. 3. Plots of resistance and propagation velocity ratio (v =c) versus
(9) frequency obtained from measurement of type 2 cable.
600 IEEE TRANSACTIONS ON COMMUNICATIONS, VOL. 49, NO. 4, APRIL 2001

TABLE II
MEASURED PARAMETERS FOR THREE TYPES OF CABLE

Fig. 5. Transmission line configuration used for simulations.

Fig. 4. Plots of resistance and propagation velocity ratio (v =c) versus

frequency obtained from measurement of type 3 cable.

where is the relative velocity, is the dc relative velocity,

is the relative velocity factor, and is the frequency in gi-
gahertz. The curve fitting schemes permit to extract values for
, , and . These values are shown in Table II for the
different cable types.
In [7], Wheeler demonstrated that the skin effect leads to in-
crements in the resistance and internal reactance of a conductor.
The internal inductance results from the penetration of the mag-
netic field into the conductor. These effects can be accounted for
by modeling the complex per-unit-length impedance as

(10) Fig. 6. Near-end waveforms for type 3 cable.

where and are the resistive and reactive contribu-
tions of the skin effect. In can be shown [7] that . reflected pulses that are separated in space as a result of the
Using the frequency-dependent model derived from the ex-
tracted values, a transmission-line simulation program is used
to predict the response of a pulse traveling along the cable. The
circuit schematic is shown Fig. 5. The magnitude of the exci-
tation pulse is 2.7 V, the rise and fall times are 1 ns, the pulse
width is 80 ns. The capacitance and inductance per unit length
, and are obtained using the characteristic impedance and
the propagation constant. Equation (10) is used for the complex
impedance per unit length. Figs. 6 and 7 show the computer sim-
ulations as well as the experimental simulations.
An examination of the near-end waveform shows incident and
delay in the cable. Since the line is open at the far end, an ideal
propagation (i.e., through a lossless cable) would have yielded a
reflected pulse of identical magnitude and shape as the incident
pulse. Thus, the attenuation observed on the reflected pulse is
a measure of the loss in the cable; the spreading and rise time
degradation are a measure of the dispersion.
It was observed that the loss due to the skin effect was a major
contributor in reducing the bandwidth of the cables. This loss
increases with cable length and operating speed or frequency.
It can also be shown that the reactive component of the skin
IEEE TRANSACTIONS ON COMMUNICATIONS, VOL. 49, NO. 4, APRIL 2001 601

IV. CONCLUSIONS
In this work, a wave propagation model was derived for signal
propagation on twisted-pair cables. From the model, frequency-
dependent parameters were measured and extracted using a net-
work analyzer. These parameters are critical in predicting signal
propagation on high-bit-rate digital subscriber lines. A method
was implemented to extract the skin effect in these lines; this ef-
fect was found to be significant and was a serious limiting factor
that reduced the bandwidth of the cables. In addition, the skin
effect caused a substantial increase in the effective inductance
of the conductors. All these characteristics were extracted from
the measurements and the resulting models were found to accu-
rately predict signal propagation on the cables.

Fig. 7. Far-end waveforms for type 3 cable.
effect as expressed by (10) was critical in the reproduction of the
reflected waveform. This reactive component models the effects
of the magnetic field induced within the imperfect conductor.
REFERENCES
[1] J. Lechleider, “High bit rate digital subscriber lines: A review of HDSL
progress,” IEEE J. Select. Areas Commun., vol. 9, pp. 769–784, Aug.
1991.
[2] J. Werner, “The HDSL environment,” IEEE J. Select. Areas Commun.,
vol. 9, pp. 785–800, Aug. 1991.
[3] S. Ahamed, P. Gruber, and J. Werner, “Digital subscriber line (HDSL
and ADSL) capacity of the outside loop plant,” IEEE J. Select. Areas
Commun., vol. 13, pp. 1540–1549, Dec. 1995.
[4] J. E. Schutt-Ainé and R. Mittra, “Modeling and simulation of high-speed
digital circuit interconnections,”, EMC Lab. Tech. Rep. 88-2, Apr. 1988.
[5] “S -parameter design,”, Hewlett-Packard Application Note 154, Apr.
1972.
[6] J. E. Schutt-Aine and R. Mittra, “Scattering parameter transient analysis
of transmission lines loaded with nonlinear terminations,” IEEE Trans.
Microwave Theory Tech., vol. 36, pp. 529–536, Mar. 1988.
[7] H. A. Wheeler, “Formulas for the skin effect,” Proc. IRE, vol. 30, pp.
412–424, Sept. 1942.