You are on page 1of 7

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/280495863

Impact of Circuit Impedance on Signal Crosstalk

Article · January 2015

CITATIONS READS

2 3,152

3 authors, including:

K.Y. See
Nanyang Technological University
245 PUBLICATIONS   1,935 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Reduction of bearing degradation in marine propulsion drives View project

EMC Practical Papers View project

All content following this page was uploaded by K.Y. See on 28 July 2015.

The user has requested enhancement of the downloaded file.


Impact of Circuit Impedance on Signal Crosstalk

Lim Jing Yun Assoc Prof See Kye Yak


Victoria Junior College Dr Chua Eng Kee
School of Electrical and Electronic Engineering
path. It can also occur via electromagnetic (EM)
ABSTRACT fields: capacitive crosstalk (crosstalk via electric
field component) and inductive crosstalk
Crosstalk is an undesirable effect whereby (crosstalk via magnetic field component).
signals from a loop can create disturbances in
another loop. Crosstalk is affected by many 1.1 Capacitive Crosstalk
factors and this project aims to investigate the
effect of circuit impedance on signal crosstalk. Capacitive crosstalk occurs when the electric
Using a setup of three pairs of wires fixed at field from the charges of one conductor couples
differing distances apart, signals were to the other conductor. The phenomenon
transmitted in the aggressor line while the loops causes a coupled voltage that is proportional to
are terminated at large impedance (R=∞) capacitance between two circuits. Consider a
moderate impedance (R=50Ω) and low setup shown in Figure 1, C1𝐺 represents
impedance (R=4.7Ω). The frequency and capacitance between conductor 1 and the
waveform of the signal was varied. A digital ground, 𝐶2𝐺 represents capacitance between
oscilloscope was used to observe the conductor 2 and the ground; and 𝐶12 represents
behaviour of the near-end and far-end crosstalk stray capacitance between conductors 1 and 2.
in the frequency and time domain. The results 𝑉1 is the signal voltage in conductor 1
show that regardless of circuit impedance, (aggressor) and 𝑉𝑁 is the capacitive crosstalk
crosstalk increases with higher frequency and from conductor 1 to conductor 2 (victim).
decreases with higher separation distance Conductor 1
Conductor 2
between the loops. At low and high impedance,
𝑅
far and near end crosstalk are similar however 𝑅
at moderate impedance near end crosstalk is
slightly higher than far end crosstalk. The time- 𝐶12
𝑉cap
domain results indicate that at low impedance
𝐶1𝐺 𝐶2𝐺
near and far end crosstalk are anti-phase. At
high impedance, far-end and near-end
crosstalk are in-phase. Thus, we can better
utilize this knowledge to reduce crosstalk in 𝑉1 𝑅
printed circuit boards and cabling, enhancing
various electrical systems.

Key words: frequency-domain, time-domain,


Figure 1 Illustration of capacitive crosstalk.
Inductive crosstalk, Capacitive crosstalk
Capacitive crosstalk between the two circuits
1. INTRODUCTION can be defined by the equation (1) below.
The use of cables for signal communication
forms part of an electronic circuit. With
increasing signal frequency and the continuous (1)
trend of product miniaturization, unwanted 1
coupling of signals between adjacent In most practical cases, 𝑅 ≪ and (1)
𝑗𝜔(𝐶12 +𝐶2𝐺 )
conductors within a cable, also known as can be reduced to:
crosstalk can cause disturbance to the original
signal. The conductor causing the disturbance 𝑉𝑐𝑎𝑝 = 𝑗𝜔𝑅𝐶12 𝑉1 (2)
is usually known as the aggressor line while the
other conductor experiencing the interference is (Ott, 1988)
known as the victim line. There are various
ways crosstalk can occur. Crosstalk can occur 1.2 Inductive Crosstalk
when two conductors share a common return
When a current flows in one circuit produces a 3 METHODOLOGY
flux that couples to another circuit, the magnetic
coupling represented by a mutual inductance 3.1 Experimental Setup
𝑀12 between circuits 1 and 2 defined as:

𝜙12
Three pairs of 1 m long parallel copper wires
𝑀12 = (3) with specific separation distance (s =1 cm, 3 cm
𝐼1
and 4 cm) are laid on a 3 mm thick aluminium
where 𝜙12 represents the magnetic flux in plate as their common return path, as shown in
circuit 2 due to the current 𝐼1 in circuit 1. Figure 3. The thickness of the conductive plate
(Johnson, Johnson, Hillburn & Scott, 1999) is chosen such that the return impedance path
Consider a setup shown in Figure 2. The offered by the plate is minimal and thus
magnetic flux generate by circuit 1 cuts through crosstalk due to common impedance is
the closed loop of circuit 2 and induces a negligible. (Paul C. R., 1992)The three pairs of
volatge in circuit 2 given by: wires are supported by a 2 cm thick Styrofoam
to ensure uniform height along the wire.
𝑉𝑖𝑛𝑑 = 𝑗𝜔𝑀12 𝐼1 (4) (Goedbloed, 1992) Figure 4 shows the
(Ott, 1988) experimental setup. A Tektronix DPO 7354
Digital Phosphor Oscilloscope was used to
observe the crosstalk. A Tektronix AFG 3022
𝑅 function generator and Agilent MXG Vector
Signal generator were used to produce the
signals needed in the aggressor line. The
𝑀12 oscilloscope, signal and function generator
𝑉𝑠𝑜𝑢𝑟𝑐𝑒 were each set to 50 Ω termination. The various
resistors used were joint to the circuits via
𝑅2 connectors instead of additional wires. This was
to minimize additional parasitic inductance and
capacitance that could affect the measurement
accuracy at higher frequencies. (ECEE
Colorado)

𝑅1

Figure 2 Illustration of inductive crosstalk


S = 1cm
To quantify the level of crosstalk in the victim
line, the crosstalk signal voltage can be
measured at the same end of the victim line as
the transmitter of the aggressor line, which is
S=3cm
called near-end crosstalk. It could also be
measured at the other end of the victim line with
the respect of the interfering transmitter, which
is termed the far-end crosstalk. S=4c
m

2 AIMS AND OBJECTIVES


This paper investigates the impact of circuit
impedance on crosstalk over a range of
frequencies. The study involves low circuit Figure 3 Parallel copper wires for the crosstalk study
impedance (4.7 Ω), moderate circuit impedance
(50 Ω) impedance and high circuit impedance
(∞ Ω). The crosstalk measured would be
analysed via frequency-domain and time-
domain experimental data.
Figure 5. Far-end crosstalk (s = 1 cm)
Oscilloscope
Copper wires

Function
generator

Figure 4 Experimental set-up

3.2 Frequency-Domain Experiment


The signal generator was used to generate a
sinusoidal signal with varying frequencies from
10 kHz to 10 MHz through the aggressor line. Figure 6. Near-end crosstalk (s= 1cm)
Far-end and near-end crosstalk was measured
When 𝑅 = ∞, there is only capacitive coupling,
along the victim line using the digital
as there is no current flows in the aggressor line.
oscilloscope. The peak-to-peak voltage of the
On the other hand, when 𝑅 = 4.7Ω, inductive
crosstalk was then recorded. This was repeated
coupling dominates since the current is large in
over all 3 pairs of wires of different spacing
circuit 1, which produces higher magnetic flux
mentioned earlier. The experiment was then
to circuit 2. (Politecnico di Torino--Technocial
repeated with the termination of both aggressor
Univresity of Turin Torino, Italy, 2010) When
and victim line varied at 4.7 Ω, 50 Ω and ∞Ω.
𝑅 = 50Ω , both inductive and capacitive
crosstalk exist. As seen from Figures 5 and 6,
3.3 Time-Domain Experiment regardless of the circuit impedance, the signal
The function generator was used to generate a crosstalk is directly proportional to signal
square wave signal with a fixed 1 MHz frequency. It indicates that both capacitive and
frequency with peak-to-peak voltage of 2V inductive crosstalk increases with signal
through the aggressor line. The oscilloscope frequency. This can be explained by the
was used to record the pulse width and equations of inductive and capacitive crosstalk
amplitude of the crosstalk observed. The given in (2) and (4). There, we see inductive
phase difference between far-end and near- and capacitive crosstalk are directly
end crosstalk of the victim line was also proportional to 𝜔 and hence directly
observed. This was done for all three pairs of proportional to frequency.
wires with the aggressor line and victim line
being terminated at of 4.7 Ω, 50 Ω and ∞Ω. 4.1.1 Far-end and near-end
crosstalk
4 Results and Discussion

4.1 Frequency-Domain Results


Figure 7 Far-end and near-end crosstalk (s = 1 cm and R Figure 10 Far-end crosstalk when R = 50Ω
= 50Ω)

Figure 2 Near-end crosstalk when R=50Ω

Figure 8 Far-end and near-end crosstalk (s = 1cm and R =


4.7 Ω)

Figure 3 Far-end crosstalk when R=4.7 Ω

Figure 9 Far-end and near-end crosstalk (s = 1cm and R


=∞Ω)

Figures 8 and 9 show that at low (4.7 Ω) and


high impedances (R = ∞ Ω), far-end crosstalk
and near-end crosstalk are of the same
magnitude. As mentioned earlier, inductive
crosstalk is dominant at low impedance while
capacitive crosstalk is dominant at higher
impedance. Hence, both far-end and near-end
Figure 4 Far-end crosstalk when R =∞ Ω
crosstalk are similar. Figure 7 shows that at
As seen from Figures 10-13, despite increasing
moderate impedance (50 Ω), near-end
circuit impedance, the crosstalk between the
crosstalk is higher than far-end crosstalk. Since
aggressor and victim lines reduces with
both inductive and capacitive coupling exist, the
increasing separation distance. When the
near-end and far-end crosstalk will be different.
conductors are moved further apart, 𝐶12
It will be apparent when we analyse the results
decreases thus decreasing the capacitive
in time-domain later.
coupling on the victim line. In the case of
inductive coupling, the magnetic flux also
4.1.2 Crosstalk with varying reduces by physical separation of the wires.
separation distance When the impedance is moderate (50 Ω), there
is a mixture of both inductive and capacitive same circuit impedance, we would expect
crosstalk. Hence, if both inductive and 𝑉𝑓𝑎𝑟 < 𝑉𝑛𝑒𝑎𝑟 .
capacitive coupling decreases with increasing (intel)
distance, crosstalk when R = 50 Ω would also
decrease with increasing distance. 4.2.1 Time-domain crosstalk
waveform
4.2 Time-Domain Experiment

Figure 14 Capacitive and inductive couplings in time-


domain

As seen from Figure 14, when a time-domain


signal with a rising edge propagates down the Figure 5 Signal source and far-end crosstalk waveforms
aggressor line, a current is coupled in the victim
line through mutual capacitance given by: Figure 15 shows a typical waveform of the
crosstalk in time-domain. The crosstalk
𝑑𝑉 waveform is a series of repetitive spikes that
𝐼𝐶𝑚 = 𝐶𝑚 (5)
𝑑𝑡 occur when there is a change in signal source
voltage. As for high impedance circuit (R = ∞Ω),
At the instant the current is capacitively coupled capacitive crosstalk dominates and Figure 15
into the victim loop. Both ends of the victim loop shows that capacitive crosstalk is negative
are terminated with the same impedance (50 Ω). during the signal’s fall time and positive during
The capacitive coupled current splits into equal the signal’s rise time, as shown in Figure 15.
halves, one travels forward to the far-end and (Paul C. R., 2006)
the other travels in the reverse direction
towards the near-end.

According to Lenz’s law, the current induced in


the victim loop is in the direction that resists the
change in magnetic flux. Thus the inductive
induced current flows in the opposite direction
of the current in the aggressor line. The
inductive coupled current flows towards the
near-end. Hence when circuit impedance is
moderate (50 Ω), the capacitive and inductive
coupled currents add at the near-end:

𝐼𝑛𝑒𝑎𝑟 = 𝐼𝐶𝑚 + 𝐼𝐿𝑚 (6)


Figure 6 Time-domain far-end (yellow) and near-end
The far-end current is the difference of crosstalk (pink) (s=1cm and R=∞)
capacitive and inductive coupled currents:
The far-end and near-end capacitive crosstalk
𝐼𝑓𝑎𝑟 = 𝐼𝐶𝑚 − 𝐼𝐿𝑚 (7) are positive and therefore in-phase, as
explained earlier. This can also be observed in
Since 𝐼𝐿𝑚 is travelling backwards towards the figure 16. When termination impedance
near-end, a negative voltage is created at the decrease (R = 4.7 Ω), inductive crosstalk is now
far-end. Therefore, far-end current is the dominant. In Figure 17, we see that far-end and
subtraction of the inductive induced current near-end inductive crosstalk are anti-phase, as
from the capacitive induced current. For the the current inductively coupled to the victim
loop travels in the opposite direction from the
aggressor signal, hence far end inductive I would like to express my gratitude and
crosstalk is negative and is thus anti-phase. appreciation to my main-supervisor Assoc Prof
See Kye Yak and co-supervisor, Dr Chua Eng
Kee, for their guidance in this project. I also like
to thank the School of Electrical and Electronic
Engineering, Nanyang Technological
University for the laboratory safety induction
training. Finally, I thank the Talent Outreach
Department of Nanyang Technological
University, Mr Alvin Pek, Head of Science
Research of Victoria Junior College and Dr Goh
Ker Liang, my school research mentor for the
administrative support and research
methodology training.

Figure 17 Time-domain far-end (pink) and near-end


crosstalk (yellow) (s=1cm and R=4.7Ω)
References
Referring back to figure 16, we see that the ECEE Colorado. (n.d.). Retrieved from
osiclloscope is terminated at 50 Ω this means http://ecee.colorado.edu/~ecen3400/Chapter%
that though capacitive crosstalk is dominant 2018%20-%20Transmission%20Lines.pdf
there is also a substantial amount of inductive
crosstalk. Since at the far-end, capacitive Goedbloed, J. J. (1992). Crosstalk. In J. J.
crosstalk and inductive crosstalk are anti-phase Goedbloed, Electromagnetic Compatibility (pp.
the voltages ‘cancel’ each other out, making the 79-102). Prentice Hall.
crosstalk at the far-end lower than that of the intel. (n.d.). Retrieved from
near-end, as shown in figure 16. psec.uchicago.edu/library/anodes/Anodes_Cro
sstalk_Overview.ppt
5. Conclusion
Johnson, D. E., Johnson, J. R., Hilburn, J. L.,
The frequency-domain experimental results
& Scott, P. D. (1999). Electrical Circuit
show that the crosstalk increases with
Analysis third edition. John Wiley & Sons, Inc.
frequency regardless of circuit impedance. At
low and high circuit impedances, the near-end
and far-end crosstalk voltages are nearly the Ott, H. W. (1988). Cabling. In H. W. Ott, Noise
same, as either inductive or capacitive coupling Reduction Techniques in electronic systems
present. At moderate impedance, where there (pp. 29-72). John Wiley & Sons, Inc.
is a mixture of inductive and capacitive
couplings, the near-end crosstalk is higher than Paul, C. R. (1992). Experiments and
far-end crosstalk. This is because at the far-end, demonstrartions in electromagnetic capability.
capacitive crosstalk and inductive crosstalk In E. C. Society, EMC Education Material (pp.
cancels each other, while at the near-end, both 5-8).
capacitive and inductive crosstalk adds. This
phenomenon is well illustrated by the time- Paul, C. R. (2006). Crosstalk. In C. R. Paul,
domain crosstalk waveforms. We can therefore Introduction to Electromagnetic Compatibility
utilize this knowledge for real-life applications (pp. 595-617). Hoboken, New Jersey: John
and use the most effective method to reduce Wiley & Sons, Inc.
crosstalk. For example, separation distance
between loops should be increased instead of Politecnico di Torino--Technocial Univresity of
circuit impedance. Also, the signal frequency Turin Torino, Italy. (26 July , 2010). Retrieved
should be kept as low as possible. Further work from
could be done to investigate the behaviour of http://ww.umc.edu.dz/vf/images/emc/PDFs/Pa
signal crosstalk at higher frequencies, so as to pers/Workshops/005_170.pdf
have a more holistic understanding of the
behaviour of crosstalk at varying frequencies.

6. Acknowledgements

View publication stats

You might also like