EECS423Lect16

1
Lecture 16
In this lecture we will discuss comparisons
between theoretical predictions of radiation and
actual measurements.
The experiments consider simple configurations to
test the validity of the prediction model.
2
Experiment 1
5V
28 Gauge
50 mils
74LS04
74LS04
10 MHz Oscillator
50 mils
inverter gate
1 meter of 3-wire ribbon cable
Figure 1 schematic of the device tested
This system was placed inside a semianechoic
chamber to measure radiation in the frequency
range of 30-200MHz. The measuring antenna was
biconical. The setup of the semianechoic chamber
is shown in the following.
.
Ribbon cable and current probe
3m
antenna
direct ray
1m
reflected
1m
ground plane
1m
image
Figure 2 physical dimensions of the measurement
site
3
A current probe was placed at the midpoint of the
cable to measure the common-mode current. The
probe was connected to a spectrum analyzer and,
as previously seen, the relationship between the
current and the measured voltage is
(1)
The electric field prediction needs to account
for the ground reflection, so by introducing a
correction factor we have
(2)
Combining equations (1) and (2) gives
(3)
4
The comparison between the predicted radiated
fields computed using equation (3) and the
measurements is shown here
STD
Figure3 Measured and predicted emissions of the
device of Fig. 1 (source Paul)
5
The theoretical values are shown with an X and
the agreement is within 3dB except at 50MHz,
80MHz and 130MHz. If one removes the active load
74LS04 at the right of Fig. 1, the common-mode
currents do not substantially change and this
result is shown below
Figure 4 Measured and predicted emissions of the
device of Fig. 1 (source Paul)
6
The conclusion from the discussion about the
previous experiment is that common-mode currents
are usually the dominant radiation mechanism for
long cable.
Common-mode current also constitute a dominant
radiation mechanism for conductor lands on a
PCB. Let us consider the following configuration
380 mils
25mils
25mils
10 MHz Oscillator
glass epoxy, ?r4.7
glass epoxy, ?r4.7
62 mils
6 inches
14 pin DIP oscillator
Figure 5 device schematic and the PCB
cross-sectional dimensions
This configuration is considerably small and
symmetric and one way be tempted to think that
common-mode currents should not exist.
7
The radiated emissions are compared with the
theoretical predictions in the following
Figure 6 Measured and predicted emissions of the
device of Fig. 5
It is apparent that the common-mode currents
provide the dominant contribution to the
radiated emissions. Again, if one removes the 330
? load of Fig. 5 the common-mode currents are
practically unchanged.
8
So far we have considered radiation models for
wires and PCB lands. These models help us
evaluate the fields that are created when these
elements carry common-mode and differential-mode
currents.
Another important consideration, from an EMC
standpoint, is the prediction of the effect of
an already existing field onto a pair of wires or
PCB lands. Hence, we are ready to start
discussing issue of susceptibility models for
wires and PCB lands.
We consider the following two-conductor line
problem
S
Figure 7 two-conductor line model
9
This model applies to any two-conductor
transmission line as long as the appropriate
per-unit-length parameters are used.
The goal is to predict the terminal voltages
and once the incident plane were
is known.
The component of transverse to the plane of
the wires induces an emf that can be viewed as
an inducted voltage source of strength
(4)
The component of transverse to the wires and
directed along creates an induced current source
given by
(5)
10
The incident fields nearby the line are related
to their source by the Friis formula
(6)
and
(7)
Using a distributed parameter model,
Fig. 8
We can write
(8)
11
Therefore, we obtain the transmission line
equation
(9)
In the special case of an electrically short line
if we neglect the per-unit- length inductance and
capacitance we obtain
Fig. 9
Hence, equations of (7) simplify into
(10)
12
(11)
This simple model applies to electrically short
lines where the termination impedance does not
differ significantly from the characteristic
impedance of the line. By introducing in this
model the per-unit-length capacitance and
inductance one can relax the conditions on the
termination impedance, while keeping the
assumption of electrically short line.