Tidbit discusses a method of measuring the relative phase of two common
mode currents on system cables and how to interpret the results. The
results of such a measurement can be useful in EMC emissions
troubleshooting as will be seen in the example presented.
Figure 1 shows a pair of matched F-33-1 current probes
on two system cables. Matching of the probes is important as the method
to be described requires the outputs of the two probes be subtracted
to give a result and the normal +/- 2 dB accuracy of the probes is not
close enough to yield good results.
A simplified diagram of the test
setup is shown in Figure 2. For this simple case, there are only three
cables: the power cable and two signal cables.
System with Three Cables
Suppose that common mode currents are measured on the two signal
cables. There are two possibilities for the current directions and
these have different mechanisms involving different "fixes" for the
problem. Take the case where the current enters on cable 1 from the
left in Figure 2, is pushed through the system by a ground potential
difference between the two signal cables, and then the current exits
on cable 2 to the right, always flowing from left to right. Driven this way, the two signal cables act
like a dipole antenna. One possible fix is to address the ground
potential difference between the two cables, if one exists. By "beefing
up" the ground reference between the two cables, the driving voltage of
the dipole will be reduced and so will the common mode current.
Another possibility is that a potential difference exists between the power
cable and the two signal cables taken together. This potential difference drives a current from
the power entry upward to exit on the two signal cables, flowing on cable
1 to the left and cable 2 to the right. In this case, changes to the
ground reference between the two signal cables will not have much effect, the
potential difference between the power cable and the signal cables
taken together must be reduced.
Which one of these two current flow possibilities is actually happening
can be determined by using two
matched current probes and either subtracting or adding their outputs by
connecting the current probes to a combiner. First, one probe is added
to a cable and the spectrum plot noted and then one notes how the
spectrum plot changes when the second current
probe is mounted on its cable. For cables whose connectors to the
system are very close, a single current probe can be used and the
effect of adding the second cable to the current probe is then noted.
An example is shown in Figure 3. I often have a 180 degree subtracting
available for other types of measurements so I will use it for this example. The use of a zero degree adding combiner is also
possible taking into account that the signals are added instead of
Be sure to put 50 Ohm feed-thru terminations (like we used to use on
older scopes to get a 50 Ohm input) at the current probes to keep the
combiner terminated in 50 Ohms. The combiner output should be connected
via 50 Ohm coax cable to the 50 Ohm input of a spectrum analyzer.
Adding the 50 Ohm termination to a current probe will decrease its
sensitivity in the flat region of frequency response to about 1/2 or
about -6 dB while extending the flat region of frequency response to
about 1/2 of the former low frequency cutoff. The effects of external
resistors on current probe response and impedance is covered in detail
in my paper "Current Probes, More Useful
Than You Think
" delivered at the 1998 IEEE EMC Symposium and in Chapter 8 of my book, High Frequency Measurements and Noise in Electronic Circuits
Figure 3. Example Spectrum Plots
We start with both current probes (with their 50 Ohm terminations)
connected to the combiner. Then one probe is attached to cable 1 of
Figure 2. The plot on the left in Figure 3 results and shows
several frequencies where
common mode current is flowing. Let's say that the current probe is
placed on cable 1 so that a current entering from the left produces a +
output. Upon adding the second probe in the same orientation to cable
2 (a current entering from the left produces a + output) the frequency
plot on the right results. Let's look at what happened at frequencies
f1, f3, and f5.
At frequency f5, the output went down to a strong null of over 20 dB
when the second probe was added. Since the probes are mounted in the
same orientation (current entering from the left produces a + output)
their outputs would be the same if current flowed from left to right
through both probes making cables 1 and 2 act like a dipole. Since
the current probe outputs are fed to a 180 degree subtracting combiner,
the output will drop as the second probe is added as happened at
frequency f5. Under these conditions, the reduction at frequency
f5 means that cables 1 and 2 are being driven as a dipole at frequency
f5 and a possible fix to try would be to "beef up" the ground reference
between the two cables and reduce the voltage driving the dipole.
At frequency f1, the output went up by 6 dB, a factor of two, when the
second probe was attached. Given the directions of the probes described
and the 180 degree subtracting combiner, the currents on cables 1 and 2
must be flowing away from the equipment under test (EUT) in opposite
directions on the two cables. This could mean there is a voltage between the power cable
and the signal cables driving a common mode current from the power
cable out through the signal cables. If so, the fix would be different
than the case at frequency f5.
What about frequency f3. Here the output went down, but only by 4 dB.
One possibility is that the current is flowing from left to right from
cable 1 to
cable 2, but there is some phase shift between the two currents,
possibly due to physical separation of the cables' connectors on the
system. The higher the frequency and the further the cable connectors
are apart, the more likely this is to happen. Another possibility is
that some current is also being supplied from the power cord, mixing
the two cases described above.
The amplitudes at f4 and f6 did not change. The currents at these
frequencies must be flowing on cable 1 but not much on cable 2.
What does it mean if the output goes up by 3 dB no matter which orientation the
second current probe is added to its cable? The answer is at the bottom
of this article.