Most traces on a
printed wiring board should not be resonant, especially digital signals
are usually driven by resistive sources and terminated in a lossy load
in some manner. Occasionally, a trace has a sharp resonance, such
as one going to a pin header that is not connected to anything. Such
is the case shown in Figure 1. The first problem is finding resonant
traces and the second is determining if a resonance is likely to cause
a problem. This Technical Tidbit will address the first problem and
touch on the second one for traces that run (at least partially) on
surface layers and are physically accessible.
In Figure 1, a one inch square shielded magnetic loop is being held
against a circuit board over a trace that goes from the flat pack IC on
the upper right to the pin headers on the lower left. Loop
construction is described in my 1999 IEEE EMC paper, Signal and Noise Measurement Techniques Using
Magnetic Field Probes (~600K pdf file).
Basically one can think of the loop as starting with a length of
semi-rigid coax of small diameter and shorting the center conductor to
the shield with solder at one end. Then the loop
is bent around to form a square (being careful not to bend the coax too
sharply at the corners) and the solder shorted end is soldered back on the coax so as to
form a square symmetric loop. A small gap is made in the shield in
the middle of the side opposite the feed line. An example of a very
small such loop can be seen in Figure 2 without a plastic housing, such as the one used to protect the loop in Figure 1.
In Figure 2 the loop was being used to inject a signal into the brass
wire to measure the resonant frequency of the circuit board and ground
plate underneath it connected by the short brass wire. Similar to what
we will be doing here, but at the system level instead of for
individual traces on a circuit board.
Figure 2. Example of a Small Shielded Loop
If a square shielded loop is held next to a trace which is resonant in
the frequency range being scanned, it will absorb energy from the loop
and the energy reflected from the loop will be reduced at that
frequency. One could either use a specctrum analyzer with a tracking
generator and external directional coupler or use a
network analyzer and just plot the reflection
coefficient looking into the cable feeding the loop. The June 2006 Techical Tidbit, Measuring Structural Resonances
how to optimize the measurement parameters of the instrument. For this
test, a shielded loop should be used to minimize errors caused by capacitance
between the loop and the structure to be measured.
The loop injects a voltage by mutual inductance (e = Mdi/dt). The
resonant frequency may be dependent on the location along the trace where
the voltage is injected so a location should be chosen near possible
sources of excitation of the trace of interest on the board. Examples include active devices or
coupling from other potentially noisy traces. One could start
looking for resonances in the unpowered state. This has the advantage
of not having board signals couple into the measurement but device
source impedances will be different from the powered state and may
result in higher Q resonances as well as affecting the measured resonant frequencies.
Figure 3 shows the
result from the test in Figure 1 as a plot of return loss relative to 50 Ohms. Notice the vertical scale is
only 1/2 dB/division. Since only a small amount of energy is
usually coupled into the circuit from the loop, this vertical scale
makes it easier to see resonances.
Wire Loop Construction
The data in Figure 3 clearly shows a resonance at about 243 MHz. Under
some circumstances this could be a problem. If a source on the board with energy at
243 MHz could couple into the trace at this point, an emissions problem
might occur. Also, coupled energy from outside of the board could
excite this resonance and since the trace is connected to a device, might
cause an immunity problem. In this example, the
PWB is a two layer design with
one layer being mostly a ground plane.
A resonance in a trace could also be caused by a missing termination,
something one would like to find. Alternately, a trace might be used as
an antenna by design, as in some wireless products. In that case, it
may be possible to check the antenna's resonant frequency using this method.
Scanning a board for trace resonances can uncover design issues.
After trace resonances are found one must decide if the resonance is a
problem or not.