Discussion: Figure 1 shows the
test setup used for this Technical Tidbit. The setup is composed of two copper clad boards connected by a wire about
four inches (~10 cm) long, a Fischer TG-EFT pulse generator, two
current probes, and a scope. The two boards are separated about an inch
(~2.5 cm) by a paperback book. One current probe, a
Fischer F-62b, was used to inject pulses and the other, a
Fischer F-65, was used to monitor the current.
A
Fischer Custom Communications
TG-EFT high voltage pulse generator was used to drive the F-62b current
probe. The TG-EFT output has a fast risetime of about 2 nanoseconds and
a slow fall time of several tens of nanoseconds. Since the F-62b has a
low frequency cutoff well above 100
MHz, only the higher frequency components of the TG-EFT are injected.
This characteristic of the F-62b is not necessary for many
investigations but can be useful as well.
Figure 2 shows the F-62b clamped around the ground wire shorting an
Agilent
1163a
500 Ohm resistive one GHz probe. The 500 Ohm input resistance of the probe
is high enough so that the probe measures the open circuit voltage that
the F-62b injects into circuits. The result is shown in Figure 3.
Figure 2. Measuring the Open Circuit Voltage Injected by the
F-62b Current Probe
Figure 3. Open Circuit Voltage Induced in the
Agilent 1163A Probe
(Vertical scale = 2 Volts/div, Horizontal scale = 20 ns/div)
In Figure 3, one can see a peak voltage of about 8 volts (for a TG-EFT
setting of about 500 Volts open circuit). Figure 4 shows the
signal expanded from 20 ns/div to 5 ns/div. The smaller peaks after the
initial one are reflections off the
series 50 Ohm termination I was using with the TG-EFT generator.
Figure 4. Open Circuit Voltage Induced in the
Agilent 1163A Probe
(Vertical scale = 2 Volts/div, Horizontal scale = 5 ns/div)
Figure 5 shows the two current probes attached to the wire connecting
the copper clad boards and Figure 6 shows the resulting waveform
measured by the F-65 current probe.
Figure 5. Close-up Showing Two Current Probes on the Wire Connecting the Copper Clad Boards
Figure 6. Current Measured Between Two Copper Clad Boards
(Vertical scale = 100 mA/div, Horizontal scale = 100 ns/div)
A
damped sinusoid is clearly seen with an amplitude of about 400 mA peak
(the F-65 has a one Ohm or one Volt/Amp transfer impedance). Figure 7
expands the waveform from 100 ns/div out to 20 ns/div where the
oscillating frequency is seen to be about 77 MHz. The capacitance of
the two copper clad boards together with the inductance of the wire
connecting them results in a fairly high Q resonant circuit at 77 MHz.
I have seen structures like this in many electronic products and such
structures can result in operational problems or failures of EMC tests.
Figure 7. Current Measured Between Two Copper Clad Boards
(Vertical scale = 100 mA/div, Horizontal scale = 20 ns/div)
Figure 8 shows the paperback book removed from between the copper
clad boards and replaced with a plastic bag so the spacing is now
around 1 mm instead of a few cm. Figure 9 shows the result and Figure
10 expands the time scale from 50 ns/div to 20 ns/div.
Figure 8. Copper Clad Boards Separated by Plastic Bag
Figure 9. Current Measured Between Two Copper Clad Boards at 1 mm Spacing
(Vertical scale = 200 mA/div, Horizontal scale = 50 ns/div)
Figure 10. Current Measured Between Two Copper Clad Boards at 1 mm Spacing
(Vertical scale = 200 mA/div, Horizontal scale = 20 ns/div)
The waveforms in Figures 9 and 10
are different in two main ways from the waveforms in Figures 6 and 7. First the frequency of oscillation is
lower, now about 34 MHz instead of 77 MHz. In addition the
amplitude has increased to nearly 600 mA, up from a little over 400 mA. These
effects are related to the increased capacitance between the boards.
Resonance measurements such as those above can be made with
magnetic loop probes,
but the use of current probes can give a wider bandwidth, greater
amplitude of the output signal, and a more repeatable measurement.