Figure 1 shows the
voltage waveform injected into a printed circuit board trace by the
method described in this article. In this case, the peak value is about
1.7 Volts with a width of a little over two nanoseconds. The amplitude
of the pulse actually injected into a circuit is a strong function of
the circuit design and the physical layout of the circuit being tested
and thus injecting noise pulses can be used to find weak points in a
The method used here injects pulses via mutual inductance as a series
voltage into the circuit to be tested equal to Mdi/dt of the output current
of a pulse generator. The ideal pulse generator would have a current
output with a fast risetime and a relatively slow falling edge. The
time derivative of this current would be a single unipolar pulse, like
that in Figure 1. The fast rising edge produces the injected pulse. The
falling edge would normally produce a negative injected pulse, but if
the falling edge is much slower than the rising edge, the negative
pulse will be so small in amplitude so to be insignificant.
There are two pulse generators I know of that have a fast rising edge
of output current and a slow falling edge with enough di/dt so as to
produce a usable injected pulse into circuits. The first is an
Electrical Fast Transient, EFT, generator that is used for the IEC 61000-4-4
immunity test. A brief description of this test can be seen on the
Thermo Scientific website by clicking here
.as well as an example of an EFT generator by clicking here
In the EFT test, a 50 Ohm generator produces an output waveform with a
5 ns rise and a much slower fall time, resulting in a pulse that is
about 50 ns wide at the 50% amplitude points. With a relatively slow
risetime of 5 ns, a large current is needed to produce enough
di/dt to inject reasonable voltages into circuits by mutual inductance.
Luckily, EFT generators have outputs to at least 4400 Volts, which
translates into a short circuit current of almost 90 Amperes. If you
already have an EFT generator or plan to get one, it will work in the
method described below. When I use this generator for noise injection,
a typical output setting would be around 1000-1500 Volts. It's output
is tailored to the IEC 61000-4-4 standard, but works well for noise
injection. One disadvantage of EFT generators is they tend to be bulky
and heavy, too heavy to carry around for field work or pack in carry-on
luggage for air travel.
Another option is a small pulse generator. model TG-EFT, from Fischer Custom Communications
that is shown in Figure 2. The pulse shown in Figure 1 was generated by
the TG-EFT coupled into a path on a board using the loop shown in
Figure 3. I use a similiar loop when an EFT generator is used as well. The loop
represents a load of about 80 nH to the end of the coax cable from the
If a loop driven by the TG-EFT is held adjacent to a similar loop
connected to a scope with a 50 Ohm input, the injected pulse delivered
to the scope has about a 300 ps
risetime. The waveform in Figure 1 was slowed somewhat by the 500 MHz
scope used and inductance in the circuit path on the board.
Figure 2. Fischer TG-EFT Transient Generator
With such a fast rising edge, much
less current is needed to produce the same di/dt as an EFT generator so
the TG-EFT generator is small by comparison and can be packed in
carry-on luggage on an airplane (although you may be questioned a bit
by the security people). Because of the faster edge, I tend to use an
output voltage (open circuit) of around 500 Volts or less when scanning
circuit boards, sometimes as low as 100 Volts. The TG-EFT will produce
up to a 2000 Volt pulse, although I have never had to use anywhere near
that amplitude in my investigations.
As mentioned earlier, I use a wire loop, such as the one in Figure 3,
to inject the pulses into printed circuit boards, wires, and even ICs.
The loop can be used to find sensitive areas of a system or board.
After some experience, one can correlate the amount of noise it takes
to affect the most sensitive part of the circuit to the circuit's
response to environmental noise like ESD or EFT.
The TG-EFT produces positive pulses, so the drop across the top of the
loop in Figure 3 will be positive to negative from left to right. The
injected voltage will have the same polarity (positive left, negative
right), but as a source not a drop, so the injected current will flow
in the opposite direction, from right to left.
A word of caution:
The ground side of the loop must be securely
connected to the shield of the coax from the pulse generator. If that
connection breaks, the whole loop will be at the open circuit voltage
of the generator.
Figure 3. Loop Used for Noise Injection
I have had a lot of fun injecting noise over the years, and in the
process have fixed a lot of designs. I have developed this method into
a very effective test procedure that is covered in detail in one of my seminars (pdf file)
An interesting note about the procedure is that it is often best done
by someone not involved in the design, such as a quality person testing
the robustness of a design.