Discussion: 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 design.

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 pulse generator.

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.