Tracking down ESD

You know something is not right. Somewhere in your system, charge builds up until there is an arc, a spark, and the resulting noise pulse compromises system performance. Tracking down the cause of this spark can be frustrating, but a trick of the trade can help you zero in more quickly.

Electrostatic discharge (ESD), electrical fast transients (EFT) and other impulse sources can cause electronic equipment to fail or induce soft errors that compromise data. Finding ways to eliminate these ESD events grows more essential as high-performance devices become more sensitive to ESD due to thinner oxide levels. Often, simply using ESD-dissipative materials is not enough. You have to find the ESD source and eliminate it.

Simply looking over an entire system to try to visually identify problem areas is time consuming and frustrating, with a low probability of success. But you can localize the search area, says consultant Douglas C. Smith (Los Gatos, CA; www.dsmith.org), an expert in electromagnetic compatibility test and design. With a few antennas and a fast digitizing oscilloscope, Smith says, you can use the time of arrival of the ESD noise pulse to zero in on the source.

Fig. 1  Multiple antennas connected to an oscilloscope allow you to triangulate on the source of an ESD event by measuring differences in the noise pulse arrival time. With some careful thought and moveable antennas, you can achieve the same result using only two.

Knowing the differences in arrival time as well as the antenna locations allows you to calculate the location of the noise source. A 4-Gsample/s oscilloscope permits you to resolve distances to within a few inches. Computational algorithms for calculating the location are described in papers by J. Bernier and D. Lin in the 1997 EOS/ESD Symposium Proceedings, available from the ESD Association (www.esda.org).

As Smith points out, however, using this method requires careful control of the cable delays from the antennas to the oscilloscope to ensure that they match. It also requires a three-channel oscilloscope that offers the needed sampling speed. The need for a third channel adds to the complexity and expense of the method; you can sidestep this need by using moveable antennas and some reasoning.

Using two antennas only gives you one time difference to work with, which results in a parabolic surface for the locus of possible source locations—not enough information to localize the source. By moving one antenna and recreating the event, however, you gain additional information. If you move antenna A 2 ft closer to antenna B, for instance, and the repeat test shows the time of arrival changes by 2 ns, you know that the source must be along the line between the two antennas. Other outcomes reduce the possible locations to an area, but the physical configuration of the equipment can help narrow that down as the source is not likely to be outside of specific regions in most designs.

If the ESD event repeats quickly enough, two antennas mounted on a cart can localize the event in real time. Rolling the cart past the equipment, you can quickly determine when the source is halfway between the two antennas. Make two such passes from different angles, and you can identify a narrow column within the equipment under test where the ESD source must lie.