Figure 2 shows a typical output from the antenna and test setup of Figure 1. The waveform in Figure 2 exceeds the range of the scope and is "off the scale" at more than 4 Volts! Note that one Volt per division is the least sensitive range of the scope. Figure 3 shows the waveform expanded out to 2 ns/div from the 10ns/div used in Figure 2.




In the first 5 ns of Figure 3 there is a very sharp peak exceeding 4 volts. The oscillations preceding it are between 3 and 4 GHz, well past the 1.5 GHz bandwidth of the scope, and reflecting the high bandwidth of low voltage ESD. It is likely that that first peak is greater than 10 Volts in reality. Just to the left of the center of the screen,  a series of peaks occur with a frequency of about 1.5 GHz and with amplitudes well over 4 Volts. Judging by the appearance of the peaks and the 1.5 GHz frequency response of the scope, I would estimate the peaks are nearly 10 Volts.

If the waveform of Figure 3 were further expanded, the plot would look jagged because some parts of the waveform are very fast and require a sampling rate well above the 8 GSa/sec of the scope used here in order to accurately display the waveform. Sin(x)/x interpolation is turned off for these measurements.

With peak voltages on the order of 10 Volts (substantially larger than the transmit signal normally used in 802.11b), there is the possibility of overloading of the receiver front end. If this occurs, receiver performance will be degraded for sometime after the event until the receiver recovers. If sensitive devices are used in the receiver front end, protection may be required depending on the likelihood of nearby ESD and the front end design. If the jingling change were held very close to the antenna, the resultant voltages would be substantially greater as well.

Another interesting case is that of a mobile phone placed in a pocket or handbag with loose change. The phone would be imbedded with the ESD events! This very close range ESD could result in large signals picked up by the phone's antenna. The amplitudes may be significantly greater that those shown in Figures 2 and 3 raising the possibility of receiver front end damage. In a related, although somewhat different situation, I know of one case where ESD occurring within a mobile phone by static induction damaged the receiver front end and reduced the receiver sensitivity permanently by a significant amount.
ESD event location techniques often use calculations based on the time of arrival of an ESD generated wavefront at 3 or 4 antennas. By using sensitive directional antennas, like the one used in this experiment, one can better define the area of interest and achieve better signal to noise ratios compared to dipoles or omni-directional antennas. Research into ESD event location by time calculations was done at Bell Labs about 8 years ago and more recently at Ion Systems in Berkeley, California.

Summary: ESD generated EMI can have strong effects on systems. Data presented here raises the possibility that such EMI may also have an effect on high frequency RF systems such as wireless LANs. RF designers may find the data presented here useful in their designs.

Discussion of wireless LAN transmit power can be found at: http://www.wi-fiplanet.com/tutorials/article.php/1428941

A short description of a patch antenna can be seen at: http://home.iae.nl/users/plundahl/antenne/patchant.htm

Additional information on this site regarding ESD effects on systems includes:

• November 1999: Transient Suppression Plane
• May 2001, Hidden Threats to Electronic Equipment
• June 2001, A Static Field Powered EMI Source
• January 2002, Cable Effects Part 1: Cable Discharge Events
• May 2002, Printed Wiring Board Coupling to a Nearby Metal Plane, Part 2: ESD Immunity
• February 2003, Crossing Ground Plane Breaks - Part 3, Immunity to Radiated EMI
• May 2003, Signal Paths Passing Through Ground and Power Planes, Effects on Immunity
• ESD Immunity in System Designs, System Field Experiences and Effects of PWB Layout (~950K)
• (2000 EOS/ESD Symposium paper)
  

Thanks to Agilent Technologies for supplying the scope for this experiment. The model used for this article was an Agilent Technologies 54845a, an 8 GSa/sec unit that is now replaced by a much faster scope, the 54853a.