Link EMI to ESD events
Don't underestimate the effects of radiated EMI on today's logic circuits.
John H. Mayer, Contributing Writer -- Test & Measurement World, 3/1/2002
Although most engineers recognize the benefits of controlling static charges that can damage sensitive devices, there's an additional reason to implement a comprehensive control program for electrostatic discharge (ESD). ESD events can also produce electromagnetic interference (EMI). Because logic devices have small noise margins, they're increasingly sensitive to EMI, which can cause hardware failures.
The EMI radiated from an ESD event can couple into a system's cables or into an open chassis where it changes to a voltage or current spike that can corrupt the operation of logic circuits. The EMI proves difficult to trace to a source because it can come from an ESD event almost anywhere in the room housing the electronic equipment. Moreover, the effect of EMI on ungrounded or unshielded conductors is often difficult to quantify.
The move by semiconductor manufacturers to use deep submicron process technologies has increased the vulnerability of ICs to EMI. Many logic devices reach logic 0 at 0.8 V or below and logic 1 at 2.0 V or above. That leaves a 1.2-V indeterminate range as a noise margin. Any induced EMI voltage that exceeds that range can "upset" logic devices (Ref. 1). Unfortunately, no standard methodology currently exists for measuring the EMI radiated from ESD events, but industry groups consider the threat serious enough to have begun examining the problem.
EMI—It's stronger than you thinkThe strength of ESD-induced EMI can be substantial. In the early '90s, Doug Smith, an independent consultant who specializes in the measurement of high-frequency pulses, looked at radiated EMI produced by common office furniture (Ref. 2). He found that the ESD generated inside some types of office chairs could radiate a series of impulse fields from the chairs' metal legs. The ESD occurred despite the application of normal ESD-reduction precautions, such as the use of wrist straps by chair occupants and the use of ESD-dissipative floor coverings. Tests showed no build-up of static charge in the chairs.
Smith developed tests to measure the EMI produced by ESD events when no one could link a specific cause to the failure of nearby equipment. The only hint of a cause: Failures seemed to occur when a person arose from a chair. Current measurements taken on system cables and on cables placed near a chair being tested showed substantial electromagnetic fields radiating from the metal legs of the chair. The source of this energy turned out to be a series of ESD events that occurred inside the chair. Within a 10-s period after a person rose from the chair undergoing testing, Smith recorded as many as 12 EMI pulses.
Calculations that Smith performed showed that an EMI field would have to reach 1 V/in. to create potential problems for office equipment. The fields he measured exceeded that value. In one instance, Smith found the EMI caused by an ESD event in a chair induced a voltage of 4 V/in. in cables up to 1 ft away.
Grab the waveformMeasuring the EMI radiated from an ESD event is not a simple task. Although you can use many types of antennas to detect EMI, most will not represent the waveform accurately. Often, these antennas have a resonant structure that oscillates, or rings, in response to energy generated by an ESD event. Moreover, many antennas have dispersive characteristics that distort fast pulses. So, an antenna may put out a waveform that indicates the presence of an ESD event, but typically the waveform doesn't accurately represent the characteristics of the radiated field.
 |
| Figure 1. A transverse electromagnetic (TEM) antenna picks up the EMI field radiated by an ESD event. Unlike other antennas, this antenna produces a signal that accurately represents the EMI field. Courtesy of D.C. Smith Consultants. |
The field strength (FS) measured by this antenna equals:
FS = Vm/D
where:
D = the distance in meters between the open end of the plates
Vm= the measured voltage (in volts) measured at the antenna terminals
It's easy to underestimate the threat that these fields pose to electronic equipment. Yet, a recent experiment (Ref. 3) shows just how powerful they can be. Investigators used a TEM antenna to measure the radiated fields generated during maintenance operations performed on a computer server. The measurements took place 1.5 m in front of and 1.5 m to the side of a server. In each case, the person who performed the maintenance was charged to 500 V.
In one test, the antenna measured the fields generated when the maintenance person inserted a power-supply module into the test server. Researchers recorded a peak voltage of 3.2 V, which translated into a field strength of approximately 22.1 V/m. That value exceeds by about a factor of 7 the maximum radiated immunity field strength (3 V/m) allowed by the requirements for the CE mark in Europe.
In a second test, researchers measured the fields generated when a person inserted a disk drive into a server, another common maintenance operation in computer centers. During this test, the electric field reached 19.3 V/m.
 |
| Figure 2. The EMI signal produced by an ESD event shows a high voltage as well as high-frequency ringing. Courtesy of ESD Association. |
Because computer centers space servers close together—usually closer than 1.5 m—an adjacent server would likely encounter more intense fields than those measured in these tests. Moreover, components within the test server would feel the effects of even higher field strengths because they're even closer to the ESD event.
The researchers also observed that the measured fields oscillated. They attributed the oscillations to the metal in the server and in the parts being inserted, as well as to the large size of the power supply and the disk-drive packages. The oscillations produce many high-frequency signal edges that further stress nearby equipment and increase the likelihood of equipment failures.
Keep in mind that the signals produced by these types of ESD events aren't generally repetitive. So, if you need to make similar tests and use a DSO to capture waveforms, the maximum sampling rate has more importance than the DSO's frequency response. In the server tests described above, researchers used an Agilent Technologies 1.5-GHz Infinium DSO (Model 54845A) with a sample rate of 8 Gsamples/s.
DSOs generally provide all the triggering capability you will need. To catch particularly elusive ESD events, use the runt-pulse triggering capabilities. By triggering on an event that exceeds one threshold but not a second, higher threshold, often you can trigger on the event you are looking for in a repetitive waveform rather than on the repetitive waveform itself (Ref. 4).
Take precautionsProperly grounding isolated conductors and placing ground planes near active conductors will help minimize the effects of EMI radiated by ESD events. As an additional precaution, you can shield potential EMI-emitting devices, but it's difficult to identify these beforehand.
It proves more practical to take into account the effects of radiated EMI when shielding devices most susceptible to damage. Whenever possible, reduce ground loops between interconnected equipment and systems by routing interconnected cables in conduits, cable trays, or raceways. Until the electronics industry develops standard methods to measure EMI radiated from ESD events, these precautions offer the best chance to minimize the damage from radiated EMI.
| References | ||
|
