Check ESD simulators first

When you are designing products for compliance with electromagnetic compatibility (EMC) standards, testing circuits for immunity to electrostatic discharge (ESD) is a must. Standards such as IEC 61000-4-2 and ANSI C63.16, which specify how to set up and perform these ESD tests (Refs. 1, 2), require that you use an ESD simulator to produce test pulses.

The standards also specify the shape and timing of the current pulse you must inject into your equipment under test (EUT), so before running an immunity test, you must verify that your ESD simulator produces a current pulse with the proper shape and rise time. You can verify a simulator’s performance by using a calibrated ESD target and a high-bandwidth oscilloscope.

Figure 1.  The current from an ESD event has a rise time of less than 1 ns.

You can model the human body with a simple series RC network (Figure 2). As the electric charge builds, the capacitor charges to several thousand volts. When the switch is flipped, this charge discharges rapidly into the EUT. Several manufacturers offer simulators that reproduce current waveforms very close to this human-body model. The wave shape these simulators must generate is specified in IEC 61000-4-2.

IEC 61000-4-2 requires that you verify the ESD simulator’s tip voltage before testing your EUT. It also requires that you verify several characteristics of the resulting current waveform, such as current peak, current reading at 30 ns, and current reading at 60 ns.

Figure 2.  An RC network simulates an ESD event from a person’s finger.

To check an ESD simulator’s output, you must measure the waveform of the resulting current across a low-impedance, high-frequency resistive shunt connected to ground. This shunt, or ESD target, emulates a discharge into a large metallic object such as an equipment enclosure.

Two styles of ESD targets: older style (top) and newer style (bottom).

When setting up your test, you must mount the target in the center of a 1.2-m² ground plane. ANSI C63.16 target specifications include a reflection coefficient of less than 0.1 (equivalent VSWR of less than 1.22) and an insertion loss of less than 0.3 dB up to 4 GHz.

To complete the test setup, you’ll need cables, attenuators, and an oscilloscope. Use good-quality low-loss cables between the target, the attenuators, and the oscilloscope. Keep the total cable length less than 1 m so you comply with the IEC and ANSI standards. ANSI C63.16 requires a double-shielded cable that prevents signal leakage from affecting your measurement. It also recommends RG-400/U cable, but RG-214/U—although twice the diameter—has half the loss and seems to work well. You can also use any gigahertz-bandwidth coax cable.

IEC 61000-4-2 also specifies that you place the oscilloscope inside a Faraday cage to shield it from ESD-induced radiated emissions. During the time the standard was developed (early 1990s), many engineers made these measurements with analog oscilloscopes. The standard specified a shield to prevent distortion in the displayed waveform on an analog oscilloscope. The shield also minimized false triggering caused by fields emitted from the discharge.

Today, most high-speed digital oscilloscopes have well-shielded input circuits, so the Faraday cage may not be required in practice. I’ve found simply mounting the ESD target in the center of a 1.2- m² sheet of aluminum prevents unwanted triggers in well-shielded oscilloscopes.

Figure 3.  Attenuators between an ESD target and an oscilloscope protect the instrument’s input amplifiers.

When executing your tests using air discharge, try to approach the target with the ESD simulator from a 90° angle and at a constant speed. You’ll maximize repeatability, but you can expect to see a lot of variation in wave shape. For contact-discharge tests, place the tip directly on the target prior to discharging the simulator.

The target-attenuator-cable chain will produce some loss of signal amplitude. Variations in loss from one test setup to another must be ±0.3 dB from DC to 1 GHz and ±0.8 dB from 1 GHz to 4 GHz. Table 1 shows that system-accuracy variations of less than 1 dB can greatly affect measurement accuracy.

Figure 4.  An ESD contact discharge, when run through a 20-dB attenuator, peaks at 3 V, but is actually 30 V.

When choosing an oscilloscope, look carefully at an instrument’s bandwidth, rise time, and noise. To accurately measure the signal without sampling errors, an oscilloscope must have sufficient bandwidth. For a Gaussian-response oscilloscope, you may need a sample rate up to six times the oscilloscope’s bandwidth, although four times the bandwidth is more typical.

With a digital oscilloscope, you must also pay attention to sample rate. A digital oscilloscope has a more flat response over its usable bandwidth, and it has a sharp roll-off above its 3-dB frequency. Thus, you need a sample rate of 2.5X the oscilloscope’s bandwidth to avoid alias errors.

In order for an oscilloscope to accurately display an ESD pulse’s rise time, it must have sufficient bandwidth and rise time. The rules for determining whether a scope’s specifications are adequate differ for analog and digital models (Ref. 4).

For analog oscilloscopes, the generally accepted rise time and bandwidth rules were:

• Bandwidth = 0.35/(rise time), or rise time = 0.35/bandwidth.
• The oscilloscope must have less than one third the rise time of the incoming signal in order to measure the rise time with an error of 5% or less.

For digital oscilloscopes, use the following calculations:

• Bandwidth ~0.43/(rise time)
• The oscilloscope’s rise time only needs to be ~0.7 times the rise time of the signal in order to measure rise time with an accuracy of a few percent.

The flatter frequency response of digital oscilloscopes enables them to produce less attenuation at frequencies below the –3-dB point than analog oscilloscopes do. Thus, digital oscilloscopes produce more accurate measurements. Secondly, the steeper roll-off of digital oscilloscopes helps reduce aliasing errors.

Typically, a human-body ESD pulse can have a rise time of less than 200 ps. The bandwidth required to accurately display this would be approximately 0.43/(200 ps), or 2.15 GHz. Some ESD simulators can produce rise times of 50 ps and thus require an oscilloscope bandwidth of 8.6 GHz.

The IEC and ANSI standards place more stringent requirements on measurement repeatability than they do on the rise time. To capture ESD, you must set your oscilloscope to “single-shot” mode. If the oscilloscope returns a range of different answers for repeated rise-time measurements, then you can’t depend on it to accurately measure the rise time on any one occasion—even if the average of many measurements is highly accurate. A major factor in single-shot repeatability is low internal noise, so compare noise specifications when you evaluate oscilloscopes for ESD testing.

The higher an oscilloscope’s bandwidth, the more accurately it will capture the ESD pulse’s rising edge. Table 2 shows that an oscilloscope’s rise time directly affects the measured rise time of an ESD pulse. For a pulse with a rise time of 700 ps, you need an oscilloscope with at least 4-GHz bandwidth to get less than 1% error. You must add this error to any system errors when you measure rise time.

To measure an ESD pulse, set the oscilloscope to single-shot mode and use a positive-edge trigger. Set the trigger level just above zero. You may need a minor trigger-level adjustment to capture the entire waveform. Set the vertical sensitivity to either 0.5 V/div or 1 V/div and set the time base to 20 ns/div. Assuming the measured signal is a triangle (for simplicity in calculations), a measured rise time of 800 ps will require a 10-Gsamples/s sample rate, which equates to 100 ps/sample, or eight samples on a rising edge, enough to accurately represent it.

Be sure to perform and document your verification test prior to any precompliance or compliance test to prove proper simulator operation. Then, once you complete verification tests, you can perform investigative or qualification testing, knowing that your ESD simulator is working properly.

Table 1. System accuracy variation causes a percentage measurement error.