Through a combination of shielding and circuit changes, you have sufficiently improved the ESD discharge immunity of your product so now it survives repeated human body model (HBM) system tests. Congratulations!
But before you can enjoy the kudos resulting from your achievement, you have one more thing to do. You must find out why a colleague in a sister division obtained different results after making the same modifications and running the same tests.
Perhaps there was a mistake or the measurement techniques differed ever so slightly. There may be a simple explanation. On the other hand, ESD system testing has been notorious for producing inconsistent results. Human error can account for some of the problems, but ESD testing also is characterized by standard test methods that are only partially defined.
For example, present standards, such as the International Electrotechnical Committee (IEC) 1000-4-2, allow a wide range of voltage-discharge waveforms which contribute to the variability of results. Two valid discharge-current waveforms are shown in Figure 1. The waveforms satisfy the rise time and 30-ns/60-ns current values specified by the standard, but they are very different in the initial few nanoseconds that cause the greatest threat to the system being tested.
ESD tests on fast digital electronics systems suggest that most coupling is caused by current induction from the near field or radiation. As a result, high-frequency components of the discharge waveform, equivalent to high values of the current derivatives, determine how severely a simulator affects electronic circuits.
In other words, the equipment under test (EUT) may react differently if tested with simulator A or B in Figure 1 because of different current derivatives. In an attempt to define the test waveform more completely, Working Group 14 (WG14) of the ESD Association is considering limits to both current derivatives: positive—5 A/ns/kV, negative—1.75 A/ns/kV.
Secondary effects are caused by the fields generated when the simulator discharges and by current flowing in the simulator ground strap. Although the ground strap is intended to carry slowly changing current, it may carry some of the faster peak current as well. This causes uncontrolled fast changing fields that influence the EUT.
The present IEC 991 standard places no limits on transient E and H fields. WG14, however, is considering limits for both fields associated with a discharge: E field—12 kV/m at 0.1 m, 1 kV/m at 0.7 m; H field—26.5 A/m at 0.1 m, 2.1 A/m at 0.7 m.
An example of a standard that has been changed and improved relates to the speed of approach associated with HBM manual discharge testing. A rapid-approach speed originally was recommended in IEC 801-2:1984 to minimize the length of the ionization path. A short path exhibits less inductance, maximizes the discharge’s peak current, and minimizes its rise time.
Of course, unless the rapid approach is mechanized in some way, its variability produces a variation in the discharge rise time and the di/dt current derivatives. The approach speed cannot be repeated accurately when a discharge is initiated manually.
To overcome this limitation, the 1991 version of the IEC standard introduced contact-discharge testing. The ESD electrode is placed in contact with the chosen conductive surface, and a high-voltage relay initiates the discharge. Not only is this method more repeatable, but also more convenient since a 4-kV contact discharge is equivalent to an 8- kV air discharge. Based upon experience gained from the 1984 standard, the 150-W resistor in the HBM simulator was replaced by a 330-W resistor in the 1991 standard.
In addition to incompletely specified test methods the tolerances of present test standards are not sufficiently narrow to avoid test inconsistencies. Measurement uncertainty is affected by the following factors:
Inaccurate current target resistance value.
Nonflat current target frequency response.
Reflections due to imperfect connectors.
Cable loss.
Test instrument amplitude error and rise-time limitations.
Attenuation amplitude and frequency-response errors.
Target Characterization
The DC target resistance can be measured using a Kelvin, 4-wire technique. ESD source impedance can be estimated by dividing the predischarge voltage by the peak discharge current. A target resistance of about 2 W causes less than a 1% error when used with a typical discharge impedance of greater than 330 W .
An additional source of error is the frequency dependency of the target current sensor’s volts/amp sensitivity. Although the average power dissipation in the target is low, peak currents of tens of amps and rise times of less than 1 ns are common. Figure 2 shows the frequency responses of three different types of targets. Figures 3a and 3b compare the responses of two of these targets when driven with waveform A from Figure 1.
Errors that appear to be small in the Figure 3a current-time plot are accentuated in the Figure 3b current derivative-time graph. Consequently, a nonflat target frequency response will contribute to overall measurement error. Perhaps more importantly, unintentional, large current derivatives from the ESD simulator will significantly increase the severity of the test.
When testing a target, be sure to test the test equipment first. Use a high-speed DSO to observe the pulse-generator output without the target in the circuit. Then repeat the test with the target included.
Determining the performance of a multigigahertz, low-impedance target is not straightforward. A preferred approach uses a special conical, 50-W transition between the 50-W generator output and the target. Because a conical section maintains rotational symmetry, the local field around the target will have less influence on its operation than, for example, simply attaching a cable between the target’s inner conductor and ground.
In addition, the tapered transition maintains the connecting cable’s characteristic impedance right up to the target. This means that the measurements of a target’s performance won’t be confused by unnecessary reflections.
Other Error Sources
Interconnecting cables can contribute significant errors at high frequencies. All cables will exhibit skin effect, so it is not uncommon to use large-diameter coax cables even for relatively short lengths—the larger the diameter, the greater the conductor areas and the lower the skin-effect resistance. Coax cables also produce dielectric loss that varies linearly with frequency. Skin effect varies as the square root of frequency. And, some types of coax cables have strong dispersion that causes ringing on step responses.
Attenuators should not change their DC value when dealing with very large pulses. If they do, this is another source of nonlinearity. However, it is very common for there to be a slight error in an attenuator’s nominal ratio. For example, a 10:1 attenuator actually may measure 10.1:1 even at DC. Consequently, your measurements may need compensation to correct for this systematic error. With regard to frequency response, use attenuators rated to more than 10 GHz to allow minimal time-domain distortion for very fast pulse edges.
Similarly, there are gain and offset errors and frequency dependencies associated with the DSO used to acquire data from the target characterization tests. You need to determine the exact response of your DSO to fast-edge pulses—something not often specified in data sheets.
The effects and precautions outlined here are second nature to ESD professionals who often make these stringent measurements. Because the measurement setup appears to be simple, test engineers may forget that the fast pulse equivalent circuit actually is much more complicated. So to get repeatable results, effects that would ordinarily be of secondary importance for slow rise times now must be considered.
“Our ESD test equipment includes cabling as part of the matched test set that is serialized and calibrated with the instrument,” said Donald Boehm, executive vice president of Novx. “This approach takes a variable out of the equation that is critical to the accuracy of the measurements.”
Making a related point about test variability and accuracy of results, Jon Barth, president of Barth Electronics, commented, “The threat from a 0.7- to 1.0-ns rise-time test pulse simulates a 5- to 10-kV discharge. Lower-voltage ESD events have lower peak currents but faster rise times. Fast current pulses of lower amplitude may produce more of a threat.
“WG14 is planning to make accurate measurements of real human ESD events over a wide range of conditions,” he continued. “WG14 also plans to perform upset and damage ESD tests on different electronic equipment to define actual threat levels.”
For the present, the best course of action is to be aware of the pitfalls of HBM system testing. Then you can take steps to avoid the major problems, knowing which variables are most likely to cause your remaining measurement uncertainties.
Acknowledgment
Material in this article was presented in: Lin, D., Pommerenke, D., et al, “Metrology & Methodology of System Level ESD Testing,” ESD Association and IEEE EOS/ESD Symposium Proceedings, Reno, NV, 1998, pp. 29-39.
ESD Test Equipment
Protection Characterization
The Model 4002 TLP (transmission line pulse) Test System is used to determine the voltage-current (V-I) characteristics of ESD protection structures in ICs. The system applies flattop 100-ns pulses at successively higher voltage levels. The leakage current is measured during the last 10 ns to 20 ns of each pulse. Many measurements are made and averaged at each level to reduce the effect of digitizer noise. The resulting V-I plot identifies the test pulse amplitude where damage to the device under test occurs. The 4002 is a fully automatic tester with upgradable software. $99,400. Barth Electronics, (702) 293-1576.
Ionizer Analyzer
The Model IPA 287 Ionizer Performance Analyzer measures decay and balance levels for any type of ionizer. It includes a charger, a field meter, a timer, and a temperature and humidity sensor and comes with a two-year warranty. The instrument meets the current requirements of ESD Association DSP3.3 for periodic verification of air ionizers. Either manual or automatic operation can be selected. A number of optional settings may also be chosen from an alphanumeric display. $995. Monroe Electronics, (800) 821-6001.
Static-Decay Meter
Model 406D is a static-decay meter used to determine the static-dissipation performance of materials. It consists of a control unit and a Faraday cage. The two-part design allows the cage to be placed in a humidity chamber for testing material under environmental conditions. The decay meter offers selectable cutoff levels of 50%, 10%, and 1% with 0.01-s decay-time resolution. In the manual mode, the operator controls all functions; the automatic mode accommodates consistent multiple tests. $9,995. Electro-Tech Systems, (215) 887-2196.
Discharge Instrument
The PESD 3000 High-Voltage Discharge Tester provides ESD immunity testing beyond that required by EN61000-4-2. It has a microprocessor-controlled base unit with a built-in power supply. This lightweight, ergonomic pistol comes with an interchangeable impedance control unit that allows the system to test to ISO TR 10605. It supplies a voltage discharge of 30 kV in the air and 25 kV by direct contact with both single and repetitive modes. $11,750. Haefely Trench, (703) 494-1900.
IC ESD/Latch-Up Tester
Paragon is a turnkey system that lets IC manufacturers test packages with up to 1,024 pins for ESD and latch-up susceptibility. Simple to complex devices, such as system-on-chip, can be tested. Paragon is custom configurable in the field. It meets various EIA, ESD Association, and MIL-STD-883D requirements for human body model, machine model, and latch-up testing. Starts at $220,000. KeyTek, (800) 753-9835.
Portable Ionizer Tester
The Periodic Verification Kit includes a test meter; a 1,200-V, dual-polarity charger; 1″ × 2½” and 6″ × 6″ test plates; a tripod; a 10-ft ground cable; and a manual. The kit meets the requirements described by ESD Association SP3.3, Periodic Verification of Air Ionizers. The battery-powered, 2.4″ × 4.2″ × 1.3″ test meter weighs 5 oz, has a 3½ digit LCD, and measures up to 20 kV with ±5% accuracy. The battery-powered, 2.4″ × 4.2″ × 1.2″ charger provides ±1,200 V ±50 V through a 109 W output resistor. Contact company for price. Simco, (800) 538-0750.
Soldering-Iron Monitor
The M20 Soldering-Iron Tip Voltage Tester lights a green LED if the tip voltage is below 1 V or a red LED if it is above. Two AAA batteries and a circuit board with the LEDs are contained in a 2.4″ × 2.3″ × 1″ box to which a remote sensor plate is attached by an insulated wire. The box is grounded by a separate wire with a spade lug. Either dampened cleaning sponges or metallic cleaner pads can be used. $75. Novx, (800) 728-6689.
Combination Tester
The STS-200F-ST Tester for shoes and heel straps features dual footplates for touch-free testing of both feet simultaneously and a touch handle for testing one foot at a time. The factory low setting is 0.5 MW with an adjustment range of 0.5 MW to 2.0 MW . The high limit is set for 50 MW with adjustability from 10 MW to 1,000 MW . The STS-200FS-ST tests single-conductor wrist straps, the STS-200FD-ST is for dual-conductor wrist straps, and the STS-200FSD-ST tests both single- and dual-conductor wrist straps. From $380. SpectraScan International, (719) 447-0170.
Wide-Range Ohmmeter
The PRS-801 Resistance System measures resistance from <0.1 W to >1014 W , stores up to 64 data points, and transmits stored data via an RS-232 port. One of 14 colored LEDs illuminates to indicate the approximate measured value. A 2-5/8″ × 1-5/8″ LCD presents test results in either a 3½-digit or an X.XE00 exponential format using 0.5″ high numerals. A 19-segment analog scale; a ×1, ×10, ×100 multiplier; the number of data points in memory; data and status messages; and <10-V, 10-V, and 100-V test voltages are also displayed. The 4" × 6" × 2" PRS-801 weighs 22 oz complete with two 9-V alkaline batteries. Contact company for price. Prostat, (630) 238-8883.
Electrostatic Voltmeter
The Model 523 Hand-Held Electrostatic Voltmeter provides surface voltage measurements that are much less dependent upon sensor-to-surface spacing than typical fieldmeter measurements. While a typical fieldmeter is only 50% accurate, the Model 523 maintains 5% measurement accuracy over a spacing range of 30 mm to 60 mm. Model 520 has similar performance over a 5-mm to 25-mm range. The measurement voltage range and resolution are 2 kV and 1 V for Model 520 and 20 kV and 10 V for Model 523. Both feature a slender sensor probe, a 3½-digit LCD, and a 2½ readings/s sample rate. Contact company for price. TREK, (800) 367-8735.
Copyright 1999 Nelson Publishing Inc.
July 1999