Frequency Approach to Analysis of ESD Pulse

Transcription

1 本文は pp Frequency Approach to Analysis of ESD Pulse Janusz Baran Non-member (Czestochowa University of Technology, Poland) Jan Sroka Non-member (EMC-Testcenter Zurich, Switzerland) Keywords : electrostatic discharges, frequency domain analysis, power spectral density, transfer impedance The paper concerns calibration of generators for simulation of the Human-Metal Electrostatic Discharge (ESD) according to the IEC standard. The analysis is restricted to ESD generator current, since only calibration of the current is mandatory in the standard. It is shown that analysis of the ESD pulse in the frequency domain can be an indication if omitting the frequency considerations in calibration of ESD guns is acceptable. The calibration setup (Fig. 1) consists of a target (current converter), attenuator, coaxial cable and a wideband, single shot oscilloscope. It is much easier to use only a low frequency model of such a measurement path setup than consider a high frequency model. If, however, a high frequency treatment is indispensable, then a frequency dependent transfer impedance of the measurement path (Fig. 2) and approximation of the oscilloscope frequency response with an infinite impulse response discrete time filter are adequate tools, because the distortion of a recorded pulse due to frequency characteristic of the instrumentation has to be taken into consideration. Comparison of power spectral densities (PSD) of theoretical pulses, measured pulses (Fig. 3) as well as the measurement path noise (Fig. 4) gives a good criterion for specifying minimal bandwidth of a setup required for reliable calibration of a given ESD gun. The main conclusions of the paper are as follows: 1) a measurement path with a 3 GHz bandwidth is sufficient to neglect ESD Gen. target (current converter) cable with attenuator OSC Fig. 1. Setup for ESD pulse calibration Fig. 2. Transfer impedance of the calibration path versus frequency Fig. 3. Normalized ESD pulses of the same ESD gun (series of 6 measurements, 6 GHz oscilloscope bandwidth) along with theoretical approximations: a) waveforms, b) PSD spectra Fig. 4. Approximately flat PSD of measurement noise calculated for initial 1 ns interval prior to start of the pulses from Fig. 3 frequency distortions for the theoretical pulse with a.6 ns rise time the fastest allowable by the IEC standard, 2) if a real ESD pulse is relatively smooth, the main component of its spectrum above 3 GHz is the noise, whose source is the gun s discharge switch. If the spectrum of a real pulse compared to the noise is limited to below 2 GHz, it is distorted in principle in the same way as a clean theoretical pulse. A too narrow bandwidth, especially less than 1. GHz, can easily result in incorrect evaluation of the metrics and erroneous verdict of verification. When a setup bandwidth is greater than 2 GHz, the main source of the measurement uncertainty becomes the pulse noise, which can be reduced by averaging over a series of pulses. Even though a 3 GHz bandwidth setup does not ensure distortion-free transmission of a pulse, further expansion of an oscilloscope bandwidth will not reduce the deterministic component of the uncertainty. Verification of guns that produce rough pulses, with steps on the rising slope and/or several smaller peaks, whose high frequency part of the spectrum above 3 GHz contains distinctive resonance peaks emerging above the noise level, requires a wider bandwidth 4 GHz or even more. The paper is a resume of previous papers of the authors, in which these issues were presented in details. -1-

2 本文は pp Measurement of Radiated Electromagnetic Field due to Low Voltage ESD with Spherical Electrode in 1-3GHz Frequency Bandwidth Ken Kawamata Senior Member (Hachinohe Institute of Technology) Shigeki Minegishi Senior Member (Tohoku Gakuin University) Osamu Fujiwara Senior Member (Nagoya Institute of Technology) Keywords : ESD, micro gap discharge, low voltage discharge, electromagnetic noise, radiation, spherical electrode 1. Introduction The very fast transients are arisen by gap discharges with low voltage ESD (electrostatic discharges). The low voltage ESD has very wideband (high frequency) electromagnetic noise source. Over the past few years a considerable number of studies have been made on electromagnetic noises of ESD from the point of view of the EMC (electromagnetic compatibility). The electromagnetic noise characteristics of the gap discharges are gradually becoming clearer. In this paper, the amplitude properties of radiated electromagnetic field due to low voltage ESD were examined by experimental study. We present an experimental system to measure the radiated electromagnetic filed in wideband region from 1GHz to 3GHz frequency bandwidth using a pair of spherical electrode and a horn antenna. 2. Experimental System and Experiments Figure 1 shows an experimental system using spherical electrodes and a horn antenna for the wideband measurements. The system consists of a high voltage D.C. power supply, a pair of spherical electrodes, high resistance lines, a horn antenna (ETS 311, 1-18GHz), an attenuator of 1dB and digitizing oscilloscope (Tektronix TDS694C, 3GHz, 1GS/s). In this experiment, four pairs of diameter electrodes were used to measure the radiated electromagnetic field. The type (a) is a pair of 12.8mm diameter electrodes, Type (b) is 19.mm, Type (c) is 2.mm, and Type (d) is 3.mm. These electrodes were made by brass ball. In the experiment, a constant voltage applies to the electrode, a gap length between the electrode decreases slowly about 1 cm/s. The amplitude of received electromagnetic field was measured by the oscilloscope when discharge occurred. The discharge voltage is from 3V to 6V. The amplitude of received electromagnetic field was expressed by a peak to peak value of received voltage measured by digitizing oscilloscope. 3. Experimental Results Figure 2 shows relationship between the discharge voltage and the received voltage. The received voltage shows an average value of more than a hundred measurements. The diamond marks indicate the relation in the type (d), the square marks are type (c), the triangular marks are type (b), and circle marks are type (a). In overall, the received voltage has proportional relation to the discharge voltage from 3V to 62V in each electrode. However, the received voltage is not proportion to discharge voltage in more than 63V discharge, respectively. And so, received voltage showed a same tendency between the type (a), (b), (c), and (d). The amplitude of radiated electromagnetic field has a widely variation of amplitude in high voltage discharge more than 63V. Furthermore, the received voltage was proportion to the diameter of the spherical electrode in the wideband experimental system. Oscilloscope Oscilloscope (Tek. TDS694C, 3GHz) Double Rigid Guide Horn Antenna ETS 311,1-18GHz Spherical Electrodes (Diameter is 3 mm) Received Voltage V [Vp-p] High Resistance kω Tek. TDS694C, 3GHz Attenuator 1dB Coaxial line (D2W, 3m) 1m Feed line. (High resistance line) High Voltage D.C. Power Supply. (1V-1V) Fig. 1. Measurement system for amplitude of the radiated EM field due to discharge using a pair of spherical electrode and a horn antenna Type (a):12.8mm Type (d):3.mm Type (c):2.4mm Type (b):19.mm Discharge Voltage [V] Fig. 2. Relationship between discharge voltage and received voltage of radiated electromagnetic field 4. Conclusion As a result, the amplitude of radiated electromagnetic field is proportion to the discharge voltage from 3V to 62V, and the amplitude of radiated electromagnetic field was proportion to diameter of the spherical electrode in the wideband system. -2-

3 本文は pp.3-3 Measurement and Validation of Transfer Impedance of IEC Calibration Current Target for Electrostatic Discharge Generators Yukio Yamanakai Member (National Institute of Information and Communications Technology, Takashi Adachi Non-member (Nagoya Institute of Technology) Shinobu Ishigami Member (National Institute of Information and Communications Technology) Ikuko Mori Non-member (Suzuka College of Technology) Yoshinori Taka Member (Kushiro College of Technology) Osamu Fujiwara Senior Member (Nagoya Institute of Technology) Keywords : electrostatic discharge generators, IEC calibration current target, transfer impedance, injection current, reconstruction The electrostatic discharge (ESD) events due to charged human bodies produce electromagnetic (EM) fields having broad-band frequency spectra, which cause malfunctions in high-tech information equipment. In this context, the International Electrotechnical Commission (IEC) prescribes the ESD immunity test (IEC61-4-2) of electronic equipment, and specifies the detailed waveform of a discharge current injected onto the IEC recommended calibration current target in contact with an ESD generator, which is designed to have a transfer impedance (rate to a sinusoidal current injected of its observed sinusoidal voltage onto the target) of 1 Ω so that the waveforms of an injected current and the resultant observed output voltage coincide at low frequencies. Due to its frequency characteristics, however, in the high frequency range, the observed voltage waveform does not correspond to the injected current one. In the present study, using a Ω tapered coaxial type adapter and a S-parameter technique, we measured the transfer impedance of a commercially available IEC calibration current target used for the immunity testing against ESDs, and thereby reconstruct the waveforms of discharge currents injected onto the target from their observed output voltages for contact and air discharges of an ESD generator. Figures 1 and 2 show a setup for measuring scattering parameters of an IEC calibration current target and the measured result of the transfer impedance, respectively. We found that the transfer Port1 N-type connecter Network analyzer Port2 Ω coaxial cable Calibration adapter Calibration target Calibration plane N-type connecter Fig. 1. Setup for measuring scattering parameters of IEC calibration current target Z T (jω) [Ω] Real Part Imaginary Part Z T (jω) 1 Ω Ω Frequency [MHz] Fig. 2. Measured frequency characteristics of transfer impedance v o (t) [V], i(t) [A] t r = 8 ps t r = ps t r = 1 ps Observed voltage voltage Reconstructed Estimated current Charge voltage = 4kV 2kV.kV Time t [ns] Fig. 3. Observed voltages and reconstructed discharge currents injected on the target for air discharges with slow approach impedance is almost 1+j Ω at frequencies below 1 GHz, while due to resonance phenomena occurring at frequencies around 2 GHz and GHz, resistive and reactive components significantly change at frequencies over 1 GHz, though their absolute values are slightly larger than 1 Ω. Figure 3 shows observed output voltages and reconstructed discharge currents injected on the target for air discharges with slow approach, which demonstrates that the reconstructed discharge currents agree well with the observed voltages with rise time of ps and 8 ps, while in the case for the observed voltage with a rapid rise time of 1 ps, the reconstructed current has a slightly small peak and a bit gentle rise time in comparison with those of the observed voltages due to the resonance characteristics of the transfer impedance at frequency over 1 GHz. -3-

4 本文は pp Measurement of Leaked High-Frequency Burst Electric Field and EMI Evaluation for Cardiac Pacemaker in Fusion Facility Yukio Yamanaka Member (National Institute of Information and Communications Technology, Jianqing Wang Non-member (Nagoya Institute of Technology, Osamu Fujiwara Senior member (Nagoya Institute of Technology, Tatsuhiko Uda Non-member (National Institute for Fusion Science, Keywords : fusion experimental facility, leaked electromagnetic field, statistical distribution, electromagnetic interference In the fusion facility, many devices such as plasma heating and discharge cleaning leak electromagnetic fields with an irregular variation ranging from several MHz to several hundred GHz. The objective of this study is to quantify the special electromagnetic environment and evaluate its safety for workers. We first measured the electric field leaked from a heating device in the ion cyclotron range of high-frequency (ICRF). The signal from the ICRF is amplified with a two-stage amplifier at frequencies of 3 1 MHz, and is then sent to a plasma load through a waveguide. Using a real time spectrum analyzers (Tektronix RSA338B) together with a bi-conical antenna, we measured the time variation and spectrum of the leaked electric field in the vicinity of the amplifier of ICRF for the plasma load. The real time spectrum analyzer can record the time waveform and the spectrum of measured electric field at the same time through a Fourier transform hardware. As a result, as shown in Fig. 1, the leaked electric field was in a burst form with a varying field level and the center frequency was at 38. MHz with most of energies ranging within ±2 MHz. Within each burst period, we extracted the amplitude probability distribution (APD) and the crossing rate distribution (CRD). Figure 2 shows an example of the possibility density function (PDF) of the leaked electric field. It is much different from a normal distribution, and consequently the mean value of electric field differs from the median value and Electric E field field [V/m] [V/m] Frequency [MHz] 電界 vs. 時間 Time [s] x 1 Time [s] Fig. 1. Spectrum and time variation of leaked electric field in one burst period the mode value. The leaked electric field itself also varied violently with time, reached 4 times in one second to cross the mode value of the leaked electric field. Yet, although the electric field varies with time in a burst form, its maximum level did not exceed 1 V/m. Under such an electric field level environment, the SAR is always much smaller than the ICNIRP safety guideline for any type of statistical variations. In addition, we also evaluated the possibility of electromagnetic inference (EMI) to an implanted cardiac pacemaker in the measured electromagnetic environment. We employed a nonlinear operational amplifier model with Volterra series representation to predict the output voltage of the sensing circuit of pacemaker which consists of an amplifier and a low-pass filter. Using the measured voltage in the real time spectrum analyzer as the input to the sensing circuit as a worst case, we found that the interference voltage induced in the output of the pacemaker sensing circuit did not exceed the threshold for a malfunction. As shown in Table 1, due to the narrow bandwidth of the leaked signal, only using the leaked electric field at the center frequency was sufficient for this EMI evaluation. Probability density [%] Density X18z data E field [V/m] Electric field [V/m] Fig. 2. Probability density of leaked electric field in one burst period Table 1. Relationship between the DC offset output voltage and bandwidth BW [MHz] Vout DC [mv] Difference.87.% % % -4-

5 本文は pp Waveform Distortion of Discharge Rising Currents caused by Built-in Inductor for Contact Discharges of Electrostatic Discharge Generators Yukio Yamanakai Member (National Institute of Information and Communications Technology, Fumihiko Toya Non-member (Nagoya Institute of Technology) Shinobu Ishigami Member (National Institute of Information and Communications Technology) Osamu Fujiwara Senior Member (Nagoya Institute of Technology) Keywords : electrostatic discharge generator, contact discharge, discharge current, built-in inductor, waveform distortion The International Electrotechnical Commission (IEC) prescribes an immunity test (IEC61-4-2) of electronic equipment for electrostatic discharges (ESDs), and specifies not the whole-waveform but only the rise time, the first peak, the levels at 3 ns and 6 ns of a discharge current injected by an electrostatic discharge generator (ESD gun) onto an IEC recommended calibration target. The rise time is required to be.8 ns ±2 %, which can commonly be realized by mounting an inductor in series with a metal electrode for current injection. Due to the impedance of the inductor, however, discharge currents should rise not smoothly but deformedly, which can affect immunity test results. In this study, to investigate the effects of built-in inductors of ESD guns on current rising waveform, we simultaneously measured discharge currents and the resultant magnetic near-fields for contact discharges of an ESD-gun, and compared their results with those for an ESD-gun with a metal cylinder in lieu of the inductor. Figure 1 shows the configuration and structure of a built-in inductor of an ESD gun and the same sized metal cylinder used in place of the inductor. Figure 2 shows an example for the waveforms of output voltages induced in a magnetic field probe for contact discharges of the ESD gun with a built-in inductor. Tip electrode for discharge With a plastic cover Without a plastic cover 22. mm 1mm Gun body 9 mm 8 mm (a) Configuration of built-in inductor 22. mm (b) Built-in inductor 9 mm 22. mm 8 mm (c) Metal cylinder (brass) Fig. 1. (a) Configuration and structure of built-in inductor, (b) dimension of built-in inductor and (c) the same sized metal cylinder of ESD gun Also shown in the same figure is the result for the metal cylinder in place of the built-in inductor. As a result, we found that the magnetic near-field waveform is deformed by the distortion of the current rising waveform for an ESD-gun with an inductor, while the ESD gun with a metal cylinder produces no distortion for the current rising waveform despite its rise time (4 ps) shorter than the IEC specification so that there is no distortion for the corresponding magnetic near-field waveform. Using the Heidler s formula for typical discharge current waveform specified in the IEC, we calculated current rising waveform and the resultant induced output voltage of a magnetic probe, which shows that these results do not agree with those for the built-in inductor, while there is good agreement for the metal cylinder. This finding implies that immunity test results can be affected by the frequency characteristics of a built-in inductor inside an ESD gun, and thus that not only rise time but also current rising waveform of a discharge current should be specified for ESD immunity testing. Measured (a) (b) Observed voltage v o (t) [V] 1 Z = [Ω] d=13 mm v o (t) Z = [Ω] d=13 mm Charge voltage = 2 kv Calculated from measured current Calculated from Heidler s current Time t [ns] v o (t) Time t [ns] Fig. 2. Waveforms of output voltages induced in magnetic field probe for contact discharge of ESD gun with (a) built-in inductor and (b) metal cylinder installed in place of built-in inductor --

6 本文は pp Measurement Method for Impulse Electromagnetic Field by Half TEM Horn Tsubasa Tachibana Non-member (Tohoku Gakuin University) Ken Kawamata Senior Member (Hachinohe Insitute of Technology) Shigeki Minegishi Senior Member (Tohoku Gakuin University) Keywords : Half TEM Horn, impulse electoromagnetic field, ESD 1. Introduction The measuring method of impulse electromagnetic field by Half TEM Horn elctoromagnetic fieid generator (EM generator) and Half TEM Horn electoromagnetic field detector (EM detector) was proposed. The impulse electromagnetic field is generated between the ground plane and element of EM generator and is detected by EM detector inserted in EM generator. As a result of the TDR experiment of EM generator and EM detector, the characteristic impedance from the input to the aperture was fixed Ω. As a result of the experiment of impulse (Amplitude 16mV, Pulse width 63ps) response, the received electric power was proportional to the cross section of the aperture of the EM detector and the received pulse width was almost invariable to the cross section of the aperture of the EM detector. This method enables the measurement which decreases the amplitude without changing the pulse width of strong impulse electromagnetic field due to the discharge such as ESD. 2. As a Result of Experiment Figure 1 shows the measurement system of the impulse electromagnetic field which is generated between the ground plane and element of EM generator and is detected by EM detector inserted in EM generator. Figures 2 and 3 show the results of experiments. The received electric power was proportional to the cross section of the aperture of the EM detector in Fig. 2. The received pulse width was almost invariable to the cross section of the aperture of the EM detector in Fig Conclusion This method enables the measurement which decreases the amplitude without changing the pulse width of strong impulse electromagnetic field due to the discharge such as ESD. (a) Top view Fig. 2. Relation between S out /S in and P out /P in at θ=3, 4, (b) Longitudinal section Fig. 1. Experimental system Fig. 3. Relation between τ out /τ in and S out /S in at θ=3, 4, -6-

7 本文は pp Fundamental Study on Mechanism of Electromagnetic Field Radiation from Electric Devices with Loose Contact of Connector Keywords : contact failure, SMA connector, common-mode current Kazuki Matsuda Non-member (Tohoku University) Yu-ichi Hayashi Member (Tohoku University) Takaaki Mizuki Non-member (Tohoku University) Hideaki Sone Member (Tohoku University) 1. Introduction When a connector is tightened with insufficient torque, electromagnetic field radiation from interconnected devices is increased. To elucidate the mechanisms of electromagnetic field radiation due a loose connector contact, we investigate the high-frequency circuit elements at the contact boundary. From the experimental results, we show the relationship between the high-frequency circuit elements and the electromagnetic field radiation from electric devices. 2. High-frequency Circuit Elements at Contact Boundary In this experiment, to investigate high-frequency circuit elements at the contact boundary of a loose connector, we employ time domain reflectometry (TDR). The measurement setup is shown in Fig. 1. The rise time of the input step pulse is 1. ns, and the voltage is 1. V. Figure 2 shows the measurement results. The peak voltage is higher for insufficient torque (dashed line) than for sufficient torque (solid line). Therefore, the inductance at the contact boundary is increased by loosening the connector. In addition, the level of reflected voltage after the peak is higher for insufficient torque (dashed line) than for sufficient torque. This finding indicates that the resistance at the contact boundary is increased by loosening the connector. The inductance and resistance at the connector boundary are both increased when the connector is tightened with insufficient torque. 3. Relationship of L and R with Common-Mode Current In this section, we show the relationship of electromagnetic field radiation with the difference, due to tightening (torque), in inductance ( L) and resistance ( R). L and R, respectively, are Fig. 1. Measurement setup for TDR Fig. 2. Measurement results for TDR Fig. 3. Measurement results for common-mode current (a) Relationship of L with CM (b) Relationship of R with CM Fig. 4. Relationships of L and R with CM the differences after tightening from L and R at the base torque (.6 N m). In this experiment, we perform electromagnetic field radiation measurement at the same time as TDR. We employ the same TDR procedure as described in Section 2. Common-mode (CM) current is measured because it is one of the main factors in electromagnetic field radiation. The measurement setup consists of a spectrum analyzer in place of an oscilloscope, and a current probe is placed at positions along a semi-rigid cable. Measurement results for CM current are shown in Fig. 3. The relationships of L and R with CM current are shown in Fig. 4: CM L and CM R. The increase in CM current depends on frequency when only the inductance at the contact boundary is increased, but not when only the resistance of the contact boundary is increased. 4. Conclusion The results show that the inductance and resistance at a contact boundary are increased by loosening the connector: specifically, we found that CM L and CM R. Moreover, the increase in inductance has a large effect on the increase in CM current. -7-