Jan Sroka

Uncertainty of ESD Pulse Metrics Due to Dynamic Properties of Oscilloscope

This paper investigates how dynamics of an HF oscilloscope affects standardized metrics (i.e., rise time and peak value) of a recorded electrostatic discharge (ESD) current pulse and evaluates the oscilloscope measurement uncertainty. A frequency-domain dynamic model of the oscilloscope, based on measurements of the input reflection and voltage gain magnitude, is derived. The complex voltage gain is approximated with a transfer function to take into account unmeasured phase shift. A high-order discrete-time filter is designed to fit the measured voltage gain magnitude not only in the passband, but also beyond it as well. Since the measured data are burdened with measurement uncertainties, the worst-case deviations of the oscilloscope input impedance and the voltage gain (in the form of envelopes of the frequency responses) are calculated, and components of type B uncertainty of the ESD pulse metrics corresponding to these deviations are estimated using the sensitivity method.

Distortion of ESD Generator Pulse Due to Limited Bandwidth of Verification Path

The paper deals with analysis of how limited bandwidth of a setup for verification of electrostatic discharge (ESD) generators affects reliability of testing. It is investigated how suppression of a high-frequency part of an ESD pulse spectrum distorts the pulse waveform and affects its standardized time domain metrics: the rise time and the peak value. Restriction of a pulse spectrum is achieved by low-pass filtering with a defined passband. The filtering models high-frequency attenuation of a measurement path, in particular that of an oscilloscope. The analysis is conducted for theoretical ESD pulses, specified by the IEC Standard, as well as for series of real noisy pulses recorded on setups with wideband 6 and 12 GHz oscilloscopes. Discrepancies of rise times and peak currents of filtered pulses with respect to an original input pulse are calculated as measures of distortion due to constraint of a pulse spectrum. It is shown that above 3 GHz, the setup circuit noise is the main factor contributing to measurement uncertainty of the metrics.

Measurement uncertainty of the target scattering parameters built in the uncertainty estimation of ESD simulator

In a previous paper by the author (see 14/sup th/ Int. Zurich Symposium on EMC, p.189-92, 2001) uncertainty estimation of current wave by calibration of ESD simulator is presented The uncertainty is defined there as discrepancy between current wave calculated for low frequency transfer impedance and for frequency dependent transfer impedance. This is termed here the basic uncertainty. Such definition of basic uncertainty is entitled because the low frequency transfer impedance is introduced as the coefficient between output voltage and input current of the target. Frequency dependent transfer impedance is burdened with uncertainty which also reflects in current wave uncertainty. This is called here additional uncertainty. A geometrical sum of basic and additional uncertainty builds total uncertainty. The formula used in the previous paper for frequency dependent transfer impedance is so complex that derivation of additional uncertainty with commonly used sensitivity coefficients between measurand (scattering parameters) and output quantity (transfer impedance) is impossible. Contribution of the additional uncertainty is however essential for the total uncertainty budget. In this paper simplified formula for transfer impedance is introduced. Simplification is achieved with replacing the actual load of the target (input impedance of the oscilloscope) with idealised 50 /spl Omega/ resistance. Suitability criterion of the simplified formula is comparison of the basic current peak uncertainty calculated for actual and simplified transfer impedance. The method is illustrated with uncertainty calculation of the current peak with different rise time varying from 360 ps to 1030 ps. The comparison of the basic uncertainty shows very good agreement specially for the rise time which is of practical interest (from 700 ps to 1 ns). The additional uncertainty dominates the basic uncertainty so strongly that the total uncertainty increases as the rise time increases.

Insertion loss as transfer coefficient for the calibration of ESD simulators. Is it sufficient to cope with

Up to now, solely the S/sub 21/ parameter of the target is used for its characterisation in electrostatic discharge (ESD) simulations. This parameter depends on the source and load impedance and on the network adapting the target input to the coaxial connector of the network analyser. The new draft of the ANSI and IEC standards require a very flat S/sub 21/ parameter of the target up to 4 GHz. Moreover, these standards introduce a low frequency transfer impedance for the recalculation of the voltage displayed at the oscilloscope to the discharge current. The paper shows that a frequency remastering method applied to the IEC target results in more than 1.6% of the error for the peak value. It is, however, negligible if the S/sub 21/ parameter of the target changes slightly with frequency.

An alternative approach to determination of the S-parameters of noncoaxial devices

Non-coaxial (non-insertable) devices are indispensable by EMC testing. Absorbing clamps belong to them. They are verified with the Vector Network Analyzers (VNA) which unfortunately have coaxial ports. The clamp can be connected to the VNA only if it is embedded in adapters. By verification the clamp must be de-embedded i.e. the contribution of adapters must be extracted from the measured S-parameters. The Thru-Reflect-Line (TRL) calibration of the VNA would be suitable tool for this task, but this function is only optional in the firmwares of the VNAs. In this paper the alternative approach consisted in calibration with the Short-Open-Thru-Match (SOTM), supplemented with the measurements of all S-parameters of the two adapters cascaded face to face is presented. Applicability of the method is verified metrologically.

Mismatch Correction by Measurements of Conducted Radio Disturbances With AMN

Conducted radio disturbances at a power port of the equipment under test (EUT) are sensed with the artificial mains network (AMN) and transmitted via the 50- coaxial system to the measurement receiver. The voltage division factor of the two-port network between the EUT port of the AMN and the receiver input is defined by neglecting a mismatching. As a matter of course, the mismatch correction factor appears. It depends on all components of the path, including the AMN. This fact is ignored in Comité International Spécial des Perturbations Radioélectriques 16-4-2 document. Due to the reflection at the receiver input, whose only the extreme is usually available in the technical specification, the mismatch correction factor cannot be determined. Its estimate is built-in into the measurement uncertainty budget. Two approaches for the estimation of the mismatch correction factor by the divided and the integrated measurement path are derived. The estimations are illustrated with numerical calculations by Monte Carlo method. The distinction of the mismatch correction factor by conducted disturbances from radiated disturbances, as well as recommendation of the approach with the integrated path, constitute the essence of the paper.

Contribution of Impedance Imperfection of the T-type ISNs in Measurement Uncertainty of Conducted Disturbances

Method of uncertainty estimation due to impedance imperfection of the T-type Impedanced Stabilisation Networks (ISNs) is presented. Elaboration is performed for two cases: for measured common mode impedance of the ISN and for assumption of tolerances imposed on this impedance. Estimation is illustrated with uncertainty calculation for the ISN designed for one pair of symmetrical, unshielded lines, so called T2 ISN. There is only one publication, namely [7] in which this uncertainty is mentioned. Numbers cited in this document are underestimated compared with what is presented in this paper.