A Note From The Author
The research references provided in this article occurred between 1978 and 1987, and are as applicable today as they were then, when they advanced understanding of ESD phenomena and through that, the ESD state of the art of knowledge.
These were the first reports to display that: ESD currents to a system do not increase in direct proportion to the initial electrostatic charge amplitude produced before the ESD event (e.g. low ESD voltages produce higher current than higher voltages); ESD spectra is greater than 1GHz with rise-times much faster than 1.0 nanosecond (e.g. 50 pSec to 200 pSec); the ESD “equivalent network” is not a single R-C network as previously thought, but rather a complex cascade of several networks, each with a different time constant; a “single ESD event” may be comprised of many events, even showing the approximate periodicity between each sub-event; suggested the initial impact of ESD was from boundary charge displacements of electric fields; detailed the ESD impulse waveforms and currents related to many common conditions (finger-tip direct, humans discharging through metal objects, humans discharging through furnishings); described how systems produce ESD amplitude-response dependencies owing to the different spectra of ESD events at different amplitudes; and, that compliance with a higher amplitude of ESD (e.g. 15kV) does not assure that adequate immunity will be exhibited at lower (e.g. 2kV) levels. During these research efforts the “human body concept”, the “human with metal object” concept; the “furnishings impact” equivalent values; the vertical and horizontal coupling planes for “radiated ESD equivalents” were all devised. The impact on the international community was sufficiently extensive starting circa 1979 that approximately five years of confirming work were required for the reports to achieve broad acceptance.
Since that time this work has been thoroughly disseminated and various standards and standard practices for ESD have been published, many (if not most) of which extend from these early research efforts. In the process of standardization some of the baseline foundational information delineating ESD mechanisms have been diffused. It didn’t help that when the IEEE converted published symposia archives to microfiche some years ago, much of the original photographic data for ESD event waveforms was lost as the contrast of the original printed publications became often degraded to black rectangles. Thanks to the invitation of The Editor of this magazine, you have the opportunity to travel with me, back to history, and back to this future to gather a broad understanding of the underlying boundary charge displacements that establish ESD, catch a glimpse of the propagational spectral mechanisms that impact systems-product performance, and to review foundational information that extends to this day.
W. Michael King
(For further reading, I suggest the book “Electrostatic Discharge” Third Edition by Michel Mardiguian, Wiley/IEEE Press, ISBN: 978-0470-39704-6, and ANSI, C-63.16-1994 “Guide for Electrostatic Discharge Test Methodologies and Criteria for Electronic Equipment.”)
ESD Amplitude Dependencies
Through the referenced research efforts, it has been recognized that the spectral bandwidth of the ESD event, considered as a continuum, is highly dependent on the electrostatic initialization amplitudes that are evident immediately prior to the displacement of the stored energy through the ESD event.
- High ESD initialization amplitudes produce relatively lower frequency spectral bandwidth distributions.
- Lower ESD initialization levels produce extremely high frequency (extending into microwave) spectral bandwidth distributions.
- Mid-range ESD levels (5 to 10 kV) develop spectral bandwidth distributions that appear to center in the general area of 100 MHz (which is approximately midrange in the spectral bandwidth envelope of the available ESD spectral probabilities).
The ESD susceptibility dependencies intrinsic to system response are obviously influenced by the conditional mechanisms described above since these factors impact various ESD response excitations of related components of the product or system (e.g. cables or casework). In and of themselves, these conditional factors can alter the susceptibility performance of system products in terms of ESD response amplitude, response mechanism, and response characteristics with a causal relationship to the conditional basis.
Suppose, for example, that the interface circuit of a desktop product has a design weakness to common-mode currents with risetimes (spectra) in the area of approximately 5 nanoseconds, but not too common-mode current components with risetimes (spectra) at 1 nanosecond (or faster) or 20 nanoseconds (or slower). According to the referenced ESD research information, ESD impulse waveforms with risetimes in the 5 nanosecond range are exhibited in the area of approximately 7 to 10 kV of ESD initialization amplitudes. When the product is placed on a metal-top desk, a distributive transfer impedance is established that will effectively bypass a significant portion of the 5 nanosecond ESD component energy through the case or cabinet structure (through the desk-to-ground), circumventing significantly the excitation of the weakness of the common-mode design in the interface.
Should the same unit described above be placed on any non-conductive tabletop, however, a larger portion of the ESD levels with the 5 nanosecond component will be available to excite the weakness in the interface design. Assuming no other product susceptibility effects, the ESD responses exhibited will be a “window response” effect where the product responds in the area of 7 to 10 kV, but only when the product is positioned on a desk or table top that is non-conductive.
Due to these fundamental propagational responses in common-mode product/system design mechanisms, so-called “mystery” ESD responses that can be encountered in various products may be understood. These responses are identified in the paragraphs that follow.
Overview of ESD Propagational Paths
Consider the ESD propagational paths that are represented in Figure 1, where:
Zs = source impedance of ESD event;
Zt = distributed transfer impedance of product case-to-ground;
Zp = impedance of external power to ground (including ground);
Zct = distributive transfer impedance of cables to ground;
Zd = directly conducted connection impedance of interconnected product to ground;
Ic = case incident impulse current;
Iif = interface cable exit current;
Ea = aperture-produced fields;
Ei = impulsive field gradient.
ESD Propagational Description
The diagram in Figure 1 generally describes the fundamental propagational ESD paths (including interface) that have been described in overview in the preceding paragraphs. The description that follows delineates the interaction and significance of various spectral ESD components that are associated with these paths.
Figure 1: Propagational Overview
Initial Impulse Effect
On application of the initial excitation ESD impulse, through its intrinsic source impedance Zs and an impulsive field displacement Ei, is instantaneously produced in the region of the charged structure that is causing the discharge. The specific size of this field displacement has a causal relationship to the size and shape of the structure causing the discharge, as well as the “load plane” (product surface) that is being subjected to the discharge as it opposes the structure being discharged.
The propagational activity of this initial impulse field displacement is very significant in the area of low ESD initialization levels, particularly below 5 kV. In that range, a surface-to-surface distribution of transfer impedance is established between the charged structure and the load plane, which results in intense impulse currents (which can range to approximately 100 amperes) and results from very severe field intensities (approximating 3 megavolts per meter). [2, 3 and 5] Further explanations are provided in Figure 2.
The initial ESD “circuit equivalent” is a spatial electric field displacement (collapse) between the charged structure (e.g. a hand) and a load plane, with the field intensity collecting at the discharge point. This displacement is between two structures, and is not dependent upon earth ground. This can be represented by distributed capacitance as a “surface distribution” that exhibits a “very low” transfer impedance function.
Figure 2a: Approximation: Initial impact of spatial field distribution yielded by review of ESD waveform characteristics exhibited in nature
Figure 2b: Block diagram overview of ESD equivalent network
Equivalent network implications for human finger-tip direct ESD and with small-metal objects intervening in the ESD path:
- The Capacitance of “surface distribution” will, in probability, vary between equivalent values of approximately 1 pfd and 20 pfd for the “finger-tip” and “small metal object” conditions, depending on the angularity of the hand with respect to the load plane as a localized field displacement.
- The “primary” Capacitance will, in probability, vary between equivalent values of approximately 100 pfd and 200 pfd for the “finger-tip” and “small metal object” conditions, since this value is presented to Earth (and space) from the human body.
- For “finger-tip” conditions the “equivalent impedance” of “skin resistance” and “arc impedance” for the “first effect” of surface-field displacement, appears to approximate 200 Ohms. In this same “finger-tip” condition, the higher-voltage “second effect” impedance appears to approximate 1,000 Ohms.
- For “small metal object” conditions, the “equivalent impedance” of “skin resistance” and “arc impedance” for the “first effect” of surface-field displacement, appears to approximate 20 Ohms. In “metal-object” conditions, the higher-voltage “second effect” impedance appears to approximate 150 to 200 Ohms.
System Response Significance: Initial Field Spatial Displacement
The significance of the initial ESD impulse field displacement on a systems product is found in two immediate paths.
First, the displacement of the localized field can result in aperture currents that develop as a result of propagation of the field and incident currents across apertures (slots or holes) in the product’s case structures. These apertures, given these excitations, function as slot antennas that produce field radials inside the product under evaluation. The effect on the product of these immediate fields, that are typically very intense in the close proximity of the slot-apertures, is dependent on the location of circuitry with respect to the slot-aperture, and the coincident match of the bandwidth of the circuitry with the effective propagational (transmitting) bandwidth of the slot-aperture incident field.
In the example above, the first interrelationship between ESD initialization amplitude and system-product response is encountered. Since it has been well established that ESD impulse bandwidths exhibit spectra well above 1 GHz at lower initialization amplitudes (in the references listed) it is to be anticipated that the size of many slots-apertures in typical products are efficient as antennas at lower initialization levels and much less efficient at higher levels due to the reduced spectral bandwidth of the higher level ESD impulse continuum. (Apart from slot apertures and owing to the limitations of various impedances in the casework at EHF, the same conclusion may be advanced when relating the general shielding effectiveness of case structures at these bandwidths.)
Assuming that in the immediate proximity of the excited aperture “antenna” there is an active circuit that has an intrinsic admittance (susceptibility-sensitivity) bandwidth complementing (coinciding with) the efficiency bandwidth produced by the ESD impulse propagating across the aperture, then a product susceptibility response is probable. In recent years, this response is typically more probable than in the past since many circuit and logic families are capable of exhibiting very high admittance bandwidths compared to older logic devices.
This affect is far less likely with the ESD initialization levels set at high levels. The simple basis for this statement is that the spectral bandwidth developed by the higher ESD initialization amplitudes are far less likely to produce efficient/coincident matches between the emission spectral capabilities of the product apertures and the spectral level distribution of the ESD impulse. Without such a match, aperture propagational efficiency is not developed, and consequently the probability of product susceptibility response becomes far less.
Second, the localized field displacement results in a case current that is propagated to ground (rather than surface-to-surface as in the previous example) through the distributive transfer impedance Zt. Here, the second interrelationship between ESD initialization amplitude and system-product response is encountered. The Zt case in impedance of the product will probably exhibit a broader bandwidth effect of response than the previously described localized service-to-surface distributions. This is because the Zt case impedance operate throughout a (probably) larger structure: the case itself! The comparatively large surface area of the case is excited not only by radiated field displacements, but by the direct conduction of current Ic. The transfer of energy across the distributive case impedance (Zt) results in the enhancement of the case-to-ground field gradient (Ei) that is driven across the case as a result of the conducted current.
Case-conducted Current Effect
The next effect in the propagational path of ESD in a product is touched on in the above paragraph. This is the development of a case-conducted current (Ic) from the product surfaces to ground. The energy of this path flows through the distributive transfer impedance (Zt) to ground, and additionally through the conduction impedances of the power lines (Zp) and the interface cables (Iif).
System Response Significance: Distribution to Earth and Cables
The propagation of the impulse case current into the distributed transfer impedance of the case to ground is paralleled by the conductive path of the power cable and the signal/data interface cable(s). These cables will exhibit “antenna efficiency” that is a development of the common-mode impedances of the cables and their respective termination points. These cables (which typically exhibit far greater inductance than the casework) will interact with the distributive transfer impedance of the case (usually capacitance-based) to result in an L-C resonance.
The inductive and resonance effects noted above suggest that another value of spectral bandwidth dependency will be produced. It is entirely reasonable that these values will be encountered at ESD amplitudes that are much higher than the lower-initialization level effects first described above. This concept is advanced due to the fact that the inductive properties of cables typically limit the efficiency of energy transfer at very high frequency spectral distributions, which is the occupancy domain of lower-initialization levels of ESD.
The probability is that the interaction of the cases and cables, as they propagate the spectral components of the ESD continuum that is efficient for them, will be responsible for susceptibility effects with varying probability above 5 kV due solely to these factors of energy transfer efficiency.
Interface Impulse Effects
The final propagational effects of system-product ESD are found as conducted currents in interface cables, including power line cables. Generally, data and signal interface cables are capable of propagating faster risetime impulse currents than are power cables, since many interface cables are shielded (resulting in lower inductance) and consist of many parallel wires which mutually combine to reduce common-mode inductance and impedance. (This is an effect that functions in the same manner as litz wire.) Power cables frequently are not shielded and do not consist of many parallel separate conductors (unlike interface cables), and accordingly tend not to support the faster impulse risetimes of ESD due to their increased inductive properties.
System Response Significance: Interface Effects
Although exceptions may be found to any rule, the probability (considering cable-conducted ESD impulse currents) is that well-shielded interface cables (with well-terminated shields) will either not contribute to significant ESD product susceptibility or, if they do, it is more likely that susceptibility will be exhibited at higher ESD initialization amplitudes, where the risetimes produce lower frequency spectra and the conducted efficiency of the cables is greater. Unshielded (poorly shielded or well-shielded but poorly terminated) data and signal cables can result in higher-frequency (i.e., lower ESD amplitude) responses because localized fields around the cables or localized high frequency transfer impedances can effectively bypass the normally-anticipated conducted energy effects that are associated with the direct cable inductance. (However, it is possible to design a common-mode loop flow architecture in a product that is sufficiently inadequate for susceptibility considerations, to the point where it may become economically or technically difficult to provide enough cable shielding without resorting to additional means, such as the utilization of lumped common-mode inductance in the interface ahead of the interface cables shields.)
Under these conditions, the cables in effect become receiving (or localized loading) antennas with effective areas (efficiencies) that will vary importantly with cable position consequently resulting in instability of the ESD performance since antenna area and efficiency is related to the bandwidth of efficiency which, of course, is related to the ESD amplitude/waveform continuum.
ESD Response Instability In Nature May Be Useful Toward Production Cost Effectiveness
Descriptions of the dynamic interactions between the ESD impulse continuum (with its related initialization amplitude-dependent spectral energy shape migrations) and various systems/products response mechanisms (operating in conjunction with product/system physical arrangement conditions) all combine to develop significant variations in the observed ESD performance of the system. These variations may be encountered both in the measured ESD threshold-amplitude of response, as well as the exhibited operational response characteristics. These variations in response and performance may be recognized as anticipatable and “normal” to the nature of the spectra of the ESD continuum.
Although absolutely normal to the physics of ESD, these variations can result in confusion and consternation in personnel who may be attempting to evaluate the performance of a system or product during ESD impulse exposure. This is particularly true because the nature of ESD dynamic physics defies the common wisdom that views the ESD phenomenon simply as an effect that proportionally ascends in difficulty with ascending ESD amplitude. The problem with this approach is that it incorrectly assumes that the higher the ESD initialization amplitude, the worse the systems response will be. Since the inverse is frequently true based on understanding the spectral and propagational influences, it is of little wonder that confusion results.
The conclusions yielded and supported through research of ESD dynamics directly contradict the traditional viewpoint both theoretically and empirically that higher amplitudes will be the worst for product/systems performance. These relatively recent research efforts affirm that the study of ESD dynamics (and the related systems/product response) is actually a study of dynamic impulsive spectral distributions wherein the excitation amplitudes have significantly large field intensities that accompany the field displacements (megavolts per meter), large impulse currents (to over 100 amperes peak) and extraordinarily fast and variable risetimes (tens of picoseconds to tens of nanoseconds). All of these components propagationally interact in various ways with specific response mechanisms of a product/system design and the installation conditions of the product or system. Understanding of the systems and product responses, along with their associated variabilities, may be gained only through understanding the dynamic interactions between the ESD continuum and the system’s spectral propagational mechanisms.
The variabilities that are natural to the physics of ESD and the propagational mechanisms of products may be viewed as instabilities by test personnel. The natural (human) reaction derived from the viewpoint of instability is to seek means of achieving stability, particularly during product validation tests. Toward this end, many attempts have been made at stabilizing the ESD test. Potentially, however, there are two major drawbacks to stabilizing the ESD test methodology (and consequently the ESD test results), especially if the means chosen to achieve the stabilization fixes the ESD pulse waveshape to the point that it cannot vary as it does in the “real world”. Drawn from conclusions based on the research, these potential drawbacks are:
- exaggerating the ESD susceptibility of the product under evaluation, resulting in exacerbated production costs to “fix” the overstated problem; and,
- understating the ESD susceptibility of the product under evaluation, resulting in inadequate performance margins in the installed base.
Exaggerated ESD Responses
Having established the importance of the ESD amplitude-waveform dependencies toward system/product performance, it may be recognized (by studying the research noted in the reference list) that if a simulated test method were to produce ESD waveforms which happened to coincide with the worst-case spectral response windows from a product, the ESD responses from the product would be exaggerated in potentially two ways.
First, the constant (probably repetitive) simulated waveshape would over-emphasize the probabilities of the response since “natural” ESD impulse shapes are highly variable at any given initialization amplitude, as shown in the listed references. Second, the characteristic severity of the response may be magnified because more impulses of a possibly worst-case nature impact the probability timeframe of a systems logic sequence. Third, many standards dramatically over-test systems for the most probable ESD condition: humans direct through fingertips.
Understated ESD Responses
In the same manner that a stabilized ESD test approach might happen to unrealistically coincide with the worst-case response window of a system, it is also possible that the stabilized waveform might not match the susceptibility admittance characteristics of a product. This raises the possibility that the “stabilized” test would not produce the excitations that would be found in the natural environment and otherwise cause the system to be susceptible. Although the initial production costs might be lower for the product, the eventual costs could be unacceptably high in terms of both finance and customer goodwill due to the potential of product/system field performance problems and the attendant retrofit need.
Given the evidence based on accepted research results, it is reasonable to suggest that the historical concept of learning the characteristics of the ESD waveform continuum as it actually exists and replicating that continuum both in waveshape and waveshape-probability during simulation tests during product evaluation may actually be the best approach, both in terms of product costs and accuracy of the simulation performance result from products/systems. This is despite the fact that it characteristically may cause a certain amount of test instability which in turn causes a certain amount of confusion and consternation among test personnel. It is possible that other simulation techniques that have been developed with a view toward simplifying and “stabilizing” the ESD test method may, in fact, be substituting test-lab efficiency for accurate and required ESD performance-measurement information that would serve as a vital predictor of product performance in the field. . Further, it is possible that the current trend observed in efforts to “stabilize” (as opposed to “standardize”) the ESD dynamic waveform continuum in the interest of product-evaluation test efficiencies will eventually become recognized as a misplaced, simplistic approach to a complicated and dynamic physics problem.
- King, W. Michael, “Dynamic Waveform Characteristics of Personnel Electrostatic Discharge”, Proceedings of the 1979 Electrical Overstressed-Electrostatic Discharge Symposium, Reliability Analysis Center, Rome Air Development Command, ITT Research Institute, Rome, NY (1979) pp. 78-87
- Byrne, William W., “Development of Design and Test Procedures to Meet Electrostatic Discharge (ESD)”, Proceedings of the 1982 MIDCON Convention Session 28/4, Dallas, TX (1982).
- King, W. Michael and Reynolds, David, “Personnel Electrostatic Discharge: Impulse Waveforms Resulting From ESD Of Humans Directly And Through Small Hand-Held Metallic Objects Intervening In The Discharge Path”, Proceedings of the 1981 IEEE International Symposium on Electromagnetic Compatibility (1981) pp. 577-590.
- King, W. Michael and Reynolds, David, “Personnel Electrostatic Discharge: Impulse Waveforms Resulting From ESD Of Humans Through Metallic-Mobile Furnishings in the Discharge Path”, Proceedings of the 1982 IEEE International Symposium, on Electromagnetic Compatibility (1982) pp. 212-219.
- Byrne, William W., “The Meaning of Electrostatic Discharge (ESD) In Relation to Human Body Characteristics And Electronic Equipment”, Proceedings of the 1983 IEEE International Symposium of Electromagnetic Compatibility (1983) pp. 369-380.
- Richman, Peter, “Classification of ESD Hand/Metal Current Waves versus Approach Speed, Voltage, Electrode Geometry And Humidity”, Proceedings of the 1986 IEEE International Symposium on Electromagnetic Compatibility (1986) pp. 451-460.
- King, W. Michael, “EMI And ESD Control In Commercial Devices”, Proceedings of the 1980 WESCON Session 29/2, Anaheim, CA (1980).
- Daout, B., and Ryser, H., “The Reproducibility of the Rising Slope In ESD Testing”, Proceedings of the 1986 IEEE International Symposium on Electromagnetic Compatibility (1986) pp. 467-474.
- King, W. Michael, “ESD Test Methodologies and Performance Criteria”, Hewlett-Packard DT Division Corp Standard (1979); Burroughs Computer Corp Standard (1982); Texas Instruments System Standard (1984); ANSI ESD Guide, 1st complete draft (1984). (*)
- King, W. Michael, “Systems Response To Electrostatic Discharge: Applications Of Impulse Waveform Research Toward Understanding of Product Performance”, Proceedings of the 1987 Electrical Overstress/Electrostatic Discharge Symposium, Reliability Analysis Center, EOS/ESD Association, Rome, NY (1987). (**)
(*) The Vertical Coupling Plane (VCP), Horizontal Coupling Plane (HCP), Incremental Level Test Requirements and incremental probability criteria were first devised and presented in these (and similar) documents.
(**) Note: This article is based on, and is an expansion of, the above published paper.
|W. Michael King
is a systems design advisor who has been active in the development of over 1,000 system-product designs in a 50 year career. He serves an international client base as an independent design advisor. Many terms used for PC Board Layout, such as the “3-W Rule”, the “V-plane Undercut Rule”, and “ground stitching nulls”, were all originated by himself. His full biography may be seen through his web site: www.SystemsEMC.com.Significantly, he is the author of EMCT: High Speed Design Tutorial (ISBN 0-7381-3340-X) which is the source of some of the graphics used in this presentation. EMCT is available through Elliott Laboratories/NTS, co-branded with the IEEE Standards Information Network.