A Guideline on Surge Voltages in AC Power Circuits Rated up to 600 V

François Martzloff General Electric Company Schenectady NY [email protected]

Reprinted, with permission, from Proceedings, 3rd International Symposium on Electromagnetic Compatibility, Rotterdam, 1979

Significance: Part 2 – Development of standards – Reality checks

Progress report to the European EMC community on the development of what became IEEE Std 587. Explains the proposition that a Ring Wave should be added to the traditional unidirectional impulses NOTE: A parallel paper was presented to the 1979 IEEE PES community at the Summer Power Meeting under the title “The Development of a Guideline on Surge Voltages in Low-Voltage AC Power Circuits”.

Filename: Guideline EMC A GUIDELINE ON SURGE VOLTAGES IN AC POWER CIRCUITS RATED UP TO 600 V

F.D. ?Aa:tz!cff General Electric Company Corporate Research and Development Schenectady, New York 12345, U.S.A.

Summary operating voltage) and having durations ranging irom a fraction of a microsecond to a millisecond. Overvoltages Surge voltages occurring in ac power circuits can be of less than two per unit are not covered here, nor are the cause of misoperation or product failure for residential transients of longer duration resulting from power equip- as well as industrial systems. The problem has received ment operation and failure modes. Because these low- increased attention in recent years because miniaturized amplitude and long-duration surges are generally not solid state devices are more sensitive to voltage surges amenable to suppression by conventional surge protective (spikes and transients) than were their predecessors. devices, they require different protection techniques. Although surge voltage amplitudes and their fre- 1. The Origin of Surge Voltages quexcy =f eccgrrence on unprotected circnits =re we!! known, their waveshapes and energy content are less well Surge voltages occurring in low-voltage ac power known. Qn the his of measurements; statltlc~:and circuits originate from two major sources: !oad switching theoretical considerations, a practical guideline for out- transients and direct or indirect lightning effects on the lining the environment for use in predicting extreme power system. Load switching transients can be further waveshapes and energy content can nevertheless be estab- divided into transients associated with (1) major power lished. The Surge Protective Device Committee of the system switching disturbances, such as capacitor bank Institute of Electrical and Electronics Engineers has been switching; (2) minor switching near the point of interest, developing such a guideline, the essential elements of such as an appliance turnoff in a household or the turnoff which are presented in this paper. of other loads in an individual system; (3) resonating circuits associated with switching devices, such as thyris- Surge voltages [ 11 occurring in ac power circuits tors; and (4) various system faults, such as short circuits rated up to 600 V can be represented by various waveshapes and arcing faults. Measurements and calculations of in an attempt to duplicate actual surge voltages. Two lightning effects have been made to yield data on what major types of surges reflecting differences in the environ- levels can be produced, even if the exact mechanism of any ment are described to represent the situation realistically. particular surge is unknown. The major mechanisms by which lightning produces surge voltages are the following: Systems located inside a building and separated from the overhead lines by some line impedance experience (a) A direct lightning strike to a primary circuit surge voltages of waveshapes and energy levels that differ injects high currents into the primary circuit, from those of the outdoor environment. Outside systems producing voltages by either flowing through exposed to direct lightning strikes or lightning-induced ground resistance or flowing through the surge surges--typically overhead lines--experience levels implied impedance of the primary conductors. L.. Wrrr ---->--A- I------I -aa-- TI.:- -,,:A-l:-- uy lccc DLalluaIUD IVI JCLUIIUaL y a1----~CJLSLJ. LlUJ 6YIUFllllr (ti) A !igh:ning s:rike that misses the !he but hit; a addresses particularly the hazards to these two types of nearby object sets up electromagnetic fields -..- rqi J,JLCIIIJ LLJ. *zhich can induce vo!?=pes on the conduc?ors of the primary circuit. Scope (c) The rapid collapse of voltage that occurs when a primary arrester operates to limit the primary The guideline presented here primarily addresses ac voltage couples effectively through the capaci- power circuits with rated voltages up to 600 V, although tance of the transformer and produces surge some of the conclusions offered could apply to higher voltages in addition to those coupled into the voltages and also to some dc power systems. Other secondary circuit by normal transformer action. standards have been established, such as IEEE 472, Guide (d) Lightning strikes the secondary circuits directly. for %Surge Withstand Capability (SWC) Tests, intended for Very high currents can be involved, exceeding the special case of high-voltage substation environments, the capability of conventional devices. and IEEE 28, Standard for Surge Arresters for ac Power (e) Lightning ground current flow resulting from Circuits, covering primarily the utilities environment. The nearby direct-to-ground discharges couples onto guideline presented here intends to complement, not the common ground impedance paths of the conflict with, existing standards, and to present a practical grounding network. proposal for the selection of voltage and current tests to be applied in evaluating the surge withstand capability of Fast-acting protection devices, such as current- equipment connected to these power circuits, primarily in limiting fuses and circuit breakers capable of clearing or residential and light industrial applications. beginning to part contacts in less than 2 ms, leave trapped inductive energy in the circuit upstream; upon collapse of Some guidance is also presented on how to proceed the field, very high voltages are generated. from the environment description to the selection of "standard" test waves. Transient overvoltages [ I I associated with the :-itching of pe~~erfacoy coyrffction cq=citors haye The surge voltages [I I considered in this guideline frequencies than the high-frequency spikes with which this & rhc* exceeding ?we pr unit (or twice the peak dnmment is concerned; Their !eve!sj at !east in the case of restrike-free switching operations, are generally less than 3. Waveshape of Representative Surge Voltages twice normal voltage and are therefore not of substantial concern here, but should not be overlooked. 3.1 Waveshapes in Actual Occurrences

On the other hand, switching operations involving Indoor - Measurements in the field, measurements in r,.rh +h,,, ,.,.-I ..,, A h.. . -,,+,,+,.. +h, 1-I.---.+ --., ....A *h ..--A +:--I --,-..,-.:--- :-A: -...- LsJL, LnCJ, D"C1I .; ,,lux y. YUUCCU YJ CYr,racrY, a 0: rnns >auv~arvt7, amw rmlsvn sr~calcalcularlulm Iltulcarc mercury switches, can produce, through escalation, surge that most surge voltages in indoor low-voltage sys- voltages of complex waveshapes and of amplitudes several tems have oscillatory waveshapes, unlike the well- *: -.-.---- *-- -L-- *L- 8 L-- TL- illllca grcairr LII~II LIE IIUIIII~I SYSLCIII VUILI~C. I IIC known and generaliy accepted uiiidiieciionai waves severest case is generally found on the had side of the specified in high-voltage insulation standards. A switch and involves only the device that is being switched. surge impinging on the system excites the natural While this situation should certainly not be ignored, in such resonant frequencies of the conductor system. As a a case the prime responsibility for protection rests with the result, not only are the surges typically oscillatory, local user of the device in question. However, switching but surges may have different amplitudes and wave- transients can also appear on the line side across devices shapes at different places in the system. These connected to the line. The presence and source of oscillatory frequencies of surges range from 5 kHz to transients may be unknown to the users of those devices. more than 500 kHz. A 30 to 100 kHz frequency is a This potentially harmful situation occurs often enough to realistic measure of a "typical" surge for most command attention. residential and light industrial ac line networks. 2. Occurrence and Voltage Levels Outdoor - Surges encountered in outdoor locations in IJnprotected Clrcljltr. have a!so been recorded, some being osci!htory !5!, others. being unidirectional. Because the overriding 2.1 Rate of Occurrence Versus Voltage Level concern here is the energy associated with these Tho rat- nf nrrnwronrn nf awooC vsrioe nvnr W~AP c,,-"-e 3 ,-,.ncc.w.,3+:,,,3 h,,+ Fa31;.-+;c. .-I-C,--;-+;~- -6 +he ...- .-.- ----..-..-- --.--" .-.--- -7-. ..A-- 2". 6L2, V LV".7LL .Y.*.L UU. L LU".,L'C "Lac1 1pL.V" Y' L"G limits, depending on the particular system. Prediction of surges can be derived from the long-established the rate for a particular system is always difficult and specified duty of a secondary arrester, as detailed in frequently impossible. Rate is related to the level of the Paragraph 3.2. While this specification is arbitrary, surges; low-level surges are more prevalent than high-level it has the strength of experience and successful surges [ 31. Data collected from many sources (Appendix I) usage. have led to the plot shown in Figure 1. This prediction shows with certainty only a relative frequency of occur- 3.2 Selection of Representative Waveshapes rence, while the absolute number of occurrences can be The definition of a waveshape to be used as repre- described only for an "average location." The "high sentative of the environment is important for the design of exposure" and "low exposure" limits of the band are shown candidate protective devices, since unrealistic require- as a guide, not as absolute limits [21, to reflect both the ments, such as excessive duration of the voltage or very location exposure (lightning activity in the area and the low source impedance, place a high energy requirement on nature nf the system! and the expos~eto stvitching sllrges the sq?pressor, with jl res"!ti~g most pem!ty tc the ed created by other loads. Such data are useful in that they user. The two requirements defined below reflect this describe the maximum levels likely to be encountered and trade-off. -.:.... -..-^ ..-a:--*.. ^S &La ....LA ..* ^^^..""^^^^ ^* "..^,. slvs ~u~nlssaun#rars YL uns ~arsul u~cut~st~sut ~ucln Indoor - Based on measurements conducted by several surges. Of equal importance is the observation that surges iiO V in the range of 1 to 2 kV are fairly common in residential indepei~deniuiganizaiions in and 240 systems circuits. (Appendix I), the waveshape shown in Figure 2 is reasonably representative of surge voltage in these power circuits. Under the proposed description of a "0.5 us x 100 kHz ring wave," this waveshape rises in 0.5 p, then decays while oscillating at 100 kHz, each peak being about 60% of the preceding peak.

SURGE CREST kV Figure 2. The Proposed 0.5 us x 100 kHz Ring Wave F~gure1. Rate of Surge Occurrence vs Voltage Level (Open-circuit Voltage)

From the relative values of Figure 1, two typical The fast rise can produce the effects associated with levels can be cited for practical applications. First, the nonlinear voltage distribution in windings and the expectation of a 3 kV transient occurrence on a 120 V dv/dt effects on semiconductors. Shorter rise times circuit ranges from 0.01 to I per year at a given location - are found in many transients, but, as those transients a number sufficiently high to justify the recommendation propagate into the wiring or are reflected from of a minimum 3 kV withstand capability. Second, the discontinuities in the wiring, the rise time becomes wiring flashover limits indicate that a 6 kV withstand longer. capability may be sufficient to ensure device survival indoors, but a 10 kV withstand capability may be required The oscillating and decaying tail produces the effects outdoors. of voltage polarity reversals in surge suppressors or other devices that may be sensitive to poiarity 2.2 Timing of Occurrence changes. Some semiconductors are particularly sensi- Surges- occur at random times with respect to the tive to damage when being forced into or out of a power frequency, and the iaiiure mode oi equipment may conducting state, or when the transient is appiiea be affected by the power frequency follow current. during a particular portion of the 60 Hz supply cycle Furthermore, the timing of the surge with respect to the (Appendix 11). The response of a surge suppressor can power frequency may affect the level at which failure also be affected by reversals in the polarity, as in the occurs 141. Consequently, surge testing must be done with case of RC attenuation before a rectifier circuit in a the line voltage applied to the test piece. dc power supply. The pulse withstand capability of many semiconduc- specified without reference to source impedance could tors tends to improve if the surge duration is much imply zero source impedance - one capable of producing shorter than one microsecond. For this reason, the that voltage across any impedance, even a short circuit. first half-cycle of the test wave must have a That would imply an infinite surge current, clearly an -,.sz;-:--* r;t,,~+;r.n JUI11LASI11 UUI,4..-..*:-.. aLIYII. I,nrsal;r+;r",,.LO..a..L a.L"-..Ymm.

Outdoor - In the outdoor and service entrance 4.2 Proposed Approach -.--..-* er~virur~~r~n~r,as we;; as in ~ocations dose to the ' Because oi the wide range of possibie source iiii- service entrance, substantial energy, or current, is pedances and the difficulty of selecting a specific value, still available. For these locations, the unidirectional three broad categories of building locations are proposed to impulses long established for secondary arresters are represent the vast majority of locations [7,8 I, from those more appropriate than the oscillatory wave. near the service entrance to those remote from it. The source impedance of the surge increases from the outside Accordingly, the recommended waveshape is 1.2 x to locations well within the building. Open-circuit 50 us for open-circuit voltage and 8 x 20 ps for voltages, on the other hand, show little variation within a short-circuit current or current in a low-impedance building because the wiring provides little attenuation 19 1. device. The numbers used to describe the impulse, Table 1 outlines the three categories of building wiring. 1.2 x 50 and 8 x 20, are as defined in IEEE Standard 28 - ANSI Standard C62.1; Figure 3 presents the Table 2 shows open-cirtuit voltages and short-circuit waveshape and a graphic description of the numbers. currents for each of the three categories. The energy deposiied in a 500 'v' suppressor has been coiiipuied and is (a) Open-circu~t shown for each of the categories. Voltage Waveform TABLE l I li Locatlon Categories

A. Outside and Service Entrance Service drop from pole to building ent Run between meter and distr~butionpa Y 'I Overhead line to detached build~ngs Underground lmes to well pumps 10. (b) Short-circuit Current (or Current in Low- B. Major Feeders and Short Branch Circuits Impedance C~rcu~t) nic+.;)...*in.. ..=..-I A-.,ir-c "La.. .""..""y"ll.1 "...LC.. I //I \ n,, 2nd feeder SYS?PmS in ind.s?ria! n~;lntr Heavy appliance outlets with "short" ronnertions to the service entrance Lightmg systems In commerc~albulldlngs C. Outlets and Long Branch Circults All outlets at more than LO m (30 ft) from Category B with wires 1114-10 Figure 3. Waveshapes for Outdoor Locations All outlets at more than 20 m (60 ft) from Cateeorv A with wlres 1114-10 4. Energy and Source Impedance

4.1 General TABLE 2 Thenergy involved in the interaction of a power system with a surge source and a surge suppressor will Ranges of Voltage and Currents divide between the source and the suppressor in accordance with the characteristics of the two impedances. In a gap- Energy Depos- type suppressor, the low Impedance of the arc after Max~mum ~ted~n a 500 V cnrarlmvar fnrrnc mnct ~f ?he enerov to di~cin~t~CI -,--..,.-... -.r-....-.-. --.--'- ...-I. 0, r---- L"L-.."am, mpu!s- $Unnrpssor elsewhere: for instance, in a power-follow current-limiting resistor that has been added in series with the gap. In an A. Outdoor and 0 10 kV 1.2 x 50 Us tor energy-absorber suppressor, by its very nature, a substan- Serv~ce h~gkmpedancec~rcults tial share of the surge energy .is dissipated in the Entrance 10 kA 8 x 20 w for suppressor, but its clamping action does not involve the low-~mpedancecrrcults 150 power-follow energy resulting from the short-circuit action B. Major Feeders 06 kV 1.2 x 50 Ps for of a gap. It is therefore essential to the effective use of and Short hlgh-impedance clrcults suppression devices that a realistic assumption be made Branch 03 kA 8 x 20 Ps for about the source impedance of the surge whose effects are Clrcu~ts ~mpedanceclrcults 40 to be duplicated. l 6 kV 0.5 )IS x 100 kHz The voltage wave shown in Figure 2 is intended to for htgh-impedance clrcults 0500 A short clrcult for represent the waveshape a surge source would produce low-~mpedanceclrcults 2 across an open circuit. The waveshape will be different when the source is connected to a load having a lower C. Long Branch 06 kV 0.5 us x 100 kHz for :--A....r^ -...A +ha Aan..^s +,. h;-h ;+ ir I,...,c.r ir LllZ UF61 FL LY.., WII1LII LL 1- ."..La La O 1UIILLI"II&,,nr+;nn L"'~Ua"CC, a1IU I c a hgh-aipd~cecaicuii. of the impedance of the source 161. 1 outlets 200 A short clrcult for low-impedance clrcults 0.8 TL- >_____A_ ___L:^L :-..^A^^^^ ' :----*....* 11rc UC~,cc LU WB~JC~IJUULCC IIII~~U~IICCi3 IIII~UL LantI depends largely on the type of surge suppressors that are used. The surge suppressors must be able to withstand the The values shown in the table represent the maximum current passed through them by the surge source. A test range, corresponding to the "High Exprsure" situation of generator of too high an impedance may not subject the Figure 1. For less exposed systems, or when the prospect device under test to sufficient stresses, while a generator of a failure is not highly objectionable, one could specify of too low an impedance may subject protective devices to lower values of open-circuit voltages with corresponding unrealistically severe stresses. A test voltage wave reductions in the currents. 5. Conclusion 2. Recordings by General Electric Company

The broad range of surge voltages occurring in low- (Data contributed by F.D. Martzloff [ 2 1 ) VO!?~~Pac Fwer circuits can be simu!ated by a limited set of test waves, for the purpose of evaluating their effects 2.1 Surge Counter Statistics on equipment. a) Three percsn: of a!! C.S. residences experience frequent occurrences (one per week or more) Field measurements, laboratory experiments, and above 1200 V calculations indlcate that two basic waves, at various open- b) There is a 100:l reduction in the rate of device circuit voltages and short-circuit current values, can failure when the withstand level is raised from represent the majority of surges occurring in residential, 2 kV to 6 kV. commercial, and light industrial power systems rated up to 600 V rms. 2.2 Typical Automatic Recording Oscilloscopes txceptions will be found to the simplification of a broad guideline; however, these should not detract from the benefits that can be expected from a reasonably valid uniformity in defining the environment. Other test waves of different shapes may be appropriate for other purposes, and the present guideline should not be imposed where it is not applicable. 6. Acknowledgments

The members of the IEEE Working Croup contributed the data, shown in Appendix I, that shaped this guideline. 0 5 20 Heipiui comments, discussions, ana reviews by ail members 10 ?O~OM~~IOOUS. . of the Surge Protective Devices Committee are also FURNACE IGNITION 24 HOUR PERIOD acknowledged. The author is particularly indebted to the contributions of Catharine Fisher and Peter Richman for the presentation and discussion of the concepts and recommendations made in this guideline. -- Append~~1 - Data Base Recordings and surge counter data have been con- tributed from several sources, in addition to the surge counter data obtained by members of the working group. Representative oscillograms and summary statistics are reprodtic& i:: this appendix, in rnppcrt cf the vc!?age !eve!s and oscillatory wave proposals. I. R~cordrngsby Bell Telephone Laboratones 2.3 Simulated Lightning Strokes on a Residential Power - Circuii iiaboratory ~odeioi Sysiem j i 9 1 (Data contributed by P. Speranza, internal report, unpublished to date) 1.1 Typical Surge Counter Statistics 120 V line at BTL facility in ester, New Jersey, during 42 months of monitoring: I46 counts at 300 to 500 V 14 counts at 500 to 1000 V 3 counts at 1000 to 1500 V 3 counts above 1500 V .Recording of open-circuit voltage !,2 Typlca! Automatic Recordine Oscilloscopes at a branch circuit outlet: 2200 V peak 500 kHz oscillations

By connecting a 130 R load at the same outlet (lA load) the voltage is reduced to 1400 V peak, with 0 20 40 60 80 100 120 ps more damping. 120 V OUTLET. LABORATORY BENCH

-.--a .:.-- F--- TL:- T--* c--:-- LUIICIU5IUIIS rIUlll 11115 lC5L XI IT5 i. A curreni of i.5 kA imoderaie for a iighining discharge injected in the ground system) raises the wiring system of the house 2.2 kV above ground. Four kiloamperes (still a moderate value) will bring this voltage to 6 kV, the typical flashover value of the wiring. 2. A natural frequency of 500 kHz is excited by a 277/480 V SERVICE ENTRANCE unidirectional impulse. 3. In this example, the source of the transient The General Electric counter statistics yield a (from the loading effect of 130 a) appears as point of 1200 V at about 1 occurrence per year (x). The General Electric clock data indicate a slope of iOO:i from i itit to 6 kv i- - - -j. The Regez data prov~dea band for the majority of 3. Statistics By Land~sand Gyr Company Inrstinnc (chnwn rrncc-hstrhorl\ with thn nvran------. .- .- . .- .. .. - . --I. .- I- .. - -, , .. . I. . ...- Surge counter data on various locations in Swiss 220 V tion of the rural location with long overhead line, systems (Data contributed by L. Regez - unpublished to which has more occurrences. date) Working Group statistics (. -- . --) indicate a less steep slope, perhaps because of the influence of outdoor locations Included in the sample (similar to the rural data of Regez). The proposed curve, which is the center of the 210 range of Figure 1, is shown In bold dashed lines (- - -). It has been drawn at the 100:l slope, passing near the Bell and General Electric points and located within the band of the Regez data.

'" 102 2 5 10' 2 5 10' 2 PEAK VOLTAGE V

-Servrce entrance, 16-famdy house, under- ground system ---Same house, outlet third floor hvmg room Same house, ourlet fifth floor lmmg room - - .- - - - . krvlce entrance of bank burldrng m Barel --- Landlr and Gyr Plant. Zug, outlet m lab. PEAK VOLTAGE - V Landlr and Cyr, Zug, outlet in furnace room Farmhouse rupplled by overhead lmes Appendix 11 - Effect of Transient Polarity Reversals oii S~iiii~~iidii~t~i~ 4. Working Group Surge Counter Statistics Breaitaown of sem~conaucrors under varlous conal- Surge counters with four threshold levels (350: 500: tlons of load and translent overvoltage appl~rat~onshas 1000, and 1500 V) were used to record surge occurrences at been 1nvest1~ated.t Ev~dence1s presented In the two various iocations. Members oi the Working Group instaiied mvestlgatlons c~tedthat a reverse voltage apphed durlng these on 120 and 240 V systems of various types, including the conduct~onper~od of the power frequency produces the following: outlets in urban, suburban, and rural lower breakdown voltages than the apphcat~onof the same residences; outlets in a hospital; secondary circuits on translent w~thno load or durmg blocking. Examples are distribution system poles (recloser controls); secondary of glven below, taken from these two mvestlgatlons, showlng padrmounted distribution transformers; lighting circuits in stat~stlcallyslgnlf~cant d~fferences In the voltage levels. an industrial plant; life test racks at an appliance manu- Average facturer; bench power supply in a laboratory. Breakdown (V) Summary Statistics of these measurements are as follows: IN1190 Diode* Transient at no load 1973 Data base from 18 locations with a total recording Fast wave under load 830 Slow wave under load 1097 time of 12 years spread over 4 calendar years, using 6 counters. Transient at no load 2056 Number of occurrences per year (weighted aver- Fast wave under load 894 ages) at ?'average iocaiion.!' -Zlnw - - .. wave. - . - ~tnrler-. .- - . .lnarl - - - .---I Inc. e 350 V: 22 Translenr appiiea ar: - peak of reverse voltage 1766 e 500 \.I: !! -25" after start of conductron 1181 0 1ooov: 7 -90' after start of conduction 906 -155' after start of conduction 1 115 0 1500 V: 3 This effect is one of the reasons for selecting an Significant extremes oscillatory waveform to represent the environment: it will One home w~thlarge number of surges be more hkely to Induce sem~conductor fa~luresthan a caused by washer operation. . unidirectional wave. Also, it shows the significance of the timing of the transient appl~cation with respect to the Four locations out of 18 never experienced a power frequency cycle. surge.

0 One home experienced several occurrences Appendix I11 - Notes and References above 1500 V, with none below that value. 1. Surge Voltage 0 One industriai iocation iswitching oi a test Deilnmons oi terms used in this guideiine are rack! produced ?housands of surges in the cc~jijtentwith !EEE Sr=&=rd !aQ-!977,Dicti~q>cf &c- 350-500 V range, and several surges in trical and Electronic Terms, 2nd ed.; however, some dif- excess of 1500 V. This location was left out ferences exist. For instance, IEEE Std 100-1977 defines a ,-.f +hn ~.,.~=,namrnnn~+=ti,-.n hn~tit nvnrnnli- sijrge as a "-..--- :--* -' --*--A:-' -- ' V' L.'.. O.L.UbL ..V...,,U.Y..V.., 1". 1. --.....,.,..- LI~II~CIIL wave ur ~urIFIIL, ~ULFIILIPI UI puwes in fies a significant extreme. the electric circuits'--a definition broader than that used From the data base cited in the preceding pages, one *Chowdhuri, P., "Transient-Voltage Characteristics of can draw the chart below, including the following informa- Silicon Power Rectifiers," IEEE IA-9, 5, September1 tion on voltage vs frequency (rate) of occurrence: October 1973, p. 582. 1. The Bell Laboratories data yield a point of 1000 V at about 2 occurrences per year (-). t F.D. Martzloff, internal report, unpublished. here. Transient overvoltage is defined as "the peak voltage 20 ps) are much longer than the travel times in the wiring during the transient condition resulting from the operation systems being considered, traveling wave analyses are not of a switching devicef'--a definition more restricted than useful here. that of the present guideline. Source impedance, defined as "the impedance 2. Amplitudes of Strikes, Worst Case presented by a source of energy to the input terminals of a The siiige voltages described in this giiideiine include device, or network?! iiEEE Standard iOOi, is a more useiui lightning eff& on ~owersystems? mostly strikes in the concept here. vicinity of a power line, or at a remote point of the power 7. Power System Sou~ceknpedance system. Tne iiterature describes rhe irequency oi occur- The measurements from wh~chFl~ure 1 was derived rence vs amplitude of lightning strikes, from the low levels were of voltage only. Little was -known about the of a few kiloamperes, through the median values of about impedance of the circuits upon which the measurements 20 kA, to the exceptional values in excess of 100 kA. were made. Since then, measurements have been reported Clearly, a secondary arrester rated for 10 kA can protect on the impedance of power systems. Bull reports that the adequately in case of a mild direct strike, or of a more impedance of a power system, seen from the outlets, severe strike divided among several paths to ground. exhibits the characteristics of a 50a resistor with 50 pH in However, a very high and direct strike will exceed the parallel. Attempts were made to combine the observed capability of an ANSI-rated secondary arrester. 6 kV open-circuit voltage with the assumption of a References: 50N5O pH impedance. This combination resulted in low energy depos~t~oncapab~lity, which was contradicted by Cianos, N. and E.T.Pierce, A Ground-Lightning Environ- field experience of suppressor performance. The problem ment for Engineering Usage, Stanford Research Institute, led to the proposed defmition of oscillatory waves as well ivienlo Park, CA 94205, A"g"5t i972. as high-energy unidirectional waves, in order to provide Bodie, D.Y., A.1. Ghazi, ivi. Syed, and 2.i. Yoods~de,Char- both :he eff?c:s of aii ~sci!kt~ij:wa'w and the high-energy acterization of the Electrical Environment, Toronto and deposition capability. Buffalo, N.Y.: University of Toronto Press, 1976. Reference: Martzloff, F.D. and G.J. Hahn, "Surge Voltage in Res~den- Bull, J.H., "Impedance of the Supply Mains at Radio tial and Industrial Power Circuits," ZEEE PAS-89, 6, July/ Frequencies," Proceedings of 1st Symposium on EMC, August 1970, 1049-1056. 75CH1012-4 Mont., Montreux, May 1975. 3. Level vs Rate of Occurrence 8. Installation Categories The relationship between the level and the rate of Subcommittee 28A of the International Electrotech- occurrence of surges is partly caused by the attenuation of nical Commission has prepared a report, referenced below, the surges as they propagate away from the source of the in which installation categories are defined. These surge and divide among paths beyond branching points. installation categories divide the power systems according Equipment at a given point will be subjected to a relatively to the location in the building, in a manner similar to the small number of high-level surges from nearby sources, but location categories defined in this guideline. However, +hara are *;"-;$;--..+ A;*$ h *L.- * to a larger number of surees from more remote sources. LB~LSL avlllL >r688AAAL'ImnL ULIICICIICCJ------UCLWCCII LllC LWU

$. T;-;nm -6 Cn,wm-c ..,;+h D-.- .-.-+ +- D-..,-.. C---..---.. concepts. First, the IEC categories are defined for a .A. ,,All6 "I -."a 6F.. w A,,, L\CJ~CLL L" I "WC, 1 ,C.,UF,IL)I Lightning surges are completely random in their "Controlled Voltage Situation," a phrase that implies the Switching presence of some surge suppression device or surge timing with respect to the power frequency. attenuation mechanism to reduce the voltage levels from siirges are iikeiy to occiir near oi after current zero, but one category to the next. Second, the IEC report is more variable load power factors will produce a quasi-random concerned with insulation coordination than with the distribution. Some semiconductors, as shown in Ap- application of surge protective devices; therefore it does pendix 11, exhibit failure levels that depend on the timing of the surge with respect to the conduction of power not address the question of the coordination of the frequency current. Gaps or other devices involving a protectors, but rather the coordination of insulation levels power-follow current may withstand this power follow with - that is, voltages. Source impedances, in contrast to this success, depending upon the fraction of the half-cycle guideline, have not been defined. Further discussion and remaining after the surge before current zero. Therefore, work toward the application guidelines of both documents it is important to consider the timing of the surge with should eventually produce a consistent set of recommenda- respect to the power frequency. In performing tests, either tions. complete randomization of the timing or controlled timing Reference: should be specified, with a sufficient number of timing coditions to ievea! the i?iojt eriticai timing. ov la ti on Coordination Within Low-Voltage Systems In- cluding Clearances and Creepage Distances for Equipment, internationai Eiectrotechnicai Commission, Report SCi8A the "classical lightn s been established as [Central Office) 5; to he puh!lshed in !979, 1.2 x 50 ~s for a voltage wave and 8 x 20 ps for a current 9. Open-Circuit Voltages and Wiring Flashover wave. Evidence has been coiiected, however, ro show that Surges propagate with very little attenuation in a oscillations can also occur. Lenz reports 50 lightning power system with no substantial connected loads. surges recorded in two locations, the highest at 5.6 kV, Measurements made in an actual residential system as well 100 to 500 kHz. Martzloff with frequencies ranging from as in a laboratory simulation have shown that the most reports oscillatory lightning surges in a house during a significant limitation is produced by wiring flashover, not multiple-stroke flash. be attenuation along the wires. Ironically, a carefully References: insulated installation is likely to experience higher surge voltages than an installation where wiring flashover occurs Lenz, J.E., "Basic Impulse Insulation Levels of Mercury at low levels. Therefore, the open-circuit voltage specified Lamp Ballast for Outdoor Applications," Illuminating Engrg., at the origin of a power system must be assumed to February 1964, pp. 133-140. propagate unattenuated far into the system, which is the Martzloff, F.D. and G.J. Hahn, "Surge Voltage in Resi- reason for maintaining the 6 kV surge specification when &n?ia! a~d!ndus$ri=! ?~I\wP~Circcif !EEE PACg_O, 6, Ju!~! going from the "B" location to the "Cut!ocation. August 1970, 1049-1056. References: 6. Surge !mpedance 2nd Source !mpedt.nce Martzloff, F.D. and K.E. Crouch "Coordination de la protection contre les surtensions dans les r6seaux basse To Drevent misunderstandina. a distinction between : ::a, .. .. source impedance and surge imped&ce needs to be made. rcl~a~vttI CXUCII urn, - rrucaearngs, is78 iEEE Canadian Surge impedance, also called characteristic impedance, is a Conference on Communications and Power, 78CH1373-0, concept relating the parameters of a long line to the pp. 451-454. propagation of traveling waves. For the wiring practices of Martzloff, F.D. Surge Voltage Suppression in Residential the ac power circuits discussed here, this characteristic Power Circuits, Report 76CRD092, Corporate Research impedance would be in the range of I50 to 300 Q, but and Development, General Electric Company, Schenectady, because the durations of the waves being discussed (50 to N.Y., 1976.