D. V. Giri, Life Fellow, IEEE, and William D. Prather ...IEEE Proof GIRI AND PRATHER: HEMP RISETIME...

8
IEEE Proof IEEE TRANSACTIONS ON ELECTROMAGNETIC COMPATIBILITY 1 High-Altitude Electromagnetic Pulse (HEMP) Risetime Evolution of Technology and Standards Exclusively for E1 Environment D. V. Giri, Life Fellow, IEEE, and William D. Prather, Senior Member, IEEE Abstract—There are many different definitions of the risetime of a transient waveform. In the context of high-altitude electromag- netic pulse (HEMP) standards, the 10–90% risetime of an idealized double exponential waveform has been defined and used for many decades. However, such a risetime definition is not strictly appli- cable to the transient voltage out of a pulse generator, since no practical switch can close in zero time. In this paper, we discuss various definitions and their applicability. More importantly, pulse power technology has evolved over five decades and the achievable risetimes have come down from 10s of nanoseconds to 10s of pi- coseconds. As a corollary, the highest achievable voltage gradient has been going upwards of 10 15 V/s. In this paper, we review the definitions of risetime, and trace the evolution of technology and HEMP Standards, exclusively for the E1 environments. Index Terms—Field sensors, HEMP standards, high-altitude electromagnetic pulse (HEMP), measurement systems, pulse rise- time, short pulse. I. INTRODUCTION A transient pulse generator is at the heart of any high-altitude electromagnetic pulse (HEMP) or a hyperband system [1] providing the required hyperband energy. Another paper [2] in this special issue discusses three basic types of HEMP simula- tors namely, 1) guided wave; 2) radiating; and 3) hybrid type that combines features of both guided wave and radiating types. In all of these HEMP simulators, the simulated electromagnetic environment E( r,t) and H( r,t) have some special relation- ship with the applied voltage waveform V(t).Consequently, the temporal and spectral purity of the voltage waveform governs the quality of the simulation. The voltage waveform has a cer- tain bandwidth and the simulator is generally expected to have a larger bandwidth to faithfully propagate all of the frequencies contained in the voltage waveform. The transient pulse gen- erator is best viewed as part of a wave launching system. It is simplistic to think of the pulse generator as merely a high- voltage device. It connects to a guiding wave structure or an antenna, and this interface between the pulse generator and the simulator has to be a “high frequency” connection to ensure no degradation of risetime. Therein lies the conflict. High voltages Manuscript received September 20, 2012; revised December 10, 2012; accepted December 13, 2012. D. V. Giri is with Pro-Tech, Alamo, CA 94507-1541 and also with the Department of ECE, University of New Mexico, Albuquerque, NM 87131 USA (e-mail: [email protected]). W. D. Prather is with the Air Force Research Laboratory, Kirtland AFB, NM 87117 USA (e-mail: [email protected]). Digital Object Identifier 10.1109/TEMC.2012.2235445 require larger standoff distances and high frequencies require shorter distances, to minimize unwanted inductances and stray capacitances. Large structures will then require tradeoffs and en- gineering compromises. Our objective in this paper is to initially review HEMP Standards in Section II and also point out how to translate the standards into specifications. In Section III, we look at risetime definitions and trace the evolution of switching technology that has permitted increasing voltages to be switched in shorter times. The paper is concluded with some summarizing comments in Section IV followed by a list of references. II. UNCLASSIFIED HEMP STANDARDS Unclassified HEMP standards are characterized by idealized double exponential (DEXP) and quotient exponential (QEXP) waveforms. The HEMP standards are derived by enveloping (in time and frequency domains) many possible waveforms. Then, a mathematical model is created that best expresses both the temporal as well as the spectral characteristics of the envelope. The measured time-domain waveforms from a high-altitude det- onation are not perfect DEXPs. The waveforms vary quite a bit depending on weapon design, altitude, etc. The DEXP is a model, and a mathematical representation of an envelope. The model is chosen as a convenient analytic expression whose frequency spectrum envelopes that of the actual HEMP from the weapon. It is analytic and convenient to use. It is a reasonable representation of the HEMP, and its time-domain properties (risetime and exponential decay) are used to design high-voltage generators that are used for testing. This is illus- trated in Fig. 1 for the DEXP and QEXP models. A. DEXP Representation The DEXP description of the HEMP waveform has been used since the early days of HEMP research and well described in [3]. The time domain expression is E(t)= E o [e α ( t t o ) e β ( t t o ) ]u(t t o )V/m E o = field intensity constant V/m α = decay constant rad/s β = risetime constant rad/s u(t t o )= unit step function t o = timeshift s (1) 0018-9375/$31.00 © 2012 IEEE

Transcript of D. V. Giri, Life Fellow, IEEE, and William D. Prather ...IEEE Proof GIRI AND PRATHER: HEMP RISETIME...

Page 1: D. V. Giri, Life Fellow, IEEE, and William D. Prather ...IEEE Proof GIRI AND PRATHER: HEMP RISETIME EVOLUTION OF TECHNOLOGY AND STANDARDS EXCLUSIVELY FOR E1 ENVIRONMENT 3 TABLE I PARAMETERS

IEEE

Proo

f

IEEE TRANSACTIONS ON ELECTROMAGNETIC COMPATIBILITY 1

High-Altitude Electromagnetic Pulse (HEMP)Risetime Evolution of Technology and Standards

Exclusively for E1 EnvironmentD. V. Giri, Life Fellow, IEEE, and William D. Prather, Senior Member, IEEE

Abstract—There are many different definitions of the risetime ofa transient waveform. In the context of high-altitude electromag-netic pulse (HEMP) standards, the 10–90% risetime of an idealizeddouble exponential waveform has been defined and used for manydecades. However, such a risetime definition is not strictly appli-cable to the transient voltage out of a pulse generator, since nopractical switch can close in zero time. In this paper, we discussvarious definitions and their applicability. More importantly, pulsepower technology has evolved over five decades and the achievablerisetimes have come down from 10s of nanoseconds to 10s of pi-coseconds. As a corollary, the highest achievable voltage gradienthas been going upwards of 1015 V/s. In this paper, we review thedefinitions of risetime, and trace the evolution of technology andHEMP Standards, exclusively for the E1 environments.

Index Terms—Field sensors, HEMP standards, high-altitudeelectromagnetic pulse (HEMP), measurement systems, pulse rise-time, short pulse.

I. INTRODUCTION

A transient pulse generator is at the heart of any high-altitudeelectromagnetic pulse (HEMP) or a hyperband system [1]

providing the required hyperband energy. Another paper [2] inthis special issue discusses three basic types of HEMP simula-tors namely, 1) guided wave; 2) radiating; and 3) hybrid typethat combines features of both guided wave and radiating types.In all of these HEMP simulators, the simulated electromagneticenvironment �E(⇀

r , t) and �H(�r, t) have some special relation-ship with the applied voltage waveform V(t).Consequently, thetemporal and spectral purity of the voltage waveform governsthe quality of the simulation. The voltage waveform has a cer-tain bandwidth and the simulator is generally expected to havea larger bandwidth to faithfully propagate all of the frequenciescontained in the voltage waveform. The transient pulse gen-erator is best viewed as part of a wave launching system. Itis simplistic to think of the pulse generator as merely a high-voltage device. It connects to a guiding wave structure or anantenna, and this interface between the pulse generator and thesimulator has to be a “high frequency” connection to ensure nodegradation of risetime. Therein lies the conflict. High voltages

Manuscript received September 20, 2012; revised December 10, 2012;accepted December 13, 2012.

D. V. Giri is with Pro-Tech, Alamo, CA 94507-1541 and also with theDepartment of ECE, University of New Mexico, Albuquerque, NM 87131 USA(e-mail: [email protected]).

W. D. Prather is with the Air Force Research Laboratory, Kirtland AFB, NM87117 USA (e-mail: [email protected]).

Digital Object Identifier 10.1109/TEMC.2012.2235445

require larger standoff distances and high frequencies requireshorter distances, to minimize unwanted inductances and straycapacitances. Large structures will then require tradeoffs and en-gineering compromises. Our objective in this paper is to initiallyreview HEMP Standards in Section II and also point out howto translate the standards into specifications. In Section III, welook at risetime definitions and trace the evolution of switchingtechnology that has permitted increasing voltages to be switchedin shorter times. The paper is concluded with some summarizingcomments in Section IV followed by a list of references.

II. UNCLASSIFIED HEMP STANDARDS

Unclassified HEMP standards are characterized by idealizeddouble exponential (DEXP) and quotient exponential (QEXP)waveforms. The HEMP standards are derived by enveloping (intime and frequency domains) many possible waveforms. Then,a mathematical model is created that best expresses both thetemporal as well as the spectral characteristics of the envelope.The measured time-domain waveforms from a high-altitude det-onation are not perfect DEXPs. The waveforms vary quite abit depending on weapon design, altitude, etc. The DEXP is amodel, and a mathematical representation of an envelope.

The model is chosen as a convenient analytic expressionwhose frequency spectrum envelopes that of the actual HEMPfrom the weapon. It is analytic and convenient to use. It is areasonable representation of the HEMP, and its time-domainproperties (risetime and exponential decay) are used to designhigh-voltage generators that are used for testing. This is illus-trated in Fig. 1 for the DEXP and QEXP models.

A. DEXP Representation

The DEXP description of the HEMP waveform has been usedsince the early days of HEMP research and well described in [3].The time domain expression is

E(t) = Eo [e−α(t−to ) − e−β (t−to ) ]u(t − to)V/m

Eo = field intensity constant V/m

α = decay constant rad/s

β = risetime constant rad/s

u(t − to) = unit step function

to = timeshift s (1)

0018-9375/$31.00 © 2012 IEEE

Page 2: D. V. Giri, Life Fellow, IEEE, and William D. Prather ...IEEE Proof GIRI AND PRATHER: HEMP RISETIME EVOLUTION OF TECHNOLOGY AND STANDARDS EXCLUSIVELY FOR E1 ENVIRONMENT 3 TABLE I PARAMETERS

IEEE

Proo

f

2 IEEE TRANSACTIONS ON ELECTROMAGNETIC COMPATIBILITY

Fig. 1. Time and frequency domain waveforms for HEMP comparing theDEXP and the QEXP models. (a) Temporal waveform. (b) Spectral magnitude.

and the frequency domain expression is

E(ω) = Eo

(1

jω + α− 1

jω + β

)

E(ω) = Fourier transform of E(t)V

m − Hz

ω = 2πf = radian frequency rad/s

j =√−1. (2)

The DEXP may be characterized in a number of ways. Theequations are defined by the three variables Eo , α, and β, whichprecisely define the DEXP waveform. The time-domain repre-sentation, however, is typically characterized by quantities moreeasily related to the measured waveform. That is, 1) Peak electricfield Ep (Note: Ep �=Eo ); 2) 10–90% risetime tr ; 3) Full-width,half-max (FWHM) or the e-fold decay time, the time when theamplitude reaches 1/e of Ep (≈ 37%). These are chosen becausethey can be read right off the measured curves.

In 1975, Bell Laboratories published an EMP engineeringhandbook [4], which used this expression to describe the HEMPwaveform. The parameters that were used in the handbook wereEo = 52.5 kV/m, α = 4.0 × 106rad/s, and β = 4.76 × 108rad/s,which means that Ep = 50 kV/m, tr = 2.2/β = 5.5 ns, thelow-frequency spectral density = 14.4 [mV/(m-Hz)], the firstbreak frequency occurs at α/(2π) = 637 kHz, and the secondbreak frequency occurs at β/(2π) = 76 MHz. It is noted that theBell Standard [4] has the widest waveform with the highest peak

amplitude. In reality, these do not occur together. The electricfield amplitude is not high when the pulse is wide. More recentstandards address this issue by considering the area under thetemporal electric field, which is related to the low-frequencycontent in estimating the pulsewidth.

The only drawback of the DEXP form is that it is discontinu-ous at t = 0, which creates a discontinuity in the first time deriva-tive. This is not consistent with natural physical processes andcreates computational difficulties. However, the simple analyticDEXP waveform has been used for many years to approximateimportant characteristics of HEMP waveforms and simulators,but it does have this limitation. As a result, another analyticalform was derived.

B. Inverse DEXP or Quotient Double Exponential (QEXP)

In order to correct for the discontinuity, another analytic formwas derived, which is the reciprocal of sum of two exponentials,sometimes referred to as inverse or QEXP [2] as shown in (3)and (4). The time domain form is

E(t) =Eo

[eβ (t−to ) + e−α(t−to ) ]V/m (3)

and the frequency domain form is

E(ω) =Eoπ

(α + β)csc

(α + β)(jω + β)

]e−jω to

V

m − Hz.

(4)This waveform has the advantage that it has continuous time

derivatives of all orders for all times. The disadvantage of thisexpression is that it extends to t = −∞ and has infinite numberof poles in the frequency domain.

The parameter t0 is used to adjust the amplitude of the sig-nal for arbitrarily small values of t < 0. We can now turn ourattention to unclassified HEMP standards. There are at leastseven unclassified HEMP specifications that are either DEXPor QEXP waveforms. Arranged chronologically, these are listedin Table I. The various parameters of the previous unclassifiedHEMP Standards are compiled and listed in Table I. A compar-ison plot is shown in Fig. 2.

In addition to the civilian HEMP standards described previ-ously, there is also a military standard [10], MIL-STD-464A,which is identical to the HEMP standard in IEC 61000-2-9. Forcompleteness, the MIL-STD-464-A HEMP waveform is shownin Fig. 3. It should be noted that we are only dealing with un-classified E1 HEMP standards in this paper.

In summarizing the unclassified civilian and military HEMPstandards, one might say that these standards are characterizedin time domain by three numbers: peak field, 10–90% risetime,and the FWHM. The ranges of the three parameters in the eightunclassified standards that we have reviewed previously are asfollows:

1) Peak electric field ranges from 50 to 65 kV/m2) 10–90% risetime ranges from 0.9 to 4.6 ns3) FWHM ranges from 23 to 184 ns.It is observed that over the last 40 years, the risetime and

duration of the HEMP in unclassified standards has come downby factors of 5 and 8, respectively!

Page 3: D. V. Giri, Life Fellow, IEEE, and William D. Prather ...IEEE Proof GIRI AND PRATHER: HEMP RISETIME EVOLUTION OF TECHNOLOGY AND STANDARDS EXCLUSIVELY FOR E1 ENVIRONMENT 3 TABLE I PARAMETERS

IEEE

Proo

f

GIRI AND PRATHER: HEMP RISETIME EVOLUTION OF TECHNOLOGY AND STANDARDS EXCLUSIVELY FOR E1 ENVIRONMENT 3

TABLE IPARAMETERS OF UNCLASSIFIED HEMP STANDARDS (NOTE: IEC 77C [6] IS THE SAME AS DEXP IN BAUM [5])

Fig. 2. Time-domain plots of the unclassified civilian HEMP standards in Table I.

Fig. 3. E1 HEMP Environment form MIL-STD-464-A [10], which is identicalto IEC 61000-2-9.

The earliest HEMP Standard [4] used the slowest and thewidest pulse calculated at that time. As we learned later, thisled to overtesting of low frequencies and undertesting at higherfrequencies.

Returning to our discussion of the rise time, the conventionaldefinitions are the exponential rise and the 10–90% rise relatedfor idealized DEXP waveform, by

exponential risetime ≡ te =t10−90%

�n(9)

∼= t10−90%

2.197∼= 0.455 t10−90% . (5)

�n(9), in the previous expression is the natural logarithm of9. In practical terms, pulse generators that aim to simulate theHEMP standards cannot be ideal exponentials. For this reason, abetter definition of risetime [3], [11] has been offered as follows:

maximum rate of rise ≡ tmrr =Epeak(

dEdt

)peak

= te(for an ideal exponential rise). (6)

In practical waveforms, reciprocal of the maximum rate ofrise tmrr appears to be a better indicator of the high-frequencycontent in the waveform. This definition has also been appliedto measured lightning current waveforms [12].

Page 4: D. V. Giri, Life Fellow, IEEE, and William D. Prather ...IEEE Proof GIRI AND PRATHER: HEMP RISETIME EVOLUTION OF TECHNOLOGY AND STANDARDS EXCLUSIVELY FOR E1 ENVIRONMENT 3 TABLE I PARAMETERS

IEEE

Proo

f

4 IEEE TRANSACTIONS ON ELECTROMAGNETIC COMPATIBILITY

The common definition of 10–90% is quite often impractical.There can be a prepulse in the transient pulser waveform, whichcan make it hard to determine the 10% value. If there are someminor ripples in the waveform near the peak, there can be morethan one 90% value. The definition of tmrr in (6) gets aroundthese issues.

For the QEXP in HEMP standards of Table I, the interrela-tionships of all three risetimes (exponential, 10–90%, and max-imum rate of rise) are also given by (5) and (6). Although theHEMP Standards and even some natural lightning standards arecharacterized by DEXP waveforms, practical pulser outputs arebetter modeled by the following expression [13]–[16]:

V (t) ={

Vo e− β t

t d

[(12

)erfc

(√π |t| /td

)]t < 0

V0 e− β t

t d

[1 −

(12

)erfc

(√πt/td

)]t > 0.

erfc (z) is the complimentary error function given by

erfc(z) = 1 − erf(z) =2√π

∞∫z

exp(−t2)dt

V (ω) =V0td

(β + jωtd)e[

14 π (β+jω td )2 ]. (7)

This analytical model of the pulser is still characterized bythree numbers and has continuous derivatives. This model canbe explained as follows. Consider a Gaussian waveform. Anintegrated Gaussian is an s-shaped waveform. When this s-shaped waveform reaches its peak, we add an exponential decayfactor to it. Such a process is represented by the time-domainexpression in (7). Typical pulser outputs are well represented bythis model. As an example represented by this model. As an, fora pulser with V0 = 120.72 kV, td = risetime = 100 ps, and β =risetime/decay time = 0.005, the resulting decay time = 20 nsand the maximum rate of rise for this pulser is

(dV/dt)max = 1.2 × 1015V/s and

tmrr = 120 kV/1.2 × 1015 = 100 ps.

Having reviewed the unclassified HEMP standards, it is clearthat all of the standards are expressed in the time domain. Asa result, it is very typical for the writers of the HEMP specifi-cations to use the standard as a specification. Time and again,we have seen the procuring agencies state the specification interms of simulating time-domain electromagnetic fields over acertain volume of space with specified uniformity. There are al-ways major differences between idealized waveforms in HEMPstandards and simulated waveforms in reality. We show a com-parison in Fig. 4. The use of the derivative waveform in definingthe risetime as per (6) is shown in Fig. 5.

In Fig. 4, t50i and t50d are the times at which the waveformreaches 50% of its peak on the initial rise and decay portions,respectively. Similarly, E50i and E50d are the correspondingamplitudes at these two instances. E with a dot on top of it isthe derivative waveform.

Fig. 4. Comparison of idealized and practical waveforms.

Fig. 5. Use of the derivative to define the risetime.

In Fig. 5, we show the initial rise portion of the simulatedwaveform and its derivative. t10 and t90 are the time instanceswhen the amplitude reaches 10 and 90% of the peak. The deriva-tive waveform is used in defining the peak amplitude T . The am-plitude peak is reached when the derivative waveform becomeszero for the first time.

In addition, there is usually no reference to the spectral contentof the simulated fields in the specifications.

1) A time-domain E1 HEMP standard (classified or un-classified) is not equivalent to an E1 HEMP simulatorspecification.

A complete HEMP simulator specification should specify ac-ceptable deviations of the simulated fields over the test volume,from the ideal standard (classified or unclassified) in both timeand frequency domains.

This problem of incomplete simulator specification wouldnot have arisen if the standards themselves had specificationsin both time and frequency domains, including acceptable de-viations from the ideal waveforms. The “acceptable deviations”appears to be an issue between the procurement agency andthe supplier of the E1 HEMP simulator hardware. The stan-dards do not address issues associated with actual simulationof these environments. The VERIFY facility [17], a threat-levelsubnanosecond E1 HEMP simulator, was the first one to useproper HEMP specifications in both temporal and spectral do-mains. The specification of the spectral domain for the VERIFY

Page 5: D. V. Giri, Life Fellow, IEEE, and William D. Prather ...IEEE Proof GIRI AND PRATHER: HEMP RISETIME EVOLUTION OF TECHNOLOGY AND STANDARDS EXCLUSIVELY FOR E1 ENVIRONMENT 3 TABLE I PARAMETERS

IEEE

Proo

f

GIRI AND PRATHER: HEMP RISETIME EVOLUTION OF TECHNOLOGY AND STANDARDS EXCLUSIVELY FOR E1 ENVIRONMENT 5

simulator was prescribed as: “In the frequency range from dcto 500 MHz, the spectral amplitude densities shall not deviatemore than ± 6 dB from the theoretical spectrum of the DEXPpulse given in paragraph __, and not more than ± 12 dB in thefrequency range from 500 MHz to 1 GHz.”

The reason why spectral fields are important can be statedsimply as follows.

1) The coupling and interaction of electromagnetic fieldswith a complex test object such as an aircraft, a pieceof electronic equipment, a battle tank, a ship or a satel-lite, is a strong function of frequency, so it is extremelyimportant to have all of the right frequencies, at the rightmagnitude, present in the incident or simulated field.

2) It is entirely possible to meet the time domain specifica-tions (peak field, risetime, and fall time), but have unac-ceptable notches in the spectral domain. A good exampleof this was the deep notch at 25 MHz in ALECS [17].The notch occurred in the electric field at the geometri-cal center of the working volume, and in the magneticfield quarter wavelength away. It was almost fortuitousthat an electric field sensor was placed at the center pointone time, and the notch was discovered. It is important tonote that the spectral notch is imperceptible in the timedomain measurements, but in the frequency domain, it isvery prevalent. The reason for the notch is the presenceof a TM01 mode in the transmission line, which canceledthe electric field of the desired TEM wave at a certain fre-quency and at a certain location. If the article being testedhad a resonant response at the notch frequency of 25 MHz,it would not be excited at all and the test would lead toerroneous conclusions. A second example of a notch inthe frequency domain was in ARES where the originalVan de Graff generator had an internal antiresonant notch,so that a certain frequency never got out of the pulser.

In this section, we have reviewed the HEMP Standards andhow they have changed over four decades. The revisions aredriven by improved calculations of the radiated EM fields fromnuclear detonations. In addition, we have traced the differencesbetween HEMP standards and specifications. HEMP standardslead to simulation facilities that are essential in threat-level test-ing and vulnerability assessment. The pulse power technologyhad to cope with the changing HEMP standards and we describethe evolution of this technology in the following section.

III. EVOLUTION OF PULSED POWER TECHNOLOGY—USER’S PERSPECTIVE

We will briefly look at a few types of transient pulse genera-tors. With each of these types, there are specific components thataffect the risetime of the output pulse. Most importantly, thereis a need to minimize stray inductors and the single most criticalcomponent in determining the risetime of the output pulse is theoutput switch. This last stage switch in a pulse generator can bean oil switch, or a spark gap switch that uses a gas, examples ofwhich are—Nitrogen, Hydrogen SF6 or mixtures of gasses, etc.

Pulse generators for HEMP facilities typically have dc sourceas prime power. The time-invariant voltage has to be shaped to

Fig. 6. VERIFY [17] pulser with SF6 gas as the insulating medium.

produce a pulsed waveform that is fast rising and slowly de-caying. An early review of the pulsed power is available in [19]and [20]. In the 1970 s and 1980 s, a 10-ns risetime was practicalat 100s of kilovolt and even into several MegaVolt with ATLASI (commonly known as Trestle) as a prime example. The typesof transient energy generators are: 1) Marx generators; 2) LCgenerators; 3) stacked transmission lines; 4) Van de Graaffs; and5) pulsed transformers. Of all these types, the Marx generatorhas become the most widely used for HEMP applications. Ina typical Marx generator, several capacitors are dc charged inparallel and spark gap switches are fired to connect the chargedcapacitors in series. Thus, high amplitude of voltage is builtbefore transferring energy to a load through an output switch.An improvement to a Marx circuit involves the use of either atransfer capacitor or a peaking capacitor. High currents throughthe spark gap of the order of 200 kA are possible with an asso-ciated charge transfer of 2 C. In HEMP pulser applications, theenergy density of capacitors used in storing the transient energyis an important parameter, because weight, size, and cost areall factors governing the reliability of the system. The storedenergy capability of a capacitor is measured in Joule/cc [21]. Itis noted that the energy densities have quadrupled over a periodof 25 years, from 0.5 J/cc in 1984 to 2 J/cc in 2008, which nowpermit compact and lighter pulse power systems.

There is another important aspect of pulse power and thatis the use of appropriate insulating media. Solids, liquids, andgasses have been studied and used, for their ability to withstandhigh electric fields. Solids tend to have high dielectric strengthsand used in capacitors, in the form of Mylar, polyethylene, etc.Quite often, the use of solid insulators also requires a surround-ing fluid (oil or gas) to combat field enhancements and coronaeffects. Fluids (oil and gas) as insulating media are self-healing,unlike solids. Liquid insulators are well suited for short pulsesand used extensively in Marx generators. A recent pulse genera-tor with amplitude of ∼600 kV, risetime of 900 ps, and FWHMof 25 ns [16] has used SF6 gas as the insulating medium. ThisVERIFY pulser installed in the HPE Laboratory in Switzerlandis shown in Fig. 6. The Marx stages and the peaking capacitorare seen in Fig. 6 in a transparent container filled with SF6 . Thispulser drives a conical transmission line type of HEMP simu-lator. What distinguishes this pulse generator from many others

Page 6: D. V. Giri, Life Fellow, IEEE, and William D. Prather ...IEEE Proof GIRI AND PRATHER: HEMP RISETIME EVOLUTION OF TECHNOLOGY AND STANDARDS EXCLUSIVELY FOR E1 ENVIRONMENT 3 TABLE I PARAMETERS

IEEE

Proo

f

6 IEEE TRANSACTIONS ON ELECTROMAGNETIC COMPATIBILITY

TABLE IIBREAKDOWN FIELDS FOR VARIOUS GASSES AND GAS MIXTURES FOR

A MILDLY ENHANCED MONOCONEGAP [26]

Fig. 7. Breakdown strength of two types of liquid Galden as a function ofpressure.

is the use of SF6 gas as the primary insulating medium. SF6gas has been extensively used for decades in providing pulsepower for HEMP simulators, and also in electric power indus-try. However, there are disadvantages in its use that have beenrecently recognized [22]–[25]. Daout and Vega [23] consider ahypothetical case of a 1-MV Marx with a peaking circuit, forHEMP simulator. If the spark gap uses 1.4 L or 8.8 g of SF6 ,the equivalent CO2 production is 200 kg for a series of pulses.They claim this is equivalent to a medium sized car traveling adistance of 1340 km. Therefore, recycling and incineration ofpolluted SF6 is an option in the future, instead of releasing thegas into the atmosphere. In concluding our comments on theuse of insulating gasses, it is noted that a fairly detailed studyof breakdown fields of certain gasses and gas mixtures [26] hasbeen documented. A representative measurement of the meanbreakdown field of various gasses and gas mixtures is shown inTable II.

Voltage polarity does not seem to make a significant differ-ence and data are not available for positive polarity in the caseof SF6 . It is very common to use SF6 spark gap switches inpulse generators for HEMP simulation.

More recently, some insulation strength measurements havebeen performed on a liquid called Galden, a Perfluoropolyether(PFPE) [27]. Measured breakdown field as a function of pres-sure is shown in Fig. 7 and is comparable to gasses discussedpreviously.

In the last three years, ISL [27] have reached breakdown fieldlevel of 9.3 MV/cm with Galden HT270, at pG = 1570 kPa∼15.25 atm for a switch gap of d = 0.30 mm. Vb = 271 kV→Eb = 9.3 MV/cm. Perhaps, this opens the way to an all liquidMarx generators in the future avoiding the negative aspects ofSF6 .

Returning to the subject of risetimes, since the early work ofHEMP simulation in 1960 s and 1970 s, an order of magnitudeimprovement in the risetime has been realized. The risetimeswere ∼10 ns in the 1960 s and now VERIFY [17] HEMP sim-ulator is capable of 900 ps. Risetime of 900 ps is the fastestunclassified HEMP standard [8], although a risetime of 2.5 ns isa more commonly used waveform from other standards [5], [6],and [9]. Pulsers with voltage amplitudes up to 1 MV and rise-times of 200 ps are also now possible, albeit, they are not re-quired in HEMP applications. In the beginning, the pulser re-quirements for HEMP simulation was in the range of 100 kVto ∼5 MV with fast rising (∼10 ns) pulses lasting 100s of ns.Such voltages could not be switched out in fast risetimes, sointermediate stages of capacitors or pulse lines were required.Switching technology (risetime related) and insulation technol-ogy for high amplitudes, required in controlling the flow of pulsepower have vastly improved.

IV. SUMMARY

In this paper, we have reviewed the evolution of unclassifiedHEMP Standards. E1 HEMP Standards are seen to be idealizedwaveforms in temporal and spectral domains. It is worth not-ing that 1/f 2 decay in the frequency domain, in the standardsis artificial. The standards are ideal waveforms and E1 HEMPsimulators should not have to follow this behavior precisely. Per-haps the standard waveforms could be improved in the future.Also it can be noted that the E2 and E3 parts of the HEMP wave-form extend (and increase) the total HEMP frequency contentat low frequencies, and it is difficult to separate E1 and E2 froma theoretical point of view even though the standard waveformsdisplay a separation. The important aspect is to ensure that sim-ulators do not take extra efforts to mimic aspects of the standardwaveforms that are not precisely correct. In practice, one hasto develop E1 HEMP specifications, based on standards, to bemet by practical facilities. The specifications are not the same asstandards, but are based on the standards. Acceptable deviationsfrom the standard waveforms (temporal and spectral) becomeimportant in specifying practical HEMP simulator facility per-formance. In Table III, we trace the worldwide simulators [18]and focus on the evolution of risetimes. Of course, the risetimeof the simulated electromagnetic pulse fields is indicative of thehighest significant frequency in the waveform.

The VERIFY facility in Switzerland has the fastest E1 HEMPsimulated pulse with a risetime of 900 ps in the working volume.

V. DEDICATION

The authors wish to dedicate this paper to the memory ofDr. C. E. Baum who was instrumental in developing HEMPsimulator concepts and sensors for the measurement of electro-magnetic quantities. His ideas have been implemented in many

Page 7: D. V. Giri, Life Fellow, IEEE, and William D. Prather ...IEEE Proof GIRI AND PRATHER: HEMP RISETIME EVOLUTION OF TECHNOLOGY AND STANDARDS EXCLUSIVELY FOR E1 ENVIRONMENT 3 TABLE I PARAMETERS

IEEE

Proo

f

GIRI AND PRATHER: HEMP RISETIME EVOLUTION OF TECHNOLOGY AND STANDARDS EXCLUSIVELY FOR E1 ENVIRONMENT 7

TABLE IIIWORLDWIDE HEMP SIMULATORS, CHRONOLOGICALLY ARRANGED WITH FOCUS ON RISETIMES AND VOLTAGE SWITCH OUTS

nations of the world. He also advised pulse power developers onmany aspects and especially on the topic of interfacing a pulserto an antenna or a transmission line.

REFERENCES

[1] D. V. Giri and F. M. Tesche, “Classification of intentional electromagneticenvironments (IEME),” IEEE Trans. Electromagn. Compat., vol. 46, no. 3,pp. 322–328, Aug. 2004.

[2] J. C. Giles and W. D. Prather, “Worldwide high-altitude nuclear electro-magnetic pulse simulators,” in this special issue, IEEE Trans. Electro-magn. Compat., Mar. 2013.

[3] C. E. Baum, “Some considerations concerning analytic EMP cri-teria waveforms,” Theoretical Note 285, Oct. 1976. Available:www.ece.edu.unm/summa/notes

[4] Bell Laboratories, EMP Engineering and Design Principles. Whippany,NJ: Electrical Protection Dept., Bell Telephone Laboratories, 1975.

[5] C. E. Baum, “From the electromagnetic pulse to high-power electromag-netics,” Proc. IEEE, vol. 80, no. 6, pp. 789–817, Jun. 1992.

[6] M. Wik, “International electrotechnical commission, IEC-77C,” presentedat the EUROREM 94, Bordeaux, France, Jul. 1994.

[7] K.-D. Leuthauser, “A complete EMP environment generated by high-altitude nuclear bursts: Data and standardization,” Theoretical Note 364,Air Force Phillips Laboratory, Feb. 1994.

[8] VG95371-10 from Bundesamtfur Wehrtechnik und Beschaffung, Ger-many (replaces Edition 1993–2008).

[9] IEC 61000 2-9, “Electromagnetic compatibility (EMC) Part 2: Envi-ronment - Section 9: Description of HEMP environment - Radiateddisturbance. Basic EMC publication,” (1996). [Online] Available: http://webstore.iec.ch

[10] “Electromagnetic environmental effects requirements for systems,” MIL-STD-464A, Dec. 19, 2002.

Page 8: D. V. Giri, Life Fellow, IEEE, and William D. Prather ...IEEE Proof GIRI AND PRATHER: HEMP RISETIME EVOLUTION OF TECHNOLOGY AND STANDARDS EXCLUSIVELY FOR E1 ENVIRONMENT 3 TABLE I PARAMETERS

IEEE

Proo

f

8 IEEE TRANSACTIONS ON ELECTROMAGNETIC COMPATIBILITY

[11] D. V. Giri, W. D. Prather, and C. E. Baum, “The relationship betweenHEMP standards and simulator performance specifications,” Sensor andSimulation Note 538, Univ. New Mexico, 2009.

[12] C. Romero, M. Paolone, M. Rubinstein, F. Rachidi, D. Paonello, andD. V. Giri, “Statistical analysis of the risetime of the lightning currentpulses in negative upward flashes measured at Santis Tower,” in Proc.31st Int. Conf. Lightning Protection, Vienna, Austria, Sep. 2012, pp.1–5.

[13] D. V. Giri and C. E. Baum, “Temporal and spectral radiation on boresightof a reflector type of impulse radiating antenna (IRA),” in Ultra-WidebandShort-Pulse Electromagnetics 3, C. E. Baum, L. Carin, and A. P. Stone,Eds. New York: Plenum Press, 1997.

[14] D. V. Giri, H. Lackner, I. D. Smith, D. W. Morton, C. E. Baum,J. R. Marek, W. D. Prather, and D. W. Schofield, “Design, fabrication andtesting of a paraboloidal reflector antenna and pulser system for impulse-like waveforms,” (Invited Paper) IEEE Trans. Plasma Sci., vol. 25, no. 2,pp. 318–326, Apr. 1997.

[15] D. V. Giri, J. M. Lehr, W. D. Prather, C. E. Baum, and R. J. Torres, “In-termediate and far fields of a reflector antenna energized by a hydrogenspark-gap switched pulser,” IEEE Trans. Plasma Sci., vol. 28, no. 5,pp. 1631–1636, Oct. 2000.

[16] C. E. Baum, W. L. Baker, W. D. Prather, J. M. Lehr, J. P. O’Loughlin,D. V. Giri, I. D. Smith, R. Altes, J. Fockler, D. McLemore, M. D. Abdalla,and M. C. Skipper, “JOLT: A highly directive, very intensive, impulse-like radiator,” (Invited Paper) Proc. IEEE, Special Issue on Pulsed Power:Technol. Appl., vol. 92, no. 7, pp. 1096–1109, Jul. 2004.

[17] M. Nyffeler, A. Jaquier, B. Reusser, P.-F. Bertholet, and A. W. Kaelin,“VERIFY, a threat level NEMP simulator with a 1ns Risetime,” presentedat the AMEREM, Albuquerque, NM, Jul. 2006.

[18] IEC 61000 4-32, “Electromagnetic compatibility (EMC) - Part 4-32:Testing and measurement techniques - High-altitude electromagneticpulse (HEMP) simulator compendium,” (2002). [Online]. Available:http://webstore.iec.ch

[19] I. D. Smith and H. Aslin, “Pulsed power for EMP simulators,” IEEE Trans.Antennas Propag., vol. AP-26, no. 1, pp. 53–59, Jan. 1978.

[20] I. D. Smith and H. Aslin, “Pulsed Power for EMP Simulators,” IEEETrans. Electromagn. Compat., vol. EMC-20, no. 1, pp 53–59, Feb.1978.

[21] J. R. MacDonald, M. A. Schneider, J. B. Ennis, F. W. MacDougall, andX. H. Yang, “High energy density capacitors,” in Proc. IEEE ElectricalInsulation Conf., Montreal, QC, Canada, 2009, pp. 306–309.

[22] M. Rigby, J. Muhle, B. R. Miller, R. G. Prinn, P. B. Krummel, L. P. Steele,P. J. Fraser, P. K. Salameh, C. M. Harth, R. F. Weiss, B. R. Greally,S. O’Doherty, P. G. Simmonds, M. K. Vollmer, S. Reimann, J. Kim,K. R. Kim, H. J. Wang, E. J. Dlugokencky, G. S. Dutton, B. D. Hall, andJ. W. Elkins, “History of atmospheric SF6 from 1973 to 2008,” Atmos.Chem. Phys., vol. 10, pp. 10305–10320, 2010.

[23] B. Daout and F. Vega, “SF6 for high-voltage pulse generators, an eco-logical analysis,” presented at the EUROEM, Toulouse, France, Jul.2012.

[24] J. Blackman, M. Averyt, and Z. Taylor, “SF6 leak rates for high-voltage cir-cuit breakers – U. S. EPA investigates potential greenhouse gas emissionssource,” (2006). [Online]. Available: http://www.epa.gov/electricpower-sf6/documents/leakrates_circuitbreakers.pdf

[25] U.S. EPA, (2002, Jan.). “Byproducts of Sulphur Hexafluoride (SF6) usein the electric power industry,” [Online]. Available: http://www.epa.gov/electricpower-sf6/documents/sf6 byproducts.pdf

[26] V. Carboni, H. Lackner, D. V. Giri, and J. Lehr, “The breakdown fieldsand risetimes of select gasses under conditions of fast charging (∼20 nsand less) and high pressures (20-100 atmospheres),” Switching Note 32,Univ. New Mexico, May 2002.

[27] R. Bischoff, P. Delmote, and S. Pinguet, “HPEM research at InstituteSaint-Louis,” Personal commun., Oct. 28, 2011.

D. V. Giri (LF’12) received the B.Sc. degree fromMysore University, Mysore, India, in 1964, the B.E.and M.E. degrees from Indian Institute of Science,Bangalore 560012, India, in 1967 and 1969, the M.S.and Ph.D. degrees from Harvard University, Cam-bridge, MA, in 1973 and 1975, certificate from Har-vard Introduction to Business Program in 1981.

He has over 35 years of work experience in thegeneral field of electromagnetic theory and its appli-cations in high-altitude electromagnetic pulse, high-power microwaves (HPM), Lightning, and ultrawide-

band (UWB). He is a self-employed Consultant doing business as Pro-Tech, inAlamo, CA, performing R&D work for U.S. Government and Industry. Heis also an Adjunct Professor in the Department of ECE, University of NewMexico, Albuquerque. He was a Research Associate for the National ResearchCouncil at the Air Force Research Laboratory, Kirtland AFB, NM, where heconducted research in EMP and other aspects of electromagnetic theory. Hehas served on the editorial board of the Journal of Electromagnetics, publishedby the Electromagnetics Society. He has also served as an Associate Editor forthe IEEE Transactions on Electromagnetic Compatibility. He has coauthored abook titled High-Power Microwave Systems and Effects published by Taylor andFrancis in 1994. His second book titled High-Power Electromagnetic Radiators:Nonlethal Weapons and Other Applications has been published by Harvard Uni-versity Press in 2004. He is a coeditor of a book titled Ultra-Wideband ShortPulse Electromagnetics 9, published by Springer, 2010. He has also publishedmore than 100 papers, reports, etc. He is an electromagnetic expert responsiblefor the design and optimization of major nuclear electromagnetic pulse (NEMP)simulators in the U.S., Italy, Sweden, Switzerland, Germany, and Israel. He de-signed, analyzed, and successfully built the first reflector type of an impulseradiating antenna, in 1996. Since that pioneering work, he and others have builtmany UWB systems that are finding many applications in military and civiliansectors.

Dr. Giri is a recipient of the IEEE Antennas and Propagation Society’s 2006John Kraus Antenna Award. He is a Charter Member of the ElectromagneticsSociety, and an Associate Member of Commission B, International ScientificRadio Union (URSI), and international Vice Chairman of Commission E, URSI.

William D. Prather (M’70—SM’89) was born inOdessa, Texas, in 1947. He received the B.S.E.E. andM.S.E.E. degrees from the University of New Mex-ico, Albuquerque, in 1970 and 1975, respectively.

He has been with the Air Force Research Labo-ratory (AFRL), Kirtland AFB, NM, since 1970. Hebegan working with Dr. Carl Baum on the design andconstruction of the first electromagnetic pulse (EMP)simulators in the early 1970 s. His special interestsinclude Wideband high-power microwaves (HPM),EMP simulation, interaction and coupling, aircraft

EMP hardening design and the development of CW test methods for measuringsystem shielding. He has also been very active in writing MIL-STDs and MIL-HDBKS for protection of aircraft and Navy ships against EM Hazards.

Mr. Prather has been twice recognized by the Air Force for the developmentof hardness surveillance technology for use in aircraft assembly plants, mainte-nance depots, and operating bases. He and his colleagues have received four BestPaper Awards for applied electromagnetics, a Citation from Air Force Office ofScientific Research for outstanding contributions to basic research in HPM, andthe R. Earl Good award from AFRL in 2003 for the development of widebandtarget ID technology. He is a Technical Advisor to numerous U.S. governmentagencies on EM-related matters and represents the United States Air Force inexchanging electromagnetics research information with NATO countries. Mr.Prather is a Fellow of the EMP Society, a member of the International ScientificRadio Union Commission E and of Eta Kappa Nu.