1204

1204
COBRA Accelerator for Sandia ICF Diode Research at Cornell University
David L. Smith, Pete Ingwersen, Lawrence F. Bennett, and John D. Boyes
Sandia National Laboratories, Albuquerque, NM
David E. Anderson, John B. Greenly, and Ravi N. Sudan
Cornell University, Ithaca, NY
I. INTRODUCTION
II. R ESEARCH PL ANS
The new COBRA accelerator is being built in phases at
the Laboratory of Plasma Studies in Cornell University
where its applications will include extraction diode and ion
beam research in support of the light ion inertial confinement
fusion (ICF) program at Sandia National Labs. The flexible
4-to 5-MV, 100-to 250-kA accelerator in Fig. 1 is based on a
four-cavity inductive voltage adder (IVA) design.
In
combination with new ferromagnetically-isolated cavities and
Ion diode experi ments in suppor t of the Sandia ICF
progra m will be the first experi mental activi ty on COBRA.
The initia l single -cavity COBRA is well matche d to the
extrac tion geomet ry, applie d-B diode used on the previo us
Cornel l accele rator, LION (1.2 MV, 4 Ohm, 40 ns), since
1992.[ 1] Figure 2 shows a sketch of the LION/C OBRA
diode. We will field this diode on COBRA to contin ue ion
Plasma opening
switch
Water ISC
Assembly
spectroscopy
LION power feed
Oil Coax
Feedline
Water
PFL / OTL
Assembly
Rogowski
monitors
Marx
Generator
ion beam
diagnostics
EMFAPS anode
current feed
Oil Tank
Diode
Experiments
Cavities and
MITL Assembly
Fig. 1. The 1.3 TW COBRA accele rator at Cornel l has a
folded overhe ad pulsed-power geomet ry for compac tness.
self-m agneticall y insula ted transm ission line (MITL)
hardwa re, it includes compon ents from existi ng Sandia and
Cornel l facili ties. Those are the Marx genera tor capaci tors,
hardwa re, and power supply from the DEMON facili ty;
water pulse formin g lines (PFL) and gas switch from the
Subsys tem Test Facili ty (STF);
a HERMES -III
interm ediate store capaci tor (ISC); and a modifi ed ion diode
from Cornel l’s LION. The presen t accele rator consis ts of a
single modifi ed cavity simila r to those of the Sandia
SABRE accele rator and will be used to perfor m the first
phase lower voltag e tests. Four new caviti es will be
fabric ated and delive red in the first half of FY96 to comple te
the COBRA accele rator. COBRA is unique in the sense
that each cavity is driven by a single pulse formin g line,
and the IVA output polari ty may be revers ed by rotati ng the
caviti es 180o about their vertic al axis.
The site
prepar ations, tank constr uction, and diode design and
develo pment are taking place at Cornel l with growin g
enthus iasm as this machin e become s a realit y. Prelim inary
result s with the single cavity and short positi ve inner
cylinder MITL config uration will soon be availa ble.
Anode
Cathode
inductive monitor
Fig. 2. The modifi ed LION diode produc es an ion beam of
10 cm mean radius . This diagra m shows 25-cm axial
extent .
source studie s, partic ularly addres sing the issues of ion
specie s purity and parasi tic load with lithiu m-bearing
evapor ating metal foil anode plasma source (EMFAP S)
active anodes .[2] We will pursue innova tions in foil
fabric ation and in-dio de discha rge cleani ng techni ques begun
on LION.[ 3] Diagno stics will include: magnet icallyinsula ted Farada y cups for beam curren t densit y; Ruther fordscatte ring shadow boxes for ion specie s-resolved beam
diverg ence;
Thomps on
parabo la spectr ometer and
Ruther ford-scatt ering magnet ic spectr ometer for ion specie s
and energy compos ition; collim ated bremss trahlung
detect ors and in-ano de collec tors for diode voltag e and
curren t; and emissi on spectr oscopy and visibl e light streak
photog raphy for in-gap light emissi on.
The substa ntial data base from the perfor mance of this
diode on LION will be compar ed to result s on COBRA to
illumi nate issues of power coupli ng to the diode load on the
new accele rator. In partic ular, the diode will first be
mounte d on COBRA with a very short (75 cm from cavity
gap to diode gap) vacuum MITL. It is expect ed that with
the single -point (azimu thal) power feed to the cavity , power
flow in the MITL will be signif icantly azimut hally
asymme tric. We will diagno se effect s of this asymme try on
diode perfor mance. Our aim is to explor e the tradeo ff
betwee n diode perfor mance degradation by power flow
asymme try for a short MITL and degradation by the delay
between the vacuum-wave prepulse and the main power
pulse at the diode with a long MITL.
After the full four-cavity COBRA is in place, the
Cornell experimental program will make a transition from
the long-standing emphasis on ion diode physics toward a
more integrated development of the diode as part of a beam
generation, transport, and focusing system. We will design a
system using an extraction diode, a gas-filled transport
region, and a solenoidal focusing lens to produce a small
analog to a module of a large ICF driver such as the
Laboratory Microfusion Facility (LMF).[4] Our aim is to
diagnose and develop the accelerator, diode, beam transport,
and lens as integrated, interacting components of the beam
driver system to provide an overview of the issues involved
and to investigate tradeoffs and optimization for LMF.
III. ACCELERATOR DESCRIPTION
The requirement for a 4-to 5-MV pulsed power driver
led naturally to four 1.0-to 1.25-MV cavities that nearly
duplicate the IVA technology presently used in the
HERMES-III and SABRE machines at Sandia.[5] The
cavity-to-cavity inductive isolation, performed by ribbonwound annular cores of type 2605CO METGLAS[6]
ferromagnetic material, and the vacuum MITL allow us the
most compact machine design to maximize the available
experimental area. As shown in Fig. 3, the inner cylinder of
the MITL is tapered at each cavity output feed gap according
to the impedance requirements to best couple to the diode
load. Our choice foa a single overhead water line to charge
need two or more equally spaced feeds to optimize the flow
symmetry.
The basic pulsed-power source for the IVA consists of
one oil-insulated Marx generator, a water-dielectric ISC or
transfer capacitor, a self-breaking multi-stage SF6 gas switch,
and water-dielectric coaxial PFLs with self-closing output
water switches. The tools we used to iteratively design and
model COBRA include the
STF experiments, the
SCREAMER circuit simulation code[7] and electrostatic field
solvers like JASON[8] and ELECTRO[9] along with
dielectric breakdown and flashover criteria like that originated
by J.C. Martin.[10] The following Table I is a summary of
the accelerator design parameters, peak values generated by
the circuit models, and some of the hardware dimensions.
Negative high voltage is assumed for the inner conductors of
the coaxial lines. Note that the subscripts ‘in’ and ‘out’
typically refer to the inner and outer coaxial radii,
Table I. Cobra Accelerator Design Summary
Marx:
No.Caps = 24 ea
V ch = 90 kV
Vrated = 100 kV
Vmarx = 2.2 MV
C/Cap = 1350 nF
E ch
= 131 kJ
Ech/Emax = 81%
IMarx = 111 kA
ISC:
(HERMES-III)
Rout = 71.8 cm
Rin = 53.3 cm
Length = 130 cm
Visc = 2.7 MV
Eout =126kV/cm
Ein =170kV/cm
C isc = 19.5 nF
Zisc = 1.98 Ohm
Tisc = 38.6 ns
Teff = 200 ns
Eo/F = 61%
Ei/F = 36%
Length = 50.6cm
Gap(x18)=16 cm
Vgas = 2.7 MV
Igas = 405 kA
(@900 ns) Ediss = 9.4 kJ
OD = 44.5 cm
No.Channels<10
Lsw = 240 nH
Q gas > 83 mC
Ediss/Eout = 12%
Rout = 17.8 cm
R in = 8.4 cm
Length =76.2 cm
Vpfl = 2.3 MV
Eout = 172 kV/cm
Ein = 365 kV/cm
C pfl = 4.6 nF
Zpfl = 5.0 Ohm
Tpfl = 22.8 ns
Teff = 40 ns
Eo/F = 47%
Ei/F = 44%
Gas Switch:
Oil Filled Cavities
V.I.S.
PFLs(4):
Impedance Matching Stalk
METGLAS
Cores
Vacuum
Region
Fig. 3. The COBRA IVA consists of four radial cavities that
deliver power to the coaxial vacuum MITL.
each cavity was influenced by cost and space limitations, but
we did confirm that the MITL current flow (for negative
polarity operation) was azimuthally symmetric within about
2 ns along the vacuum coaxial line from the cavity output
gap. These tests were performed at the Sandia STF using the
same water lines and cavity that are installed for the initial
COBRA experimental series. Larger diameter cavities may
H2O Switches(4):
Channels/Sw = 4
Lsw = 66 nH
Q wat > 17 mC
Ediss/Eout = 16%
Cavities(4):
OD = 150 cm
Lcav = 20 nH
Wt/Core=51.1kg
Volt-Sec =0.077
Tr(10-90)=23.3ns
FWHM =49.2 ns
Evis = 87 kV/cm
4Eload/Ech = 38%
Gap = 4.2 cm
Vwat = 2.3 MV
Iwat = 248 kA
(@1000 ns) Ediss = 2.4 kJ
ID = 38.1 cm
Length =41.9 cm
Cores/Cav = 4
Vcav = 1.31 MV
(Matched Load) Vload = 1.28 MV
Iload = 256 kA
Pload = 328 GW
(@1050 ns) Eload = 12.6 kJ
and ‘vis’ is the oil/vacuum insulator stack. The E/F ratios
correspond to the expected electric field stress divided by the
calculated breakdown stress.
IV. PREDICTED PERFORMANCE
Our circui t simula tion proces s involv ed a number of
iterat ions as the COBRA accele rator design evolve d and
compon ents were modifi ed or better define d. All the feed
line length s, impeda nce variat ions, and major compon ent
values had to be accura tely repres ented to allow confidence
in the model predic tions. We used transm ission line models
in the SCREAM ER code to accoun t for the proper physic al
separa tions and dimens ions of the oil, water, vacuum , and
plasti c insula ted component s. These models provide a fixed
propag ation delay time and either a consta nt or a linear ly
tapere d line impeda nce. The switch models are typica lly
repres ented by a series combin ation of time-v arying resist or
and approp riate induct or with both elemen ts shunte d by a
parall el stray capaci tance. The switch es are closed by
reduci ng the initia l high resist ance expone ntially to a final
low resist ance.
The expone ntial time consta nts were
determ ined from estima tes of the resist ive and induct ive
phase contributio ns to the switch ing action .[11] The choice
for the final resist ance is critic al for determ ining the energy
dissip ated by the gas and water switch es, which in turn
affect s the forwar d going pulse shape. Figure 4 shows the
result ing voltag e wavefo rms of one circui t simula tion that
corres ponds to the parame ters listed in Table I. This circui t
1.4E+6
VCAV
1.2E+6
VLOAD
1.0E+6
V
O
L
T
S
8.0E+5
6.0E+5
4.0E+5
2.0E+5
0.0E+0
9.00E-7
9.20E-7
9.40E-7
9.60E-7
9.80E-7
1.00E-6
1.02E-6
TIME (SECONDS)
Fig. 4. SCREAM ER genera ted these simula ted voltag e
wavefo rms at the cavity input and output .
model did not include a “crowb ar” switch in the output
water transm ission line (OTL) nor a satura ble magnet ic core
model which could signif icantly affect the pulse shape
depending on the core materi al loss proper ties. Since
accura te ion diode models are still being develo ped for
SCREAM ER, the only load we have modele d is a consta nt
matche d resist ance. The load voltag e wave shape will
defini tely be sensit ive to the impeda nce histor y of dynami c
ion diodes .
V. CONCLUSIONS
With this paper we are announcing a new terawatt class
accelerator intended to further the light ion ICF program with
research and development of diodes, beam transport, and
possibly beam focusing. COBRA is the result of a major
cooperative effort between a university and a national
laboratory (Cornell and Sandia) and, hopefully, may set a
precedent for other similar endeavors.
It represents
technology currently being applied at Sandia and should be a
robust, reliable research tool.
VI. ACKNOWLEDGMENTS
The authors wish to recognize the long term efforts of
Don Cook, Doug Bloomquist, John Maenchen, and Juan
Ramirez of Sandia Labs and Dave Hammer at Cornell
University. Without their support this project would never
have gone beyond a paper exercise. Special thanks are also
due to Dan Jobe and Pat Ryan, Ktech contractors supporting
Sandia, and Leo Brissette in Cornell whose enthusiastic
support included the preparation,shipping, and assembly of
the COBRA hardware at Cornell. This work was supported
by the United States Department of Energy under Contract
DE-AC04-94AL85000.
VII. REFERENCES
[1] J.B. Greenly, et. al., “Extraction Ion Diode Studies
for Optimized Performance: Divergence, Ion Species, and
Parasitic Load,” Proc. 10th Intl. Conf. on High Power
Particle Beams, San Diego, CA, June 20-24, 1994, p. 398.
[2] G.D. Rondeau, Ph.D. Thesis, Cornell University,
1988.
[3] C.K. Struckman and B.R. Kusse, J. Appl. Phys.
Vol. 74, No. 6, p. 3658, 1993.
[4] J.J. Ramirez, et. al., “Design Issues for a Light Ion
Beam LMF Driver,” Fusion Technol., Vol. 15, p. 350,
1989.
[5] J.J. Ramirez, et. al., “The HERMES-III Program,”
Proc. 6th IEEE Pulsed Power Conf., Arlington, VA, June
29-July 1, 1987, p. 294.
[6] METGLAS is Allied Corporation’s registered
trademark for an amorphous alloy of metals.
[7] M.M. Widner and M.L. Kiefer, “SCREAMER - A
Pulsed Power Design Tool, User’s Guide,” Sandia Nat. Labs,
Albuquerque, NM, Apr 25, 1985.
[8] S.J. Sackett, “JASON - A Code for Solving General
Electrostatics Problems, User’s Manual,” CID-17814,
Lawrence Livermore Nat. Labs, Livermore, CA, 1978.
[9] “ELECTRO Users and Technical Manual,” Version
3.9, Integrated Engineering Software Inc., 46-1313 Border
Pl., Winnipeg Manitoba, Canada, R3G 2X7.
[10] J.C. Martin, “Nanosecond Pulse Techniques,”
Internal Report SSWA/JCM/704/49, AWRE, Aldermaston,
Berkshire, England, April 1970.
[11] J.P. VanDevender and T.H. Martin, IEEE Trans.
on Nucl. Sci., Vol. 22, No. 3, p. 979, 1975.
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF

advertisement