Glass Meas QNDE 2001

download Glass Meas QNDE 2001

of 7

Transcript of Glass Meas QNDE 2001

  • 8/17/2019 Glass Meas QNDE 2001

    1/7

    See discussions, stats, and author profiles for this publication at: http://www.researchgate.net/publication/252523238

    Determination of glass thickness using laser-based ultrasound

     ARTICLE · JANUARY 2001

    DOI: 10.1063/1.1373771

    CITATION

    1

    READS

    80

    4 AUTHORS, INCLUDING:

    Frank Shih

    Seattle University

    9 PUBLICATIONS  130 CITATIONS 

    SEE PROFILE

    Bruno Pouet

    Bossa Nova Technologies, USA, Culver City

    53 PUBLICATIONS  264 CITATIONS 

    SEE PROFILE

    Marvin Klein

    Intelligent Optical Systems, Inc.

    158 PUBLICATIONS  2,112 CITATIONS 

    SEE PROFILE

    Available from: Marvin Klein

    Retrieved on: 28 September 2015

    http://www.researchgate.net/profile/Marvin_Klein?enrichId=rgreq-282bb74f-e7bf-43b4-8a5c-39417f478617&enrichSource=Y292ZXJQYWdlOzI1MjUyMzIzODtBUzoxMDM0Mjc3MjgyMTYwNzVAMTQwMTY3MDUwMDk2OA%3D%3D&el=1_x_7http://www.researchgate.net/profile/Frank_Shih?enrichId=rgreq-282bb74f-e7bf-43b4-8a5c-39417f478617&enrichSource=Y292ZXJQYWdlOzI1MjUyMzIzODtBUzoxMDM0Mjc3MjgyMTYwNzVAMTQwMTY3MDUwMDk2OA%3D%3D&el=1_x_4http://www.researchgate.net/profile/Frank_Shih?enrichId=rgreq-282bb74f-e7bf-43b4-8a5c-39417f478617&enrichSource=Y292ZXJQYWdlOzI1MjUyMzIzODtBUzoxMDM0Mjc3MjgyMTYwNzVAMTQwMTY3MDUwMDk2OA%3D%3D&el=1_x_5http://www.researchgate.net/profile/Frank_Shih?enrichId=rgreq-282bb74f-e7bf-43b4-8a5c-39417f478617&enrichSource=Y292ZXJQYWdlOzI1MjUyMzIzODtBUzoxMDM0Mjc3MjgyMTYwNzVAMTQwMTY3MDUwMDk2OA%3D%3D&el=1_x_5http://www.researchgate.net/?enrichId=rgreq-282bb74f-e7bf-43b4-8a5c-39417f478617&enrichSource=Y292ZXJQYWdlOzI1MjUyMzIzODtBUzoxMDM0Mjc3MjgyMTYwNzVAMTQwMTY3MDUwMDk2OA%3D%3D&el=1_x_1http://www.researchgate.net/profile/Marvin_Klein?enrichId=rgreq-282bb74f-e7bf-43b4-8a5c-39417f478617&enrichSource=Y292ZXJQYWdlOzI1MjUyMzIzODtBUzoxMDM0Mjc3MjgyMTYwNzVAMTQwMTY3MDUwMDk2OA%3D%3D&el=1_x_7http://www.researchgate.net/institution/Intelligent_Optical_Systems_Inc?enrichId=rgreq-282bb74f-e7bf-43b4-8a5c-39417f478617&enrichSource=Y292ZXJQYWdlOzI1MjUyMzIzODtBUzoxMDM0Mjc3MjgyMTYwNzVAMTQwMTY3MDUwMDk2OA%3D%3D&el=1_x_6http://www.researchgate.net/profile/Marvin_Klein?enrichId=rgreq-282bb74f-e7bf-43b4-8a5c-39417f478617&enrichSource=Y292ZXJQYWdlOzI1MjUyMzIzODtBUzoxMDM0Mjc3MjgyMTYwNzVAMTQwMTY3MDUwMDk2OA%3D%3D&el=1_x_5http://www.researchgate.net/profile/Marvin_Klein?enrichId=rgreq-282bb74f-e7bf-43b4-8a5c-39417f478617&enrichSource=Y292ZXJQYWdlOzI1MjUyMzIzODtBUzoxMDM0Mjc3MjgyMTYwNzVAMTQwMTY3MDUwMDk2OA%3D%3D&el=1_x_4http://www.researchgate.net/profile/Bruno_Pouet?enrichId=rgreq-282bb74f-e7bf-43b4-8a5c-39417f478617&enrichSource=Y292ZXJQYWdlOzI1MjUyMzIzODtBUzoxMDM0Mjc3MjgyMTYwNzVAMTQwMTY3MDUwMDk2OA%3D%3D&el=1_x_7http://www.researchgate.net/profile/Bruno_Pouet?enrichId=rgreq-282bb74f-e7bf-43b4-8a5c-39417f478617&enrichSource=Y292ZXJQYWdlOzI1MjUyMzIzODtBUzoxMDM0Mjc3MjgyMTYwNzVAMTQwMTY3MDUwMDk2OA%3D%3D&el=1_x_5http://www.researchgate.net/profile/Bruno_Pouet?enrichId=rgreq-282bb74f-e7bf-43b4-8a5c-39417f478617&enrichSource=Y292ZXJQYWdlOzI1MjUyMzIzODtBUzoxMDM0Mjc3MjgyMTYwNzVAMTQwMTY3MDUwMDk2OA%3D%3D&el=1_x_4http://www.researchgate.net/profile/Frank_Shih?enrichId=rgreq-282bb74f-e7bf-43b4-8a5c-39417f478617&enrichSource=Y292ZXJQYWdlOzI1MjUyMzIzODtBUzoxMDM0Mjc3MjgyMTYwNzVAMTQwMTY3MDUwMDk2OA%3D%3D&el=1_x_7http://www.researchgate.net/institution/Seattle_University?enrichId=rgreq-282bb74f-e7bf-43b4-8a5c-39417f478617&enrichSource=Y292ZXJQYWdlOzI1MjUyMzIzODtBUzoxMDM0Mjc3MjgyMTYwNzVAMTQwMTY3MDUwMDk2OA%3D%3D&el=1_x_6http://www.researchgate.net/profile/Frank_Shih?enrichId=rgreq-282bb74f-e7bf-43b4-8a5c-39417f478617&enrichSource=Y292ZXJQYWdlOzI1MjUyMzIzODtBUzoxMDM0Mjc3MjgyMTYwNzVAMTQwMTY3MDUwMDk2OA%3D%3D&el=1_x_5http://www.researchgate.net/profile/Frank_Shih?enrichId=rgreq-282bb74f-e7bf-43b4-8a5c-39417f478617&enrichSource=Y292ZXJQYWdlOzI1MjUyMzIzODtBUzoxMDM0Mjc3MjgyMTYwNzVAMTQwMTY3MDUwMDk2OA%3D%3D&el=1_x_4http://www.researchgate.net/?enrichId=rgreq-282bb74f-e7bf-43b4-8a5c-39417f478617&enrichSource=Y292ZXJQYWdlOzI1MjUyMzIzODtBUzoxMDM0Mjc3MjgyMTYwNzVAMTQwMTY3MDUwMDk2OA%3D%3D&el=1_x_1http://www.researchgate.net/publication/252523238_Determination_of_glass_thickness_using_laser-based_ultrasound?enrichId=rgreq-282bb74f-e7bf-43b4-8a5c-39417f478617&enrichSource=Y292ZXJQYWdlOzI1MjUyMzIzODtBUzoxMDM0Mjc3MjgyMTYwNzVAMTQwMTY3MDUwMDk2OA%3D%3D&el=1_x_3http://www.researchgate.net/publication/252523238_Determination_of_glass_thickness_using_laser-based_ultrasound?enrichId=rgreq-282bb74f-e7bf-43b4-8a5c-39417f478617&enrichSource=Y292ZXJQYWdlOzI1MjUyMzIzODtBUzoxMDM0Mjc3MjgyMTYwNzVAMTQwMTY3MDUwMDk2OA%3D%3D&el=1_x_2

  • 8/17/2019 Glass Meas QNDE 2001

    2/7

    DETERMINATION  OF  GLASS THICKNESS USING LASER BASED

    ULTRASOUND

    Frank J. Shih

    1

    , Bruno F. Pouet

    1

    , Marvin B. Klein

    1

    , Andrew D. W. McKie

    2

    ^asson Technologies, 6059 Bristol Parkway, Culver City, CA 90230, USA

    2

    Rockwell Science Center, 1049 Camino Do s Rios, Thousand Oaks, C A 91360, U SA

    Abstract.  Thickness  measurements

      of

     glass plates

      an d

     glass bottles using laser-based ultrasound

     LBU) a re described. Ultrasound in the glass specimens w as generated thermoelastically with either

    a pulsed CO

    2

     laser, or a Q-switched Nd:YAG laser in the

     case

     o f colored glass filters.  Th e detection

    of

      ultrasound

      wa s

      accomplished

      by one of the  following

      methods;

      a

      spherical Fabry-Perot

    interferometer   system  or a  photo-refractive  interferometer  based  on  two-wave  mixing.  A  self-

    interference  effect,  utilizing

      the

     partial reflection from

      the

      front

      an d

     back  faces

      of a

     glass plate

      wa s

    also demonstrated  to have  sufficient sensitivity under certain conditions.  Th e  thickness  of the glass

    plates  and  colored glass bottles  wa s  determined using  the

      fundamental

      reverberation

      frequency

    obtained

      from

      the

      time-domain

      waveform

      data.

      LB U

     results were compared

      to

     physical thickness

    measurements

      and

     showed excellent agreement.

    INTRODUCTION

    To  avoid breakage during transport,  the walls  of  glass containers, such  as bottles

    and

      beakers, must typically meet

      a

      minimum thickness tolerance. Therefore,

      it is of

    interest

      to

     measure

      the

      container wall thickness, especially during

     the

     production process,

    to  ensure that specifications  are  met.  To  this end,  we  explored  the possibility  of  using

    laser-based ultrasound LBU) techniques

      [1] for

     measuring

      the

      wall thickness

      of

      glass

    specimens

      of

      varying degrees

      of

      curvature

      and

      color.

      An

      in-line laser-based thickness

    measurement technique using laser-optical triangulation is  currently available [2].  In this

    technique,  a laser beam  is passed through the glass surface  at a known angle;  a portion  of

    the

     beam

     is

     reflected

     off the

     front

      surface,

     while

     a

     portion

     of the

     same beam

     is reflected  off

    the

     back surface

      after

     transmission through

      the

     glass.

      The

      spatial separation between

      the

    tw o reflections

     is

     then used

     to

     calculate

     the

     plate thickness. However, this technique

     is not

    well-suited  for glass containers since it works poorly with curved or non-parallel

      surfaces.

    Given that

     the

     second beam

     has to reflect  of f the

     back

     surface  after

     going through

     the

     plate,

    transparency

     of the

     glass

     can

     also

     be an

     issue.

      In

     this situation

     a

     tunable probe laser would

    be desirable to handle various colored glasses, which introduces complexity and cost.

    EXPERIMENTAL

    Our

      approach

      is to use

      laser-based ultrasound

      to

      measure glass wall thickness.

    Ultrasound

     in the

     glass specimens

     was

     generated thermoelastically with

     a

     pulsed

     CO

    2

     laser,

    or a  Q-switched Nd:YAG laser  in the  case  of a  colored glass

      filter.

      The  detection  of

    ultrasound

     is

     performed

     by one of the following

     methods:

      a

     spherical Fabry-Perot

    CP557,

     Review  o f  Progress

     in  Qua ntitative  Nond estructive

     Evaluat ion   Vo l .

      20 ed. by D. O. Thompson and D. E . Chimenti

    ©2001

     American

     Institute

     of Physics

      l-56396-988-2/01/ 18.00

    28 7

  • 8/17/2019 Glass Meas QNDE 2001

    3/7

    Wii

     

    FIGURE  I Glass specimens with varying degrees

     of curvature.

    00   §

    FIGURE

     2.

      Experimental setup

     of the spherical

     Fabry-Perot

      interferometer.

    interferometer  detection system  coupled with  an  argon-ion  laser, a  photorefractive

    interferometer  based

      on

      two-wave mixing,

      and a

      self-interference  technique

      that can be

    used

      for

      glass plates

      with

      parallel  faces.

      The bulk wave velocity in

      glass,

      for use as a

    reference,

      was

      established

      using a

      conventional

      immersion pulse-echo

      system.

      The glass

    wall thickness

     of various container specimens was

     then

     measured both

     by LBU

      technique

    and by a

     dial-caliper.

    Th e types o f glass

     specimens used

     are shown in Figure  1.  First, a flat

      window

     glass

    is tested.  The

      front

      and back faces of the glass plate are

     nearly

      parallel. This is the type of

    specimen that

     can be handled by the current

     triangulation

      technique. To

     assess

     th e

     merit

     of

    laser-based

      ultrasound  on

      glass with

      curved

      surfaces,

      we

     tested

      a

      1000-mL beaker,

      whose

    diameter is  about  4  inches.  Measurement was then made on several colored glass beer

    bottles.  Th e measurements were made at the lower portion of the

     bottle,

     where the shape is

    cylindrical  and the

     diameter

      of

     this

     region was  about  2 inches.

      Attempts

     were  even made

    on a classic Coca-Cola bottle,

     where

     th e front  and

     back surface

     at any given point is highly

    irregular in shape.

    Th e spherical Fabry-Perot setup

     that

     w as

     used

     for

     each

     specimen is shown in Figure

    2. The  excitation laser  was a pulsed CO2

     laser (10.6  jum).

      Th e

      COi  laser

      is a suitable

    choice

      since

      the

      glass

      is opaque at the operating

      wavelength

      and

      results

      in

      improved

    ultrasonic generation  efficiency  compared with other available  generation  lasers.  Th e

    energy of the excitation beam was ~5

      mJ/pulse, resulting

      in a thermoelastic excitation.

    Thus,

      upon

      inspection,  no damage was observed on the

      glass

      specimens

      after

      LB U

    measurement.  The probe

     beam,

     an

     argon-ion

     laser, is

     then

     directed to the

      same spot

     as the

    excitation

      laser

      beam.

      Th e  reflected

      light

      from  the  specimen  surface  is  collected,

    transmitted

     through the

      spherical

     Fabry-Perot cavity, and

     measured with

     a

     photodetector.

    28 8

  • 8/17/2019 Glass Meas QNDE 2001

    4/7

    U- ^  \

    FIGURE

     3.

      Experimental setup

     of the

      photo-refractive

      interferometer

     based

     on

     two-wave mixing.

    Another

     technique w as

     also explored

     to look at the same

     problem.

      In

     order

     to use a

    Q-switched

     Nd:YAG

     generation

     laser that was available at

     Lasson,

     a KG3

     color glass filter

    was

      used

      as a  specimen.  The detection

      system, shown

      in Figure 3, is a  photorefractive

    interferometer based on two-wave mixing.  A probe laser  (CW diode-pumped,

      frequency-

    doubled

      Nd:YAG  at 532  nm),  is  reflected

      back

      from  the  specimen  as the  signal beam,

    carrying

      information about the

      surface  motion, while

      a reference beam is

      reflected

      off a

    mirror. The

     signal

     beam and the

     reference beam then interfere

      in a

     photorefractive

      crystal,

    producing  a

      real-time

     hologram.  Th e  signal  beam and the  diffracted  reference  beam  are

    then combined

     a t a

     photodiode used

     for

     coherent

     detection.

    RESULTS

      ND

     DISCUSSION

    The

     time-domain

     waveform  for one of the parallel glass plates is shown in

     Figure

     4 .

    Th e

      measurement

      was done by spherical

      Fabry-Perot

      interferometer  system. The

      laser-

    induced ultrasonic

      wave  reflects

      back and

      forth

      between the two  faces,

      building

      up a

    reverberation,  which  causes  both  faces  of the  glass

      plate

      to  move  in and out at a

    characteristic

      frequency.  This frequency/is

      inversely

      proportional

     to the thickness of the

    plate. Specifically, we have/=v/2L,

     where

     v is the P-wave velocity and L   is the thickness

    of the  plate. From

      Figure

      4, the periodicity of the

      time-domain

      waveform  is  plainly

    visible.

      Th e

      resonance

      is best  analyzed  in the  frequency

      domain. Using

      the

      P-wave

    velocity  of

      5.81

      mm/jas

      obtained

      from  an  immersion pulse-echo measurement,  the

    thickness

     of the

     plate

     is

     determined

     to

     be2.2

     mm. The experiment is then

     repeated

     for the

    FIGURE 4 Time-domain

     waveform

     of the flat parallel

     glass

     plates.

    28 9

  • 8/17/2019 Glass Meas QNDE 2001

    5/7

    15 2 25 3 35 4

    Time   ( M S )

      Frequency MHz)

    FIGURE  5 .

      Time-domain

     waveform and the corresponding  frequency spectrum of the 1000-mL beaker.

    1000-mL beaker. Both

     the

     time-domain

     waveform  and the corresponding frequency

    spectrum for the

     beaker

     are plotted in Figure 5. It can be observed that in glass with curved

    surfaces,

      the signal

     strength

     in the time-dom ain decays rapidly.

      However,

     the

     fundamental

    frequency, determined by the

     glass

     wall thickness, is still

     distinctly

     isolated in the

    frequency  spectrum.

    Th e

      next  specimens  were

      glass

      beer bottles,

      which

      were

      the  main

      subject

      of

    interest.

      These  mass-produced  glass

      bottles

      are

      formed

      by  blowing molten

      glasses  into

    molds. While

     the outer dimension of these

     glass bottles

     are determined by the shape of the

    molds,  and are relatively  well controlled, the inner  dimensions  of the

      glass

      bottle are  less

    uniform.

      Th e  specimen

     used

      in the

     next

      series of experiments is a

      Heineken beer

      bottle,

    which  has a

     green color.

      Th e

      same

      experiments

     were also

     p erformed  on

     clear

      and brown

    bottles

      and

      yielded

      similar  results. Two locations on the opposite

      sides

      of the

      Heineken

    bottle were  marked  for  thickness

      measurements.

      Th e

      time-domain

      waveform  and the

    corresponding

     frequency

      spectrum

     for

     locations

     one are

     p lotted

     in Figure 6 .

    Th e glass

     beer bottle

     was then

     sectioned  off just

      above the marked

      location,

      and a

    physical

      thickness

      measurement was made

      using

      a dial caliper. Using the  reference  P-

    wave velocity

     of 5.81

     mm/jas described earlier,

     we calculate the

     wall thickness using LBU,

    and comp ared the two sets of measurements.  The results are tabulated in Table 1. It can be

    seen that

     the two sets of

     numbers

     are in excellent agreement.

    Q3 5

    Q3

    025

    02

    Q15

    Q1

    Q0 5

    0

    0

    10

    0 5 10

    Time

     (micro-second)

    FIGURE 6 Time-domain

     waveform

     and the corresponding frequency  spectrum of the glass beer bottle.

    TABLE 1 Summ ary of the LBU and physical measurement results of the glass beer bottle.

    Glass

     bottle

     wall

      thickness

    LB U

    1.8 mm

     (1.37

     MHz)

    2.1 mm (1.61

     MHz)

    Physical Measurement

    1.8 mm

    2.0 mm

    290

  • 8/17/2019 Glass Meas QNDE 2001

    6/7

    FIGURE 7 .

      Time-domain waveform

     and the corresponding

      frequency

      spectrum of the Coca-Cola bottle.

    The   same  approach  was  then  applied to a  classic  Coca-Cola bottle  to test the

    limitations

      of this  technique. The

      interior

      and

      exterior

      surfaces  of

      these classic

     beverage

    bottles

      are highly  contoured and irregular in shape. Both the

     time

     domain  waveform  and

    its   corresponding

      frequency

      spectrum, are given in

     Figure

      7.

      Note

      that the

      signal strength

    is

      extremely

      small,

      on the order of 0.2 mV. The

      measurement

      was

      achieved

      only

      after

    signal averaging a nd

     some careful

      alignment.

    Another

     set of LBU

     measurements

     was made with a photorefractive interferometer

    that was

     based

      on

      two-wave

     mixing. The

      experimental

      setup  is shown in

     Figure

      3. For

    this

     set of

     experiments,

     a 2 mm

     thick

     KG3 color glass  filter was u sed as the specimen.

      This

    filter  has very  high

      absorption

      at the

      1064

      nm

      wavelength

      of the  Q-switched

      Nd:YAG

    generation

      laser.

      The time-domain

      waveform

      and its

      corresponding  frequency

      spectrum

    are plotted in Figure  8. The

     fundamental frequency

     of the  reverberation  is 1.25 MHz.  The

    fundamental  peak

      is

      clearly

     distinguishable and can be used to

      determine

      the  thickness  of

    the glass plate.

    Much

     to our

     surprise,

     when the reference

      beam

     was removed, there was still a very

    strong  signal.

      Th e

      experimental  setup, time-domain

      waveform  and the

      corresponding

    frequency  spectrum

     are

     shown

     in

     Figures

     9 and 10. The

     frequency  spectrum

     gives the

     same

    fundamental frequency  as in

     Figure

      8.  Therefore,  the

      thickness

      can also be detected

     with

    jus t a photodetector.

    Frequency Spectrum

    Fundamental  frequency

    gives

     thi kness

    4 6 8 TO

    Frequency   MHz)

    FIGURE

     8

    Time-domain waveform

     and the

     corresponding frequency  spectrum

     of the

     2-mm

     thick KG3,

    obtained using a two-wave mixing scheme.

    291

  • 8/17/2019 Glass Meas QNDE 2001

    7/7

    Probe

     Laser

    Speckled

    Signal Beam

    Detector

    Color Glass

      Filter

    FIGURE 9 Experimental setup in which the

     reference

     beam was removed from the two-wave mixing setup.

     

    0 . 0 -

    a

    > 0 . 2 ~

    -0.4-

    40

    30

    20-

    10

    o r̂—

    -

    ——

    — — —

     

    — — —

     — — —  

    r i mu3? u  ofjf̂ viy u

    i

     1 1

    ^

    i l «

    '*

    i<

    *

    M f m S

    °^^

      „

      ^ \^

    0 2 4 6 8 10

    0 5 10 IS

    Time

     0/S)

    rf€Mt|piftey

      M H z )

    FIGURE 1 0 Time-domain waveform and the corresponding  frequency  spectrum of the  2 - m m   thick  K G 3 ,

    obtained using just a photodetector.

    Th e  same

     procedure

     w as

      repeated

      on a

     piece

     of

     window glass.

      A

      pulsed

      C C > 2

      laser

    was used as the excitation laser, while an argon-ion laser was used as the probe laser.  Th e

    result

      was

     similar.

      The  frequency-doubled  CW Nd:YAG

      probe laser  5 3 2

      nm

    actually

    transmits fairly well in KG3 glass as

     does

     the argon-ion

     laser

     in plain

     window glass.

      Since

    the probe

     beam

     transmits  in the glass

     itself,

      there is a

      self-interference  effect,

      in which the

    glass acts as its own low-finesse etalon.

    Specifically, the excitation laser

     generates

     an ultrasonic resonance in the glass plate,

    in

     which

      the thickness  changes periodically. The probe laser partially  reflects  back and

    forth

      in the

     glass, causing  multiple-beam  interference.

      The

     reflected  wave

      is

     then picked

    up by a photodetector,

     resulting

      in a plot like the one shown in Figure 10. The advantag e

    of having  such

      a

      system

      is its

      low-cost

      and simplicity. We also

      attempted

      to use

      this

    technique  on  curved

      surfaces,

      but it was

      less

      successful.  For best results

      with this  self-

    interference  method, the two faces of the glass

     have

     to be reasonably parallel.

    CONCLUSION

    W e have demonstrated the capability of accurately measuring the

     thickness

     of

     glass

    of  varying degree  of

      curvature

      using  L B U .  Th e

      described

      self-interfering  etalon  effect

    works

     well  in

      transparent

     materials with nearly

     parallel

      surfaces, but is

     more

     difficult  to

    implement  in

      curved  surfaces.

      In-line determination of glass bottle

      thickness

      using LBU

    appears

     feasible.

    R F R N S

    1.

      Scruby,

      C. B.,

      Drain,

      L. E.,

      Laser-Ultrasonics: Techniques   and

      Applications

    Adam Hilger, Bristol,

    2.

      McCullough,

     R. W., Bondurant, P. D., Doyle, J. L.,

     Ma terials   Eva lua t i on

      53, 1338-1345 (1995).

    3.  McKie,  A. D. W.,  Addison,  R. C., in Rev i ew   o f   P rogress

      in

      Quant i ta t ive Nondestruc t ive

      Evaluations

    edited by D. O. Thompson, Plenum

     Press,

     New York,  1 9 9 5 ,  Vol 14, pp.

      523-528

    4.

      McKie,

     A. D. W.,

     Addison,

     R. C.,  Ultrasonics 32, 333 (1994).

    292