(descriptions excerpted from [SIMPSONETAL1992A])
 
  Instrument Overview
  ===================
    The KET consists of two separate boxes, SIM3A and SIM3B,
    mounted on the other side of the spacecraft platform with
    respect to the SIM1 and SIM2 units.  SIM3B is the telescope
    portion of the KET, and SIM3A contains the analogue and digital
    electronics for the KET.
 
    The KET is designed to measure electron fluxes between 2.5 and
    6000 MeV, and to determine energy spectra in the range 7 - 170
    MeV.  The telescope also provides measurements of the proton
    and alpha-particle fluxes in several energy windows between 3
    and > 2100 MeV/nucleon.  In addition, two low-energy electron
    and proton channels provide anisotropy information in 8
    sectors.
 
    The determination of electron spectra and their variation with
    distance from the sun and solar latitude provides vital
    information on the solar modulation of interstellar electrons
    as well as on acceleration and propagation of interplanetary
    and solar electrons.  The measurement of protons and helium
    nuclei with the same telescope enhances confidence in the
    calibration and response of the telescope, and permits
    investigation of the dependence on particle species of the
    effects of solar distance and latitude.
 
 
  Detectors
  =========
    To reduce mass and complexity of the instrument, the KET
    consists of two separate parts: the sensor (SIM3B) and the
    electronics box (SIM3A), both mounted beneath the spacecraft
    platform.  The telescope is mounted to view perpendicular to
    the spin axis and has an acceptance angle of 44.6 degrees full
    cone with an auxiliary field of view of 106 degrees.
    Accordingly, for nuclei the geometric factor varies between
    0.72 cm^2 sr for particles reaching D[2] and 6.5 cm^2 sr for
    particles stopping in D[1].  A more sophisticated evaluation of
    the geometric factor taking into account energy dependence and
    effects of scattering and shower production for electrons is
    shown in Table 10.  The instrument incorporates several
    techniques in order to identify the particles and their
    energies: electron- photon cascades, Cherenkov thresholds,
    dE/dx versus E method, and discriminator settings.
    Functionally, the detector system divides naturally into two
    parts, consisting of an entrance telescope and a calorimeter
    surrounded by a guard counter.
 
    The entrance telescope is composed of a silica aerogel
    Cherenkov detector C[1] inserted between two surface-barrier
    semiconductor detectors D[1] and D[2].  Together with the guard
    counter A, it defines the geometry and selects singly-charged
    particles of high velocity ([beta] > 0.938) to discriminate
    between electrons and protons.  In conventional designs the
    velocity discrimination is performed using a high-pressure-gas
    Cherenkov detector.  The availability of a low- density solid
    material like silica aerogel, which can have a refractive index
    as low as 1.02 to 1.2, which is translucent and which shows a
    negligible scintillation contribution to the signal, allows us
    to circumvent the disadvantages of a gas Cherenkov detector,
    which include a significant weight penalty and additional
    material in the acceptance cone.  For use in the KET, we have
    chosen an aerogel with a refractive index of n = 1.066.  Since
    no use is made of the directionality of the Cherenkov light,
    the block of aerogel is placed in a diffusion box with
    millipore walls.  The light signal of C[1] is viewed by the
    phototube PM1 through a hole in the guard counter, A.  To
    prevent particles from hitting the photocathode of PM1
    directly, and thus simulating an aerogel detector response, a
    scintillator disc S[1] has been introduced in front of PM1 as a
    veto counter.  S[1] gives a signal for minimum ionizing
    particles 6 sigma above the signature of relativistic alpha
    particles in aerogel.
 
    The calorimeter consists of a lead-fluoride crystal C[1] in
    which the electron shower develops, and a scintillator cup S[2]
    to detect particles not absorbed in C[2].  C[2] has a thickness
    of 2.2 cm, corresponding to 2.5 radiation lengths.
 
    Lead-fluoride was chosen for its short radiation length (X[0] =
    6.6 g/cm^2), its high density ([rho] = 7.7 g/cm^3), and its
    convenient refractive index (n = 1.885).  Furthermore, it does
    not scintillate and is thus relatively insensitive to the RTG
    background.  The Cherenkov light of C[2] is viewed through a
    hole in S[2] using a diffusion-box design.  To prevent
    particles from escaping undetected through the hole in S[2],
    the planar part of the cup has been extended.
 
    The scintillator A not only helps to define the sensor
    geometry, but also protects from background caused by nuclear
    interactions produced in the telescope by cosmic rays and by
    neutrons from the RTG.  If experience in flight shows that
    shower development in the calorimeter reduces the effective
    geometry factor to an unacceptably low value, the discriminator
    threshold of A can be raised by telecommand for penetrating
    electrons only (channel E[300]).
 
    To compensate for a possible phototube gain loss during the
    long mission lifetime, the high voltages can be stepped up by
    telecommand in eight steps.  In the case of complete failure of
    a detector, it can logically be switched off by telecommand
    using the onboard failure reconfiguration logic.  Stimulation
    of the onboard electronics for checkout purposes is also
    provided.
 
    The data produced by KET consist of pulse height analysis data
    and coincidence counting rates corresponding to broad energy
    ranges for incident particles.  The ranges and time resolutions
    for the counting rates are shown in Table 10.  Pulse height
    analysis is performed on a sample of the incident particle
    events to provide good energy resolution (between 60 and 100 %
    for electrons) and clear identification of particle species.
    Single-detector counting rates and housekeeping data are also
    telemetered to monitor instrument performance.
 
 
  Calibration
  ===========
    The KET instrument measures cosmic ray electrons with energies
    from a few MeV to several GeV, which have a flux in the
    ecliptic plane 3 to 4 orders of magnitude lower than the proton
    flux.  The instrument response was therefore extensively
    calibrated with both electrons and protons.  The most important
    goals of these calibrations were:
 
        - to determine the response of the individual detectors as a
          function of particle species and energy
        - to determine the detection thresholds for the individual
          detectors, especially the Cherenkov thresholds of C[1] and
          C[2]
        - to determine the energy thresholds for the coincidence
          counting rate channels
        - to provide correction tables for the counting rate data
          (response matrix)
        - to determine the electron detection efficiency of
          individual counting rate channels and of the complete
          instrument as a function of energy
        - to determine the proton rejection rate for electron
          channels
        - to use the calibration data to verify and adjust the
          mathematical model of the instrument
        - to determine the background in the various counting rate
          channels due to the RTG
 
 
    Results of proton calibrations
    ------------------------------
      The main goals of the proton calibrations were to determine
      proper settings for i) the Cherenkov thresholds of C[1] and
      C[2], and ii) the discriminator threshold values C[10],
      C[20], and S[20], this last one being actually determined by
      the range of the particles in the telescope.  In addition,
      these calibrations provided measurements of key parameters
      such as angular response and sensitivity of the
      anticoincidence which define the effective geometry factor of
      the instrument.
 
 
    Results of electron calibrations
    --------------------------------
      These calibrations resulted in the measurements of: i) the
      response of the calorimeter C[2] as a function of energy, the
      monoenergetic response of C[2] being of fundamental
      importance to determine the electron spectrum from the flight
      data, and ii) the variation of the efficiency and response of
      the telescope as a function of energy and incidence angle.
 
      It was possible to determine relative variations of the
      response of the detector with energy and incidence angle of
      the beam.  However, it proved to be extremely difficult to
      measure the absolute efficiency for electrons, particularly
      at low energies, where the use of external counters to
      monitor the beam is impossible because of the large
      scattering they induce.  Therefore, only the relative
      efficiency could be reliably measured.
 
 
    Results of the RTG calibration tests
    ------------------------------------
      The RTG calibration of the KET was performed in 1985 at Mound
      Laboratories using the Galileo Flight RTG (F[3]) and in 1986
      and 1990 at KSC during the RTG compatibility tests as part of
      the two launch campaigns.  The main results can be summarized
      as follows:
 
      - RTG induced background in the instrument leads to a
        deadtime increase from 0.003 % to 0.5 %, which is a
        tolerable value. The main contribution to the dead time
        stems from the scintillation guard counter with its large
        field of view.
      - From a scientific point of view, only the channels P[1] and
        E[4] are adversely affected, as the RTG background is of
        the order of the cosmic ray quiet flux in these KET
        particle channels, e.g. 1 x 10^-3 counts/s in E[4]. They
        will only produce meaningful data during solar flares and
        near Jupiter. Use of Cherenkov detectors and multifold
        coincidences guarantees the low susceptibility of all the
        other KET scientific channels.
      - Between the RTG tests in 1986 and 1990 the KET observed an
        increase of the RTG background by about a factor of 2. This
        is attributed to time variations in the plutonium decay
        chain.
      - Flight data measured during quiet times confirm the same
        background level in the RTG-sensitive channels P[1] and
        E[4] as observed during the 1990 RTG-test on ground.
 
 
    Simulation of the telescope
    ---------------------------
      In order to determine the absolute efficiency of KET for
      electrons, it was necessary to use a Monte-Carlo simulation
      of the instrument.  The GEANT program, developed and widely
      used by particle physicists, was adapted for this purpose.
 
      In a first step, the different parameters of the telescope,
      namely thresholds and detector resolutions, were adjusted in
      order to reproduce the calibration data obtained at the
      different accelerators, with a parallel beam in the axis of
      the instrument.  An excellent agreement between the
      simulation and calibration results was achieved.  This
      agreement ensures that we can reliably determine the
      efficient geometry factor of KET for an isotropic flux, which
      was done according to the method given by Sullivan (Sullivan
      J.D., Nucl. Inst.  and Meth.  95, 5, 1971).
 
 
    In-flight calibration
    ---------------------
      On COSPIN switch-on the currents and high voltage values were
      checked and found to be nominal.  The in-flight test
      generator was used to check the amplifier gains and the
      functioning of the coincidence logic and of the failure mode
      reconfiguration.  All of these were found to be nominal.
 
      To measure the performance of the detectors, the linearity of
      the analogue chains, and the discriminator thresholds, it is
      necessary to use the cosmic ray data themselves.  Protons and
      helium nuclei with an energy above 2.1 GeV/n can be used for
      this purpose since they are minimum ionizing particles that
      penetrate the telescope while producing almost no nuclear
      interactions in the detectors, and are above the Cherenkov
      threshold C[10].  Their large flux allows collection in a few
      days of a sample with sufficient statistical accuracy to
      monitor all the detector responses and amplifier gains except
      for C[1] (for which no PHA value is registered).  For P[4000]
      the in-flight responses are identical within a few percent to
      the responses derived from calibrations with a beam of 5
      GeV/c protons.  The position and width of the helium response
      for these detectors are also compatible with the expected
      values derived from the proton response.
 
      The C[1] PHA are registered only in the E[4] and E[12]
      channels, and therefore can be monitored only in these
      channels.  The C[1] response for E[12] particles is
      compatible with the calibration results for electrons of 7.5
      and 10 MeV.
 
 
  Measured Parameters
  ===================
    Table 10.  Electron telescope data channels.
 
    Name        Primary   Energy    Geometric Avg. Time  Sectors  PHA
    Particle  Range     Factor    Resolution
    Type     (MeV(/n))  (cm^2sr)     (s)
 
    K1(P1)      proton    2.7-5.4    6.5         128       --      --
    K21-28(P4)  proton    5.4-23.1   6.5         128        8      --
    K3(P32)     proton    34.1-125   0.72        128       --      P1
    K34(P116)   proton    125-320    1.2         128       --      P1
    K12(P190)   proton    320-2100   1.7         128       --      P0
    K10(P4000)  proton    >2100      1.7         128       --      P0
 
    K2(A4)      He        6.0-20.4   6.5         128       --      --
    K33(A32)    He        34.2-125   0.72        128       --      P1
    K29(A116)   He        125-320    1.0         128       --      P1
    K31(A190)   He        320-2100   1.4         128       --      P0
    K30(A4000)  He        >2100      1.4         128       --      P1
 
    K13-20(E4)  electron  2.5-7      0.26        128        8      P1
    K11(E12)    electron  7-170      0.40        128       --      P3
    K32(E300)   electron  >170       0.38        128       --      P2