Instrument Overview
  ===================
    Both the probe and orbiter were equipped with ultrastable
    oscillators (USOs) for the purpose of generating/measuring a very
    stable telemetry signal from the probe to orbiter.  The orbiter
    measured the probe radio signal frequency every 2/3 seconds and
    stored the frequency data on the tape recorder.  Half-resolution
    data (4/3 second) were also stored in solid state memory.
 
 
  Scientific Objectives
  =====================
    The Doppler Wind Experiment was designed to extract the zonal
    (east-west) motion of the Galileo entry probe during its descent
    into the atmosphere of Jupiter on December 7, 1995.
 
 
  Calibration
  ===========
    During cruise, the probe USO was powered-on and frequency
    stability measurements were made three times.  The in-flight
    tests were the SFT (System Functional Test) and the MST (Mission
    Sequence Test).  Although these tests had different goals and
    performed under different protocol, for the purposes of the DWE,
    the tests were essentially identical.
 
    The measured drift rates of the probe USO were (all fractional
    frequency drifts for 30 minutes following a warmup of 5.5 to 6
    hours)
 
    Project (pre-launch): 4.3e-10
    1989 SFT: 1.09e-9
    1990 SFT: 1.49e-9
    1992 MST: 1.2e-9
 
    From our analysis the best estimate of the fractional frequency
    drift at the time of the probe mission is 1.73e-9 (30 minutes)
    with an uncertainty of +/-.93e-10.
 
    The magnitude of the orbiter USO drift was at least one or two
    orders of magnitude less than the probe USO, so the
    characteristics of the orbiter USO are almost completely
    irrelevant for purposes of the analysis.
 
    High Temperature effects:
 
    Following the probe mission of 7 December 1995 it was discovered
    that the probe interior reached temperatures well in excess of
    the qualification and acceptance level limits, and pre-launch
    calibration testing levels.  Therefore, in late 1996 at NASA
    Ames, high temperature tests of the USO stabilities were
    performed on the flight spare oscillators at temperatures and
    temperature rates experienced by the probe during the Jupiter
    descent mission.
 
    After the post-flight high temperature testing of the flight
    spare USOs was completed, two simple thermal models of the probe
    response to the changing temperatures were made.  These two
    different models 'best fit' the temperature-time curve in
    different ways.  In the first (moderate thermal corrections) the
    temperature-time curve for the USO thermal test was matched to
    the actual measured probe USO data for the period 20 minutes
    after entry to about 50 minutes after entry.  However, for this
    model, the temperature peak was reached too early.  For the
    second model (maximum thermal correction) the time of the
    temperature peak was more consistent with the time of loss of
    probe signal, but was several degrees below measured probe
    temperatures from 20 minutes to 50 minutes after entry.  For this
    model (maximum thermal correction) the temperature peak was
    reached about 2.25 minutes earlier than the moderate thermal
    correction model.
 
    Under the assumption that it must behave like the flight spare
    unit, the high temperature calibrations below were applied to the
    actual probe data.
 
    timee is time after probe entry (seconds) and delf_f is the 30
    minute fractional frequency drift.
 
    Maximum thermal corrections - drift model peaks at 59 min after
    entry.
 
                if(timee(i).gt.2394.)delf_f=-.161e-8
                if(timee(i).gt.2503.)delf_f=-.593e-8
                if(timee(i).gt.2599.)delf_f=-.966e-8
                if(timee(i).gt.2703.)delf_f=-.122e-7
                if(timee(i).gt.2799.)delf_f=-.123e-7
                if(timee(i).gt.2900.)delf_f=-.994e-8
                if(timee(i).gt.2999.)delf_f=-.619e-8
                if(timee(i).gt.3099.)delf_f=-.427e-8
                if(timee(i).gt.3199.)delf_f=-.983e-8
                if(timee(i).gt.3299.)delf_f=-.308e-7
                if(timee(i).gt.3399.)delf_f=-.769e-7
                if(timee(i).gt.3499.)delf_f=-.157e-6
                if(timee(i).gt.3599.)delf_f=-.278e-6
 
    Moderate thermal corrections - drift model peaks at 61.5 min
    after entry.
 
                if(timee(i).gt.2398.)delf_f=-.565e-9
                if(timee(i).gt.2497.)delf_f=-.211e-8
                if(timee(i).gt.2593.)delf_f=-.538e-8
                if(timee(i).gt.2698.)delf_f=-.880e-8
                if(timee(i).gt.2794.)delf_f=-.108e-7
                if(timee(i).gt.2894.)delf_f=-.110e-7
                if(timee(i).gt.2994.)delf_f=-.911e-8
                if(timee(i).gt.3094.)delf_f=-.601e-8
                if(timee(i).gt.3194.)delf_f=-.425e-8
                if(timee(i).gt.3294.)delf_f=-.831e-8
                if(timee(i).gt.3394.)delf_f=-.246e-7
                if(timee(i).gt.3494.)delf_f=-.613e-7
                if(timee(i).gt.3594.)delf_f=-.127e-6
                if(timee(i).gt.3694.)delf_f=-.228e-6
                if(timee(i).gt.3816.)delf_f=-.400e-6
                if(timee(i).gt.3901.)delf_f=-.543e-6
 
    At times earlier than 2390 seconds the USO drift rate is assumed
    to be the nominal 30 minute fractional frequency drift rate of
    delf_f = +1.73e-9.  The total drift in Hz (offset) is then found
    from
 
    ddt=timee(i+1)-timee(i)
    offset=offset+delf_f*ddt*f0/1800.0
 
    Note on 30 minute drift rate: To make meaningful comparisons of
    the probe performance under different operating conditions, it is
    useful to have a common baseline.  All of the probe pre-flight
    tests and calibrations are quantified in terms of 30 minute drift
    rates, and this is the reference definition.  Therefore, at any
    instant of time, we define the 'instantaneous' 30 minute drift
    rate to make references to the probe USO specs and predictions
    easier.  Under severe operating conditions and a rapidly changing
    environment, the 30 minute drift rate may well change minute to
    minute.
 
 
  Operational Considerations
  ==========================
    High temperature effects - see above.
 
    The probe radio signal frequency was measured on board the
    orbiter every 2/3 seconds.  The data were stored in a buffer in
    the Relay Receiver Hardware (RRH) prior to delivery to the
    orbiter.  Due to a slight timing mismatch between the RRH and
    orbiter clocks, occasionally the orbiter would request a
    frequency from the RRH just prior to a frequency measurement.  In
    these cases the orbiter would measure a frequency of zero and
    results in a discontinuity in the frequency-time profile of about
    7 Hz.  To correct for this timing mismatch and the corresponding
    null frequency measurements it is necessary to average through
    the discontinuities.  This process, described in more detail
    below, results in a frequency measurement period that is slightly
    longer than 0.666 seconds.  For example, suppose N frequency
    measurements (f_1,f_2,f_3,...f_N) are made at the N times
    (t_1,t_2,t_3,...  t_N) where t_x - t_(x-1) = 0.666 seconds.  If,
    due to a timing mismatch between the RRH and orbiter, frequency
    f_k is discovered to be zero, then the remaining N-1 frequency
    measurements are assumed to be uniformly distributed in time over
    the time period t_1 to t_N.  In this case the times of the
    remaining N-1 frequencies f_1, f_2, f_3 ...  f_N (with f_k
    removed) are t_1, t_1+deltat, t_1+2*deltat, t_1+3*deltat, ...
    where deltat = (t_N-t_1)/(N-2).
 
    This is equivalent to averaging through the discontinuities.
    Approximately 10 null measurements were experienced during the
    reception of the probe data.
 
    Additionally, there were several short periods when the tape
    recorder data were not received.  Most of the gaps in the data
    were filled in by the half-resolution data from solid state
    memory.  In all, out of a total possible of 5173 frequency
    measurements covering a period of 57 minutes, 28.66 seconds, 5015
    frequency measurements were made.  The total missed data were 159
    points (3.316%).  None of the missed points were consecutive.
 
 
  Electronics
  ===========
    The frequency measured was the frequency of the RRH Numerically
    controlled oscillator (NCO).  The NCO frequency is related to the
    link frequency according to
 
    f_link=57*f_uso + f_nco + 1024
 
    where f_uso is the frequency of the orbiter USO (nominally
    24.325553 MHz), and f_nco is the nco control word (frequency),
    nominally 442455 Hz.
 
    Due to the finite word length in the RRH buffer (24 bits), the
    frequency measurement has an inherent digitization error.  This
    finite word length leads to a digitization error of about .18124
    Hz.