/* FILE_CONTENTS     = CASSE calibration description */

LABEL_REVISION_NOTE  = "2012-08-01, H.-H. Fischer (DLR)"
LABEL_REVISION_NOTE  = "2016-05-12, K. Seidensticker (DLR), version 1"
LABEL_REVISION_NOTE  = "2016-05-23, H.-H. Fischer, M. Knapmeyer (DLR), K. Seidensticker,
                        version 2"
LABEL_REVISION_NOTE  = "2016-05-24, W. Schmidt (FMI), version 3"
LABEL_REVISION_NOTE  = "2016-05-24, H.-H. Fischer, version 4"


      Purpose and Limitations
      =======================
      This document describes the nominal acceleration calibration of the SESAME-CASSE time
      series delivered in the SESAME-CASSE level 3 data set using the level 2 data set. As
      CASSE was designed primarily to measure time differences, when sounding from foot to
      foot, not much effort was made before launch to calibrate the amplitude of the received
      signals. Thus for example, the sensitivity factor S(sensor, axis) of the accelerometers
      as a function of temperature as well as the amplifier gain as a function of frequency are
      poorly known. It is possible that the calibration described below might be replaced by
      an improved version in the future.


      General information
      ====================
      Transfer functions (frequency response and absolute value) are known for the accelerometer
      channels only. Transmitters were rarely used as receivers and only for qualitative
      measurements. A few measurements during the Rosetta cruise phase indicate that an external
      signal in ADC units (Analog-Digital-Converter) recorded by a transmitter is about twice as
      large as the same signal recorded by an accelerometer.

      The manufacturer Bruel & Kjaer determined the sensitivity factor S[mV/(m s**-2)] for each
      accelerometer axis at 159.2 Hz. The frequency response of the sensitivity factor is flat
      (within +-10%) from 1 Hz to 6 kHz (x-axis) and from 1 Hz to 3.5 kHz (y- and z-axes).

      Information about the time calibration and the equivalence of physical channel and
      accelerometer axis, needed to calculate the timing information of level 2 data, can be
      found in the document "SESAME Flight Software FM-3 TC / TM Formats" (Ref. RO-LSE-UG-3404,
      version 5.5a, 2012-08-01) that is part of this data package.


      Nominal Calibration
      ===================
      Several steps are necessary to calculate acceleration from a time series sample value:

       a1) All modes but Stacking Mode
           Invert the non-linear ADC characteristics and convert the sample value adc to
           voltage UADC [mV] by using

           if (adc > 96) UADC = (825/64)(4 * adc - 256);
           else if (adc > 64) UADC = (825/64)(2 * adc - 64);
           else if (adc < -96) UADC = (825/64)(4 * adc + 256);
           else if (adc < -64) UADC = (825/64)(2 * adc + 64);
           else UADC = (825/64) * adc;

       a2) Only Stacking Mode
           Divide the sample value, which is actually the sum of linearized samples from
           several measurements, by the number of measurements nMeas and perform voltage
           conversion

           UADC [mV] = (825/64)* sampleValue / nMeas.

           In a few operations, several of the nMeas measurements failed and erroneously
           delivered the same value for all samples ("flat lines"). This behavior can be
           detected by checking the channel statistics (defined in file CASSE_STAT_FM3.FMT)
           of each measurement of an operation: the minimal sample values equal the maximal
           sample values for all time series of that measurement. The stacked sampleValue
           must then be corrected by subtracting the constant value(s) from sampleValue to
           yield sampleValue#. UADC is then obtained by dividing by the number of
           successful measurements

           UADC [mV] = 825/64* sampleValue# / (nMeas - nFail).

        b) Convert the voltage UADC to the output voltage of sensors by considering the
           adjusted instrument gain.

           Uout[mV] = UADC [mV] / gain.

           The gain is calculated using the value agc of the Amplifier Gain Control (AGC)
           register (object "AGC" in file CASSE_SEQ_PARAM_BURST_L2_FM3.FMT) The status of
           the four bits in agc determine which of the four amplifier stages were active.
           Note that a cleared bit activates the corresponding amplifier. Any combination
           of amplifier stages is possible. The nominal total gain can be calculated
           according to

           gain = 1;
           if ((agc AND 0001b) = 0) gain = gain * 3.13;
           if ((agc AND 0010b) = 0) gain = gain * 2.13;
           if ((agc AND 0100b) = 0) gain = gain * 4.55;
           if ((agc AND 1000b) = 0) gain = gain * 5.55;

        c) Convert the output voltage of accelerometer channels to acceleration. The nominal
           acceleration a is given by

           a[m/s**2] = Uout[mV]/S(sensor, axis).

           The sensitivity factor S in [mV/(m s**-2)] of each axis of the tri-axial
           accelerometers as determined by Bruel & Kjaer is listed below:

           ACC foot -Y (Serial number = 2317546)
             x-axis = 9.97
             y-axis = 9.73
             z-axis = 9.64
           ACC foot +X (Serial number = 2317544)
             x-axis = 9.99
             y-axis = 9.89
             z-axis = 9.90
           ACC foot +Y (Serial number = 2317545)
             x-axis = 9.93
             y-axis = 9.78
             z-axis = 9.95."