16 th CAA Cross-Calibration Workshop IRAP, Toulouse, 6-9 November20121 Removing strong solar array disturbances and telemetry errors from DC magnetic field measurements with a dual fluxgate technique Harri Laakso and Tomasz Klos ESTEC, Noordwijk

16 th CAA Cross-Calibration Workshop IRAP, Toulouse, 6-9 November20122 TC-1 Magnetic Field Data  Two critical issues that decrease the quality of the DC magnetic field data:  spikes in full-resolution data due to telemetry errors  magnetic interferences caused by incorrect wiring of the spacecraft power system (solar arrays and the shunting system) Inboard sensor Outboard sensor

16 th CAA Cross-Calibration Workshop IRAP, Toulouse, 6-9 November20123 Cleaning Procedure Overview  Spike recognition and removal  Identify the sudden changes of the magnetic interference: they are caused by changes in the spacecraft power system (shunting) due to the variations in the operations of the spacecraft sub-systems, including payload  Cleaning procedure of the interferences contains the following steps:  1 st step (Bz only): Interference is approximated by a dipole at the center of the spacecraft and aligned with the spin axis  2 nd step: Empirical model for the variation of the interference within a spin; model is developed for each component, for each shunting level and for each FGM range  3 rd step: Since the current of the shunting system can vary continuously in some conditions, the final cleaning is done with FFT where two larges components [spin modulation +2 nd harmonics] are removed; this is applied to each component

16 th CAA Cross-Calibration Workshop IRAP, Toulouse, 6-9 November20124 Removing Spikes  Spikes in IB and OB can occur at the same time or not, they must be handled individually  First, large spikes are removed individually from each sensor  Second, steps in difference between IB-OB are studied and smaller spikes are removed  Removed IB data points are replaced with interpolated values while the removed OB points are not because final data products are based on OB while IB is only used for determining the spacecraft effects Inboard sensor Outboard sensor

16 th CAA Cross-Calibration Workshop IRAP, Toulouse, 6-9 November20125 Step-like Changes in Spacecraft power  Two kind of step-like changes appear in the OB/IB data that need to be identified  SAD step-like changes  FGM Range Changes  Sudden SAD changes  Changes in the spacecraft power handled with shunting system  Best detected in Bz  Appears as a change of the difference between the spin-averaged IB and OB  SAD changes are detected only for ranges 3-5 (in range 6-7 digitalization too large, range 6 from ~2000 nT with ~1 nT digitalization)  There can be up to 30 SAD steps per day Spin-axis component Spin-plane component

16 th CAA Cross-Calibration Workshop IRAP, Toulouse, 6-9 November20126 Detecting SAD Steps in Bz  IB: full-resolution in green, spin-averaged in blue  OB:full-resolution in red, spin-averaged in purple  Clear change in the difference between the two spin-averages -> indicates the SAD step

16 th CAA Cross-Calibration Workshop IRAP, Toulouse, 6-9 November20127 SAD Steps in Spin-Plane Components  IB: full-resolution in green, spin-averaged in blue  OB:full-resolution in red, spin-averaged in purple  The difference between spin averages remains constant and nearly the same  There is a clear change in spin shape (Bx shown) at the SAD step  Some SAD changes too small in Bz but they are detected in the change of the Bx envelope SAD Step

16 th CAA Cross-Calibration Workshop IRAP, Toulouse, 6-9 November st Step - SAD Removal from Bz  Spin-axis component:  the net effect of SAD is non-zero (due to azimuthal current system of the SA and additional shunting system)  Changes of the SAD (due to shunting system) are well visible in spin-resolution data, and their effect in full-resolution (and spin-resolution) data disappear using the empirical function B corr = B ob – 1.3 * (B ib – B ob )  This equation is based on a dipolar approximation of the source at the center of the spacecraft and the distance of the sensors at the subsolar point -> factor 1.3  Spin-plane components:  the net effect of SAD is zero in full-resolution data  SAD steps are not visible in spin averages (but can be seen in amplitude of full-resolution disturbance)  Only outboard sensor data are used to provide cleaned spin-plane components

16 th CAA Cross-Calibration Workshop IRAP, Toulouse, 6-9 November20129 Removal of SAD in Bz  With a simple dipole model, the step-like SAD changes disappear in Bz  Remaining signal (blue) contains the real field and a high-resolution SAD component; the remaining interference is a few nT

16 th CAA Cross-Calibration Workshop IRAP, Toulouse, 6-9 November  As in previous slide, but for a longer interval Removal of SAD in Bz

16 th CAA Cross-Calibration Workshop IRAP, Toulouse, 6-9 November Spin Variation of SA Interference  Spin-modulation model is developed and applied to each component:  Empirical SAD model = variation of solar array interference during spin  Model validity: interval between two changes, either due to solar power steps or range changes  Creation of model: Each spin is divided into 100 parts in order to compare data at the same spin phase (3.6º resolution in phase)  Data from each spin is interpolated to these 100 points  Median value at each spin phase point is used for the generation of the model  Actual measurements are not used here but the difference of the full resolution value and the spin-average value  Usage of model:  The value of the SA interference is calculated by interpolating the model values at the time of OB data and at the exact spin phase angle  The found interference is removed from the measurements

16 th CAA Cross-Calibration Workshop IRAP, Toulouse, 6-9 November Finding the Spin Modulation of Interference  red – data:full-resolution data (after removing dipole term, for Bz)  blue – trend:sliding spin-averaged value in each point  difference = data – trend: this is used for finding the SAD model  Note: difference contains both interference and real high-resolution variations

16 th CAA Cross-Calibration Workshop IRAP, Toulouse, 6-9 November Interference from Power System  Difference between full-resolution and spin-averaged values  difference contains both interference and real high- resolution variations  Bx  Interference envelope changes when shunting level changes  A new model is needed between any two changes  Bz  Interference is quite constant because the major component was removed with a dipole term  However, a new model is developed between any two changes Spin-axis component Spin-plane component

16 th CAA Cross-Calibration Workshop IRAP, Toulouse, 6-9 November Non-steady Disturbance  Spacecraft power system currents can sometimes vary continuously, e.g. during charging of the batteries  After removal of a model, a spin modulation up to 1 nT can appear  quality flag indicates this  This is corrected by an FFT process where the largest two frequencies are removed from data: spin period and half spin period  Quality flag indicates when this is done  Resulting components show noise whose amplitude is less than 0.2 nT

16 th CAA Cross-Calibration Workshop IRAP, Toulouse, 6-9 November Final Cleaned Magnetic Field Data  red – full-resolution de-spiked OB data  black – corrected OB data

16 th CAA Cross-Calibration Workshop IRAP, Toulouse, 6-9 November Summary  Double Star spacecraft have highly corrupted magnetic field vector measurements  wrong wiring of the solar power system (panels & shunting)  frequent telemetry errors causing random spikes in the data  With help of dual fluxgate measurements and signal processing:  First, data have been cleaned from the spikes  Then, solar power interferences are removed resulting in an accuracy of better than 0.2 nT for each component for most of the time  Basic steps of the cleaning the magnetic interferences:  First, dipole approximation for the spin-axis component  Second, empirical model for the spin variation of the interference for each component: this is applied to any given FGM range value and the given level of the shunting current  Third, when necessary, FFT-method is finally applied when strong spin modulation (~1 nT) occurs: this is caused by time varying current of the shunting system  Note: the developed procedure is applicable to a spinning spacecraft, and it would be very hard/impossible to remove such interferences on a 3-axis stabilized spacecraft