The magnetic field data from the GOES satellite must be corrected on the ground for the contribution from the roll and yaw torquer coils which comprise important elements of the spacecraft attitude and orbit control system (AOCS). The correction may be calculated from measurements of the torquer current which are telemetered at the same rate as the magnetic field data or from torquer input command data. The baseline plan was to allow torque current changes only at infrequent intervals of once per 20 minutes since it was suspected that torquer switching noise might be a problem. However, for improved navigation and registration performance in the radiometric instruments, the actual mode of operation has been to calculate the torquer current drive more frequently and to rate limit the current changes to one commanded increment per step. The magnetic signature of the 3.2 ma step is an much as 1.2 nanoTeslas. Both the current level and the telemetered value change essentially instantaneous when compared to the frequency response of the magnetometer, which incorporates a 0.5 Hz bandwidth five- pole Butterworth anti-aliasing filter in each data channel. Straightforward correction of the data using the current telemetry therefore results in objectionable noise spikes in the derived bandwidth-limited output data. The problem is complicated by the fact that the clock which governs the AOCS current commands to the torquers is not synchronized with the telemetry unit clock which governs magnetometer data sampling. Fortunately, it has been possible to incorporate data in the spacecraft telemetry which provides improved information for determining the time of a current step in the telemetry unit time base as well as the torquer input command data mentioned as one method of compensation. This paper describes the development of a numerical filter used in ground processing of the telemetered current to simulate the five-pole Butterworth response of the magnetometer output filter. Since the timing information is determined to an accuracy of 1 part in 8, one of eight sets of filter weights is selected based on the telemetered timing information. Examples are given of the improvement in output data quality resulting from the technique.
|