Telemetry Decommutation

The data were split into attitude telemetry, marked by spacecraft time, and science and engineering telemetry, marked by instrument time.

• Packet Cleaning: The team examined PACKET_ID and PACKET_TYPE in steps of 2 bytes and chose only those packets that contained the whole 1,024 bytes without truncation. The team made a new MMS file containing clean packets.

• Orbit Splitting: First, the team applied a median filter to TIME_SYNC data to remedy a checksum error, and then assigned a real-time OBS_TIME to each PACKET_TIME using the two OBS_TIME records stored at the beginning and end of the observation. The orbit number was identified by comparing the observation schedule against OBS_TIME and the observation time as inferred from the total number of packets. Occasionally, when there were too many packet errors, the orbit number for the data packet was found manually by examining the data.
  

• Merging: The team made a single MMS data file for each orbit by gathering the best packets from multiply-downloaded data packets. In other words, when new telemetry was received, the team identified its orbit number and compared with the previously downloaded data; if the newly-downloaded data had error-free packets (with no checksum error) corresponding to previously-downloaded packets with error, then the old data were replaced.
  

• Science Data Splitting: The team classified the data for each orbit into science data or housekeeping data, according to the PACKET_TYPE.

Stimpulsor Correction

The location of the photon signal recorded on the detector slightly drifted due the electric, electronical, and thermal causes. To correct this effect, periodic signals were made to a prescribed 4 corners. Using this signal, the photon location on the detector plane was corrected.

Four stimpulsors were wired to the SPEAR anode and fired at a total nominal rate of 45 Hz.  Fiducial centroids were determined from ground based data obtained during thermal vacuum testing at predicted flight temperatures. Each flight photon was shifted in detector X and Y coordinates based on the difference between the fiducial centroids and the measured 100-second running average centroid of each stimpulsor. Basic filtering was applied to the telemetry stream to remove problematic events. Typical event shifts were about 10 pixels, and the resulting stimpulsor centroids showed an RMS spread of 1.5 pixels.

To prevent digitization effects caused by binning floating point numbers, each X and Y value had a random number (ranging from -0.5 to 0.5) added to it at this stage. This ensured that the digits to the right of the decimal point were fully populated. This technique did not add any additional noise to the data.