The figures available through the links tabulated on the Gain Calibration webpage, show primarily the positions of the Fe (6 kev) and Cd (22 keV) lines from the four calibration sources on each unit, in terms of linearized onboard channels from the ADC. We use this information to calculate the gain correction factor (PHA channels per keV) that is needed to turn the raw energy values of each science event (PHA channels) into a physical energy scale (keV).
Uncorrected data shows results obtained without the use of any required IC gain history table. All results without the word `uncorrected' indicate results produced using the IC table instead of the automatically produced table.
To verify the energy calibration of the instruments we look at the position of the Xe line found in the instrument background of the science data that has been corrected with the Cd/Fe data. This line arises from fluorescence in the detector gas, and hence can be seen all over the detector plate. The line should appear at 29.6 keV. Acceptable values of the Xe position for individual Science windows, have a scatter lying between 28.8 and 30.4 keV, symmetrically about the 29.6keV level. Revolutions with a Xe line more than 3% away from 29.6 keV either have problems with the gain fitting and smoothing (seen on the Cd/Fe figures) or have experienced unusual conditions in the detector. The latter category is very rare. These revolutions are flagged with red/pink or green backgrounds and are corrected by the use of IC tables.
Xe Line shows the fit to the Xe line using all the corrected events from the entire revolution. This figure also shows the peak value of the fitted line. For a well fitted revolution this is between 29.2 keV and 30.0 keV. Originally this was the only Xe analysis we performed and so the SCW-by-SCW analyses described below may be missing for the earlier revolutions of the mission.
Xe Evolution This shows the Xe line position in the corrected data for each science window in the revolution. Fitted widths to these lines are also shown as an indication of the reliablity of the fit. Widths less than about 1 keV and greater than 3 keV generally indicate a poor or misleading line position due to a very weak Xe signal and can be disregarded.
Xe histograms These figures show the spread of Xe line positions and widths for the individual science windows in the revolution.
Other background lines are available for gain verification. Molybdenum and copper from fluorescence in the collimator structure have lines at 17.4 keV and 8.04 keV. They are considerably weaker than the Xe line and whether they can be seen or not depends strongly on the source spectra in the field of view. The copper line generally lies considerably higher than expected as a result of interference from a sillicon line.
Other data about the revolution is also tabulated in the last column of the table. This can vary depending on any problems with the gain correction and whether public or survey data is available for further analysis. `Other data' includes:
Corrected gain history tables needed to perform correct analysis of the data in those cases where automatic calibration has failed. These are simply copies of the tables delivered to ISDC and available as part of their Instrument Characteristics package.
Corrected smoothing or Uncorrected indicate figures to show how the gain history and gain calibration has been improved by the use of the gain history table produced by offline analysis.
GCDS and GPS surveys Lists of sources found using only the JEM-X units where these survey data are public.
Full analysis Link to analyses of unsual gain behaviour of the instrument in those cases where something really strange/interesting has happened. These are fairly rare.
Average gain This figure shows the average gain of the instruments in PHA channels per keV, derived from averaging the positions of the useable calibration sources. This figure gives an indication of the gain factor used to calibrate the instrument during the revolution. Whether the gain factor is high or low has important consequences for the electronic efficiency of the instrument and whether the Xe line can be reliably fitted.
TemperatureThese values are taken directly from the Housekeeping data, and show how the instrument changed temperature during the revolution. This will depend strongly on the solar aspect angle of the instrument during each pointing, and whether the satellite was in eclipse during perigee.
Event ratesThese values are also taken from the housekeeping data, and show the total number of events transmitted by the instrument, including particle and background events. Here you can see if there was a high background, and whether the satellite was performing a dithering pattern over a strong source.
Any other data we think might help a user to understand how they're getting the results they are, and whether these are real scientific discoveries, or just annoying instrument fingerprints.
The value of TSTART is given in Integral Julian Days (IJD) and refers to the start of the calibration spectra integration period. Each set of calibration spectra is integrated for 256 secs. Therefore each point in the gain history figures covers about 4 minutes. A gain history table for a single revolution usually has about 840 entries for each source, giving about 60 hours of data for each 72 hour revolution.
The sources are identified using the following colours:
JEM-X1 Source 1 (Fe) data: Black; model: Red Source 2 (Cd) data: Yellow; model: Pink Source 3 (Fe) data: Blue; model: Green Source 4 (Cd) data: Purple; model: Orange JEM-X2 Source 1 (Cd) data: Black; model: Red Source 2 (Cd) data: Yellow; model: Pink Source 3 (Cd) data: Blue; model: Green Source 4 (Cd) data: Purple; model: Orange
The model value, is the smoothed mathematical model used to determine the instantaneous gain correction factor by j_cor_gain. The average of the four smoothed curves is used to determine the gain calibration for the entire microstrip plate. If the model does not describe the raw data well, the gain calibration will not be good for the time period where model and data differ. To ensure that the different data sets do not lie on top of one another, the actual plotted values are:
JEM-X1 Black/red Source 1 peak position Blue/green Source 3 peak position - 40 Yellow/pink Source 2 peak position Purple/orange Source 4 peak position + 50 JEM-X2 Black/red Source 1 peak position Yellow/pink Source 2 peak position -100 Blue/green Source 3 peak position +100 Purple/orange Source 4 peak position +120The label `No Fe Lines' appears in the third column for revolutions where only JEM-X2 was active. JEM-X2 has only cadmium calibration sources and hence does not produce iron calibration lines.
A quick glance through the Gain Calibration link table will show you that the vast majority of the revolutions processed are well-behaved and well-processed. However, those lines with pink/red or green back grounds indicate revolutions with some gain calibration problem. Due to gain aging, all revolutions after 949 must be corrected using an IC gain history table.
Pink and red colours indicate revolutions which have problems even using the latest software. The darkness of the colour colour indicates the severity of the problem. Pale pink can mean that an entire revolution is affected very slightly, possibly to an extent not even noticeable to the uninitiated, or it could indicate a noticeable problem for a small part of the revolution. Dark red indicates a problem so severe that the data cannot be used. So far this has only occurred during a couple JEM-X2 warm-up revolutions, where the NRT data was processed with the wrong software.
Shades of green indicate revolutions that had problems being processed with old versions of the software, most notably, sparse gain history tables produced using j_calib_gain_fitting 7.1. However, with newer software these problems are corrected. It is important for some revolutions, usually those with heavy grey filtering, that the gain history table is generated with verion 7.2 not 7.1. This is not easy for the non-specialist to determine. Be sure to download the latest revolution file data before you begin processing, and start the JEM-X Science Analysis at level COR.
Corrected science data is archived in the ISDC repositories, but this is invariably a reflection of the processing and instrument model as it was at the time of Consolidated processing and will not reflect the latest understanding of the instruments. If you want the best possible results, be sure to have the latest instrument characteristics tables, the most recently generated revolution files, the latest ISDC software release and start all you processing at analysis level COR.
CAO 9/02/2011