What 'gotchas' exist in the spectral package?
- In PLOT_COUNTS, the left-most pixel is off the plot.
- You MUST use PI for output in QPSPEC, NOT PHA. PROS now defaults to pi bins for spectrum extraction.
- For multiple spectral components, only the FIRST normalization is in real units; the remainder are specified as a percentage of the first. ALL normalization factors are free parameters, however.
- The lowest level bin should NOT be used in ANY fit, as there is no response for this bin. This occurs because the LLD (Lower Level Discriminator) occurs at an unknown location in the lowest PI channel due to conversion from PHA to PI binning and gain variations. In addition, the next three lowest bins are known to be useless, due to soft x-ray contamination (solar-scattered x-rays, auroral x-rays, ...). Therefore, use bins 3-34.
- There is a problem in PROS due to improper handling of the precision. This problem appears as the error message column already exists when trying to do table creation with the search_grid task. The column names in the table are the actual grid values, so a lack of precision aborts the table creation because the software believes a column already exists. This problem arose before, and the PROS people thought they had fixed it.