Background subtraction using many non-local off-source observations

The procedure to generate an universal background model is identical to that in section 3.2, except for the following points:

1. A large number, typically more than 10, off-source observations are required: there must be sufficient observations to adequately sample the 37-day cycle. The observations can be spread over several months, but the best results will be obtained if some are before the source observation and some are after.

2. The time counters, describing the induced radiation, should be inserted into both the source and all the off-source data cubes using the program bgdsaa37. The program bgdsaa37 differs from bgdsaa as the decay parameters it creates vary on the 37-day cycle.

3. Checking the quality of the background fit by the methods described in section 3.1 and 3.2 can be a prohibitively long and tedious procedure because of the number of off-source data files involved. Inspection of the CONSTANT parameter files (bgd_constant00m.phad) can frequently help locate regions of poor quality data as only small differences due to variations in the diffuse cosmic X-ray background should be present.