The product of the transmission efficiency of the optical filter stack and the detector efficiency defines the effective wavelength passbands and the nominal out-of-band rejection performance of the detectors. This is described in Section II.C.5 and Tables II.C.4 and II.C.5. The detector passbands were verified before launch at the component level, but not at the system level. In-flight tests were performed to test the consistency of the passbands and out-of-band rejection within each band and to verify the nominal passbands. There was no direct in-orbit verification of the effective transmission characteristics of the detector/filter combination.
Numerically, the approach was to measure, in each wavelength band, the flux Si,j from test source i on detector j. The average of Si,j over the detectors in the wavelength band, <Si>, then gives the best estimate of the flux from the source in that wavelength band. The statistical distribution of the ratio
(IV.B.1)has, by definition, a mean of one if all detectors respond equally to the test source and NGC 6543. The test sources (Table IV.B.1) were four stars and two cold sources, Neptune and the galaxy UGC 11348.
The stars probe the short wavelength rejection and the short wavelength transmission cuton of the detectors in all bands. The two 50 K sources probe the longwave cutoff and longwave rejection in the 12 and 25 µm bands and the inband transmission of the 60 and 100 µm bands. Significant changes in the band pass characteristics of an individual detector relative to the band average transmission would be detectable as a statistically abnormal member of the Ri,j data set.
The results of this test indicate that deviations in the flux estimates from the mean are typically less than 10%, with a worst case value of 16%. Table IV.B.2 gives the typical and worst case value of the absolute deviation of Rij from unity in each band for the stars and the cold sources. In spite of their wide temperature range, there was no difference between the hottest and the coolest star.
The nominal response of the filter/detector combination and results of preflight tests at the component level are summarized in Section II.C.4. In-flight verification was required for two reasons. First, it was not possible to test fully the detector/filter combinations before the flight even at the component level. This applies in particular to the spectral response of the 100 µm detectors. Furthermore, a number of problems could have resulted in a degradation during system integration or during the flight.
Potential problems with the spectral response can be divided into two categories, those that represent systematic deviations from the nominal characteristics of all detectors in a band and those that result in a random scattering in one or several detectors in a band from the band average. Problems likely to result in random scattering are cracking or delamination of the filters, photon leaks around filter mounts etc. Extensive in-flight tests, described in the previous section, indicate that the detectors had rather similar passbands, thus eliminating random, but not necessarily systematic, effects.
An attempt was made to address the problem of systematic deviations of the nominal inband and out-of-band transmission of the four bands in a semi-quantitative way, by using observations of stars with widely different temperatures, asteroids and the planet Uranus to probe different portions of the passband.
|Hot test source||Cold test source|
|Vir,||B1 IV||Te=24000 K||Neptune||Te=55 K|
|CMa,||A1 V||Te=10000 K|
|Car,||F2 II||Te=6900 K||UGC11348||Te=50 K|
||(Ri,j - 1)||
|µm||Typical||Worst Case||Typical||Worst Case|
|100||0.10 #||0.13 #||0.06||0.16|
|* not detected||# low signal-to-noise ratio|
The following hardware related issues affect the various bands: