Detector dark noise and dark current are
kept in calibration G files 21_1, 21_2, 21_4 and 21_8 ,
corresponding to the 1, 2, 4 and 8 second reset times.
Units are 
. Note that 8 second resets were only used early in the
mission.
Table 5.1 shows average detector dark noise for an entire band
for a one second reset interval. The detector noise given is that if some
processing has been carried out on the data, namely throwing away samples
affected by glitches . For comparison, the pre-launch values of
detector noise were 0.6 
for bands 1, 2, 3 & 5 and 0.9 for
bands 4 and 6.
| Detector | Detector | Detector |
| Array | material | Noise |
| | ||
| 1 | InSb | 0.7 |
| 2 | Si:Ga | 1.5 |
| 3 | Si:As | 3.0 |
| 4 | Ge:Be | 2.0 |
| 5 | Si:Sb | 1.1 |
| 6 | Ge:Be | 3.0 |
Note that all detectors in a band are not equally good, and which ones are or are not good can change. For example, at the start of the mission band 3 had two poor detectors, numbers 5 and 10. However, from revs 250 to 500 the worst detector in that band was number number 12, with detectors 5 and 10 improving. In general a bad detector has either a high dark current or/and a high noise.
Cases where the continuum is weak can cause problems for dark current subtraction, especially when memory effects are included. When the amount of flux falling on the detector is small, there may be cases where the output of the detector from (incident flux plus dark current plus noise) is less than the detector from (dark current plus noise), the end result being negative fluxes in the AAR product. The limiting flux for this is of the order of a few Jy for band 2 .