Noise performance of the Herschel-SPIRE bolometers during instrument ground tests

The flight model of the SPIRE instrument underwent several test campaigns in a test facility at the Rutherford Appleton Laboratory (RAL) in the UK. A final dark campaign, completed in March 2007, provided an environment virtually free from optical radiation. This allowed re-determining the fundamental model parameters of the NTD spider web bolometer detector arrays in the new environment. The tests reported in this paper produced a fairly homogeneous dataset to investigate white noise and 1/f noise at different bias voltages, bias frequencies, and bath temperatures. We find that the white noise performance is in excellent agreement with the model predictions, once we correct the low frequency signal variations that are due to temperature fluctuations of the thermal bath at about 300 mK. The temperature of the thermal bath (detector array base plate) is measured by thermistor pixels that are part of the bolometer arrays. A residual 1/f component beyond those variations is hardly detected. This unexpected stability is very welcome and will positively impact photometer scan maps, the most popular observing mode of SPIRE.