flats into the flight reference frame, and to monitor any gross…

flats into the flight reference frame, and to monitor any gross changes in response; however, recent measurements show that the
changing thermal conditions during the time that the ground flats were taken will make this impossible. Instead, flat fields will
be derived from in-flight measurements of bright white dwarfs.

         Figure 2 shows several regions of one of the ground flat field images for segment 1A. The ground-based flats show a
complex structure containing dead spots due to blocked pores, brush marks from cleaning of the MCPs before assembly,
multifiber bundle hex boundaries, and moiré features6 due to a beating between MCPs. These same features appear, along with
a stretch and shift due to changes in detector electronics temperature (to the limits of the S/N of the measurements) on orbit,
which is consistent with previous missions using similar detectors. The fixed pattern noise, particularly the moiré with a
periodicity of about 50 µm, limits the ability to make very high signal to noise measurements without dithering, or moving the
spectrum on the detector during an observation. The two dimensional detector format means that a given spectrum falls on




Figure 2 A reversed image of two regions from the ground-based flat field on detector segment 1A, showing typical features.
The image on the left shows several large dead spots with bright rims, along with brush marks. The image on the right shows
the chicken wire pattern from the MCP hex boundaries. Shadows of the grids are seen as faint horizontal stripes.
millions of pixels. Since it is not possible to determine the flat field all of those pixels to high S/N, a simple shift and add, or
averaging approach is being taken for observations where the highest signal-to-noise is required. This allows a given part of the
spectrum to sample multiple regions on the detector and nearly recovers the S/N expected from Poisson statistics alone. Using
this method, a signal-to-noise of 120 per spectral resolution element was demonstrated during the instrument checkout period.

3.2. Background

          The apparent detector background, from a combination of residual radioactivity in the MCPs7, cosmic rays, other high
energy particles, and scattered light, is approximately 0.8 counts/cm2/sec on all four segments; reasonable pulse height thresholds
can reduce this number by a factor of two. Before launch, when the particle background was not present, typical rates were ~0.35
counts/cm2/sec. Because there is no shutter on the FUSE instrument, the detectors are constantly collecting photons, primarily
from airglow when there is no target in the apertures. This contamination means it is not possible to get an accurate background
measurement from the detectors in the region where the spectra fall. Thus, unused regions of the detector are used to measure
this effect. Since there appears to be no identifiable structure to the background, no spatial variation has been assumed.

3.3. Single Event Upsets

          The only significant detector anomaly discovered on orbit was the sensitivity of the electronics to single event upsets
(SEUs). When the detector electronics were powered on after launch, the detector data processing unit began reporting errors
in the memory which stores the code controlling the detector. Further investigation revealed that the memory was being corrupted
by high energy particles as the satellite passed through the South Atlantic Anomaly (SAA). These SEUs, which now occur
roughly once every three days on each detector, have no effect on the science data, but are a potential detector health and safety
issue, since corruption of the executing code could cause unpredictable behavior. Although it is not possible to decrease their
frequency, their effect has been minimized by developing a procedure by which the instrument flight computer reloads the
detector code whenever an SEU is detected. Rarely, about once per month, a potentially more vital part of memory is corrupted,