IRAS Explanatory Supplement
IX. The Low-Resolution Spectra
Table of Contents | Index | Previous Section | Next Section
- Optical Properties
- Effects of the Zero-Clamp
- Summary of Instrumental Characteristics
Figure IX.A.1 Relative location and width of the image of the
exit slit in the field-of-view of the telescope, and the resulting
spectral resolution are given as functions of wavelength. The
width and resolution have been calculated taking into account
the diffraction by the telescope and the electronic filtering.|
A.1 IntroductionThe IRAS survey instrumentation included a low-resolution spectrometer which covered the wavelengths between 8 and 25 µm. The spectrometer operated during the entire survey, providing spectra of the brighter point sources. This section briefly describes the instrument, dealing primarily with those aspects of its optics and its signal-handling electronics that have a direct bearing on the interpretation of the spectra. The spectra are presented in two forms -- a tape version, henceforth called "The Catalog of Low Resolution IRAS Spectra", or simply "The Catalog", and a hard-copy version, designated as "The Atlas of Low Resolution IRAS Spectra", containing graphical representations of the spectra.
Because the survey function required a passive instrument, a slitless design was selected. This design is equivalent to an objective prism spectrograph, oriented in such a way that the dispersion was aligned with the scan direction. Obvious penalties of this design are a degraded resolution for extended objects, sensitivity to spatial confusion, and short integration times.
The detectors were sampled continually during the mission, and the data were received on the ground together with the other survey data. The spectra were extracted during data processing on the basis of point source detections from the survey array.
A.2 Optical Properties
Figure IX.A.2 The optical layout of the spectrometer is shown
schematically. SO is the entrance aperture of the spectrometer
in the focal plane of the telescope. The field mirror M1, and
curved prism P1 and the exit slit S1 comprise the primary spectrometer.
M2, P2 and S2 from a secondary system that provides most of the
dispersion at the short wavelengths.|
The spectrometer had a rectangular aperture mask in the focal plane that measured 6' in the dispersion direction and 15' across. Although the cross-scan width of the focal plane aperture was only half that of the survey array width, the overlap between adjacent scans in the survey strategy ensured full sky coverage.
Two overlapping wavelength ranges were scanned simultaneously, one extending from 7.7 to 13.4 µm and the other from 11.0 to 22.6 µm. The scan length was nominally 6' internal alignment obscured the extreme low end of the short wavelength range. The resolution was primarily determined by the exit slit width of 15", although diffraction at the telescope aperture and electronic filtering caused significant smoothing. Figure IX.A.1 shows, respectively, the aperture location of a source, the effective monochromatic image size, and the resulting spectral resolution as functions of wavelength.
Several detectors were used in each of the two wavelength ranges to reduce confusion problems. Three short wavelength detectors each covered 5' of the aperture width. At the longer wavelengths two detectors each covered 7.5'.
The optical layout of the spectrometer is shown in Fig. IX.A.2. A back-reflecting KBr prism with curved surfaces served the three functions of culmination, dispersion and refocusing in the long wavelength spectrometer. A field mirror imaged the telescope pupil onto the prism. The short wavelengths also passed through this system, but were given additional dispersion in a second spectrometer section with a curved NaCl prism, using a field mirror adjacent to the first exit slit. field optics immediately behind the exit slits refocused the telescope pupil on each of the detectors.
Figure IX.A.3 The AC-coupling and zero clamping circuit. Zero
clamping occurred when the input signal was decreasing, resulting
in an output signal of zero.|
An important aspect of the data handling was the encoding of signals with a large dynamic range into an 8-bit format. The output voltages were digitally encoded on a logarithmic scale with increments of 3.5% limiting the signal-to-noise ratio for single samples to less than 100. To avoid loss of precision, the baselines were kept at low positive levels by AC-coupling and "zero-clamping". Occasionally, the zero-clamping affected estimates of the baseline, as discussed in the next section. Fortunately, the occurrence of zero-clamping was rare among the spectra selected for the catalog, since the rejection of confused spectra tended to eliminate those with zero-clamps.
Further detail on the spectrometer electronics is contained in the instrument description by Wildeman, Beintema and Wesselius (1983).
A.4 Effects of the Zero-Clamp
The effects of AC-coupling and zero clamping on flat spectral scans
with sloping backgrounds. From top to bottom: input signals, output
signals and reconstructed input signals. Absolute levels are relevant
only for the output signals. During reconstruction, the possibility
of zero-clamping is ignored. Within a spectral scan, wavelength
decreased with time.|
Figure IX.A.3 shows a schematic representation of the circuit usedfor the AC-coupling and zero clamping. The voltage across the coupling capacitor equaled the difference between output and input signals and thus constituted an error signal. As long as the output signal remained positive, the circuit behaved like a high-pass R-C filter with a time constant of 10 seconds (a spectral scan lasts 1.5 seconds). The fact that the error signal changed at a rate proportional to the output signal was easy to correct. However, as soon as the output tried to drop below zero, the capacitor was instantaneously recharged to prevent any further drop; the output signal was clamped to the zero level. The error signal was now the inverse of the input and could rise rapidly, while the output indicated no rate of change. The clamp remained active as long as the input signal continued to decrease. As soon as the input signal rose again, the clamping ended and linear behavior was restored. The output signal again tracked the rate at which the error signal changed.
Normally, zero clamps occurred just after spectral scans or during negative noise excursions on "empty sky." However, a steeply decreasing background could continually clamp the output at zero, except during a spectral scan. In such a case, the presence of a sloping baseline would go unnoticed. Because of the direction in which the spectra were scanned the result is then an underestimate of the signal at the shorter wavelengths of both spectral ranges. A rising background would only raise the baseline in the output signal with no worse effect than a reduced precision of digitization. Figure IX.A.4 illustrates the effects of background slopes.
A.5 Summary of Instrumental Characteristics
The design of the instrument and the planning of the IRAS mission resulted in the following global characteristics of the spectra:
- spectral coverage from 8 to 25 µm in two overlapping ranges;
- a spectral resolution for unresolved sources varying from 20 to 60;
- a sky coverage identical to the coverage of the IRAS survey;
- typically 2 or 3 observations per spectrum, together providing about 0.15 seconds of integration time per spectral element per source;
- slit widths corresponding to 15" and instantaneous fields of view of 30 square (short wavelengths) and 45' square (long wavelengths);
- quantizationsteps of 3.5%, resulting in an rms digitizationerrors of 1%;
- potential problems from negative backgrounds gradients.
Table of Contents | Index | Previous Section | Next Section