Subsections

• 2.4.1 Design of the signal chain
• 2.4.2 Processes in the signal chain

2.4 Signal Detection

2.4.1 Design of the signal chain

The SW and LW gratings are associated with 4 different detector arrays of 12 elements each, and the 2 SWS Fabry-Pérots are associated with 2 double detectors (only one of each pair being used to gather valid data). More information on the detectors is given in Table 2.1. All detectors are operated simultaneously. The InSb, the Si:As (BIBIB) and the Si:Ga detectors are operated at a temperature of 4 K, which is the temperature of the spectrometer unit. The Si:Sb detectors for the FP are heated to 10 K and the Ge:Be detectors are cooled to 2.5 K by a thermal strap to the helium tank of the cryostat.

Except for the InSb photo-diodes, the detectors are photo-conductors that require a finite bias voltage. The detectors are connected to discrete integrating-amplifier chains. The first amplifier stages are heated-JFET buffer amplifiers located close to the detectors. The remaining electronics are located in the warm amplifier box in the service module of the satellite.

Figure 2.6: Simplified scheme of the SWS electronics showing all components known to affect the integrated signal from the detectors.

A schematic diagram of the detector electronics is shown in Figure 2.6. The detector current charges the small capacitor formed by the finite capacitance of the JFET gate and of the electrical leads between the detector and the gate. The resulting voltage ramp is amplified and digitised, to be read out `non-destructively'. Every 1, 2 or 4 seconds (selected by the AOT logic) the pre-amplifier is reset by removing the charge from the input gate. The residual charge after a a reset is minimised by a compensation pulse immediately following the reset.

The heated-JFET pre-amplifiers of each grating detector array are combined in a single integrated circuit. A side-effect of this arrangement is that stray capacitance between the input leads of the pre-amplifiers causes significant cross-talk between neighbouring array elements. Since the only involved impedances are capacitive, the cross-talk percentage is independent of frequency and can be accounted for in the data analysis.

The pre-amplifier resets are sufficiently compensated to avoid overloading of the first stages of the warm amplifiers. To avoid dynamic-range problems due to remaining offsets, the signals pass through an AC filter that removes the DC offset. This filter affects the shape of the ramps in a predictable and correctable fashion.

The filter outputs are sampled 24 times per second, and then they are multiplexed. They pass through an amplifier with selectable gain factor of (1, 4 or 16) times 225. The gain for FP detectors is an overall factor of 2 higher. Finally they are digitised into 12-bit numbers (covering a 20-V range). At the highest gain setting, the digitisation is precise enough to sample the pre-amplifier noise. At the lowest gain setting, the gain of the warm amplifiers is 225 for the grating detectors and 450 for the FP detectors.

Figure 2.7 gives a data example in which the individual datapoints from the non-destructive readouts and reset pulses can be seen. The plot starts with a reset pulse, where the capacitor is short-circuited causing the bit readout to spike high. The system stabilises after about four readouts, and as light falls on the detector the charge on the integrating capacitor slowly increases. Careful analysis shows the system to require more than four readouts to stabilise. All the slopes are slightly curved, due to the limited time constant of the AC filter - see Chapter 7. The second slope suffers from a glitch approximately half way along its length, where the measured voltage suddenly jumps instead of following the (curved) slope. Some extra charge is dumped on the capacitor due to e.g. a cosmic ray hit. The rest of this readout is not affected by the glitch as the slope just continues on its curved path, as can be seen by comparing it with the other four slopes.

Data examples for each AOT are shown in Sections 3.3 to 3.6. The automatic data processing chain (OLP) is discussed in Chapter 7.

Figure 2.7: Example readouts, from part of an SWS06 observation. Note the glitch approximately half-way through the second slope, where the output suddenly increases. Also note the curvature in the slopes, caused by the AC filter. Both these effects are removed in the OLP processing chain.

2.4.2 Processes in the signal chain

In this section all processes that can be identified in the 24 Hz data or in the extracted spectrum induced by the detectors and electronics will be briefly described. This is adapted from Lahuis et al. 2001, [25].

Detectors: detector cross-talk
The cross-talk between adjacent grating detectors amounts to between 10 and 15%. The correction for this is discussed in Section 7.2.7.

Detectors: memory effects
The SWS Si:Ga, Si:Sb and Ge:Be detectors all suffer from memory effects or transients, whereas the InSb and Si:As detectors are both free from transient effects.

Transient effects were observed in the laboratory prior to launch and have been taken into account in the design of the observing templates used for all general observations. The templates were designed such that the influence of transient effects were minimised. There is sufficient redundancy in each observation so that transient effects can be recognised in the time domain and possibly corrected for. Memory effects are discussed further in Sections 7.3.3.2 and 9.2.

Detectors: de-biasing
During the integration of the detector current, the voltage rise across the detector reduces the bias voltage so that the responsivity of the detector drops slightly during the reset interval. This effect is predicted for all detectors except the InSb photo-diodes. It has been seen only for the Ge:Be grating detectors, and only at very high input signals ( few Jy).
This effect is under study. Currently no correction is available.

Detectors: particle impacts
All SWS detectors suffer from the impacts of charged particles. When a particle impact occurs, the additional charge released in the detector causes a jump in the 24-Hz readout signal, constituting a `glitch' in the data. A secondary effect of the particle impact might be a transient behaviour after the impact where, on time scales of a few seconds, transient tails are observed in particular for the Ge:Be and the FP detectors.
These jumps are recognised and corrected for in the pipeline, see Section 7.2.8. The transient tails, however, are not corrected. For more information on glitch tails see Section 9.2.4.

Reset: saturation
The application of the reset/compensation pulse is never perfect. A fraction of the compensation pulse is fed into the first few samples and these may go into saturation in the amplifier chain, e.g. in the operational amplifier in the high-pass filter. See Figure 2.7, the points at 4095 bits. The reset time of this saturation is fast, less than a quarter of a second. The direct consequence however is that the first samples of the integration ramp cannot be used in the slope fit.

Reset: reset-pulse aftermath
After the reset/compensation pulse has been applied a charge can be left before the input gate of the JFET. This charge will decay over the combined capacitance of the cold electronics and be added to the charge built up by the integration of the detector current. This is seen as an additional exponentially decaying signal onto the integrated 24 Hz signal, usually referred to as the pulse-shape effect. The pulse-shape is an additive effect and only important for low-flux cases. The correction for this effect is discussed in Section 7.2.6

Readout: high-pass filter
The high-pass filter which removes DC offsets has a typical time constant of 2 seconds. Individual time-constants were derived for all 52 detectors from laboratory data. The correction for this is discussed in Section 7.2.5.

Readout: gain amplification
Accounting for the selected gain setting is trivial. The gains are selectable per detector array to values of 225 times (1, 4 or 16). The FP gains are an overall factor of 2 higher. No deviations from the nominal gain ratios have been detected.

Readout: A-D conversion
At the end the analogue signal is digitised. The amplified analogue signal ranges from V, and is converted to a bitrange from 0 to 4095 corresponding to V. Any analogue signal outside the V range is therefore set to 0 or 4095 and flagged as out of limit in the first stage of the processing - see Section 7.2.2.

Readout: saturation
At very high signal levels (V/s), saturation effects come into play. First, parts of the ramps become saturated and at even higher signal levels the complete ramp can become saturated. In the last case the ramp will be outside the bitrange already after the first sample.

Unless otherwise noted, these effects are corrected for in the pipeline in the reverse order in which they occur, to undo their effects and to reconstruct the incoming signal. See Section 7.2.