Ionizing radiation has a strong disturbing effect on the performance of the detectors. The high flux of protons and electrons trapped in the earth's radiation belts makes operation impossible. At about 2^h30^m after perigee passage the temporary radiation damages of the detectors are cured by a combination of heating, illuminating and bias increase (see section 4.2.4). Outside the radiation belts, during the 16 hours science window, the ionizing radiation remains a limiting factor to detector performance. Not only direct hits on the detector, but also secondary electrons released by satellite materials after a hit cause disturbances.

**Figure 4.6:** Example of the effects of glitches during a measurement. The measurement is at a very low flux level (about 5 times the dark signal) so that even ``weak'' glitches can clearly be distinguished. Upper panel (A): the ``raw'' CRE voltage level as a function of time, each dot is a non-destructive readout. The measurement consists of 32 integration ramps of 4 sec reset interval, each ramp contains 128 readouts. Most of the ramps show one or more significant voltage jumps between consecutive readouts caused by ionizing radiation. Based on the average increase in voltage between the readouts in a ramp these glitches can be removed. Lower panel (B): the resulting signals after deglitching at CRE voltage level. Despite this deglitching the signals still show disturbances. The glitch which occured around t=50 sec caused a variation in detector responsivity lasting about 10-20 sec. The dashed line is the final mean signal level after removing signal outliers and after applying a drift recognition algorithm. The corrections presented here are also used in Derive_SPD
$\begin{figure} \begin{center} \begin{tabular} {c} {\epsfxsize=10cm \epsfbox{glitchexample.eps}} \\ \end{tabular}\end{center}\end{figure}$

Ionizing radiation consists of high energy protons and heavier nuclei. It causes several effects in the photoconducting detectors:

A glitch can be recognised by a step on the integration ramp between two readouts. The influence of low level glitches to low signal data is illustrated in Fig. 4.6.

Note that the energy of the particle passing the detector is not directly correlated with the (measured) glitch energy. The glitch energy depends on the amount of electrons freed by the high energy particle along its path through the detector. Higher energy glitches can cause ramps to saturate or can even cause a transient for several seconds. Low level glitches become more apparent in measurements with low in-band power. Examples are faint sources on a low background, PHT-S spectra of faint targets, observations with small apertures using detectors P1 or P2.

The glitch rate is a function of the detector size and the energy deposited. Calibration measurements have been carried out to determine the particle hit rate as well as the signature of the glitches as a function of orbital position of the satellite with respect to perigee passage. On average, a clear glitch is detected every 10 sec for standard science observations.

**Figure 4.7:** The in-orbit glitch energy deposition spectra of detectors P3 (squares), C100 (diamonds), C200 (crosses). The spectra were obtained on the basis of dark signal calibration measurements collected throughout the mission, see [4].
$\begin{figure} \begin{center} \psfig {file=lwglitch_spectrum.eps,height=7.5truecm,width=11.5truecm}\end{center}\end{figure}$

Analyses of glitches have shown that they have a continuous energy distribution (Fig. 4.7). Detectors P3 and C100 are more vulnerable to cosmics. These detectors exhibit a temporary responsivity variation for several seconds. Although special algorithms have been developed to remove these disturbances they can determine the sensitivity limit of these channels.

The slow responsivity increase caused by cosmic rays in the 16 hours science window is about 20% for the silicon doped detectors P1, P2 and PHT-S. It can reach a factor 2 for the P3 and C200 detector. These variations are calibrated out by FCS reference measurements which are equally affected as the celestial targets in the same observation sequence. Moreover, to minimize the disturbance, an extra curing procedure has been introduced for the detectors P3, C100 and C200 to be carried out near apogee when the satellite is handed over between the two ground stations.

The radiation effects described here become worse during periods of higher solar activity. Also measurements carried out near the end of the science window when ISO is approaching the radiation belts, suffer from increased radiation effects.

The tuning of the parameters controlling the OLP deglitching algorithms has been performed empirically after evaluation of many datasets. Descriptions of the algorithms are presented in section 7.3.4.

4.4 Effects of Ionizing Radiation