A limitation of the sensitivity of the LW array is due to the responsivity variations and glitches due to the impact of charged particles. There are two separate problems: i) the response variation after the perigee passage due to the very high radiation dose coming from trapped particles in the van Allen belts, and ii) the glitches due to the impact of single particle from galactic cosmic rays all along the good part of the orbit, i.e. outside the van Allen belts.
Extensive radiation tests were performed with gamma-ray sources, protons and heavy ion accelerator beams to simulate the conditions in the van Allen belts. The ionizing radiation induces a responsivity increase which relaxes in a few hours. This effect is minimized if the photoconductor is polarized and exposed to a high infrared flux. During the perigee passage, since the experiment is switched off, a specific power supply keeps a bias voltages on the photoconductors, and the camera is left open to light. With this procedure, our test results show that the responsivity variation should remain below 5% in the good part of the orbit.
During the good part of the orbit, the main perturbation will come
from galactic protons and 
particles. Their fluxes are
defined within a factor of 2, depending on the solar activity. In
addition to these external particles, the antireflection coating of
the lens contains Thorium, which generates a flux of low energy
particles. This flux depends on the solid angle of the lens
viewed from the array, and is maximum for the 12'' lens. All these
sources are summarized in Table 
.
Table: Glitch rates for the whole array
Less frequent, but more damaging, will be the heavy ions, at an expected rate of about 1 every 20 minutes on the array. Each ion will affect about 50 pixels, and will generate a glitch followed by a decrease of responsivity.
Proton and heavy ions glitches can be easily removed from the observation by a filtering technique. The worst problem is the responsivity variation after the glitch, especially for heavy ions. A detailed modelling of these effects is under investigation, but it will have to be tuned up with real flight data.
A minimum of four exposures is required for each observation (and 16 if the integration time is 0.28 sec), to be able to properly deglitch the data.