Four of the 19 SPI Germanium detector have ceased to function properly and they are now disabled. The failed detectors also affect the response of their neighbors as the double events involving before a neighbor and the ``to die'' detector are now seen as single events in the neighbor detector (all multiple event are affected similarly). Currently, we have four independent instrumental responses, one for each period in between the different detector failures.
| Det ID | UT | IJD | Pointing | Revolution |
| #2 | 2003-12-06T09:58:30 | 1435.41635 | 014000040010 | 140 |
| #17 | 2004-07-17T11:00:00 | 1659.46000 | 021400600010-021500000012 | 214-215 |
| #5 | 2009-02-19T12:00:00 | 3337.50000 | 077500610010-077600000012 | 775-776 |
| #1 | 2010-05-27T16:00:00 | 3799.66740 | 092900810010-092900870010 | 929-930 |
Our software cannot currently handle a time dependent response, and the easiest way is to analyze the three possible cases independently and to combine the results later on. It is possible however to analyze different mixtures of data together using one of the three responses as they are not so different. Great care should be taken in this case anyway. For example, the single event background in the neighbors of a dead detector will increase sharply at the time of the failure, while the other detector backgrounds will remain approximately constant. As a consequence, it will no longer be possible to assume that the background follow the same time evolution in all detectors (as you do when using the saturated Germanium event count-rate as a time model, for instance).
It is no longer necessary to specifically exclude dead detectors from the analysis. The software does that automatically.