Subject: JEM-X Monday Meeting 2006-11-06
X-Keywords: 

Participants: SB, CAO, JC, CBJ, NL, NJW.

JEM-X status (SB, CBJ): Still in eclipse season implying slow gain relaxation
after switch-on at belt exit. Temperature dependent gain: 2.5% per degree
detector temperature (as derived from revols. 488 and 496). In the start
of the mission the value was around 1% but an increase has been noted before.
higher than it used to be.
Similarly a gradual increase has been seen in the calibration jitter.
The implication seems to be that the increases in temperature sensitivity
and the local jitter in gain are strictly proportional to the  overall gain
increase (which by now for JEM-X1 is a little more than a factor of 3).

Anode status (JC): Anode RAWX = 127 is nearly dead now. 196 is dead.

j_ima_src_locator (NJW): Progress relative to report at SDAST meeting. Inclusion
of variance or significance maps if present.

Old and new selection criteria (SB, CBJ): JMX2 had higher efficiency with old
rejection critera than with the new ones (after revolution 170). The background
above the rejection cut has also been lowered, though.
The cut below which we have open criteria was determined using JEM-X2 to 
be PHA 150.
The (faulty) argument was then to also open JEM-X1 at the same energy. 
However, JEM-X1 was running with much lower gain at the time, so the cut 
in PHA corresponded to 115.
However, we are dealing with electronic efficiency, so energy is not 
really relevant. We should have opened JEM-X1 up to 150 as well. This 
would have made the pronounced (gain dependent) wiggle in sensitivity 
much less pronounced (it is still there, also in JEM-X2, but then at PHA 
150).
Conclusion is: use as is (for now), because no matter what criteria are 
used a gain dependent ARF must be used to produced reliable spectra in 
the 3-10 keV range, as the gain can only be kept with in a +-10% (or 
even wider) range.

j_ima_iros with few events (NL): The previously reported error seems to have been
removed with other changes. On the other hand the hot spot detection algorithm went
wrong for a low number of counts where a single count is a large deviation from
the average. The variance map has a strange appearance with few events since it is
based on backprojection. This is cured by making a distribution of fractions of
counts in the shadowgram. It is investigated, however, if this method has some
sideeffects.

Gain fitting (CAO): Randomized event positions in DETX and DETY have caused changes
in the results of j_src_spectra and j_src_lc. The best solution seems to be to go
back to the previous j_cor_position in OSA6. In the long run we can have two extra
columns in the event list with randomized positions e.g. named as RDETX and RDETY.

In principle the Xe-line can be used for normalization of the gain correction except
in a few cases where the line is too weak (or there are too few counts).
But perhaps doing it per science window gives too large statistical uncertainty.
Can it be done per revolution? CAO will look into that.

On-request telemetry (SB): High Speed Line dumps and software parameter - integer
and float - can be extracted from the *.IPF files and converted into FITS.

/NJW & SB