Analysis of the startup anneal sequence in Nano-SV (Campaign 33a)
 
Abstract:  This report applies to the start-up sequence
conducted during the Nano-SV phase between IRS Campaigns
33 and 34.  Only anneal data were obtained for a quick
comparison to previous campaigns.  The arrays are all
behaving in line with previous campaigns, with SH, LL 
and LH below their read-noise limit and SL above.  All
of the post-anneal SUR results are within one standard
deviation of the mean from the last three campaigns,
except LL, which shows less noise than before.
 
Analysts:  Lei Hao, Sergio Fajardo-Acosta and Greg Sloan
Procedure version:  2.7
Campaign 33a, startup anneal
YEAR 2006 DOY 243
CE1 used for all data collection
AORKEY =   20408576  for startup
Number of files = 64 for each
No known errors.
 
Steps 2 through 5 run OK.
 
Unruly.pro results for startup (IRS_001):
 
    Noise characteristics of the modules (in DN)
 
            SUR data                SRS data
            noise   pre     post    noise   pre     post
    Module  limit   anneal  anneal  limit   anneal  anneal
 
    SL     8.8     94.7      9.2    15.0      8.0     12.4
    SH     8.8     10.1      7.7    15.0     12.7     13.0
    LL    14.5     26.5     13.5    15.0     12.4     13.0
    LH    12.0     16.2     11.9    15.0     12.5     12.8

    Average noise SUR read-noise, Campaigns 31-33, post-anneal

    Module  mean   std. dev.   min   max
    SL       9.60  0.93        8.8  12.5
    SH       7.47  0.45        6.8   8.5
    LL      14.70  0.65       13.6  16.1
    LH      12.01  0.57       11.3  14.0

    The analysis applied the PIRAT procedure unruly.pro to
    FITS files generated by MIPL.
 
    For SUR data, the quoted values are the standard deviation
    of the read noise map.  For SRS data, these are the median
    of the standard deviation map.
 
    Comments:  The SRS data all look normal and well below
    the read-noise limit.  The SUR data, however, are a
    bit more problematic, with elevated noise levels in the
    pre-anneal sequences in all modules except SH.  Ignoring
    these data for the moment, the post-anneal results are
    in line with the post-anneal results from the three
    previous campaigns, indicating that the arrays are
    behaving as expected.

    The pre-anneal SUR read-noise results are higher than the
    maximum seen in the previous three campaigns for all of the
    modules.  In the case of LL and especially SL, they are
    especially severe.  We suspect the cause is a lack 
    of sufficient time before the anneal data are taken. 

    A careful look at the SL and LL ramps in the pre-anneal
    data indicate that they remain flat for the first several
    samples.  Then the ramps look more normal.  Good SL
    examples are EXPID 0 DCENUM 0, pixel (23,63) or pixel 
    (66,111) (IDL coordinates), both in the illuminated orders.
    For LL, we looked at EXPID 8, DCENUM 0, pixel (79,69).  In 
    fact, the entire column around that pixel looks depressed, as 
    if it were a latent.  These results are consistent with an
    insufficient pause before obtaining the anneal data after
    start-up.