New target class strategy for TRACE
                 automatic exposure control (AEC)

                    version 1.2, 1999 May 18
                        C. C. Kankelborg


                         **** NOTE ****

  Recent AEC modifications are described in a companion document:
    http://icarus.physics.montana.edu/TRACE/AEC/AEC_latest.txt

                         **************


There are two fundamental dilemmas:

(1) stability vs. responsiveness. 
        AEC can be made 'stable' against oscillations, and fairly
        resistant to radiation. Alternatively, AEC can be made 
        'sensitive', so that it will react quickly to a small area
        brightening event.

(2) long vs. short exposures
        AEC over/under exposure thresholds specify the DN exposure 
        level to converge to.

Taking these into account, there are four possibilities, which map
nicely to our four target classes:

stability/exposure      target class     typical AR   typical QS
----------------------------------------------------------------
stable/normal          class0 (top1)        20s          55s
sensitive/normal       class1 (top2)        17s          46s
sensitive/short        class2 (top3)         8s          23s
stable/short           class3 (top4)        10s          27s

The 'typical' exposures above are estimates for AEC convergence in 171A
passband. 

By 'sensitive' vs. 'stable' I mean:

Sensitive (modeled after our present top2/class1):
(1) Overexposure criterion based on the brightest 1/8 % of the pixels
(sensitive to small-area brightenings)
(2) Rapid exposure adjustments:
        6 steps down (a factor of 2.8) for overexposure.
        3 steps up (a factor of 1.7) for underexposure.

Stable (based on last week's experimentation with top1 & top4 [1]):
(1) Overexposure criterion based on the brightest 1% of the pixels.
(2) Moderate exposure adjustments:
        2 steps down (a factor of 1.4) for overexposure.
        1 step up (a factor of 1.2) for underexposure.



AEC parameters
--------------
Of greatest interest to science planners are the default exposures. The
other parameters listed in the tables are defined below.

Class0 (top1) -- stable AEC, normal exposures

Wavelength   Default   Up/Down   Overexposure   Underexposure
 171ao        16.4 s    1/2        1%  >256       1%  <191
 195ao        19.5 s    1/2        1%  >256       1%  <191
 284ao        55.1 s    1/2        1%  >256       1%  <191
 1216          3.4 s    1/2        1%  >512       1%  <383
 WL            6.4 ms   1/2        1% >2048       1% <1535
 1600          1.2 s    1/2        1%  >512       1%  <383
 1700          4.9 s    1/2        1%  >512       1%  <383
 1550          9.7 s    1/2        1%  >512       1%  <383


Class1 (top2) -- sensitive AEC, normal exposures [2]

Wavelength   Default   Up/Down   Overexposure   Underexposure
 171ao        11.6 s    3/6       .125%  >352     1%  <159
 195ao        13.8 s    3/6       .125%  >352     1%  <159
 284ao        39.0 s    3/6       .125%  >352     1%  <159
 1216          2.4 s    3/6       .125%  >704     1%  <287
 WL            4.8 ms   3/6       .125% >2336     1% <1343
 1600          0.86 s   3/6       .125%  >768     1%  <255
 1700          3.4 s    3/6       .125%  >704     1%  <287
 1550          6.9 s    3/6       .125%  >928     1%  <223


Class2 (top3) -- sensitive AEC, short exposures [2]

Wavelength   Default   Up/Down   Overexposure   Underexposure
 171ao         5.8 s    3/6       .125%  >192     1%   <95
 195ao         6.9 s    3/6       .125%  >192     1%   <95
 284ao        19.5 s    3/6       .125%  >192     1%   <95
 1216          1.2 s    3/6       .125%  >384     1%  <159
 WL            4.8 ms   3/6       .125% >1664     1%  <959
 1600          0.43 s   3/6       .125%  >384     1%  <127
 1700          1.7 s    3/6       .125%  >384     1%  <159
 1550          3.4 s    3/6       .125%  >544     1%  <127


Class3 (top4) -- stable AEC, short exposures

Wavelength   Default   Up/Down   Overexposure   Underexposure
 171ao         8.2 s    1/2        1%  >128       1%   <95
 195ao         9.8 s    1/2        1%  >128       1%   <95
 284ao        27.6 s    1/2        1%  >128       1%   <95
 1216          1.7 s    1/2        1%  >256       1%  <191
 WL            4.8 ms   1/2        1% >1632       1% <1215
 1600          0.61 s   1/2        1%  >256       1%  <191
 1700          2.5 s    1/2        1%  >256       1%  <191
 1550          4.9 s    1/2        1%  >256       1%  <191


Definition of parameters
------------------------
Default - default exposure
Up/Down - number of steps to adjust exposure when under/overexposed
        (one step is a factor of 2^0.25 ~ 1.2).
Overexposure - overexposure criterion, written as PAH >INH. For
        example, .125% >256 means that if the top 1/8 % of pixels
        are all greater than 256 DN above pedestal, then the
        image is deemed overexposed. [3]
Underexposure - underexposure criterion, written as PAL <INL. For
        example, 1% <191 means that if the dimmest 99 % of pixels
        are all less than 191 DN above pedestal, then the image 
        is deemed underexposed. [3]


NOTES 
-----
[1] The means of stabilizing AEC for TRACE was originally articulated
in the following document: 
        http://icarus.physics.montana.edu/TRACE/AEC/aec_plan.html

[2] Ordinarily, the AEC may be stabilized by constraining the ratio of
overexposure to underexposure thresholds as follows [see 1]:

            INH/INL > 2^(0.25*UP) = 1.19  for Class0 & Class3
                                    1.68  for Class1 & Class2
                
However, this assumes PAH = PAL. Since the 'sensitive' AEC settings have
PAH = .128% and PAL = 1%, the above factor must be multiplied by a
typical ratio of pixel values (or rather, DN above pedestal) for the
.128% and 1% bins. Empirically, this ratio is found to depend strongly
upon wavelength:

        wavelength        DN@PAH / DN@PAL   INH/INL (Class1 & Class2)
          EUV (8x8 binned [3]) 1.19              > 2.00
          WL                   1.03              > 1.73
          1216                 1.44              > 2.42
          1600                 1.76              > 2.96
          1700                 1.38              > 2.32
          1550                 2.54              > 4.27

The above values were derived from active region images taken in the
northern activity belt by CDS synoptic, 1999 March 14. Note that the
above correction does not guarantee stability, because the DN ratios
depend on solar conditions.

[3] For EUV images, AEC uses the dosimeter image (binned 8x8).