Lens wheel config, CD023

The only change made between CD022 and CD023 was swapping the ZnSe lens with a CaF2 blocker. The pupil images from the ZnSe hadn't been giving us any new information that we didn't get from the KRS5 lens but regardless I needed the mounting rings as much as the space.

Angle     Optical Component                          
-120 deg  mirror                                     
-60 deg   1% neutral density filter                  
0 deg     25mm KRS5 lens                             
60 deg    Open                                       
120 deg   CaF2 window                                
180 deg   38mm KRS5 lens

Proposed Filters for CD23

This is how I want to load the FW/DW for the next cooldown. Anyone see issues?
I tried to take into account cycle 4 awardees and engineering tasks. 
Spaces where a change needs to be made are in bold. The purpose for each filter is noted with a comment with either the GI name or "engineering".

F1 OPEN            
F2 BBP5500  --- Sargent, Indriolo
F3 Reading 6.2, tilted,  [Remove 18a]--- Engineering
F4 Reading 6.2, flat--- Indriolo, Engineering
F5 N04679 --- Engineering
F6 N07466 --- Encranez, Tsang, Aoki
F7 BP14100 --- Orton/Engineering
F8 Soifer --- Harper
F9 CO2 filter from UT,  [remove BP4990] --- engineering
F10 N07053,  [remove BP6000] --- Indriolo
F11 N07901, [remove OCLI S] --- Indriolo, Orton
F12 N05340 --- Goto

D1 BP13465 
D2 BP6000, [N07053 to FW]
D3 BP6600
D4 N05340

There will be space in CD24 for reinstalling 18a & OCLI S for Rangwala's science. Are there other projects that could be carried over? Of course I am aware that not everything will be scheduled in March, but we can optimize again before shipping.


UPDATE 11/19/15

Only one change was made, adding the tilted Reading 6.2um filter. The wedged washers were made by E. Quigley with a 2.5 degree tilt. Loaded with same side facing out as for the flat reading filter +MgF2 in slot 4. 

post flight series TAAS info

After our Sep flight series, Mark and I got EXES on the TAAS.  After a YH visit and realizing we needed to "home" the thing, we saw the FCLS spot.

We used the FCLS 200um aperture and a temperature of 150 K.  The robofocus was at 6800.  We didn't play with that.  EXES was imaging through the N04679-4 filter and the ND filter.  We ran with N0 D0 T0 pattern.

We first put the FCLS at the HIPO Boresight Position (more recent of the two dates - not much difference between them).  We imaged with open/open/open and then the slit inserted.  We then moved the FCLS to put the spot on the slit.  This establishes our boresight in FCLS coordinates.

We then adjusted the filter/decker/slit wheels to make things rounder.  This ended up being a 2.5 degree motion.  We saw 4 degrees motion was too much, but there wasn't much difference between 2 and 3.

With the new f/d/s position, we again imaged the field, the slit, and then moved the FCLS spot to come through the slit.

The pictures below document these different cases.  Each image is a combination of open and thru-slit so that the slit can be seen relative to the full field.

I used a circle drawing routine in IDL to quantify the radius and center of our circular field.  I'd guess this is accurate to about 5 pixels.  It'd be nice to come up with an automatic routing to do this.

Finally, the IDL journal file I used is in eE/analysis/cd22/taas/getimages.jou and the data are all in cd022_2015sep/5.


First image has FCLS at HIPO Boresight: 45.75, 31.112 (yellow cross).  This lands (for this echelle angle) on EXES pixel 462, 445.  The green circle roughly encompassing the EXES Field is centered at 570, 498 and has a radius of 369.

Second image has FCLS on our slit.  FCLS coordinates are (x,z) = (48.35, 24.4) and the EXES pixel is (x,y) = 596, 513.  

Third image is after we moved slit/decker/filter by +2.5 degrees (-2.5 for decker).  It looked like 2 and 3 degrees changes were equally round, but 4 degrees was starting to clip.  The FCLS spot is at the same place (x,z) = (48.35, 24.4).  The spot on EXES detector is one pixel different: (x,y) = 597, 513.  The new center of the circular FOV is at EXES pixel (x,y) = 491, 482 and the radius is 398 pixels.  Unclear how much of the radius could be caused by different level for where the circle starts.

This last image has the FCLS at (x,z) = 45.95, 27.9 and the spot comes in at the EXES detector (x,y) = 512, 493.  

Some Distances:
Between FCLS spot on slit in flight configuration and after moving s/d/f:

In FCLS coordinates:  del_X = 2.4, del_Z = -3.5, => 4.24 (mm) => 17.7" (4.16" per mm from TAAS to SOFIA).
In EXES coordinates: del_X = 85, del_Y = 20; => 87 pixels => 17.9" (use 0.206" per pixel)

Not sure if this was all the boresight motion we had or not.   If we had the notes from before other flight series, we'd be able to check compared to FCLS.

Between the center of the circle in flight and after moving s/d/f:

In EXES pixels: del_X = 79.4, del_Y = 15.7, => 81 pixels => 16.7"

How to start quicklook

We currently run Quicklook on Zion under the EXES account.  It requires IDL.  We also have to have the data disk mounted by zion.

From finder, use Go -> Connect to server 

server address: smb://10.4.2.2

usually choose all volumes that are available.  data is the most important.

Start an xterm using Xquartz

load in the IDL set up variables

prompt% source .profile

change to the quicklook directory

prompt% cd Desktop/quicklook/ver2.0      (for today)

start IDL

prompt% idl

within IDL, start the quicklook

idl>  @startup

a long string of procedures compile and windows open.  Look for the dialog box to pick the data directory.   Can type directly into the top field.  Something like

Directory
/Volumes/data/CD022/cd022_2015aug/25

Can also hit return after entering a part in the Directory field to see further options.  Last directory is clicked in the "file" field.

The main display will have a mostly white screen until you click on something other than atmo.  I'll try to fix this at some point.

Minimize the scan window and open the dels.  I find it useful to open a couple others for editing procedures and operating the sandbox (by hand analysis)

idl> $xterm &

These will be in the data directory.  cd one to the quicklook subdirectory.

arrange the windows as you like.  Here's what I do.

EXES startup procedure

Comments will be used to update this post. Any reason not to put in passwords?



Turn on the CWR power, and EBOX power supply. Lakeshore 12 mW power, to setpoint, 7.000K

Turn on bryce and zion mac minis on the PI Rack.

After bryce is at the login prompt, you will probably be unable to move the mouse or type with the keyboard. If this is the case go pull out and reinsert the mouse/keyboard USB from the back of zion.

passwords
bryce: m*****s
zion: R*******

On bryce, open up CORD terminal. Go to gus, ipadd:10.xxxxxx. The login is gus, 4s****.

On CORD/gus, open the labview program: Desktop: reed/Labview229/____.vi

At this point, go to CORD menu and open a full screen session- you'll have to log in again, but sometimes this is necessary to get labview to initialize.

On the CORD/gus desktop, double click on the ACQ icon (Mighty Green Triangle)


On bryce, open ~/Desktop/EXES_GUI/startgui.??

Should open within ~5 sec. If it takes 20 sec, then it hasn't linked up to labview. In which case it will inform you that it found all motor mechanisms out of their last positions.

It will often find SLIT/DECKER in slightly off positions. This is due to some voltage drifts/winding in the mechanisms and pots. Maybe. but it is normal.


From the GUI, go to the initialization panel, and press "connect", and then "machwinit". If machwinit is successful, it should tell you that it loaded 80 opts tables or something similar. "no echo from hardware" could be disconnected Ebox, or fiber to the ebox ... something with the ebox.

Pressing "Update" on the inits page will tell you the next file index and directory will be.

From the instrument control panel, the waveno0 and order fields will be empty on startup. If you know the echelle order that the system was left in you can enter it in the spinner and the hit the refresh button. Otherwise, you can enter a new waveno0 and order and hit "setup" for the GUIs recommended mechanism positions for that combo. Make adjustments and hit "Move all".

Setup files are loaded from the Obs Setup screen with a button at the bottom.
Hit load, the directory will be something like "~/setupfiles_jul-aug-2015" and then directories are listed by flight number. Pick the one you want and it will import all the mechanical positions and observation parameters (OTs for flat, nod, NODN, NINTs).

It will not know about necessary changes to: dispersion driver, frametime, subarray modes, V_bias.

Hi-lo settings for CD22 flights

Testing the 3 HIGH_LOW settings for 03_0126 (Lee).

Summary: A similar setup for all three settings:

1. Slit6 (3.2")
2. Echelle order +2
3. 14-pix BP13465
4. Filter: OPEN

On the 260K blackbody, N0 D0 will work without issue (<3000 ADU).
The GUI setup in Waveno0 has an angular offset -- to get the WN you want at the center of the detector offset the waveno entered in the GUI by -5.6 cm-1.

Setting 1: 756.9cm-1, 2nd order, HIGH-LOW

Instrument Setup: Slit 6 (3.2"), 14-pix decker dash BP13465, FILTER=OPEN, EDEG=22.5611 degrees
On room: OT= N0 D0 T0, Frametime=1.0s

The waveno0 is off by about 5.6cm-1. We will have to adjust to best setting.
flux(room)/flux(260 K) ~ 1.6. Therefore, ND pairs on the blackbody will be comfortably below 3000 ADU.

756.9 cm-1, 2nd order, gas cell spectrum;
Real waveno0 appears to be ~ 762.56 cm-1

Histogram of pixel values in 11000-frame(*,*,1) of ND on ROOM.

Setting 2: 768.5 cm-1, 2nd order, HIGH-LOW

Instrument Setup: Slit 6 (3.2"), 14-pix dash BP13465, Filter" OPEN, EDEG= 22.1781 degrees

On room: OT= N0 D0 T0, Frametime=1.0s 

The waveno0 is off by about ~6 cm-1. We will have to adjust to best setting.

flux(room)/flux(260 K) ~ 1.65. Therefore, ND pairs on the blackbody will be comfortably below 3000 ADU. 

768.5 cm-1, 2nd order, gas cell spectrum;
Real waveno0 appears to be ~ 774.xx cm-1

Histogram of pixel values in 11000-frame(*,*,1) of ND on ROOM.

Setting 3: 745.6 cm-1, 2nd order, HIGH-LOW

Instrument Setup: Slit 6 (3.2"), 14-pix dash BP13465, Filter" OPEN, EDEG= 22.9468 degrees
On room: OT= N0 D0 T0, Frametime=1.0s

The waveno0 is off by about ~6 cm-1. We will have to adjust to best setting.

flux(room)/flux(260 K) ~ 1.65. Therefore, ND pairs on the blackbody will be comfortably below 3000 ADU.

Histogram of pixel values in 11000-frame(*,*,1) of ND on ROOM.

Gas Cell pressure broadened?

We may be unable use the available DAOF roughing pump to get the gas cell to low pressures necessary to see the true spectral focus. 

My usual procedure for pumping the cell is to put in C2H2, look at 729.5cm-1, 4th order spectra and pump down until the C^13CH2 lines disappear. I did so, and I've been finding disappointing R values for the spectra compared to in CD21. 

For this experiment I turned on the pump and opened the gas cell valve and took exposures at various times after starting. At the start the gas cell had already been pumped down (for 10 minutes at least), but was left untouched for an hour or so while I analyzed data. These are resolutions of fits to lines Q11-Q15 with the 1.9" slit. (Wavenumbers taken from default wavemap from pipeline. I didn't try to touch up the centering.



R (line1) R (line2) R (line3) R (line4) R (line5) time elapsed (min)
67312       71392       67032       73580       67660 0.0 min
72102       75955       70719       78602       71164 4.2 min
71317       76556       70699       77111       70998 6.5 min
71466       76373       70787       77540       70613 10.9 min
72578       75171       71248       78622       71367 17.0 min
73657       75134       71792       80275       72272 23.5 min

I think this suggests we haven't hit the true spectrograph width yet. We've never had to pump down this long ...

If we find another pump and try again that would be good. Regardless, I think we have to live with the results, since my earlier tests showed that we were at best focus on both high-med and medium modes (allowing for the compromise position we adopted in CD21).

UPDATE : 8/3/15-  Yes, the C2H2 lines were suffering pressure broadening.

MMcK used the turbo pump on the gas cell and discovered that the brass adaptor was leaking. 

He replaced it and took a new sequence of Q branch spectra as the cell was pumping down. 

Using file gascell.lab.00200.fits and fitting the lines in the same way as above, I got :

R (line1) R (line2) R (line3) R (line4) R (line5) time elapsed (min)

82205       82327       79504       83283       83683

83107       84126       79668       82536       82248 BEST R***

There were many files to choose from in the sequence, but there's a tradeoff for the quality between line strength, resolution and pumping time. Later I can check to make sure this is the ultimate resolution on slit4, but I think this proves the point that our performance hasn't changed since CD21.

***Update: 8/4/15  I checked the linewidths for all the files from Mark's pumpdown data. Of the ones with good line fits, the best R is file gascell.lab.00202.fits. Therefore, we can expect about R=83000 with slit4. 

Medium Resolution Settings for CD22 flights

N. Rangwala's requested medium mode settings have particularly high fluxes. We will have to take special measures to observe them while staying efficient.

The slit length needs to be at least 60" long in order to nod on source. This is about 300 pixels. The angle of the spectrum to the detector rows is arctan(0.095), so that an additional ~96 rows must be read out to get the slit length specified. Therefore, we'll work with 400 rows for these settings.
The command is:
1 1 152 200 0 1032 400 ectpat
1 ectgopat
(0 ectgopat returns to 1032x1024 full frames)

Setting 1: 
NGC 253 S(1). med.  2.4" wide slit.  wn0 = 586.55 cm-1.  order 3.  nod on slit for stepped map as above.

tested setup: 

587.0 cm-1, order 3, 18a, decker open, 1.9" slit (change from request)

Frametime = 0.5 sec, detector bias= 1.0 V (DC12, VDET= -3.5 V), OTPAT: D0 or D0 T0

Caption: 587cm-1, 3rd order, looking at room (295K); gas cell with 584.71 cm-1 H2O line.

Data must be taken with OTPAT: D0 (or D0 T0). Probably with many repeats, so that we spend a reasonable time per step. There is no detectable image difference between D0 or D0 T0. We will have to make sure 0.5s D0s save to the computer without glitches. 

Here is the histogram of pixel values on the room (295K). I subtracted the data frames from a D0 frame with the lens mirror in (dark). 

Histogram of pixel values for 587cm-1,3rd, looking at room.
We expect 1.5x less flux on 260K blackbody.

Using mplanck.pro, we should expect the 260K blackbody to have 1.5x less flux than the 295K room.

Setting 2: 
NGC 6240 S(1).  med.   2.4" wide slit.  wn0 = 573.00 cm-1 order 3 nod on slit.

tested setup: 

573.0 cm-1, order 3, 18a, decker open, 1.9" slit (change from request)

Frametime = 0.5 sec, detector bias= 1.0 V (DC12, VDET= -3.5 V), OTPAT: D0 or D0 T0

 

Caption: 573cm-1, 3rd order, looking at room (295K); gas cell with 571.29 cm-1 H2O line.

As above, this must be taken with OTPAT: D0s. 

Here is the histogram of pixel values on the room (295K), subtracted from a D0 frame with the lens mirror in (dark). 

Histogram of pixel values for 573cm-1,3rd, looking at room.
We expect 1.5x less flux on 260K blackbody.

Setting 3: 
NGC 253 S(2).  med.  2.4" wide slit.  wn0 = 813.74 cm-1.  order 4.  nod on slit for stepped map.  Plan is full slit width steps on either side of start position.  9 total steps

tested setup: 

814.0 cm-1, order 4, OCLI S, decker open, 1.9" slit (change from request)

Frametime = 0.5 sec, detector bias= 1.0 V (DC12, VDET= -3.5 V), OTPAT: D0 or D0 T0

OTPAT: D0, 814cm-1, 4th order, looking at room.
There are no water lines in this setting, so I do not show the gas cell spectrum.

This must be taken with OTPAT D0. Really counting on the 1.71x lower factor when we are using the 260K blackbody. 

Here is the histogram of pixel values on the room (295K), subtracted from a D0 frame with the lens mirror in (dark). 

Histogram of pixel values for 573cm-1,3rd, looking at room.
We expect 1.71x less flux on 260K blackbody.

Hi-med settings for CD22 flights

Some of the instrument settings required for the science in the Aug 2015 flight series have very high flux. We have to choose the best way to handle the flux level from the blackbody by : reducing the frame time (default 1.0s), selecting a narrower slit or using subarray mode.

The goal is to get the peak flux of the brightest parts of the order to ~8000 DN (3000 DN below 11000 dark level).

Setting 1

wn0 = 811.32 cm-1, order 4, high-med, Slit 6 (3.2")

for 03_0065 (Rangwala) NGC 1068 S(2)

instrument configuration

slit6, ddeg ~36.5, OCLI S, edeg=47.798 deg

811.3 cm-1, 4th order, looking at 295K room

For frametime=1.0 we get >6500 counts in N0 D0

For frametime=0.5 we get 4000 counts in N0 D0

Histogram of net pixel DN (11000-lastframe) for Frametime=1.0 (red) and Frametime=0.5s (green)

The scaling from 295K room to 260K is a factor or 1.72.

Solution: FT=0.65 sec, SLIT6, N0 D0

We want to keep slit6 because the sources are so faint. Straight Ds might compromise the data quality. Not known yet, so too risky. 

Also, maybe we can actually linearize up to 3500? Allowing us use a longer frametime.

Setting 2

wn0 = 584.8 cm-1, order 3, high-med, Slit 6 (3.2")

for 03_0065 (Rangwala) NGC 1068 S(1)

instrument configuration

slit6, ddeg=41.567, 18a, edeg=50.561

584.8 cm-1, 3rd order, looking at 295K room

Histogram of net pixel DN (11000-lastframe) for Frametime=1.0(red), Frametime=0.5 (green)

For frametime=1.0 we get 5600 counts in N0 D0

For frametime=0.5 we get 3000 counts in N0 D0

The scaling from 295K room to 260K is a factor or 1.5.

Solution: FT=0.75 sec, SLIT6, N0 D0

Setting 3:

wn0 = 1888.06 cm-1, 11th order, (03_0067, Goto).  2.4" wide slit (slit5)

instrument configuration

slit5, ddeg=32.95(+), N05340, edeg=62.0191

OTPAT: ~3000DN for N3 S22 N2 D0 while looking at room. Estimate 108 seconds needed on 260 K blackbody.

1888.0 cm-1, 11th order, slit5, gas cell with H2O lines

Setting 4 (GTO):

wn0 = 1778 cm-1, 10th order

instrument configuration

slit TBD, ddeg=32(+), BBP5500_6000, edeg=58.295

1778 cm-1, 10th order, gas cell with H2O lines

Setting 5 (GTO):

wn0 = 1748 cm-1, 10th order

instrument configuration

slit TBD, ddeg=32.7, BBP5500_6000, edeg=60.020

Setting 5 (GTO): 

wn0 = 1620 cm-1, 9th order

instrument configuration

slit TBD, ddeg=32.6, Reading 6.2um, edeg=57.120

Quick timing from Feb/Mar syslogs

Using Mike's CSV file from the syslogs from Feb/Mar, I looked a little at the numbers.  The CSV is in EE/analysis/cd21/timing/ along with my journal file and Mike's python.

DONE was returned 25% of the time faster than 5.156 seconds, 50% of the time faster than 6.281 seconds, and 75% of the time faster than 7.828 seconds.

We had a time out of 4 seconds.  Comparing the total of all the waits (what would have been elapsed time with no time out) and what we should have experienced (any wait longer than 4 seconds gets replaced with 4.0), find that

• total of all waits was 405 minutes over 5 flights
• total of our time-out adjusted waits was 222 minutes
• Difference is roughly 3 hours of waiting time.
• without a time out, the waiting time was 1.82X longer

Labeled Filter and Decker Wheel for CD22

Filter Wheel

F1 OPEN
F2 BBP5500+ CaF2
F3 18a + ZnSe
F4 Reading 6.2 + MgF2
F5 N04679+MgF2
F6 N07466+CaF2
F7 BP14100
F8 Soifer
F9 BP4990 + MgF2 (from ISP)
F10 BP6000 + MgF2
F11 OCLI S + BaF2
F12 N05340 + MgF2 (From ISP)

Numbers F 1-12 are etched in wheel underneath retaining plates.


Decker Wheel

D1 BP13465 + BaF2
D2 N07053 + CaF2
D3 BP6600+MgF2 (ISP)
D4 BBP5030+MgF2 (ISP)

Decker wheel filters loaded for CD22

Numbers for D 1-4, scratched into Al of decker wheel

Updated Throughput Plot

With some help from Kristin and Curtis I have made some changes to the throughput plot Kristin posted in December 2013.  In the plot below, I use her data, scaled for the factor of ~2.8 difference between the a-omega value used in the previous plot and the telescope-design value, which I believe is closer to the effective acceptance angle based on Bob's earlier post.

Besides that change, I've eliminated throughput values for each filter that fall below 60% of the highest throughput value calculated for that filter.  The lower values tend to be for angles far off the blaze in weaker orders.  My original method was to limit results to echelle angles +/- 10 degrees from the blaze, but I haven't been able to locate all the data needed for that approach so that will have to wait for later.

I've omitted the filter labels and color variations in the data point for clarity, but these can be added back as desired.  I hope to use this plot or something like it in the commissioning report slide deck.

Throughput calculations

Bob McMurray posting from Mark's account.

The throughput calculations performed by the IDL journal file for the various filters appears to have an error in calculating the A-omega value. The easiest way to see it is looking back up from the pixel area, where the A is a 25 micron square and the omega is the internal f10 beam. The file uses the pixel area and the externally measured angle, which is variously F18 or F16.66 or something in that range. If you use the f10 value here, and assume your FOV is filled by the blackbody, you can take the emissivity of the BB per cm^2-sr and just multiply by the conserved A-omega to get flux in photons/sec-Hz from the BB formula. This gives a factor of 3-4 more light entering the instrument, and therefore a throughput (detected electrons/photon) that is 3-4 times smaller than had been calculated. I think these numbers are in line with what we expect from the various efficiencies of the optics.

Resolving Power

I looked at a series of sky frames with different slits from the last flight.  The central wavenumber is around 746 cm-1.  The pair I eventually chose are at 749.4 cm-1 and separated by 5.4 km/s.  The weaker line is actually a double (strength 2.5:1, separation 1 km/s), but I chose to ignore that.

If perfect on an unresolved line, our resolving power should be 1.5, 2, 2.67, 3.5, 4.5, and 6 km/s.  Measuring the FWHM of Gaussian fits to the lines and trying to correct for the slit angle relative to rows/columns, I found values of 3.1, 3.1, 3.6, 4.0, 4.3, and 5.4 km/s.  Not certain if diffraction, or intrinsic line width, or errors in our optics are the cause for not reaching slit limits.   Note that we repeated the first setting, Slit 4, at the end of the series as we were climbing.  It was slightly narrower at higher altitude, suggesting there could be some altitude effect.

Will add something on the gas cells later.   They should be definitely unresolved.  Titan may be able to help, too.

Sky frame with slit 4.  Pair of ozone lines used for further analysis in box.  Line cut for fits at row 150. 

zoom on the two 03 lines used for fits.  Lines are 5.4 km/s apart.  Right hand line is actually a pair with strength ratio of 2.5:1 and 1 km/s separation.

2 Gaussian fits to the cross cuts of each slit width.  Note that we repeated the first setting (slit 4 - not shown) twice and I found that the second time (lower left) narrower.  Looking at lower right, where the Gaussian FWHM equals the line separation, I'd say the FWHM does pretty well - the lines are just barely resolved at that point.

Pupil Montage from Allan Meyer

Allan Meyer took our pupil images from lineups when we held the door fixed and adjusted the angle to see when we vignette.  This was from the second night of line ops.  The image and copy/paste of Allan's email are below.

Made a simple 3x3 montage of the 9 images, after subtracting the initial 'centered' image, and rotating each ~85 degrees counterclockwise.

The numbers added on are the difference between the TA elevation angle and the door elevation angle (fixed at 30 degrees).

However I should check the TA LOS position while we were doing this.  It was probably near center, but if not could have introduced a bias of a degree or two.

Assuming small LOS, looks like the limits are +4 degrees and -5 degrees with respect to the door.
So at lower door limit of 24 degrees TA can go to 19 degrees, and at upper door limit of 57 degrees TA can go to 61 degrees.

Detector Hamlet: To Throwaway or not to Throwaway

Curtis noticed while looking at the blackbody and gate valve that the detector often exhibited lower response after we filled the wells appreciably.  The first diagnostic was that for a fixed grey-scale level, the first frame of the second integration when we were looking at appreciable flux would have "white patches" at the pixels that saw the most flux.  The OTPAT below is N3 S1 N2 D0.

Frame0 during first integration with greyscale = 8000  to 11500.  Note pixels near center where flux is starting to accumulate

Frame0 during 2nd integration with same greyscale.  Note pixels with most flux (black above) are white here indicating fewer electrons in the wells than above or even general pixels.

Added a throwaway to the end of the OTPAT (N3 S1 N2 D0 T0) and saw that the behavior went away.

Took a series of 8 stares at blackbody (260K) with the N3 S1 N2 D0 pattern (file bb.sci.20412.fits) and with N3 S1 N2 D0 T0 pattern (bb.sci.20413.fits).  You clearly see the same offset in the D0 set for every integration after the first and you clearly see more normal behavior in the D0 T0 set.  However, if I do the ND subtraction as we typically do while observing for each of the 8 integrations in both of the sets and compare how these NDs vary depending on OTPAT, I see that only the first integration is different.

Top Panel: Row 600 for 8th integration with N3 S1 N2 D0 pattern (white) and with N3 S1 N2 D0 T0 pattern (red).  Clear difference especially at the highest flux level.  Bottom Panel: Each integration minus the 8th integration for Row 600.  White is N3 S1 N2 D0 pattern 1st minus 8th.  Green is for integrations 2 through 7 minus 8th.  Red is for N3 S1 N2 D0 T0 pattern 1st through 7th minus 8th.  

Since adding a throwaway costs us a frame each time through the OTPAT and we are always going to be doing like-for-like subtractions when doing the nod differences, I think we probably don't need the T0 patterns even though something funny is going on.  Flat's might be a different matter since the photon flux between black and sky is so different.  Comments?

Labeled photo of Filter wheel before changing CD19 configuration.

I think the title says it all.

CD 21 SDF Configuration (Commissioning II)

SDF Configuration for CD 21

Filter Wheel:
1) 18a + ZnSe (no change)

2) BBP5500 + CaF2 (no change)
3) Open (no change)
4) Reading 6.2 + MgF2 (no change)
5) N04679 + CaF2 (no change)
6) N07466 + CaF2 (no change)
7) BP6600 + CaF2    -- MgF2 blocker with BP6000

8) Soifer (no change)

9) BP4990 + MgF2

10) BP6000+ MgF2  -- (OCLI O removed and stored+ CaF2 paired with #7)

11) N07904 + CaF2 (no change)

12) BP14100 (no change)

Decker Wheel:

1) BBP8000 + BaF2 (no change)

2) BBP4700 + blocker(?)

3) N07053 + CaF2

4) BBP5030, no blocker

(please see CD 19 filter configuration post from 12/14/14 for additional information)