Commissioning 1 Flight 2 log analysis

All these data are on Dropbox "everything EXES/docs/commissioning/LogAnalysis/Flight2Notes.* as well as pdf's of each graphic below.

1.  Summary of entire flight as best can be reconstructed from notes, logs and integrated comments from MCCS.

Description	Start	End	Duration
Flight	04/10/14 02:25 AM	04/10/14 11:40 AM	09:15:06.13
Setup leg	04/10/14 02:56 AM	04/10/14 03:18 AM	22:14.87
Turn to leg 5	04/10/14 03:18 AM	04/10/14 03:19 AM	00:46.67
Leg 5, Alpha Cma	04/10/14 03:19 AM	04/10/14 04:29 AM	01:09:46.99
Turn to Leg 6	04/10/14 04:29 AM	04/10/14 04:31 AM	01:56.29
Leg 6, Alpha Tau	04/10/14 04:31 AM	04/10/14 05:17 AM	46:11.26
Turn to Leg 7 then 8	04/10/14 05:17 AM	04/10/14 05:27 AM	10:03.84
Leg 8, Ceres	04/10/14 05:28 AM	04/10/14 06:30 AM	01:02:07.46
Turn to Leg 9	04/10/14 06:30 AM	04/10/14 06:35 AM	05:23.03
Leg 9, Jupiter	04/10/14 06:35 AM	04/10/14 07:38 AM	01:02:56.57
Turn to Leg 10	04/10/14 07:38 AM	04/10/14 07:43 AM	04:49.33
Leg 10 or 11, Alpha Lyr	04/10/14 07:43 AM	04/10/14 09:06 AM	01:23:26.73
Turn to Leg 13	04/10/14 09:06 AM	04/10/14 09:19 AM	12:55.69
Net FAIL & Recover	04/10/14 09:19 AM	04/10/14 09:40 AM	20:50.53
Leg 13, Delta Vir	04/10/14 09:40 AM	04/10/14 09:58 AM	17:45.51
Turn to Leg 15	04/10/14 09:58 AM	04/10/14 10:04 AM	06:11.88
Leg 15, AFGL 2591	04/10/14 10:04 AM	04/10/14 11:24 AM	01:19:36.19
Turn to home	04/10/14 11:24 AM	04/10/14 11:40 AM	16:02.71

I believe we set the time to GMT on the PC for flight 2, which should also be UTC time.

2.  Summary of each non-dead flight leg and how much time spent between events such as MD signaling we are "On Leg" and TO signaling we are "Ready".  Alpha Lyr, Ceres and AFGL 2591 the start of flat field file is actually a couple or three files of data into the leg. Those are usually "stares", my guess is that they are used to make sure we are at the proper wavelength.

TOL	Total Time On Leg
TLR	Time between On Leg and Ready
TRF	Time between Ready and Taking Flats
TD	Taking Data
TRS	Time between Ready and First Science

Leg	TOL	TLR	TRF	TD	TRS
Setup leg	1335	749	0	706	0
5: Alpha Cma	4187	84	109	3736	302
6: Alpha Tau	2771	93	37	2523	302
8: Ceres	3727	79	8	3581	127
9: Jupiter	3777	112	26	3593	319
11: Alpha Lyr	5007	113	776	4007	1019
13: Delta Vir	1066	2026	137	819	411
15: AFGL 2591	4776	163	58	4383	299
Average	3330.75	427.38	143.88	2918.5	347.38
SD	1483.88	684.35	259.86	1431.84	300.39
median	3752	112.5	47.5	3587	302

3. For each non-dead leg, show time elapsed for all peakup, flats or science data. Other includes files labeled in the EXES Notes files as adjusting focus as well as files that were taken between end of one leg and official first flat of the next leg.

TOL	Total Time On Leg
TS	Total Time doing Science
TF	Total Time doing Flats
TP	Total Time doing Peakup
Other	E.g. focus

Leg	TOL	TS	TF	TP	Other
Setup	1335
5: Alpha Cma	4187	1476	128	232	2351
6: Alpha Tau	2771	1010	226	763	772
8: Ceres	3727	1487	412	838	990
9: Jupiter	3777	2853	84	0	840
11: Alpha Lyr	5007	1964	398	769	1876
13: Delta Vir	1066	273	233	183	377
15: AFGL 2591	4776	1549	810	1165	1252
Average	3330.75	1516	327.29	564.29	1208.29
SD	1483.88	795.08	246.05	426.49	685.35
median	3752	1487	233	763	990

4. These two go together. One shows the elapsed time between the close of a file in one wavelength configuration and the opening of the next file in the new wavelength configuration.  The other shows the "delta wave0".  The table shows the wave0 transition values (from / to).

I think these plots are useless or at least confusing.  For example, the transition between 717.06 and 1548.56 happens in one or two steps. One of those steps consists of "move 2 402363 7500 2000" (echelle) which according to what I can tell in the logs takes almost exactly 1 minute co complete and all the other motor moves of that same configuration file load takes about 1 minute 50 seconds to complete.  So maybe these "statistics" are too messy.  To really understand it has to be analyzed on a case by case basis based on the moves of all motors between the transitions.

Leg	Time (seconds)	Wave difference	From wave	To wave	Transition time per inverse cm (seconds)
setup to acma	251	-831.5	717.06	1548.56	0.3
atau to ceres	221	818.26	1548.46	730.20	0.27
atau to ceres	135	-0.83	730.20	731.03	162.65
ceres	229	-573.86	731.03	1304.89	0.4
jupiter	176	573.86	1304.89	731.03	0.31
jupiter to alyr	370	-902.09	731.03	1633.12	0.41
alyr	55	-3.37	1633.12	1636.49	16.32
alyr to dvir	2281	88.03	1636.49	1548.46	25.91
dvir to afgl 2591	628	-88.03	1548.46	1636.49	7.13
afgl 2591	150	150.04	1636.49	1486.45	1
afgl 2591	192	770.51	1486.45	715.94	0.25
Average	426.18				19.54
SD	633.42				48.21
median	221				0.41

Dropping longest time to change per unit (2281/88 and 135/0.83) and two shortest transitions (55 sec and 68 sec)

Average	280.86				1.4
SD	168.69				2.54
Median	221				0.4

Noise, Signal v. Bias in DAOF PIF

Just before Line Ops I ran a data series at varying bias and illumination levels to verify that we hadn't somehow encountered excess noise in the transition from Ames to DAOF.  Overall, the results look pretty good.

Here is a plot of raw noise in a set of Fowler-4 frames taken at different bias levels.  At each bias, the Fowler-4 set was repeated five times, and a standard deviation was calculated for each pixel.  The plot shows two sets of results.  In the higher line, all five repeats are included in the calculation.  In the lower line, the first and last repeats are omitted.

I'm looking into what the reference pixels can and can't do to reduce the disparity related to tossing the first and last scans.  For present purposes I think the lower curve is useful in demonstrating the noise floor of the overall system.  Absent any photon noise contribution we can get down to about 24 electrons rms in a Fowler-4, which would scale to 17 electrons in a Fowler-8.  The MIRI instrument team has reported a Fowler-8 number under similar conditions of about 15 electrons, so the noise floor of the EXES electrical system is right in the ballpark.

If you feel like messing around with the data, I've placed the script that generated the values for this plot in EE/software/idl_stuff.  It's called batchfowlernoise.pro and has switches for different ways to handle reference pixel subtraction, reference pixel location, and whether or not to skip the first and last scans. Let me know if you want to use it and have questions.

Latency in reset frames for some flight settings

In some ramp sequences, areas that received moderate levels of flux would reset to a background value *higher* than the usual background of ~11000 ADU.

Here's one example:

gl2591.flat.40101.fits

This is a flat field at a hi-med setting at 1636cm-1, taken with the pattern "N6 S8 N5 D0".

On blackbody, the first frame of the ramp looks normal, with the background level around 11000, and some small accumulation of bb flux subtracting from that value.
On a bright part of the blaze, median(image(410:450,600:700))= 10860 ADU

At the start of the ramp: median(subim)=10860

At the end of the 1st ramp, median(subim)=8710

At the end of the ramp, frame #14, the counts in the box = 8710 ADU

After reset following ramp #1, median(subim)=11164

And then on reset, frame#15, the median counts in the box are=  11164 ADU