Engineering 2011JanFeb
Daily Log:
2011 January 25:
FB: began pumping on CHAMP. Set up plan with Rafael for the alignment tool. Restructured the CHAMP directory on wolverine and updated the wiki instructions for CHAMP quickstart. Tracked power failure for CHAMP mounts to a disconnected cable on the MAXO box.
2011 January 26:
FB: there is an annoying gtk setting issue which makes running any gui remotely on anything other than OS X a problem. I've tried modifying the gtkrc on wolverine but to no effect. Worked more on the Alignment GUI, cleaned up the code to remove deprecated function calls and warning from IRCam gui. No more need to press enter to activate the channel (R1, L2, ...) selection.
2011 January 27:
FB: cooled the dewar, final pressure 6e-7. Started working on astropci code. Gone to JDM presentation at Caltech in the afternoon.
2011 January 28:
FB: refilled dewar, modified IRCam GUI to add automatic channel and quadrant selection. Started alignement of CHAMP table. Moved jdm_opds into a configuration file.
2011 January 29:
JDM + FB : debug of the spooler + saving data through rt_pipe. Needed to use sudo ./spooler... Found internal fringes on W1W2 B1B2, using the ESP RETRO and then delayline retros. See Champ manual for opd values.
2011 January 30:
JDM: worked on generating opd map.
JDM: replaced a bad PZT cable for PZT 3 (on beam 5) with a new one from piezojena, we still have one spare in the mirc cabinet.
FB: Test of the repeatability of the pickoff mirror positions. Fringes initially at 3.3931 on ESP. GUI says 210175 offsets on both IR1 and IR2 (doubtful). Refresh does not seem to work...
Homing all. Flux drops from 100+ to 20... Moving back to 210175. Nothing at +/- 0.3 mm.
Homing all the mounts, Home = against the platform.
(in all the following, closure BC = 0)
IR1 0 IR2 0 : Fringes at 3.50 mm.
IR1 100000 IR2 100000 : fringes at 3.4539 mm
Test of ESP repeatability:
Home + Back to 3.4539 : ok Go to 4.45 then 2.45 then 1.45 then 3.45:ok
TESTS OF IR1 AND IR2 USING DIRECT "MOVE TO"
Test of IR1 repeatability (waiting ~ 10 seconds between each move)
From 100000, move to 50000 and then back to 100000: ok
Move to 200000 then back to 100000: ok, within the fringe range.
50000, 200000, 100000: ok
Test of IR2 repeatability (waiting ~ 10 seconds between each move)
From 100000, move to 50000 and then back to 100000: ok
Move to 200000 then back to 100000: ok
50000, 200000, 100000: ok
REHOME IR1 + 100000 : ok
REHOME IR2 + 100000 : ok
TESTS OF IR1 AND IR2 USING PADDLES
Step size = 10 microns.
IR1: Relaxed clicking down to 95000 then back to 100000: ok
IR1: Nervous clicking down to 95000 then back to 100000: ok
IR2: Relaxed clicking down to 95000 then back to 100000: ok
IR2: Nervous clicking down to 95000 then back to 100000: ok
Home All (IR1+IR2) + Back : ok
CONCLUSION: IR1 and IR2 seem to be ok.
With IR1 and IR2 platforms aligned with the mounts:
Offset IR1 175910
Offset IR2 172160
Fringes ESP 3.1540
FB: tried to align on B3-B4 but had a hard time moving offsets to get the spot in.
2011 January 31:
FB: goal = generate opd maps for B3-B4 and more
Used offsets row =8 and col = 20.
Homed pickoff mirrors + retro.
Manually aligned the mounts with pickoff mirrors, result:
IR3 173123
IR4 172835
Found fringes at 3.056 mm (double spot for R4, updated wiki).
Recorded data for opd maps, with R4 and L4 fringes.
Repeatability tests:
Home IR3 and IR4 and back: slightly off by 0.03 microns.
Moving only IR3, ESP RETRO to the max amplitude fringe, home then back: ok
Moving only IR4, ESP RETRO to the max amplitude fringe, home then back: ok
IR3 50000, 200000, current : off by 8 microns (0.004 x 2)
IR4 50000, 200000, current : off by 4 microns
IR3: 20000-200000-20000-200000-20000-200000-current: ok
IR4: 20000-200000-20000-200000-20000-200000-current: ok
Nervous paddling test: IR3 ok, IR4 ok.
CONCLUSION: IR3 and IR4 ok
Now doing B4B5.
IR4 172835
IR5 210102
All beams involved in R5 and L5 have multiple peaks.
Procedure I followed for alignment:
B4, R5, T1H bottom spot
B5, L5, Dichroic 5 brightest spot (at the right)
B5, R5, BC4-5 bottom spot
Found fringes at 5.70 mm. Chose the brightest spots (bottom spots for R5).
IR5 repeatability test failed: no fringes when coming back. The platform stayed stuck somehow, and the 'mount and platform aligned position' was not physically possible anymore. Rehoming manually + using the gui seems to have fixed this, however the fringes moved to 5.49 mm. Something to remember when using IR5.
Now doing B5B6
Need to be wary of CHARA cables, lying around in the beams... B6 was partially obscured.
Found 4 spots (2x R6 + 2x L6) from the start. All beams involved in R6 and L6 are very clean.
IR5: 210017
BC: 0 ? Right choice ?
Found fringes (after a long search) at 10.3610 mm.
2011 February 1:
Now doing B6B1
IR1 175910
BC: 58440 (mount/platform aligned)
B6 R1 T2L top spot
B1 L1 Dichroic 1, spot looks messy but single peak, I've aligned on the top of it
B1 R1 BC61, elongated spot, top peak, brightest
Scanned 10 mm, did not find the fringes.
Retrying with BC at 0. Scanned 10 mm and still could not find fringes
Moving on to B1B2
Got fringes at 3.15 but only on R2. Had to realign to get all 4 spots in the fov.
Now I've got the spots, but fringes have disappeared.
Found fringes later at 3.159 mm but different spot than noted previously.
B1 R2 top spot
B1 L2 top spot
B2 R2 top spot
B2 L2 top spot
Test of RETRO search (MUCH faster than manual). Using minimum acceleration = 0.1, vel=0.05 is too fast, vel=0.02 works. Role of acceleration is difficult to quantify, so I tried different intervals. Fringes were in 3.15. 3.00 to 3.30, ok, 2.0 to 4.0 ok. This makes scanning by intervals of 2.0 possible, each such interval takes 100 seconds.
Moving on to B2B3
IR2 and IR3 in fiducial positions (see above for values).
Found fringes at 6.67 mm.
B2 R3 bottom spot
B2 L3 was a single spot
B3 R3 bottom spot
B3 L3 bottom spot
2011 February 2:
Started working on writing data with rt_pipe.
Put all the current working software under ~/CHAMP/dev on devel@wolverine.
Cleaned up some of the directories (put backups in a Backup directory). Updated the CHAMP manual wiki page.
Started looking into creating camera profiles for IRCam as well as defining flux scaling/contrast controls.
2011 February 3:
Finished cleaning up, committed code to chara CVS and UM SVN.
Finished the function to write data through pipe, using some buffering too. This may require more testing (what happens if the user tries to save overlapping chunks of data ?).
Debugged the state machine "on paper", need to confirm this through actual practical testing tomorrow.
Added the display average and variance of FT estimators to the FT gui, tested on fake fringes.
Started testing the IR5 actuator: a small grain of something (sand ?) was sometimes blocking the platform. New tests shows that the repeatability is ok. As a consequence the value for the fiducial position of IR5 has to be revised: I find a value similar to the others, around 174000.
2011 February 4:
More work on data writing as there seem to be a bug somewhere. Also reorganized it. Fixed part of the spooler code, which was expecting frames and not dithers. Commented a lot of lines in the spooler as some pointers to structures were not allocated but still used !
In some cases the spooler claims to receive correctly the data but rt_pipe says otherwise. Probably a problem due to the use of an incorrect sizeof() somewhere. Strangely does not occur every time...
Also prepared the layout for the wraparound (unfortunately most of the code written previously has to be tweaked).
Now solved the spooler problem. Installed Fv on lothlorien to check fits files. Analysis of the files revealed a few problems (needed to start a next dither, unsigned char pointer conversion, etc.). Mostly fixed this tonight.
2011 February 5:
Found yet another bug in the spooler, but could not locate the precise place where it is happening. Looks like a buffer overflow of some sort. As this overwrote my table names, I've made the table larger (hack). Will need to work on this back at Michigan as the full code needs to be reviewed. I suspect a problem when failing to contact sockman. Unfortunately a lot of the spooler code has this kind of bugs (DHB had to hack things badly to make it work).
Began testing the new opd map John made from the results we got before. Realigned for B1B2. Fixed few oversights in the code (values at wraparound pixels uninitialized). Began tracking and recording data. Added other tables to the fits. Tried to correct the display for the average opds over time (scaling bug). Carts got lost somehow but got them back by homing. Finally reached a stable state overall where I can begin to record nice fringes.
Plan for tomorrow: get quantitative values on every performance.
Test search speed with the new automatic threshold
Try to improve/adapt the transitions numbers
Record opd estimation performance (+ phi and parabola delays) in coherencing, cophasing modes, with and without coherent integration.
Maybe work on the gain of the OPD/DL ?
2011 February 6:
B1 became disaligned during the night. Ended up redoing the full alignement. Fringes at 3.1 mm.
I concentrated on debugging the FT gui this morning as the bugs make working with it difficult.
New opd map seems to have a bug. Will post screenshot later. This makes the amplitude threshold code fail.
Flux vs time shows a peak reoccuring, maybe due to the bad wraparound code ? Cannot see any peak on IRCam.
Phidev vs time makes the ft gui crash. Investigating.
FT gui just outputs too much text to the terminal, slowing it down. Reduced the amount.
Selecting no plot crashes the gui.
Fixed a lot of small issues. OPD maps seem to be skewed.
