Door open again (2016-10-16)
After a period of inactivity at Obs. Vermes we could manage to get all the components up and running again.
One of the reasons was my winter over stay at Dome Concordia in Antarctica as a member of DC 11. I took care of the infrared telescope IRAIT/ITM and of the camera AMICA, wrote a new RTS2 driver for the mount and acquired data for a pointig model and of course I focused IRAIT/IT with rts2saf. Living and working at an altitude of 3227 m or better under a pressure of 630 hPa, temperatures as low as -81.4 degC and a longer period of almost complete darkness was tough. Fortunately the DC11 crew teamed up and made it easy to endure the nine months of complete isolation. Living on almost nothing but a flat ice shield may appear boring but to me the opposite was the case, there were so many things to discover.
All had begun at the conference Autonomous Observatory Software RTS2 Lijiang, China, 2014 where I met again Jean-Marc, the PI of IRAIT/ITM, from Aosta Valley. After a short conversation and after talking with my wife it was clear that I'd go to Antarctica.
While the weather in Switzerland is still bad, I took the chance to visit Bootes-2 near Málaga to install and test rts2af package. Last night I took several samples. Adjustments to SExtractor parameters were necessary in order to receive enough objects on each image. Done that, rts2af reproduced the manually chosen best focus position (1645 ticks) as 1642 ticks. The FWHM was 7.2 pixels.
Polar alignment measurement with rts2pa (2012-09-28)
On 2012-09-17 the Internet connection was interrupted since obsvermes.org did not reboot properly. After manual boot the observatory resumed operation.
In the evening the script rts2pa_measurement.py took five images and the result of the polar alignment calculation is
KingA dx=-5.064 dy=-17.1044996475 arcsec KingA dtau=0.135778986508 arcsec RA= 19.7235488889 HA=2.93587219088 KingA HA=350.333610323 deg lambda=131.378103146 arcsec A=-32.5488058658, k=129.5
The HA axis is tilted about 2 minutes to the East above the north celestial pole (NCP). Within 11 years, the date of the last measurement was 2001-09-30, the HA axis moved about by 60-80 arcsec mostly to the South.
The original images are available here.
Downs, ups, rts2af, rts2pa, (2012-08-05)
The inverter of the door motor died while the door was open. That required a taxi ride on 2012-07-09 at 23:30. This device has been replaced and since 2012-08-03 the observatory is fully operational again.
Since early June I support Baker Observatory in getting the rts2af autofocuser package up and running. Main of the enhancement tasks were better test features. It is now possible to test the whole chain with existing images] together with a running RTS2 system. A new feature is the fall back calculation for best focus based on weighted mean of FWHM and flux.
Polar alignment can now be measured with the help of rts2pa package.
For both packages installation, testing instructions as well as man pages describing the general operation and configuration are available.
Obs. Vermes operational (2012-07-07)
After a long repair of FLI PDF and update of the computer infrastructure Obs. Vermes is now operational again.
During the past weeks the gateway obsvermes.org reported that its disk was dying. The Alix host was reconfigured to run OpenSuSE 12.1 and strongSwan replaced Openswan. It turned out that strongSwan itself is easier to configure and to debug than Openswan, but to resolve the interplay with openSuSE's SuSEfirewall2 was still an undocumented adventure.
Obs. Vermes down (2012-03-24)
Again the FLI PDF focuser ceased its operation. Fortunately(?) this time the motors don't have enough torque to drill out the threads.
Obs. Vermes up (2012-03-17)
After inspection it turned out that the steel screws destroyed the threads and a complete replacement of the focuser was necessary.
Obs. Vermes down (2012-02-08)
The Obs. Vermes is down due to a jammed FLI PDF focuser. This is the second time that this product caused a failure.
Extended temperature model to -16.2C (2012-02-04)
Due to the exceptionally cold weather (outdoor temperature -22C) the focuser temperature model has now a data range from -16.2C to 25C. The whole observatory operated all night without any malfunction.
Obs. Vermes exiled (2012-01-29)
As Speacweather.com announced on 2012-01-24 evening we saw extra ordinary northern lights (aurora borealis,G1-class geomagnetic storm) in Norway near Jøvik (Longitude: 19.722473 E, Latitude: 69.544480) having a clear sky.
Typical settings for NIKON D70s:
ExifTool Version Number: 8.40 File Name : DSC_0019.JPG Camera Model Name : NIKON D70s Orientation : Horizontal (normal) Exposure Time : 15 F Number : 4.5 Date/Time Original : 2012:01:24 23:36:05 Focal Length : 18.0 mm ISO : 800 Quality : Basic White Balance : Preset Shooting Mode : IR Control Sensor Pixel Size : 7.8 x 7.8 um Lens ID : AF-S DX Zoom-Nikkor 18-70mm f/3.5-4.5G IF-ED Create Date : 2012:01:24 23:36:05.80 Circle Of Confusion : 0.020 mm Field Of View : 67.3 deg (13.31 m)
Storm front Joachim (2011-12-17)
Storm front Joachim interrupted the power supply for about 12 hours. UPS went low and NUT shut RTS2 and all computers down. Resuming operation revealed that there was no IP connection to the equipment on the dome. The power supply of Linksys WET54G died and we replaced it.
Obs. Vermes up again (2011-11-26)
Today we installed the new frequency inverter and since the weather was good started immediately the observations. Out of the box the FWHM was 4.2 pixels
2011-11-26 18:56:17,085 INFO rts2af_fwhm.py: FWHM: 3115.0, 4.2803565365, 589, X, /scratch//images//que/20111126/20111126175506-148-RA.fits
for a white light image and after the first focus run
2011-11-26T19:22:52.289 CET EXEC 4 rts2af_acquire: focus line: FOCUS: 3426.943141, FWHM: 2.393442, TEMPERATURE: 4.94444417953C, OBJECTS: 64 DATAPOINTS: 11
the focus was found at 3427 and the FWHM was 2.4 pixels. Subsequent white light images have a FWHM varying between 1.8 ... 4.2 pixels (indicating room for improvements).
Obs. Vermes down again (2011-11-11)
Since 2011-10-22 Obs. Vermes is out of order because of a total failiure of dome azimuth frequency inverter. The new inverter arrived today.
Obs. Vermes resumed operation (2011-10-01)
After two months Obs. Vermes started its normal operation. It's pitty we had so many good nights. Update 2011-10-02: and it got the focus again.
Obs. Vermes down (2011-07-31)
The ssd650v frequency inverter went off line last Thuesday (2011-07-26) and the site is not reachable since last Saturday. A visit this week will be necessary.
Update 2011-08-07: obsvermes.org had a erroneous file system preventing normal boot process. The inverter SSD650v communicates not reliably over RS 232 anymore. It will be replaced.
Update: The first frequency inverter arrived on last Friday (2011-09-23) in bad condition (see image) and the second on Monday (2011-09-25). But it seems that the software changed or it does not accept the command to set the setpoint (247).
Mean of the fitted functions (2011-07-23)
In order to further clarify the open issue mentioned in the next paragraph the analysis has been extended and the mean of the parameters of the fitted functions have been determined. On the graph at the lower left the green lines indicate the individual and the red curve the mean of the fitted functions. In the region of the minimum (+/-100 ticks) the FWHM variation is about .2 pixels and hence a slightly off focus position can not explain the growing differences below.
In order to compare the fitted polynomials the minimum has been determined first. Then the FOC_POS_reduced has been calculated as (FOC_POS- FOC_POS_MIN). p0r, p1r, p2r are the parameters of the fitted even 4th degree polynomial.
Up to now all focus runs were done in the white light. To reduce the errors only objects near the center with a radius less than 800 1x1 pixels have been used for this analysis. To find the parameters of the fitted function it might be better to use a (narrow) filter in the optical train. Comparing filter - filter offsets should yield higher precision, but then still the offset to the white light needs careful attention.
More experience (2011-07-10)
The problems with setting the focuser position have been sorted out and only in rare cases the FLI PDF does not reach the target position. Most of the time the focuser was operating in relative mode. Relative mode means that the FWHM of an image is analyzed and used to decide to schedule a run if it is above threshold. The normal procedure should be that a focuser run occurs in the maximum once per night (e.g. at the beginning) and then temperature compensation takes over the control. The night of 2011-07-05 was not particularly clear so the explanation for the increasing FWHM could indeed be denser and denser clouds. The sudden FWHM drop at 03:12:07 and the more or less constant fitted FWHM stand in contrast to this idea but it can not be decided without the narrow angle cloud sensor.
2011-07-05 00:53:22,444 INFO rts2af_fwhm.py: FWHM:4.31583333333 2011-07-05T01:01:17.359 rts2af_acquire: FOCUS: 3680.055732, FWHM: 2.358561, TEMPERATURE: 16.2777786255C 2011-07-05 01:03:16,463 INFO rts2af_fwhm.py: FWHM:2.36775862069 2011-07-05 01:09:07,437 INFO rts2af_fwhm.py: FWHM:4.37724137931 2011-07-05 01:14:13,156 INFO rts2af_fwhm.py: FWHM:3.75842105263 2011-07-05 01:19:19,785 INFO rts2af_fwhm.py: FWHM:3.81375 2011-07-05 01:24:23,093 INFO rts2af_fwhm.py: FWHM:3.8525 2011-07-05 02:04:33,777 INFO rts2af_fwhm.py: FWHM:5.77862068966 2011-07-05 02:06:22,956 INFO rts2af_fwhm.py: FWHM:5.51906666667 2011-07-05 02:11:20,868 INFO rts2af_fwhm.py: FWHM:5.85329113924 2011-07-05T02:16:47.523 rts2af_acquire: FOCUS: 3702.626771, FWHM: 2.490576, TEMPERATURE: 16.2777786255C 2011-07-05 02:17:31,968 INFO rts2af_fwhm.py: FWHM:5.68388349515 2011-07-05 02:18:55,127 INFO rts2af_fwhm.py: FWHM:3.075625 2011-07-05 02:30:45,343 INFO rts2af_fwhm.py: FWHM:4.22466666667 2011-07-05 02:39:29,658 INFO rts2af_fwhm.py: FWHM:3.664 2011-07-05 02:50:07,742 INFO rts2af_fwhm.py: FWHM:4.89428571429 2011-07-05 03:01:07,111 INFO rts2af_fwhm.py: FWHM:5.215 2011-07-05T03:07:03.303 rts2af_acquire: FOCUS: 3603.294323, FWHM: 2.500838, TEMPERATURE: 14.8333330154C 2011-07-05 03:12:07,213 INFO rts2af_fwhm.py: FWHM:1.94508571429 2011-07-05 03:25:24,375 INFO rts2af_fwhm.py: FWHM:3.63823529412 2011-07-05 03:36:20,648 INFO rts2af_fwhm.py: FWHM:3.33786885246
The hot end (2011-06-29)
We had a few days with a maximum temperature above 30 degC and during the night it did not drop below 16 degC. This was a good occasion to extend the range towards the "hot" end. For the graph on the right side the selection criteria were slightly modified. It turned out that a fitted FWHM below 2.3 pixels contribute overly to the tails of the residuals.
Looking at the data the absolute resolution of the the focuser is about 100 ticks or 2 degC over a period of more than six months.
At the RTS2 repository is a temperature compensating version flitc.cpp available for the FLI PDF. It is not yet stable since there are still unresolved issues concerning the setting of the focuser position (2011-07-10: most bugs in rts2af_acquire removed).
Temperature compensation for FLI focuser (2011-06-18)
Temperature model for AP Astrofire F/7 (2011-06-03)
Retrieving the data from the local meteo database yielded a significant improvement of the temperture dependency of the focuser position. The constraints
- more than 10 objects
- a chi2 of less than 5.
- a fitted FWHM of less than 4.0 pixel (1.41 arcsec/pixel)
are moderate. The analysis is still in progress.
Initial temperature model for AP Astrofire F/7 (2011-05-29)
Due to the large FWHM variation as a function of ambient temperature frequent focus run were observed. To compensate the temperature variation the primary aim is to establish a model for the AP Astrofire F/7 refractor which can be used by the focuser to predict a reasonable position.
The developed software infrastructure uses the existing rts2af package. Two new scripts rts2af_model_extract.py and rts2af_model_analyze.py have been released to the repository. The first scans a given base directory for focus runs of a given filter (filter X, that is the clear optical path), uses rts2af_offline.py to fit the focus and writes the result to file. The second reads this file and plots the data using matplotlib.
Last night rts2af failed to fit and set the focuser far off the correct position. Now rts2af.py includes the chi2 in its decision and if it is too large it returns the weighted mean of the distribution as focus position. This position is only a good estimation.
Initial installation and RTS2 operation of the apache1 telescope at Astronomical Observatory of the Autonomous Region of the Aosta Valley (OAVdA) (2011-05-20)
After the introduction given by Petr and Markus and Petr's modification of the telescope dirver and the addition of the device driver for the Microfocuser the observatory was ready for the first observation in the evening. At arround 9 o'clock the sky was partly clear and apache1 started making flats and entered state night at about 21:30. After rough adjusting the secondary mirror focuser the effective focus was achieved interactively by rts2-xfocusc. A first trial of rts2af autofocuser package didn't provide a focus because the current position was far away from the real minimum, but anyway it completed. The gallery of the workshop including the first RTS2 move movie of apache1 telescope can be downloaded here (all images).
First complete rts2af autofocus cycle (2011-05-10)
If rts2af_fwhm.py detects a FWHM above a given threshold it triggers a focus run executing rts2af_exec.py meaning queueing target number 5 named OnTargetFocus in the queue
next of service EXEC. EXEC then executes the associated script rts2af_acquire.py (see postgres database table
scripts), which acquires a series of preconfigured images and hands them over to rts2af_analysis.py
as soon as they are stored on the file system. As soon as all images arrived rts2af_fit_focus.C fits the FWHM and FLUX and the result is fed back to rts2af_acquire.py which sets the focuser device variable FOC_DEF.
The after exposre image analysis started this morning at 05:00:40 and the focuser was set within about six minutes at 05:06:59. The first image taken right after the focus run completed, confirms the FWHM of about 2.5 pixels (3 arcsec).
A typical CCD exposure for the clear optical path is in the order of 5 seconds, a read out of the 2x2 binned 2000x2000 windowed frame takes 8 seconds and the focuser has a speed of 100 tick/second. So about 3 minutes of the ellapsed 6 minutes are used to perform the whole analysis of the 13 images.
The rts2af package is not Obs. Vermes specific. It can be used by all observatories which use RTS2. In addition rts2af.py, this is the place where the classes are defined, operates solely on FITS images. So it can be reused in other projects as well.
The motivation behind having a good focus is twofold. There is a large temperature induced focus shift from the beginning to the end of the night which yields a FWHM as bad as 10 pixels. Having a refractor with a diameter of "only" 0.18 meter the photons should be put in as few pixels as possible.
The current release includes now error bars (standard deviation) for the FWHM data. During the past week the configuration has been tweaked and now a single focus run is done in about four minutes. It is still an open question how to treat flux. Sextractor's output FLUX_APER/FLUXERR_APER might be a better approach.
Visitors of High Altitude Research Stations Jungfraujoch and Gornergrat (2011-05-06))
On Friday we welcomed the guests Prof. Dr. Hans Balsiger, Prof. Dr. Erwin Flückiger and Dr. Rolf Bütikofer of the fondation High Altitude Research Stations Jungfraujoch and Gornergrat at our observatory. We made a presentation about the construction and the present state.
rts2af Python autofocus package released (2011-04-24)
After thorough on- and offline testing the rts2af autofocus has been released to RTS2 svn repository.
Main new features:
- There is now a script which can be executed by rts2-executor as a regular target script.
- The fitted focus position of the optical clear path is fed back to the focuser (configurable).
- Each image is analysed immediately after it has been stored in the file system. A few seconds after the last file has been written, the position is available.
- The configuration of the filters has been changed: now relative offsets are stored in the configurable array (you can always dump a configuration file like it is done in SExtractor).
- After the focus position of the clear path has been set the filtered images are done at the correct positions given you have reasonably defined filter offsets.
- It is fast: analysing 20 4096x4096 Fits images is done in less than 10 seconds (Intel(R) Core(TM)2 Duo CPU E7400 @ 2.80GHz).
Focus - Fit loop closed (2011-04-22)
After several attempts this week finally the rts2af.py based scripts managed to fit and set the focus.
2011-04-22T23:42:33.042 CET EXEC 4 rts2af_acquire: got fitted focuser position at: 3650 2011-04-22T23:42:33.042 CET FOC_FLI 4 changing focuser position to 4450.000000 2011-04-22T23:42:33.611 CET FOC_FLI 4 focuser moved to 4450.000000 2011-04-22T23:42:37.045 CET FOC_FLI 4 changing focuser position to 3650.000000 2011-04-22T23:42:45.090 CET FOC_FLI 4 focuser moved to 3650.000000 2011-04-22T23:42:57.065 CET EXEC 4 rts2af_acquire: current foc_pos: 3650.0
Basically rts2af_acquire.py is executed by EXEC as target 3, and it spawns rts2af_analysis.py as a subprocess for each filter in use. The analysis subprocess reads the fits file names on its standard input and processes each file immediately. So no time is wasted. As soon as the last file appeared the fitting is done with rts2_fit_focus.C which is built on CERN's root package. Having now experience with several 50 focus runs rts2_fit_focus.C proved to be a reliable component.
Seminar on Autonomous Observatories (2011-04-17)
Markus Wildi will give a seminar on autonomous observatories at Astronomical Observatory of the Autonomous Region of the Aosta Valley (OAVdA) on May 20 through 23.
Cirrus clouds (2011-04-02)
Last night there were high clouds in the first half of the night. Usually the image background intensity is about 3200 units, while on the image on the left it is around 4000. The reason for the 4 bad images: one has no stars on it and in case of the other three the mount did not track.
The reason for the pronounced asymmetry of the FWHM has to be investigated. Nevertheless the fitted minimum represented the focus last night.
From 2011-04-01T01:00:00.000 CET to 2011-04-03T13:00:00.000 CET Number of nights 2 observations nights 2 (100.00%) Images 107 good images 104 ( 97.20%) Date Images % of good Observations Good Images 2011-04-01T01:00:00.000 CET 13 42 42 100.00 2011-04-02T01:00:00.000 CET 14 65 62 95.38
Six nights in a row (2011-03-26)
The past six nights were clear and the observatory worked with the exception of Monday seamless until it closed the door in the morning. The ratio of good images varied between 2, at the beginning of the week, and 97%. The main source of the losses were incomplete focus settings in older target scripts.
Over 900 images were acquired with exposure times of typically 300 seconds. In addition 380 darks and 350 flats were exposed during twilight. The FWHM of the stars on the images was on the average 4 pixels which is about twice as much when it is really focussed.
miyo:/usr/local/src/rts-2-head # rts2-nightreport -S -f 2011-03-20 -t 2011-03-27 From 2011-03-20T00:00:00.000 CET to 2011-03-27T00:00:00.000 CET Number of nights 7 observations nights 7 (100.00%) Images 908 good images 470 ( 51.76%) Images Good Images % of good Date Observations 2011-03-20T00:00:00.000 CET 62 105 4 3.81 2011-03-21T00:00:00.000 CET 34 213 5 2.35 2011-03-22T00:00:00.000 CET 21 83 58 69.88 2011-03-23T00:00:00.000 CET 31 109 103 94.50 2011-03-24T00:00:00.000 CET 45 195 104 53.33 2011-03-25T00:00:00.000 CET 39 133 128 96.24 2011-03-26T00:00:00.000 CET 16 70 68 97.14
There are still minor issues with the overhauled flats.py, thanks to Petr it has now a separate configuration, not sending an email report in the evening. During the past weeks the troubles with the newly introduced enforcement of the BLOCK EXPOSURE could be sorted out and
rts2-dome-door-vermes as well as
rts2-cupola-vermes signal their states now correctly.
Power outage during observation (2011-03-06)
In the night of 2011-03-04 we had a power outage lasting for about 15 minutes.
After the analysis of the log file it is sure that the door closed and after the failiure was over the observation continued.
Five nights in a row (2011-02-11)
After a long period awful weather and a lot of snow we had five clear nights in row. The image of M31 on the right side was taken with the focus settings of the previous night. The average of images having an astrometrical calibration (using astrometry.net) was over 90% indicating a smooth operation of all soft- and hardware components.The processed image with the full resolution can be found here as well as a version without Halpha filter.
The web server of the observatory has been updated and displays now the images of two dome web cams. In case you see nothing turn on the light. Be sure not doing that during clear nights.
The first production release of rts2af_offline.py and rts2af.py were released on 2011-01-22.