The RoboDIMM.

On the 27th of August 2000, a new seeing monitor was installed on Cerro Tololo to replace the old Carnegie monitor in use since 1995. This DIMM (Differential Image Motion Monitor) is based on similar devices built first by ESO and more recently by University of Washington. Our instrument has been robotized (hence it is called "RoboDIMM") and works autonomously every night. On the picture below, the DIMM is located in the narrow tower in the middle of the mountain top background. The seeing is calculated for a wavelength of 500nm and zenithal direction, the measurements are stored every minute in a database and displayed in a web page (see here for intranet online version).


How does the DIMM work?
Characteristics of our DIMM
Acknowledgements and links





How does the dimm work?

The first DIMM was developed by M. Sarazin and F. Roddier (Sarazin, M., Roddier, F., "The ESO differential image motion monitor", 1990, Astron. Astrophy. 227, 294-300). Refer to this paper for more complete information, especially on the theory.

Image quality through a telescope is directly related to the statistics of the perturbations of the incoming wavefront. The DIMM method consists of measuring wavefront slope differences over 2 small pupils some distance apart. Because it is a differential method, the technique is inherently insensitive to tracking errors and wind shake. In practice, starlight goes through 2 small circular sub-apertures, cut in a mask placed at the entrance of a small telescope. One of the sub-apertures contains a prism in order to create a second image of the star on the detector. The dual star images obtained exhibit a relative motion in the image plane that represents the local wavefront tilts, which can be expressed in terms of an absolute seeing scale according to the approximate formula:

FWHM =l-1/5(s2cosg)3/5

where l is the wavelength, s2 is the variance of the differential image motion and g is the zenith distance of the star. Actually, two seeing measurements are made (which improves the accuracy): a 'longitudinal' one in the direction of the sub-aperture axis, and the 'transverse' one that is perpendicular. Both measurements are then averaged to yield the final value displayed on the web page.

Sources of error:

Pixel scale: the FWHM varies as the 6/5 power of the standard deviation of the motion, which is measured in fractions of pixels. The pixel angular scale is determined typically with a 1% accuracy, leading to a 1.2% error in the FWHM.

Instrumental noise: the accuracy of the centroid algorithm, measured in laboratory on 2 fixed spots, corresponds to an equivalent random error of about 0.03 arcsec rms.

Statistical errors: it decreases with the square root of the sampling (number of images used). In our case, the variance of image motion is obtained from typically 250 short exposures per minute in each direction (i.e.. 500 in total), which leads to an accuracy of 3.8% in the image size.

Exposure time: the error caused by the finite exposure time is minimized by using very short exposures that can freeze the motion of the atmosphere in most conditions. We implemented the 5ms-10ms (the minimum CCD frame transfer time is 1ms) interleaving technique and calculate (and log) the extrapolated seeing for a virtual integration time of 0ms (we know from ESO that 5ms is freezing the image motion 99% of the time in Chilean sites).

Back to top

Characteristics of our DIMM

Telescope: Meade LX200 10",f/10, alt-az mount. Plate scale is 76.9"/mm and CCD FOV is 4' x 3'.
Detector: ST5C SBIG CCD camera. No filter. Mean wavelength of QE is 670nm.320x240 pixels. 10 micron pixels, 0.769"/pixel. 1ms minimum exposure time (custom firmware).
Sub-apertures: 90mm of diameter, 150mm center-to-center.
Integration time: interleaved 5ms and 10ms.
Software: Visual C++ with MFC. Libraries: SBIGUDRV 4.03.10, MySQL 3.92,cfitsio 2.420
PC requirements (tested): 300MHz, 64 MB RAM, 2 GB HD.
Software features: the seeing measurements are stored in a database server using a MySQL DLL.The star catalogue contains 400 stars. The weather conditions are checked to command the dome opening or closing.
Dome and location: the dome opening and closing is controlled using a QuickSilver motor with RS-232 interface (manual control is available for emergency use). The system is located on a 6m-high tower at the North edge (upwind of all other buildings in the prevailing direction) of CTIO's mountain top.

Back to top


(The most of this following sections apply to the LX200 telescope. Current development of CTIO DIMMs is using the new version LX200GPS telescope)

More detailed description about software and computer systems can be found here.


A.1Components and connections:

Meade telescope and hand paddle

ST5C SBIG Camera (CCD Head and Controller)

Laptop Sony VAIO PCG-Z505(?): 1 Ethernet port, 1 parallel port, 1 serial, 1 USB port

Clam-shell cover operated by Quicksilver dc servo motor and limit switches.

UPS Best Power Patriot 600

See the corresponding block diagram with all the connections.



Telescope cover with sub-apertures must be aligned vertically with the wedge on the top. Visually, the axis of the pair of stars in the image should be horizontal (maximum of 5 lines difference between each spot). If ever it is not the case, adjust the cover rotation rather than the CCD's (untight slightly the 3 screws holding the cover, rotate, and tighten).

CCD head must have the flat side on the right side when. This is important to respect that orientation for the guiding pulses.

Do not unplug the cable between the CCD Controller Box and the CCD Head if the CCD Controller Box is powered, it could result in serious damage to the CCD. Turn off the power to the Controller Box first, then unplug the CCD Head.

Dome counterweight must be sliding properly in its guiding rail.

Laptop lid must be near to be closed (be careful because if it is closed the laptop will be turn off).

Cable hanging down from the CCD head must be free to move. Use a cable tie to do this.

!!OJO!!Telescope pointing will be lost in any of the 2 following cases: the telescope itself is bumped (so work with lot of care when you are around the telescope), and/or the telescope loses power when it is not in its park position at the celestial pole (the latter condition is normally avoided thanks to the UPS, but if you need to unplug the telescope power, follow the instructions in the Operation Manual).

The PC system time must be configured in Universal Time.

Check if network is up. To do this you can open a console window (MSDOS window) and try a ping to other machine.

Be sure the time synchronization software Tardis is running and connecting with CTIO Network Time Server (contact technical staff people if you need help). The system time should be adjusted every minute.

If the dome isn't open/close properly, use the dome hand controller to close it, turn off the motor power box and turn on again (this action will reset the initial position of the QuickSilver motor). If the problem persist or the dome doesn't open enough, try changing the options in the robodimm configuration file.


B.Basics of the automatic observing process

Tardis (time sync. program, runs in Windows background to regularly (every minute) update the PC's time to the exact UT gotten from CTIO Network Time Server. The main executable calculates day-by-day the twilight-begin time and twilight-end time. At twilight-begin time, the system checks the weather conditions from provided by the site meteorological tower and decides whether it is safe to open the dome or not. It will open if : RH < 83%, and wind speed < 25mph and atmospheric pressure > 777mb. The system is not yet able to detect clouds (nor precipitation with an ambient RH<83%, which can happen with an isolated cloud loaded with rain).

If the weather parameters are satisfactory, the dome is opened and the telescope is pointed to the star nearest to the LST+HAcrit position (HAcrit can be change in the configuration file. Typically HAcrit=2 hours), chosen among the ones available in the catalog. An image is taken and if no star is found, the system starts taking a frame spiral around the position where the star was expected to be located (all images have some overlap, the R.A. steps are 3' and 2' in Dec.) centered on the nominal position. This entire search spiral will take about 20min. If ever no star is identified in this spiral (which can be due either to clouds or bad pointing), the telescope will park itself toward the celestial pole 10min after a spiral failure, the system will try again on the next star in the catalog (unless it has detected a weather shut-down condition meanwhile). Whenever a (dual) star is found, the telescope stops the search and takes a new image for confirmation. It then centers the pair of stars on the CCD, and when it get it centered within 40 pixels, it launches the seeing calculation. During the seeing measurement, the system will define a rectangular region of interest ('roi') around the pair of star and draw it. The rectangle indicates the only region of the ccd which is read. Then it will draw a circle around each psf centroid whenever it finds one. These graphical features are useful to control visually that all is behaving well. The star will be change if another star is near to the LST+HAcrit position. Every time the telescope finds the selected star, it will synchronize its pointing parameter to the position found, thus updating regularly the pointing model. Usually, only a few minutes are lost when changing stars.

During the night, the system will consult the weather tower every 5min to make sure the environmental conditions are satisfactory. The system can be stopped anytime (i.e. end the seeing measurement, park the telescope and close the dome), by sending the REMOTE STOP command from a machine in the 4m console. Such a command will activate a lock file in the robodimm directory in the PC, which will inhibit it until it receives a REMOTE START command. The REMOTE STOP command is useful for example if a dark cloud loaded with rain approaches the mountain (!), or if the night is completely clouded (with none of the 3 weather conditions activated) to prevent the robodimm trying desperately to look for a star during all night. More details about this functionality can be found here.

At dawn, the system will shut down whenever the sky background gets too high or, at latest, when the sun comes up to 5° below the horizon. The telescope will park at the celestial pole and the dome will be closed. It is not necessary to send a REMOTE STOP command to the system at the end of the night (the REMOTE STOP command is only for emergency cases).

C.User interface

At present time the robodimm software version is 2.5 (this version includes MySQL to update seeing database).

Below is a copy of the current GUI running on CTIO seeing tower.

The automatic operation is configured by setting option "Telescope ON" in the configuration file. Then the checkbox "Automatic Measure Enable" must be turned on.

At the upper left side of the GUI is the Time subsection. This group of information is updated avery 2 seconds and it includes the time when the automatic measurement should be start and stop.

At the upper right side a window will show the "stars" when the measurement are running.

At the center left the subsection "Last seeing" will show the last measurement including other useful information like time of the last measured seeing, the two seeing values for T1 and T2, number of images, total average flux and noise contribution.

Just under "Last seeing" subsection a group of diagnostics values are shown: the values for dx and dy(instantaneous horizontal and vertical separation between the "stars"), and the instantaneous maximum intensities.

At the lower left side of the GUI, in the "Last Strehl Ration" subsection, after every measurement the values of Strehl Ratio for the "stars" are shown. This values are useful to diagnostics purposes: both values must be close to equal and typical value is 0.4 (the theoretical optimum is 1).

In the center left side, the GUI is showing the last scintillation values (in percentages).

At the lower left side a subsection is showing the last meteorological information.

Finally, at lower side, the status is shown. Some procedure status is displayed in the left box. Hardware status is displayed in the rest of the boxes.

The GUI bottons are useful in manual operation or under test conditions. If measurement need to be terminated, just uncheck the check box "Auto Measure Enable" and the measurement will finish within 2 minutes and the dome will close.


D.What to do in case of problems?

D.1. !!OJO!! In any case, before unplugging the power of the telescope, press 'Park polar' (uncheck the check box "Auto Measure Enable" first if the instrument is in measurement mode) in order to leave the telescope in a known position : when starting up again the internal Meade controller will assume that it is pointed toward the celestial pole. The pointing should be preserved if the telescope is parked correctly.

D.2. Focus: in some situations the telescope focus would be require adjustment. Check the "dx" value and move the focus knob until the typical value is reached. A more complex procedure includes the use of CCDOPS (sbig ccd software) as shown below.

How to focus?: the 12mm reticled eyepiece is closely parfocal with the ccd so having a sharp image in the eyepiece (you should see the Airy rings) usually puts you close to best focus but best ccd focus is obtained with a simple specific method described now. Close the robodimm.exe application. Open the CCDOPS icon in Windows main screen to run SBIG's software. Double click on the Focus button, select an exposure time of 0.5sec (make sure 'Clear filter' is selected in the top menu). Then, select through the finder any star from mag 1 to 4 (as long as it doesn't saturate the ccd) and center it in the eyepiece. Flip to the ccd position and you should see the 2 images of the star on the CCDOPS focus display. Images can be saved in fits format to be check with image analysis software like SAO for windows.

D.3. If the number of images acquired (see the seeing.log file) is much lower than 200 (this value can vary depending of PC capabilities, however a typical good number of images is always greater than 100), it means that too many images were rejected, which can be due to one of the following reasons:

Telescope is out of focus and no centroid can be found (the star is a donut).

After taking images during 60sec, maximum and minimum counts are determined to realize a last selection based on standard deviation of flux (must be <=3) and minimum threshold which must be >= (minflux + Deltaflux/10).

D.4. If the system doesn't start up alone at the appropriate time, it could be for one of the following reasons:

The program executable (see the program icon) didn't launch automatically: start it by double-clicking the icon and select "Auto Measure Enable: the system will start working alone after initialize components and check weather.

The sky is clear and the telescope can't find the star after completing the spiral: there is a pointing problem and an operator needs to do a manual triangulation to recalibrate the pointing parameters. This is how to proceed:

!! LX200 TELESCOPE ALIGNMENT PROCEDURE !! (when pointing is bad)

** Quit the robodimm program.

** The telescope is equipped with a flip mirror that can divert the light either to the CCD (direct) or to the 12mm reticled eyepiece (after reflection from the flip mirror when it is IN). Selection knob on the flip mirror case is turned ccw to select ccd and cw to select eyepiece (see label). So you don't need to ever remove the CCD. Also DO NOT change the telescope focus unless it is really, really required (for now it is not possible to reach focus with the eyepiece but it is still quite easy to center the out-of-focus star). The 12mm reticle eyepiece has a 11' FOV for precise centering. There is also a 26mm eyepiece which yields a 32' FOV if ever you can't find the star with the 12mm eyepiece. Precise alignment of the telescope has to be done with 2 stars properly centered in the 12mm eyepiece field of view.

** IMPORTANT: Before any alignment, check the clock of the hand paddle (press 'Mode' until you get to the Local/Sidereal time display), it must be accurate within a second for good pointing. If you need to readjust the time, do the following: A/ press and hold 'Enter' until you see a blinking cursor on the time stamp; B/ look at the HH:MM:SS clock panel in the PC and set up the time in the hand paddle to the next round minute (press the number keys to select numbers and press East key and West key to move the cursor right and left respectively) ; C/ once the PC's time (automatically adjusted by Tardis) coincides with the time you selected on the hand paddle, press 'Enter' to start the hand paddle clock ; D/ press 'Enter' twice to validate the GMT and date displays (unless they need to be changed too). All this procedure for Time adjustment (as well as Tololo's coordinates, the other important parameter for good pointing) is fully explained in the Meade blue manual.

** On the Meade hand paddle, press 'Mode' until you get to the 'Telescope/Object library' menu. Press 'Enter' and select 'Align' with the 'Next/down arrow' key then press 'Enter'. When the unit displays 'Altaz/polar' press 'Enter' (i.e.. valid altaz). Next select the 'Star alignment' option by pressing '2'. Press 'Enter' to agree that the base is leveled (check the bubble level on the telescope base) and 'Enter' twice again to agree with picking up a first star and get access to the bright star list. Use the 'Prev/up arrow' and 'Next/down arrow' keys to scroll the list and select a first star visible above you head (you will need to select 2 stars: ideally the first one away from zenith and the second one near the zenith ; they need to be preferably at least 90degrees apart from each other). Press 'Enter'.

** The display will ask you to center the star. Press the 'Slew' key to use fast motion and the N-S-E-W motion keys to point the telescope. Use the 8x50 finder (aprox 5° FOV) and its reticle to center the star as best as you can. Press 'Cntr' to select a slower speed and center the star in the eyepiece. Press 'Enter' to valid the position.

** The display will request you to pick a second star. Press 'Enter' and select another star in the list. Repeat the previous steps for centering the star (go back to "Slew' speed then "Cntr' speed again for fine tuning). When the star is centered press 'Enter'. The alignment procedure is done.

** In order to check that the pointing is good, test on a few stars: go back to the first star, the second, any third bright star present in the sky. To execute a goto: press "Mode' to get to the 'Telescope/Object Library' menu, then press '6/Star'. The unit will display 'Star name:'. Press 'Enter' twice and you should get to the bright stars list. Select and star with 'Enter' and you will read some basic star information (mag.,...). Press 'GOTO'. Check that the stars are very close to the eyepiece's FOV. If not, redo the entire alignment procedure with more care. If you fail several times, something else goes wrong and you should call for help. Meade claims a rough alignment should yield a pointing accuracy of 5' or less, a precise alignment (the one we do, i.e. entering the latitude and longitude of the site) should yield 2' or less (it is usually true in practice and therefore the first ccd frame of 4'x3' will show the target star).

** Now park the telescope toward the celestial pole : press 'Mode' to get to the "RAxxx, DECxxx" menu, then press "GOTO" for 1sec or more until a blinking cursor appears. Enter 00:00:00 for the R.A. then "Enter', and -90:00:00 for the Dec. then 'Enter'. The telescope will move to the celestial pole.

** Position the flip mirror in the CCD position (so that light gets to it!)

** Open robodimm executable and check the "Auto MEasure Enable" checkbox, the automatic operation will follow.


E.Some useful software details

Usually the PC is configured with a desktop icon linking to the last version of robodimm software. Check this shortcut to see the name of the executable program and where is the directory where all files are living.

In the robodimm directory must/should exist the following files:

1.robodimm2X.exe (required, version of the robodimm software).

2.robodimm2.rc (required, configuration file. See here the typical automatic configuration for this file) (required, star catalog file)

4.cfitsio.dll (required,fits DLL)

5.libmysql.dll (required,database DLL)

6.lastweather.log (weather info file, required only if weather info is used, see configuration details below)

7.seeing.log (local archive for final seeing values, is created if it doesn't exist)

8.seeingt1t2.log (local archive for temporal seeing values, is created if it doesn't exist)

9.robodimm.log (application log file, created if it doesn't exists)

More detailed info can be found in the postscript document.


F.Other useful tricks

Bright stars used (and visible from the dome) for the alignment of the telescope : Achernar (aEridani), aCrux, ??Alphard (aHydra), Antares (aScorpio), Canopus (aCarina), ??Diphda (aCetus), Fomalhaut (aPiscis Austral), Hadar (bCentaurus), Sirius (aCanis mayor), ??Spica (aVirgo). The ??stars might be occulted by the dome because of their declination.

If ever you need to close robodimm2x.exe while it is executing a spiral, you need to kill the application in Windows Task manager (Alt-Ctrl-Del).

Precise plate scale must be determined within 1% accuracy by pointing to a know double star (a Centauri and Albireo are good candidates) and measuring the separation.

Clam shell cover must be clamped in rotation with a U-shape block on the eastern side. The southern half stays closed during observation (it is fixed anyway): tests have shown that it does not interfere in the seeing measurement and is a good wind barrier.

In addition to that document, for further information on each sub-system, you should find on the PC desk a copy of other dedicated resources : Meade telescope manual, CCD manual (2 volumes). Further information for the Quicksilver motor can be obtained online at

Useful accessories to buy: Meade 12mm illuminated reticled eyepiece with cord, Apogee 8x50 finder with right angle mirror, Meade 1812 DC electronic adapter (18V<->12V), Meade visual flip mirror ref 644, which you need to machine according to that picture so that the ccd -mounted with the shortest neck possible so that it clears the fork- and the eyepiece can be made parfocal. Machine the 2 eyepiece holding tubes: the one moving when you focus up and down and the one fixed on the flip box. This latter one must be shorten to 12mm high. That tube doesn't come out so it must be done by putting the whole flip box on a machine tool, taking care and protecting the inside mirror.



Accessories used: 1/Meade 644 CCD flip mirror: you need to machine down 0.6" off the eyepiece holder -inner and outer tubes- in order to put the eyepiece parfocal with the ccd head screwed directly -without the 1"1/4 neck- onto the ccd port through the thinner adapter provided (1/4" thick). These changes are necessary to ensure that the ccd will not bump into the alt-az fork mount when going through the zenith; 2/ we use the Meade 12mm corded illuminated reticle eyepiece in the flip mirror; 3/ we use a right-angle 8x50 finder from Apogee with a Meade 25mm illuminated reticle eyepiece to make faster and easier the alignment process for one's neck (make sure you turn the reticle off after alignment is done)

Cables handling: there are 3 cables that rotate with the mount: the ccd, the dec motor and the centering eyepiece. The ccd cable is a bit heavy and can be attached to the telescope handle in the back of the cell to avoid pulling on the ccd, which could loosen it and make it rotate sometimes.

The schematic diagram of the dome robotics (motor, limit switches, hand controller) can be found here.

Cleaning the optics: telescope corrector and prism. Use dry air in can. Call help for stronger cleaning (washing).

Realigning the prism in the cover: the prism axis of tilt should be indicated by a + mark (=thick side) on the glass edge. This mark should be aligned with the axis formed by the 2 sub-apertures. The best way to check whether this is the case is to run CCDOPS in focus mode, defocus the telescope (in and out) and watch for the 2 images: if they don't move along a line (when you turn the focus knob, their separation will vary, ie. they will go away from each other or come together) but rather seem to rotate, it means that the prism is not positioned properly. Unscrew the prism holder from the cover, carefully remove the prism from its cell, turn it a little, put it back and check. Iterate until you see the proper 'linear' behavior. The position (toward the inside or the outside) of the + mark along the sub-aperture axis is indifferent, only the prism axis should be exactly parallel to the sub-aperture axis.

Examples of correlation fwhm_transverse vs. fwhm_longitudinal. This is a sanity check of the system : the time correlation between these 2 measurements should be 1.

Examples of correlation RoboDIMM vs. Carnegie monitor made in August 2000. We have seen perfect time and amplitude correlations of both instruments (they have a completely different measurement process) down to the 0.4" level several times for periods of several hours. Most of the time, the time correlation was very good, whereas the amplitude correlation would vary a lot for reasons that we don't fully understand. Example of a perfect night ; example of curious night.

Correlation of CTIO RoboDIMM vs. ESO/Cornell DIMM ran together during 4 nights on Cerro Honar in the Chajnantor Conicyt Science Preserve (mission from 23-30 of October 2000): follow the link in the 'Work in progress' then 'Site Testing' sections of the CTIO home page.

Back to top

Acknowledgments and Links

Nicolas Long (3rd year computing engineering student at EMA) has done an outstanding work to get the DIMM up and running in 3 months! He is the author of all the programming in its first version.

We owe a great debt to Armin Rest from University of Washington for his assistance in first sharing his source code (written in Pascal) and then in helping us to develop our own modified version. The DIMM he developed operates at Apache Point Observatory and details can be found here.

Mark Sarazin from ESO has been very helpful in answering questions whenever we encountered difficulties. His excellent astroclimatology web pages can be accessed here.

You will find all you ever wanted to know about Meade telescopes and how to fix them at the MAPUG web site.

Back to top


Last revised: Nov. 20 2002,

Original version: Maxime Boccas