GROND operations

This document describes some issues of the operation of GROND, including the preparation of OBs with p2pp.

News

• 27. Mar. 2014: GROND is operational again.
• 16. Jan. 2014: We plan the next GROND maintenance intervention during the following period: 14.-30.3.2014. GROND will be warmed-up, and will not be available for observations. If all goes well, we might get back operational earlier, but this cannot be guaranteed.
• 11. Sep. 2013: Do NOT use p2pp version 3.3.3 for creating any GROND OBs. ESO has introduced two new parameters which will make your OBs non-executable! Stay with version 3.3.1 or older.

OB preparation and data handling

• OB preparation with P2PP: You should prepare the OBs for your observations with P2PP prior to arrival on La Silla. Detailed description on how to do this is given below. For this, ESO has provided you with a unique user-ID and password. You will need this to create the OBs, but also in La Silla to install the OBs on the w2p2dhs machine!
• Before your first observing run with GROND it is advisable to ask one of your collegues for an introduction. In some cases, there will be a member of the GROND team available during the MPI period, but this is not guaranteed. They will help you in case of problems, but are not responsible in introducing you into the system, and the specialties of GROND. Please, understand that they are not on La Silla to service you - they have their own tasks and deadlines. Of course, they are willing to help in case of problems, also on short notice in the middle of the night.
• If you are interested in data reduction on La Silla, the following procedure is suggested: (i) Login into our so-called offline machine wgrondoff and create a directory according to your preferences. (ii) Copy your data to wgrondoff (via scp from wgrondoff). (iii) Our standard GROND data reduction pipeline is installed, so the processing should have a very high success rate when certain parameters are tuned. Again, our GROND team members can help in the proper usage in case you have not trained this at your home institution (which is a good idea to do). These scripts will usually be fast enough to provide you a combined image of one OB within a few minutes (griz) up to 10 min (JHK).
  Due to the limited space on wgrondoff, please be warned that data older than 3 days can be deleted without notification. If you want to keep your processed data, copy them to an USB device (from wgrondoff).
• The Real-Time Display (RTD) shows 10-sec images for JHK, and individual integrations of the griz bands (typically 1 min for a 4 min OB, and 4 min for a 20 min OB).
• According to the agreement with ESO, data should NOT be ftp'ed from La Silla to Germany - and most likely even cannot be ftp'ed: per night, you can gather up to 25 Gb data!
  Starting Oct. 1, 2013, all GROND-data are transferred via the ESO data handling system (DHS) to the archive at ESO Garching, in nearly real-time. Access is controlled via the User-Portal identification. CALIB and STD data are freely accessible for every GROND-user.

Rules for preparing your OBs:

• The OB definition in GROND is according to the exposure in the NIR. The actual execution time is substantially longer (see table below). In order to avoid either the VIS or the NIR channels sitting idle, one of the following exposure combinations is strongly recommended.
  Other combinations may work, but are not tested in the pipeline processing!
  It is your responsibility to use proper values. If you decide to use non-standard values, the OB may crash, or leave the observing system in an undefined state. In the worst case this requires a reboot of the wgrond instrument workstation, and loss of at least 30 min observing time!

    
    OB           exposure    actual        IR      OMODE  Numbers for OB (per telescope position)
    mnemonic     in NIR    duration    IntegrTime        #of_IR_INTs  i/z & g/r expo
    ---------------------------------------------------------------------------------
    4m4TD      4 min      6.2 min     10 sec     Slow     1            35.4 sec
    4m4TDf     4 min      6.2 min     10 sec     Fast     1            66.0 sec
    8m4TD      8 min     11.5 min     10 sec     Slow     2           114.9 sec
   20m4TD     20 min     27.3 min     10 sec     Slow     5           369.0 sec
   40m4TD     40 min     53.8 min     10 sec     Slow    10           799.0 sec
   
    6m6TD      6 min      9.3 min     10 sec     Fast     1            66.0 sec
   12m6TD     12 min     17.2 min     10 sec     Slow     2           114.9 sec
   30m6TD     30 min     41.0 min     10 sec     Slow     5           369.0 sec
   60m6TD     60 min     80.8 min     10 sec     Slow    10           799.0 sec
  
   10m8TD     10.7min    17.3 min     10 sec     Slow     1            62.0 sec
   12m6TDf    12 min     17.2 min     10 sec     Fast     1            66 sec
   48m6TDf    48 min    ~60   min     10 sec     Fast     1            66 sec
    

• In principle, the g/r and i/z exposure could be different: The only presently incorporated option is to have 2 i/z images for 1 g/r image. Thus, the settings would be: DET2.NGR=1, DET2.NIZ=2, DET2.UITGR=65, DET2.UITIZ=22 for fast readout. Note, that this feature has not yet been tested in the pipeline processing software. Other i/z vs. g/r ratios are not implemented, and OBs with other settings will abort with the error message "Problem during execution of GROND_img_obs_exp => incorrect NGR/NIZ combination requested"
• The fast read-out mode provides 10-sec readout instead of 40 sec, but at higher noise level (12 e- instead of 5e-). So far, this mode is only implemented/tested for the 4 min and 6 min OBs.
  The 10min8TD is less efficient than the 12min6TD (for the same total duration), but allows better background subtraction in crowded fields.
• The two OBs 12min6TDfast and 48min6TDfast have 6*10sec NIR and 66 sec griz exposure at each of the 6 telescope dither pointings, but then repeat the same dither pattern a second time (12min6TDfast) or 8 times (48min6TDfast). If you want to design other OBs of that kind, please note the following definitions:

     NTP = Number of Telescope Pointings
     NTD = Number of Telescope Dithering Positions
     

  The OB 12min6TDfast has NTP=12 and NTD=6.
• For very accurate relative photometry or to avoid saturation, there is a parameter TEL.TARG.FOCOFFSET. Values in the range 100-300 will defocus (via M2 movement) the image to a degree which still leads to Gaussian-like PSFs. A value above 450 will produce pupil images, i.e. you will see a donut-shaped image. Note that this parameter appears twice in the OB: once for the preset, and once of the actual exposure.
  Note also that this defocusing also defocuses the guide camera, so depending on your guide star the accuracy of guiding may be compromised!
• Other OB parameters:

    "Combined offsets" should stay on, so guiding continues during telescope offsets. 
    "TARGET.NAME" and "TEL.TARG.TARGETID" can be identical
    "Run sequence number" should be a running number per target, one for each OB.
    "Run ID" should be a running number, one per target PER NIGHT. 
             If you observe the same target the next night, increment this #. 
     Example:
               Target  Run-ID   Sequence-ID
      night1    tar1      1          1
      night1    tar1      1          2
      night1    tar2      1          1
      night2    tar1      2          1
      night2    tar1      2          2
      night2    tar2      2          1
      night3    tar1      3          1
    

  In GROND_img_acq, no input is required for the guide star.
  For the rest, please use the default values.
• You should NEVER have any empty space in the keyword for the target name. If you do, the OB will crash (ESO software feature; we cannot do anything against it).
• You should prepare enough OBs - there is no upper limit. For each target, you should fill in the RunID (one number per night) and the SequenceNumber (one for each OB). If you execute an OB with an identical pair of RunID and SequenceNumber, the data will not be processed by the pipeline data reduction system!

How to export/import your OBs to La Silla and use them?

• Export your OBs on your home computer, and bring the resulting *obx files with you to La Silla.
• Copy the OBs (*obx files) onto the DHS machine: w2p2dhs.ls.eso.org:/home/service/p2pp-impex/GROND/ user-ID=service, pwd=.... You can also create a sub-directory under ../GROND/ and put your files there.
• Login to the w2p2dhs machine, kill ANY running P2PP, re-open P2PP with your Run-ID number and password. Click "Import" to load the *obx files you just copied.
• At this stage, you can duplicate or change your OBs again. Make sure that while doing so, you obey the above rules, in particular the Run-ID and Sequence-Number scheme!
• On the Bob-Panel on the wgrond machine, switch the "Environment" to OH=w2p2dhs, and process=p2pp (both in first line). Only then your OBs can be fetched.
• In the p2pp panel on the w2p2dhs machine: highlight the OB you want to execute. If you click the uppermost button on the Bob panel on wgrond, you can fetch that highlighted OB. You can look at it for double-checking (in "Interface", "Engineering" rather than "Observation" mode), and execute it.
• If a GRB occurs, your running exposure will be stopped, data read-out and saved, and a pop-up window appears with the request to "Preset" or "Abort". Note, that at this stage your exposure IS ALREADY stopped. So you cannot save your running OB by aborting the RRM (Rapid Response Mode) request. In general, there are strict rules to be obeyed by ESO (and you) for aborting RRM: basically only technical or weather conditions count as an allowed reason. Up to 15% of the total time may be lost for you this way. Except for the rare cases of night-time GRBs (happens only for 15% of all GRBs due to SAA constraints), a scheduling page exists (wgrpipe.ls.eso.org:9222/gp/night.html; UID: twoptwo, passwd is written above the DHS terminal) on which you can check the RRM-schedule for the upcoming night.
  Once the RRM is finished (typically 3-4 hrs), you NEED to switch in the Bob-Panel on the wgrond machine the "Environment" back to OH=w2p2dhs, and process=p2pp (this will have been changed by the RRM)!!

Typical limiting magnitudes

And the following table gives you relative magnitudes in the seven filter bands (all AB magnitudes, for New Moon, airmass 1, seeing 1, and a 8m4TD exposure):

g'	r'	i'	z'	J	H	K
23.0	23.0	22.5	22.2	20.6	20.1	19.4

Log of technical or operational changes

• 9-21 Dec. 2008: M1 re-coating, and correction of the misplacement of M1 from the last re-coating. This removed the bad image quality since the last re-coating. This may also have led to a systematic offset of the pointings, as the pointing model was not updated.
• 24-30 Jan 2010: Replacement of the old/broken/test-K band detector by a new one, GROND vessel taken off, and re-mounted with alignment done with red CCD and red laser. Alignment accuracy about 2"; pointing test towards zenith reveals 10" (instead of 8") offset.
• 17-22 Mar 2011: Maintenance of the CCC coldhead; GROND vessel not removed.
• 04 May 2011: z-band CCD shows strong noise during fast read-out. ESO Garching suggests several tests, but obvious cause is not identified.
• 8-12 Aug 2011: M1 re-coating.
• 10 Aug 2011: Change of all IPN addresses of all GROND-related computers.
• 30 Oct 2011: New settings of bias levels for CCDs (by P. Sinclair). This implies the need to use new bias frames for the reduction of science images (calibation OBs will do this); most importantly, one should not use old bias frames with new science data, or vice versa.
• 28 Nov 2011: New wiring/grounding of TCCD (Guide camera), and the 24V voltage supply moved from basement to M1 cell: now only 7-8 e-rms (instead of 16!)
• 04 Jan 2012: Due to the persisting problem with noise in the z-band CCD during fast readout, we have switched back to normal readout mode for 4m4td OBs.
• 03 Feb 2012: Now all optical CCDs affected by strange-pattern noise in fast readout mode. ESO is working on the problem.
• 02 Apr 2012: CCD (z-band) noise has been fixed - was due to one of the preAmp boxes. The bias levels had to be adjusted, so a new master bias should be used from now on.
• 10 Apr 2012: CCD noise is back, now in all four CCDs. Still only in fast (225 kHz) readout, not in normal readout. Completely unclear, how this could happen. ESO Garching has been informed.
• 08 Sep 2012: Faulty disk on wgrpipe RAID substituted.
• 28 Dec 2012: New pointing model installed at the 2.2m telescope, incorporating a dynamic change of three individual coefficients for each instrument (ID, IH, CH) whenever an instrument is changed (in the past, CH had to be changed manually).
• Early Jan 2013: Meteomonitor works again, but seeing values don't make it into FITS headers. Change Request at ESO issued.
• 20 Jan 2013: Faulty disk on wgrccd substituted.
  Back-ups of system-software disks of wgrccd, wgrond, wgrdcs
• 26-28 Feb 2013: Video board in FIERA died. P. Sinclair replaced it with the spare available (#86). Thereafter, the bias levels had to be re-adjusted. Be careful to use the new CALIB files for any observations after 1 March 2013.
• 1 May 2013: After tests of switching between ENDESA/Transelec and La Silla generator power, GROND FIERA cannot be rebooted anymore. GROND is not operational.
• 15 May 2013: Despite many tests and investigations, the underlying problem could not be found. Unexpectedly, swapping boards to WFI-FIERA and back made GROND-FIERA re-think and come online. However, two channels on Video board 0 in FIERA died, leaving the northern halves of the i and z band detectors dead. It is suggested to apply an 1 arcmin offset to the nominal pointing direction to move the target into the lower halves of the chips.
• 8 Jul 2013: The FIERA problem has been identified, and solved. A video board spare from Paranal has been inserted, and our damaged video board sent to Paranal for repair (and planned re-insertion later on). The read-out noise is higher with the new board, and the bias levels have been adjusted, so be careful to use the proper calibration frames.