Highlights of the Commissioning of FEROS at the ESO/MPI 2.2m telescope

Commissioning I -- September 21 - October 17, 1998
Commissioning II -- November 11 - 30, 1998
Commissioning III -- October 22/23 - 25/26 2002
Commissioning IV -- November 13/14 - 14/15 2002
Commissioning V -- March 08/09 - 10/11 2003

Commissioning III -- October 22/23 - 25/26 2002

These are the main actions carried out during Commissioning-III. Weather and technical problems did not help, so only 3 out of the 7 Commissioning and 3 Technical nights could be use for scientifically useful observations (plus IS had to spend 1 day in bed with fever...).

The general idea is to follow the guidelines of the 1998 Commissioning. Please find the Target list.

Summary

FEROS is functional, but with limited capacity. The efficiency is down by a factor or ~3 (compared to to the benchmark of 15% set at the ESO-1.52m).

Log

• The new adapter has been installed at the Cassegrain focus and the calibration unit in the FEROS room.
• Attempt to optimize the M2-M3-fiber head-TCCD configuration: M2 is set to the nominal value for WFI at the current ambient T, then the fiber head and TCCD must be moved simultaneously until a good focus of the star and the fibre head (as judged by the machining ring around the fibre apertures) is obtained. This can only be done when WFI is not installed.
• Currently the telescope focus must be adjusted manually in the TCS panel. Guiding will be made with the WFI tracker chip using the Besssel R filter. The focus equation is:
  
  Focus = 23740 - 66×T_LS + 2×ZD + O_Filter
  
  Where T_LS is the Temperature of the Long Surririer, ZD is the Zenith Distance of the telescope and O _Filter is the "Filter Offset", O_R = -50.
• Telescope balance setups are determined for both FEROS/WFI Adapter with and without WFI configurations.
• When an object is in the fiber head FOV, it is ca. 1 arcmin South of the WFI center, but this is liable to change.
• The measured TCCD scale is 0.60 arcsec/px, the FOV is ca. 4×2.9 square arcmin. Note: TCCD used during Comm-III is a standard ESO small format TCCD.
• The fibre apertures are approximately 3 pixels in diamter on the TCCD, i.e. 1.8 arcsec.
• In the default orientation, the TCCD has N up and E to the right (but the TCCD-RTD orientation can be set to the "normal" view)
• The object fiber is the Western fiber (the one on the left when the TCCD-RTD starts up).
  
• [ ] TBD: With the pickup mirror inserted, a circle ca. 3/4 the size of the WFI vignets the chip.
• Preliminary measurement of the difference between the best focus for the WFI and the best focus for the FEROS fiber head: +80 e.u. when using the R filter, based on FWHM of image on the TCCD. Later, the calibration is +115 e.u, done with repeated exposures of FF. It will be important to establish correct simultaneous focus of the Fibre Head and WFI tracker chip during Commissioning-IV.
• Patterns are seen on the EEV CCD bias images (science CCD). Apparently due to video board of BIAS PC. Made by Copenhagen, don't have spares at the moment. Patterns come and go during following days.
• The RSM position has been balanced in order to illuminate evenly the 2 fiber entrances with FF light. Position is 7176 e.u. This needs to be re-established each time the alignment of either the science or calibration fibres is adjusted. It must also be established for the ThAr lamp.
• Lamps were not running at their old voltage/current values. New FF bulbs were installed. Adjusted in the following days. Values are:

  	Old		New
  Lamp	I	V	I	V
  halogen 1 ( red)	1.08 mA	3.2 V	1.2 mA	3.3 V
  halogen 2 (blue)	3.80 mA	13.0 V	4.2 mA	12.0 V
  ThAr/Neon	12 mA	N/A	12 mA	N/A
•  First initialization of DRS on Oct 20, with 5x60sec FFs + 3x120sec ThAr. RMS is ca. 20 mAA, used to be ca. 5mAA, not too bad given the bias noise patterns. Blue orders clearly lost.
• The calibration lamp flux could be lower because the beam is no more projected in-axis with the science fibers. The loss in the blue could be due to excessive length of cal fibers. Have to shorten cal fibers (AG proposes 5m), or buy better optical quality ones. Also affected by mis-alignment of science fibres.
• Fiber flux ratio computed with twilight sky flat field, and @@ fiberratio. Fiber 1 (object) gets 1.8 more flux than fiber two (sky) (see Fig below, left panel). Number confirmed during the night by taking a spectrum of the same star with the two fibers (and taking the ratio) (right panel of Fig below). (of course there is a problem in merging orders #29 and #30, however, we are not yet in the best conditions in terms of cal lamps and electronics noise)
  
• FCU opened and FF beam inspected, looks very blue until the entrance of the cal fibers
• "First Light" at the 2.20m achieved 2002-10-24T00:20:26:00...
  
  
  
• Oct 25/26: Clear problem with missing blue orders in cal images (just compare extension of science and cal spectra in OC mode), so it must be the cal fibers. DRS loses a few blue orders, fortunately the interesting spectral regions to compare with the old commissioning are in redder regions.
• [ ] TBD: ADD TO CALIBRATION PLAN: check focus equation during periods of _stable_ seeing by running efficiency test with differenct focus offsets
• Calibration of NDF wheel with FF images. "Density" is a linear dimming factor, so dens=2 means you can expose twice the good ThAr time without saturation. Tested with 51 Pegasi 10 min exposure.
• Fixed FCU panel software to auto set of NDF position based on the object/ThAr exposure ratio, however density function yet to be coded in software.
  
• During last night (Oct 25/26), several objects of original commissioning observed, see below comparisons
  

Commissioning 1998 vs. Commissioning-III 2002

De-blazing

Above we show a DRS-reduced, 180 sec spectrum of tau Ceti, taken in OC mode. In particular, orders #17 and #18 around Hbeta are plotted (see Table 1 for the wavelength coverage vs. order number). The S/N of the two spectra is comparable, though the DRS did a slightly better job in deblazing the original spectrum.

Spectral coverage

SMC R50 was used to explore the spectral coverage of FEROS, as it was done originally. Below we plot the spectrum of that star, in four wavelength ranges. For each panel, the left graph comes from the original commissioning, the right graph is that obtained in October 2002. The original observation is a 5400 sec exposure, our observation is an average of 3x1800 sec exposures. Evidently, our spectrum has lower S/N, however the seeing was oscillating around 1.5" fwhm, and we now have a 1.8" fiber diameter on the sky.

Region of Hdelta

Region of HeI 5876 (in the right panel, the spike at the center of the HI line is a cosmic)


Region of Halpha


Region of Paschen 13

Maximum S/N

A 15 sec spectrum of beta Orionis (V=0.12) gives S/N=120, it was S/N=700 during CommI. After the commissioning was completed the fibre head was examined and it was found that the alignment of the fibre apertures, ball lenses and fibres was far from optimal. So at the moment it is beleived that this is the main cause of the low efficiency, and of the low relative fibre efficiency since the alignmnet for the sky fibre
is noticably worse than for the object fibre.

Efficiency

The ca. 15% peak efficiency measured at the 1.52m was never reached, due to the poor quality of the night, and the non-optimal alignment of the fiber-entrance--ball microlens--fiber pupil (see above). A maximum around 8% was measured in the morning of Oct 26, and that is plotted below. The standard star HR1544 was used. The seeing was down to around 1" fwhm, but the airmass was 1.4. Moreover, a wrong effective area was used in the computation, so the real peak efficiency is around 6.6%. See this email by AK, and my reply.

Clocks

The FITS header of a few images was checked. On Oct 26, the coordinate info (both apparent and from catalog) were correct, and so were the airmass, ZD, and hour angle. The UTs were a more mixed situation, as follows:

Starting an exposure at UT 6:22:36

in the header it was:

TM_START = 6:22:17
TCSUTC = 6:22:34

so TM_START was late by 19 seconds, TCSUTC was about right.

Apropos the workstations, I found

UT(CERME) - UT(bias) = 4 sec
UT( off) - UT(bias) = 2 sec

Tables

Table 1: FEROS wavelength coverage vs. order

Order #		wavelength  [Å]
0001	--	3528-3661
0002	--	3584-3720
0003	--	3643-3781
0004	--	3703-3844
0005	--	3766-3909
0006	--	3831-3976
0007	--	3898-4046
0008	--	3967-4118
0009	--	4039-4193
0010	--	4114-4270
0011	--	4191-4351
0012	--	4272-4434
0013	--	4355-4521
0014	--	4442-4611
0015	--	4533-4705
0016	--	4627-4803
0017	--	4726-4905
0018	--	4828-5012
0019	--	4935-5123
0020	--	5047-5239
0021	--	5165-5361
0022	--	5288-5488
0023	--	5416-5622
0024	--	5552-5762
0025	--	5694-5910
0026	--	5844-6065
0027	--	6002-6229
0028	--	6168-6402
0029	--	6344-6585
0030	--	6531-6778
0031	--	6729-6984
0032	--	6939-7202
0033	--	7163-7434
0034	--	7401-7682
0035	--	7657-7946
0036	--	7930-8230
0037	--	8224-8535
0038	--	8540-8863
0039	--	8881-9217

Commissioning IV -- November 13/14 - 14/15 2002 + November 15/16 - 24/25 half nights of reserved time...

Summary

FEROS is now fully operational, though the efficiency is still down compared to the ESO-1.52m by a factor of ~1.5.

Log

• Based on efficiency measurements as a function of telescope focus the FEROS Fibre Head (FFH) position was optimised such that maximum efficiency coincides with the WFI Besssel R filter (BB#Rc/162_ESO844) being in focus. This is a tedious process as each adjustment of the FFH position requires to dismount WFI!
• AutoFocusAdjustment has been implemented within the TCS-Control panel. To optimise focus a WFI focus sequence should be made to determine the instantaneous value of the filter offset (O _Filter) for WFI Besssel R filter (BB#Rc/162_ESO844) and hense FEROS. This offset can then be entered into the TCS-Control panel and a threshold set such that when the current focus setting differes from the theoretical one by more than the threshold the focus is automatically preset to the theoretical value:
  
  Focus = 23740 - 66×T_LS + 2×ZD + O_Filter
  
  Where T_LS is the Temperature of the Long Surririer, ZD is the Zenith Distance of the telescope and O _Filter is the "Filter Offset". Experience indicates that O _Filter should be redetermined for each "significantly" different pointing of the telescope.
• The measured pixel scale for the large format TCCD scale is 0.51 arcsec/pixel, the FOV is ca. 5.0×3.6 square arcmin. Note: TCCD used during Comm-IV is a standard ESO large format TCCD.
• The Rotating Selection Mirror (RSM) position has been determined in order to balance the FF flux arriving at the FEROS CCD for the 2 fibres: position is 7152 e.u.
• The Rotating Selection Mirror (RSM) position has been determined in order to maximise the ThAr flux arriving at the FEROS CCD for the 2 fibres: position is 3562 e.u. Balancing the flux would mean a reduction in the flux recieved by each fibre and thus increase exposure times beyond the already long ~300sec!
  
• Lamp current/voltage settings were fine tuned:

  	New
  Lamp	I	V
  halogen 1 ( red)	1.2 mA	3.3 V
  halogen 2 (blue)	4.16 mA	12.2 V
  ThAr/Neon	12 mA	N/A
• Loss of blue flux in the ThArNe calibration lamp was found to be due to misalignment of the BeamSplitter and Infra-Red Cutoff Filter. Both were aligned but difficulties in aligning the Infra-Red Cutoff Filter (which is believed to be an interference filter, hense its angle to the light beam determines the cutoff) mean that this filter should be replaced with a different system.
• Calibration of Neutral Density Filter Wheel (NDFW) with FF images. Parameters of the density equation are:
  
  • Gain == 0.00081604673
  • Offset == 0.431688720
  • Min == 0.0
  • Max == 2.5
  • Ref == 0.0
    
  The attenuation factor (A) is related density (D) according to A = 10^D. So the practical range of attenations factor is 1 to 10^2.5 = 316.23. Therefore given that the minimum reasonable ThArNe exposure time is ~100sec the maximum Object exposure time in Object Calibration mode is 31600sec ~ 8.8hrs. Exposure times shorter than the exposure time used for ThArNe calibration images will be underexposed compared to the exposures acquired for DRS initalisation.

Commissioning 1998 vs. Commission-IV 2002

Efficiency

The ca. 15% peak efficiency measured at the 1.52m has still not been achieved. It is beleived that the remaining misalignment between the Entance Aperture in the fibre head and the MicroLens+Fibre assembly is resposible. The maximum efficiency measured thus far is shown below, with a peak efficiency of ~11.5% which is to be compared with 6.6% from Commissioning-III, i.e. an improvement by a factor of ~2 resulting from the improved alignment of the MicroLens+Fibre assembly achieved during the time between Comm-III and Comm-IV. Raw Image FEROS.2002-11-16T01:35:34.000.fits, object HR9087, exposure time 60.0sec.

Relative Fibre Efficiency

The Relative Fibre Efficiency computed with twilight sky flat field, and @@ fiberratio. Fiber One (object) gets ~1.2-1.5 times more flux than fiber Two (sky). Raw Image FEROS.2002-11-24T22:29:47.000.fits, object Twilight sky, exposure time 60.0sec.

Maximum S/N

A 5 sec spectrum of beta Orionis (V=0.12) gives S/N=200 implying 15sec would give S/N~600 however 15sec resulted in some saturation. It was S/N=700 during Comm-I and 120 during Comm-III. Raw Image FEROS.2002-11-15T08:11:57.000.fits, object ß Orionis, exposure time 5.0sec.

Resolving Power

The nominal Spectral Resolution (R) is 48000. This corresponds to a FWHM of the lines of ~2.2 pixels. The following pair of plots shows that the nomianl FWHM is achieved and moreover the spectrograph is better focused than it was at the ESO-1.52m with a flatter distribution of focus values over the dispersion direction of each order AND over the full spectral range of the spectrograph.

De-blazing

Above we show a DRS-reduced, 180 sec spectrum of tau Ceti, taken in OC mode. In particular, orders #17 and #18 around Hbeta are plotted (see Table 1 for the wavelength coverage vs. order number). The S/N of the two spectra is comparable, though the DRS did a slightly better job in deblazing the original spectrum.

Spectral coverage

SMC R50 was used to explore the spectral coverage of FEROS, as it was done originally. Below we plot the spectrum of that star, in four wavelength ranges. For each panel, the left graph comes from the original commissioning, the right graph is that obtained on November 17 2002. Both the original and the 17/11/2002 observations are 5400sec exposures. Raw Image FEROS.2002-11-18T02:16:35.000.fits, object SMC R50, exposure time 5400.0sec.

Region of Hdelta


Region of HeI 5876 (in the right panel, the spike at the center of the HI line is a cosmic)


Region of Halpha


Region of Paschen 13

Clocks

The clocks on the BIAS PCs are supposed to be synchronised with UTC using the NTP daemon. However for unknown reasons this service has not been performing correctly. Instead as an interrim solution we synchronise the BIAS PC clock to the 2.20m time server w2p2tcs using ntpdate every ten minutes (via a cronjob). The result is that the clock then drifts by only approximately 0.033±0.002secs by the end of any given 10min period.

Commissioning V -- March 08/09 - 10/11 2003 + Reserved nights March 15/16 - 17/18 2003

Summary

Efficiency slightly improved compared to prior to intervention but still must be restored to nominal 15-20%.
This (mini) commissioning was required by the following changes carried out during the March-2003 FibreHead Upgrade Mission:

1. FibreHead Upgrade
   
2. Shorter Calibration Fibres (were 20m, now 15m): Will possibly improve throughput at shorter wavelengths.
3. FCU Fibre input unit lenses moved closer together: Should improve calibration lamp intensities as the lenses will now be closer to the center of the beam comming from the FCU.
4. NDFW in convergent beam: Should eliminate fringing
5. NDFW moved to new motor assembly with higher resolution encoder.
6. FCU Fibre input unit shutter now on former NDFW motor assembly.
7. Small Format TCCD coupled with new FFH-TCCD Optics
   

Log

• The measured pixel scale for the large format TCCD scale is 0.768 arcsec/pixel, the FOV is ca. 5.0×3.6 square arcmin.
• The focus offset between WFI  Besssel R filter (BB#Rc/162_ESO844)and FEROS is measured to be O_FEROS = O_BB#Rc + ~130e.u. This should be corrected for during the time the adapter is off between March 25th and the beginning of Period-71 at the 2.20m telescope.
  
• The Rotating Selection Mirror (RSM) position has been determined in order to balance the FF flux arriving at the FEROS CCD for the 2 fibres: position is 7268 e.u. Exposure times of 60sec give adequate S/N for DRS initialisation.
• The Rotating Selection Mirror (RSM) position has been determined in order to balance the ThAr flux arriving at the FEROS CCD for the 2 fibres: position is 3636 e.u. Exposure times of 300sec give adequate S/N for DRS initialisation.
• Lack of blue flux in the FFs was investigated and found to be due to the fact that the Halogen lamps simply do not produce flux at such short wavelengths. Deuterium lamps will be investigated.
• Calibration of Neutral Density Filter Wheel (NDFW) with FF images. Parameters of the density equation are:
  
  • Gain ==TBD
  • Offset ==TBD
  • Min == 0.0
  • Max == 2.5
  • Ref == 0.0
    

  The attenuation factor (A) is related density (D) according to A = 10^D. So the practical range of attenations factor is 1 to 10^2.5 = 316.23. Therefore given that the minimum reasonable ThArNe exposure time is ~100sec the maximum Object exposure time in Object Calibration mode is 31600sec ~ 8.8hrs. Exposure times shorter than the exposure time used for ThArNe calibration images will be underexposed compared to the exposures acquired for DRS initalisation.

Efficiency

The maximum efficiency measured thus far is shown below, with a peak efficiency of ~13% which is to be compared with 6.6% from Commissioning-III and 11.5% for Commissioning-IV. Raw Image FEROS.2003-03-11T08:52:35.000.fits, object HR4963, exposure time 180.0sec.

Relative Fibre Efficiency

The Relative Fibre Efficiency computed with twilight sky flat field, and @@ fiberratio. The throughput of fibres is almost exactly equal now. Raw Image FEROS.2003-03-16T22:04:39.000.fits, object Twilight sky, exposure time 120.0sec.

Resolving Power

Nominal Spectral Resolution (R) of 48000 is maintained.

Clocks

The clocks on the BIAS PCs now accurately synchronised with UTC on w2p2tcs using NTP.