Michelle Programs in Semester 04A

Michelle Programs in Semester 04A

The Michelle software has been substantially reworked for Semester 04A. In particular, there are changes to the acquisition sequence, which are reflected in the Observing Tool library’s details for Michelle. This document shows some Michelle specific items not covered in the normal user guide to the OT. It assumes that you have read that generic guide. It also discusses a Michelle-specific acquisition method – peak up after imaging acquire.

At the summit, MSBs are now executed via the “Observing Queue,” detailed in the execution manual


The OT library for Michelle is divided into several subsections, as shown in Figure 1.

Figure 1 – The Michelle library, showing the imaging, spectroscopy, and polarimetry subsections.


Figure 2 – An imaging sequence

Instrument component: A typical imaging sequence is shown in Fig. 2. Note that the target and instrument components are kept outside the standard and target sequences. Since exposure times in the mid-IR are mostly dependent on the sky brightness rather than the target’s magnitude, the Michelle component can be applied to both the standard and target.  Therefore, the Michelle component is inherited by both the standard and target sequences in this case. If your target is faint, increase the number of repeats.

For imaging mode, all you have to do in the target component is to select the appropriate filter. If you want to shorten or lengthen the time taken in each nod beam, you can also alter the observation time. The default 20 seconds for observation time actually means 20 seconds total in each chop beam (including array readout overheads). Therefore, the default setting will give approximately 40 seconds in each nod beam, or 160 seconds total for a nodded ABBA sequence. The target component also provides the detector duty cycle in percent. Use this number to estimate the actual exposure time in each chop beam. E.g., if the detector duty cycle is 80%, then approximately 16 seconds out of the 20 second observation time is spent exposing the source. 

NB. Exposure times are sky-dependent and rarely need to be changed. On the odd occasion that an exposure time needs reducing because the target is extremely bright or carrying out a classically scheduled run in poor conditions, please refer to the list of working exposure times to choose a more appropriate exposure time. 

Figure 3 – A typical image of a point source

Target information: Both the chop position angle (degrees east of north), and throw (arcsecs) are set in the target component, as this is an input to the telescope system rather than the instrument. The default chop angle is 0 degrees (north-south), with a throw of 15 arcsecs. The maximum chop throw while guiding is 21.7 arcsecs, the unguided maximum throw is about 30 arcsecs, but depends on the top end’s temperature. Chop frequency is determined automatically by the software and cannot be changed by the observer.

DR recipe: For bright point-like targets, you should use the NOD_CHOP_APHOT data reduction recipe. For fainter targets, the NOD_CHOP recipe is probably more appropriate. For extremely faint targets, use the NOD_CHOP_FAINT recipe. This last recipe assumes the target falls in the same place on the array as the standard star.

Repeats: To calculate how many times you should repeat a typical ABBA pattern, please refer to the tables in “Estimating the number of repeats required in a Michelle observing sequence.”

Offsets: Typical offsets for point source imaging are perpendicular to the chop angle and half the distance of the chop throw. This means that the source will always appear on the array even in the negative chop position, and the resulting ABBA pattern will produce a “square” chopping and nodding pattern, as shown in Fig 3.

Be sure to leave the final offset iterator in the sequence, i.e., the one after the observation. This moves the telescope back to your 0,0 position and is also required by the data reduction.

Flat fields: The actual benefit of taking flat fields in the mid-IR for imaging programs is questionable, although you can set them up in your program if you want to. Bear in mind that the imaging data reduction recipes will not use any flat fields your produce and that this will have to be done offline. To set up a flatfield sequence, copy one of the normal imaging sequences and replace the observation component with a flat component. Ensure you open the flat component and hit default (the flat source should be “sky”). You can offset the telescope as you would a normal observation, or stare at one piece of sky – the choice is yours. Remember that if you want to use a particular piece of sky, you can enter the target component’s coordinates. Alternatively, if you want to point the telescope to the east, for instance, and take flats at airmasses of 1, 1.5, and 2, you can specify azimuth and elevation angles in the target component.

Overheads: Typical overheads for Si filter imaging are 100%, i.e., 10 minutes on source requires 20 minutes overall. Because of the short exposure time needed, broadband filters result in much larger overheads of approximately 200 to 250%. 


Figure 4 – A spectroscopy sequence

Instrument component: As is the case for imaging, the instrument component typically remains outside each sequence in the MSB and is therefore inherited by all sequences in the MSB. In spectroscopy mode, the instrument component will allow you to select the grating, wavelength, order (for the echelle), detector sampling, the slit width (mask), and position angle of the slit. NB. If you want to chop along the slit, which is the default for point-like sources, then make sure that the position angle of the slit and the position angle of the chop in the target component are the same. 

Although the observation time in this component has the same definition as in imaging, it is more complicated due to detector sampling. The observation time gives the total time spent in each chop beam (including overheads) per sample position. Therefore, you may consider reducing the observation time to 10 seconds if you are using sampling of 1×2 (which moves the detector a distance of one pixel to remove bad pixels on the array) as this will give you approximately 40 seconds in one nod beam, or 160 seconds for an ABBA sequence. If you decide to select the default observation time of 20 seconds and use 1×2 sampling, you will spend 80 seconds in each nod beam, or 320 seconds for the ABBA sequence. With 2×2 sampling, you should almost certainly reduce the observation time to 10 seconds or less. With the observation time set to 20 seconds, 2×2 sampling will result in a nod position that lasts for 160 seconds, and you will almost certainly see problems due to thermal instability.

The instrument component will also provide an estimated detector duty cycle, which is the percentage of time actually spent exposing a source in each .chop beam. Use this number to estimate the total exposure time on your target.

Target component: We recommend putting the standard star information in the target component at the top of the MSB and your target information in the target component within the science target sequence. The telescope will then slew to the standard when doing the flatfield and remain there for the standard star observation. It will slew to the target once these two observations have been made.

The chop throw and position angle are also set in the target component. As mentioned above, if you want to chop along the slit, make sure that the chop position angle in the target component is the same as the slit position angle in the instrument component.

Figure 5 – A typical group spectral image of a bright source.

DR recipe: Select the STANDARD_STAR recipe for your standard star and POINT_SOURCE for your target. The POINT_SOURCE recipe can also handle data taken on extended sources. There are other flavors of the recipes. e.g., there are *_NOFLAT recipes if you decide to take data without first taking a flatfield.

Repeats: To calculate how many times you should repeat a typical ABBA pattern, please refer to the tables in “Estimating the number of repeats required in a Michelle observing sequence.”

Nod iterator: One major change we have made to the sequences compared to 2001/2002 is the use of nod iterators in spectroscopy sequences, as opposed to offset iterators. This makes setting up your spectroscopy sequence simpler in that you do not have to specify telescope nod offsets. The nod iterator allows you to choose from two options, AB or ABBA. The default is ABBA, which gives four observations for each sequence. Use the repeat option to increase the number of times the pattern is repeated to increase the signal to noise. The nod iterator uses the chop throw and angle settings to nod the telescope along the slit with the same throw as the chop. Typical resulting data are shown in Fig. 4.

The offset iterator can still be used if you want more flexibility for the telescope nod positions and are by default used in the MSB for nodding and chopping off the slit, and also in the echelle MSBs since the echelle does not utilize chopping.

Overheads: Expect total overheads of 100% for low-resolution spectroscopy and perhaps as low as 50% for medium to high-resolution spectroscopy. For bright sources, target acquisition should add two or three minutes to the total observation time.

Arcs: Arc lamp spectra are not currently used with Michelle; skylines are used instead. These come for free in your individual chop beams, so there is no need to set up anything specific for wavelength calibration. The data reduction will also produce sky frames, which you can use for offline wavelength calibration.

Target Acquisition

Figure 6 – The target acquisition component

The acquisition sequence has been modified to enable UIST-style imaging acquisition. This is the function of the “Target Acquisition” Observe (aka Acquisition Eye) shown in both the standard star and target observations within the MSB.

The user has to hit the default button in this component. Since only one filter is used for target acquisition, the exposure time is fixed and cannot be changed by the user. However, if the target is relatively faint, the number of coadds can be increased. At the time of writing, this acquisition mode has not been fully tested, so the limiting magnitude for acquisition is unknown. However, 30 coadds should be enough to detect a 650 mJy source sufficiently well for acquisition. Adjust the coadds as necessary depending on your source’s brightness, but do not drop the coadds below about 10 since this might result in poor chop subtraction.

At the telescope, the observer also has another chance to do a traditional peak up through the slit after target acquisition to ensure that the source is well positioned on the slit. If you are want to make sure the source is as accurately positioned as possible, you should mention in your observing note in the OT that the observer should check the peak up position.


Figure 7 – Imaging polarimetry

Imaging polarimetry and spectropolarimetry are covered in one section here. For specific information about imaging and spectroscopy setups, please refer to the sections above.

Instrument component: The only differences between imaging/spectroscopy setups and those for polarimetry are that 1) the polarimetry box at the top of the component should be ticked and 2) the exposure times will be longer in most cases due to the reduction in throughput caused by the waveplates. The exposure times will automatically default to the correct values if the polarimetry box is ticked.

Target componentThere is no difference in the target component when doing polarimetry. However, you may wish to limit the chop throw for point sources as the waveplates offer a limited field of view of approximately 40 arcsecs in Michelle’s current mode.

IRPOL iterator: For consistency within the OT, Michelle uses the IRPOL waveplate iterator in its sequences. However, Michelle has its own polarimetry module in the calibration unit, and the iterator will control this rather than the IRPOL unit itself. There is no need to specify which polarimetry module is used, and the observing software does this automatically. Please leave the final IRPOL iterator in the sequence (after the final offset of 0,0 in imaging mode) as this will move the waveplates back to their 0 angle position.

Programme checklist for PIs and support scientist vetting

  • Make sure that chopping is enabled in the target component of all sequences – even echelle programs. Even in spectroscopy modes that don’t require chopping (e.g., the echelle), chopping needs to be turned on in order for target acquisition to work correctly.
  • Check the chop throw and angle is correct, and that the chop angle is the same as the position angle of the slit for spectroscopy programs if you want to chop along the slit.
  • Is the polarimetry box in the target component ticked for polarimetry sequences?
  • For imaging programs, make sure that the final stand-alone offset iterator is in the sequence. This should contain one offset of 0,0.
  • In spectroscopy programs, make sure you have hit the default button in the flat components.
  • Please do not include arcs in spectroscopy sequences. Sky lines are used for wavelength calibration.
  • In spectroscopy programs, make sure you’ve hit the default button in the target acquisition component, or have selected an appropriate number of coadds for your target’s brightness.
  • Check you have the correct number of repeats to achieve your desired signal-to-noise.
  • While Michelle is on the telescope, we will be using the gold-coated dichroic which has reduced emissivity but also a reduced optical throughput to the guider. Through the central region of the dichroic (defined by the green dichroic square in the OT’s position editor) the faintest guide star you should select is about V=13. Outside this region, the limiting magnitude is around V=17.
  • Remember to include a site quality component in your program, otherwise the program cannot be successfully submitted to the database.
  • Make sure your observer notes are complete and that the “show to observer” box is ticked; remember other observers will be observing your targets for you.