11.1 Instrument Overview

The instrument sensitivity and resolution summaries are provided to permit estimating feasibility of scientific investigations. The HIPO performance summaries show the expected system performance for Full Operational Capability, which may differ from that during commissioning.

11.1.1 Instrument Design

The HIPO optical system is reconfigurable to meet its varied requirements. It incorporates two dichroic beamsplitters, one to divert the infrared beam to FLITECAM (if mounted) and one to split the red and blue sides of the HIPO optical paths. Either or both of these may be removed if desired. It is also possible to move either CCD so it is placed directly at the optimal SOFIA focal plane for highest spatial resolution and throughput. The optical design is described in detail in Dunham (2003, SPIE, 4857 62).

The 8-position filter wheels are located near the pupil image formed by the collimator optics. Two positions in the red CCD's filter wheel are normally used for Shack-Hartmann lenslet arrays, but these may be replaced if necessary for a particular observation. We have contemplated adding grism capability to HIPO but have not yet carried this out.

The region between the mounting flange and the gate valve on the telescope can be evacuated to reduce image degradation due to density fluctuations in this region of the optical path. When this is done a window is installed at the location of the gate valve.

HIPO provides a number of CCD readout modes as described in Dunham et al. (2008, SPIE, 7014, 70144Z). The most commonly used are the single frame mode and a frame transfer time series (basic occultation) mode with readout frequency up to 50 Hz. A variety of readout rates are available allowing the observer to optimize the subframe size, speed, noise, full well, and linearity tradeoff for a particular event. This is also discussed together with HIPO data format details in the "HIPO Data Cycle System Software Interface" document.

11.1.2 Angular Resolution

The primary HIPO detectors are e2v CCD47-20 1024x1024 pixel frame transfer CCDs with plate scales of 0.33''x0.33'' pixels at low resolution and 0.05''x0.05'' pixels at high resolution. The HIPO field of view (FOV) is a 5.6' square, the 8' diagonal of which corresponds to the 8' diameter SOFIA FOV. Pixels will normally be binned to best match the seeing blur size and to reduce the effect of read noise. The high resolution mode includes no re-imaging optics. It is possible to replace one or both of the CCD47's with CCD67's having half the field of view, twice the pixel size, and much faster imaging operation.

The HIPO image quality is dominated by seeing and image motion effects. The red curve in Figure 11-1 is the nominal image quality expected at first light for SOFIA, based on the expected shear layer seeing, the as-built optical performance, and 2'' rms image motion. The blue curve represents the ultimate combined optical quality and image motion requirement (80% encircled energy in a 1.6'' diameter circle) convolved with the expected shear layer seeing. Also plotted are representative photometry aperture diameters likely to be used for processing occultation frames under both conditions described above. The image motion assumed is larger than will be experienced when observing at high frame rates. Occultation photometry will be extracted from data frames using effective aperture sizes comparable to the 80% enclosed light diameter plotted here.

HIPO FWHM beam diameter as a function of wavelength

Figure 11-1: Plotted here is the expected instrument FWHM beam diameter as a function of wavelength.

11.1.3 Filter Suite

Wavelength range: 0.3 - 1.1 µm. HIPO currently includes standard Johnson UBVRI  and Sloan Digital Sky Survey u' g' r' i' z' filters. Occultation observations will normally be unfiltered for events involving faint stars or will use specialized filters such as the narrow-band methane filter (λc ~ 0.89 µm) for events with bright stars. Additional custom filters will be added as needed for specific events.

HIPO uses a dichroic reflector to separate its blue and red channels. Two dichroics are currently available with transition wavelengths of 0.575 and 0.675 µm respectively. Other dichroics will be added as necessary for specific events. The 0.575 µm dichroic transition wavelength is close to the boundary between the Sloan g' and r' filters and the 0.675 µm transition is close to the boundary between the Sloan r' and i' filters.

HIPO has been upgraded recently with the addition of deep depletion CCDs having a multi-layer antireflection coating. This provides higher quantum efficiency across the board with much higher quantum efficiency at longer wavelengths. It also nearly eliminates fringing at longer wavelengths. The figures below reflect this improvement.

Below are two plots (Figures 11-2 and 11-3) of the HIPO total system throughput including atmospheric extinction, SOFIA telescope throughput, and instrument throughput for each of the available bandpasses. The first figure shows the total system throughput for the Johnson and methane filters with the 0.575 µm dichroic. The second figure shows the Sloan filters with the 0.675 µm dichroic. These figures assume that the FLITECAM dichroic beamsplitter is not installed. The very low total system throughput for HIPO is due to the dichroic tertiary mirror in the telescope. The long-planned aluminized tertiary will increase our throughput by a factor of nearly 3, on average, when it becomes available. The corresponding throughputs are presented in Figures 11-4 and 11-5.

HIPO total system throughput 0.575 µm dichroic

Figure 11-2.

HIPO total system throughput 0.675 µm dichroic

Figure 11-3.

HIPO throughput 0.575 µm dichroic

Figure 11-4

HIPO throughput 0.675 µm dichroic

Figure 11-5

11.1.4 Imaging Sensitivities

Plotted in Figure 11-6 below are HIPO first-light sensitivities for several representative cases. These figures assume that the FLITECAM dichroic beamsplitter is not installed. The upper figures correspond to occultations by Pluto or Triton while the lower two are for the case of a very faint occulting object. The left and right figures are for 0.5 sec and 50 ms integrations, respectively.Each figure shows S/N for no filter (dichroic only) and for the 0.675 um dichroic plus standard Sloan filters. The dichroic transition is assumed to occur from 0.57 and 0.67 μm.

HIPO sensitivities

Figure 11-6: Plotted here are the anticipated sensitivities for HIPO for several representative occultation scenarios.

The deviation of S/N from a square root dependence is mostly due to shot noise on the occulting object in the upper left figure, about equally shot noise on the occulting object and read noise in the upper right figure, about equally shot noise on the sky and read noise in the lower left figure, and read noise in the lower right figure. Once the aluminized tertiary mirror is available read noise will only be significant in the lower right figure. The improved final SOFIA pointing stability will increase sensitivity for sky-limited events and improve discrimination from nearby bright objects (e.g. Neptune for a Triton occultation).

11.1.5 HIPO Sensitivity Spreadsheet

A spreadsheet exposure time and SNR estimator for HIPO is available here. This spreadsheet accounts for transmission of all optical components from the atmosphere through the telescope, and inside HIPO, ending with the quantum efficiency of the detectors. It accounts for shot noise on the target, a possible background object (important for occultation work), and sky brightness as well as read noise. The wavelength dependence of the stellar flux is simply modeled as blackbody radiation, with the effect that there are systematic errors in the predictions that vary with wavelength and spectral type. The spreadsheet has been checked against a number of spectrophotometric standard stars with calibrations traceable through Vega to blackbody sources. Observed signals of these standards with the deep depletion CCDs is typically ~10% higher than predicted but in the Johnson I and Sloan i' bands the observed signals are ~30% higher than predicted.


The spreadsheet is operated from its first tab, the Control Sheet. The next four tabs deal with optics that are common to both sides of HIPO (e.g. the atmosphere and telescope optics), optics peculiar to the blue or red sides, and the special case of the bare CCD. The last two tabs deal with blackbody and sky fluxes and the final combination of all the previous calculations that eventually appear back on the Control Sheet.

The spreadsheet is operated by filling in the yellow (most commonly used) and orange (less commonly used) cells to account for the star brightness and effective temperature, observing circumstances, CCD operations (integration time, binning, photometric aperture diameter), optics selections, and CCD choices. Results are returned in the light green cells. The main results are the signal levels from target, sky, and background object and the various noise contributions modeled by the spreadsheet. Scintillation noise is very difficult to measure and its value is not well determined yet. Finally the overall S/N ratio and fractional error are given. An estimate of the peak pixel signal in a star image is given in cells G34-H36. This is highly dependent on focus state and image jitter, and somewhat dependent on altitude so should be only used as a guide. The peak signal for the target star is also given in units of full well (assumed to be 100,000 electrons) in cells G35-H35. Note that full well is not affected by binning in a CCD47-20 since this CCD has no summing well. It is sometimes convenient to know the flux-weighted effective wavelength of a given filter/star combination. These are returned near the top center of the Control Sheet for the blue and red sides. For convenience the names of the beamsplitters, filters, and windows, or an indication if they have been removed, is given in the green cell above the selection cell for each optical element. The same applies to the CCD selection cells.

The optical elements modeled, in order, follow:

  • Atmosphere: Atmospheric extinction is a combination of ozone (assumed to be 300 Dobson units) and Rayleigh scattering appropriate for 41,000 foot altitude. This matches observations quite closely. The usual sec(Z) airmass correction is applied.
  • Telescope: The condition of the aluminum coatings on the telescope optics can be changed using cells J43-L43. At present the only tertiary mirror available is the dichroic tertiary and cell L43 has no effect. Cell B26 allows selection of either the dichroic tertiary or the not-yet-available aluminized one. If the aluminized tertiary is selected the aluminum coating condition in L43 does have an effect. The SOFIA coatings are maintained in excellent condition so cases 0 and 1 are the only reasonable ones to use.
  • Fore optics: If the tub is to be evacuated the gate valve window must be installed. Whether this window is installed or not is modeled by selecting 1 or 0 in cell C26. The dichroic reflector used in the FLIPO configuration is controlled by cell D26. It can be left out (0), be the new Reynard dichroic (1), or the old Lick dichroic (2). The Reynard dichroic has about half the emissivity of the Lick one and similar optical performance so is likely to be preferred in most cases.
  • The dichroic beamsplitter internal to HIPO is controlled by cell E26 and can be left out (0) or be one of the two presently available ones (1 or 2). If it is left out no light reaches the blue side CCD.
  • Filters for the blue, red, and bare CCD cases are controlled by cells F26-H26. There is a short explanation of what the various codes mean to the right of the cells used to select the various optics. New filters are certain to be added from time to time.
  • CCD selection is made using cells F30-H30 with 0 being the original standard silicon CCDs and 1 being the newer deep depletion CCDs.The deep depletion CCDs have substantially higher QE and will be preferred in most cases in spite of their greater cosmic ray sensitivity. For UV work the thin blue CCD has better sensitivity than the deep depletion CCD.

    A helpful figure showing the total signal per micron of bandwidth as a function of wavelength is shown in the upper right of the Control Sheet. It gives a graphical representation of the total signal accounting for everything from the star and sky through the CCD. The vertical axis on this graph autoscales.

    An informational box related to the systematic error introduced by the blackbody approximation for stellar fluxes is given in the lower right of the control sheet. Errors at the ends of the wavelength range for very red stars can be ~50% but are 30% or less for most cases.

    Another informational box at bottom center provides a crude estimate of effective temperature given either spectral class or B-V for a target object.

    Finally the revision list is given at bottom left. This spreadsheet will be updated from time to time as optics or detector selections change, or as additional information becomes available. The sky brightness estimate is currently a weak area and may be changed relatively soon.