Two focal plane scales are available for direct imaging ; the fast camera with 0.5''/pixel, and the slow camera
with 0.25''/pixel. The instrument can be configured for either of
these by typing the commands FAST or SLOW. Direct
imaging in the 3-4 
m window must be done with the 0.25''/pixel
scale to avoid saturation on the background flux. Images can also be
obtained at M using the 0.25''/pixel scale and the red leak in the
P 
narrow band filter to reduce the background intensity to a
manageable level. Broad band filters are background noise limited at
both scales, while some of the narrowest narrow band filters are
marginally dark current/read noise limited.
Direct imaging is performed using the Direct_Imaging observing mode which is set by typing:
CASPIR/MODE=DIRECT_IMAGING
The preferred readout method for direct imaging is method 2. This method forms a difference between the end of the integration ramp and the reset level (see §4.3.2), so is immune to DC voltage drifts. However, the reset pedestal pattern is imprinted on the data in these method. It is essential to record dark and bias frames to permit linearization and remove the pedestal pattern.
It is often necessary to use readout method 1 for direct imaging in
the 3-4 m band and at M. This is the fastest readout method
where the data are referenced directly to electrical ground (see
§4.3.1). Method 1 is susceptible to DC voltage
drifts because it does not perform a voltage difference. This is
manifest as a column striping with a four column period which changes
from frame to frame. The pattern can be characterized in clear sky
regions and subtracted during data reduction, but a better alternative
is to use readout method 2 when background levels permit.
Readout method 3 differences the signal level at the end and beginning of the integration ramp (see §4.3.3). Reset pedestal is not imprinted on the data in this method, but it is susceptible to DC drifts between the beginning and end of the integration interval. This is removed in readout method 4 (see §4.3.4) by additional samples of the reset level at the beginning and end of the integration period. Further performance characterisation of these methods is required.
In reducing your data (§6, §8, and §9), you will need bias frames and dark frames recorded with the same exposure time as each object frame and with the same readout method. These can be obtained by typing, e.g.:
CASPIR/DARK/TIME=5.0/CYCLES=10/METHOD=2
The minimum integration times in each readout method are listed in §4.3 (0.2 sec in method 1, 0.3 sec in method 2, 0.4 sec in method 3).
Three broad band K filters are available (see Table 19 and Appendix K). Kn is the preferred broad band K filter to use from SSO. Its short wavelength edge is similar to the original K filter, but its long wavelength edge is tailored to exclude much of the thermal emission in the long wavelength end of the original K band. K' was designed with the same goal for observation at Mauna Kea Observatory (Wainscoat & Cowie 1992, AJ, 103, 332). It extends to short wavelengths with poor transmission from SSO, so in practice it has a similar band pass to Kn from SSO, but the short wavelength edge is defined by vagaries of atmospheric transmission rather than the known properties of the filter. The K filter is the original broad band K filter. It is less sensitive than Kn due to the thermal contribution to the measured background flux, and should only be used when it is necessary to accurately reproduce data on the original broad band K system. The following transformation between K and Kn has been determined from model stellar atmosphere energy distributions and measured filter transmission curves:
It is not possible to use the broad band L filter without saturating
on the thermal background at the shortest integration times and
smallest pixel scale. The 3.6 m continuum filter is used as a
narrow band L filter substitute and is referred to below as nbL.
The red leak in the P narrow band filter requires it to be used
with a PK50 glass blocker. This is not mounted with the P
filter so that it can be used as an attentuator with the M filter.
If the P filter is selected by typing:
CASPIR/FILTER=PBETA
the PK50 blocker in the Lower Filter Wheel is automatically selected.
If the P filter is selected by an explicit upper filter wheel
command, this will not be the case.
The AAO [Fe II] filter is centered slightly redward of the standard [Fe II] filter (see Appendix K). It should also be used with a PK50 blocker that is automatically selected if the filter is requested by typing:
CASPIR/FILTER=AAOFEII
Standard stars for direct imaging are listed in Appendix
F. The IRIS standards in Table
21 are recommended for JHK imaging. The
standard values are listed on the Carter SAAO system. Care should be
exercised in selecting these standards as many of them may saturate
the array in good seeing using the 0.5''/pixel scale. Fainter
equatorial JHK standards can be found in the UKIRT list (Table
22). The original MSSSO standards (McGregor
1994, PASP, 106, 508) are listed in Table 23.
These are recommended for the 3-4 m region and at M.
Sky subtraction is most demanding in the 3-4 m band and at M.
Since most of the background is thermal emission from the telescope,
it is advisable to perform these observations away from the zenith
where field rotation is slower. This is so that the view of the
telescope structure seen by CASPIR changes only slowly during the
observation sequence.
Direct imaging is usually performed as a dithered mosaic using the
CASPIR/DO=filespec command (see §5.2). The
actions performed during the execution of a DO file are controlled by
two parameters, the TIPTILT parameter and the STAGE_OFFSET parameter.
The values of both parameters are displayed in the MISC Status Display
selected by typing:
CASPIR/DISPLAY=MISC
The value of the TIPTILT parameter is set by the
CASPIR/TIPTILT=... command and the value of the STAGE_OFFSET
parameter is set by the CASPIR/STAGE_OFFSET=... command (see
§5.2).
If the TIPTILT parameter is set, the Tip-Tilt image correction system is operated during the execution of the DO file or in NOD mode. DO file parameters specify the type of correction to be used. If the STAGE_OFFSET parameter is set, the IMB X-Y stage is moved in response to each telescope offset to bring the specified reference star onto the Tip-Tilt sensor. The default action is to leave the correction subframe fixed with respect to the Tip-Tilt sensor. The geometry of the offset pattern is then set by the accuracy of the IMB X-Y stage settings, rather than the offsetting ability of the 2.3 m telescope which is the case if the Tip-Tilt system is not used. If the STAGE_OFFSET parameter is not set, the IMB X-Y stage is not moved in response to telescope offsets but the correction subframe is moved with respect to the Tip-Tilt sensor to reacquire the reference star. The former mode is suited to mosaics using large offsets. The latter may be preferred where more accurate offsets of small amplitude about the science object are required, since it is not subject to IMB X-Y stage setting errors. The operation of the Tip-Tilt system is more fully described in §3.13.
Before using the Tip-Tilt system in this way, it is necessary to calibrate the scale of the X-Y stage motions. Center a star in the CASPIR array and move the X-Y stage to center the same star in the Tip-Tilt sensor display by typing:
IMB/XY_INCREMENT
and using the keypad cursor keys to move the X-Y stage. Exit from
this by typing a Q. Then zero the X-Y stage offsets at this
position by typing:
IMB/XY_ZERO
Move the X-Y stage 20 mm East by typing:
IMB/Y=-20.0
and drive the telescope west to recenter the star in the Tip-Tilt sensor display. Record the position of the star in the CASPIR Idle Display. Repeat the procedure with the X-Y stage at Y=20.0 mm and determine the X-Y stage scale by adopting the nominal value for the CASPIR image scale (0.5''/pixel or 0.25''/pixel). Determine the appropriate correction factor to the nominal X-Y stage scale (5.0''/mm) by dividing by 5.0, and input this correction factor by typing:
CASPIR/XY_SCALE_FACTOR=factor
The X-Y stage scale factor is normally about 1.055. The X-Y stage should now accurately define the requested mosaic pattern, and it should be possible to register the data frames using the adopted array pixel scale and the OFFRA and OFFDEC offsets recorded in the FITS file headers.
Tip-Tilt DO file commands are described in §5.2. In the simplest application, it is necessary to enable tip-tilt correction on object frames and disable tip-tilt correction when measuring sky frames. This is done by adding a tiptilt or notiptilt DO file command to each DO file line; tiptilt enables tip-tilt correction and notiptilt disables it. The type of Tip-Tilt correction must be specified at least in the first run of the DO file using the tt_mode command (typically tt_mode=correct).
Guide star acquisition procedures are described more fully in §3.13 and the guide star acquisition hardware are described in §4.5.