CFH12K - Observing Overheads


Telescope overheads
When moving from a target to another, several actions take place: slewing the telescope to the new position while rotating the dome, then selecting a guide star. On a given target, small offsets usually take place between exposures.

The following list contains the exact timing of these actions:

When slewing to a new position, the dome rotation is the time limiting factor in most of the cases.

A camera readout can take place during a slew and dome rotation as long as the prime focus bonnette X-Y stage (guide camera pick-up mirror) is not activated. Otherwise extra noise is injected on the CFH12K images. For example, this property is used by the twilight flat-field sequencer: the telescope is slewed back to zenith during the readout to maximize the time spent collecting photons in this time critical period.

The prime focus bonnette Z stage (focus) can be moved during a readout with no impact on the CFH12K images noise characteristics.


Camera overheads
CFH12K actions also require time: selecting a new filter, preparing the acquisition system to take an exposure, readout out the CCDs.

An important feature of DetCom (the CFH12K data acquisition system) is its capabilities in reducing the camera overheads by running (safely) multiple actions in parallel. For example, as soon as all the pixels are read into memory a new exposure can be started while the previous image is still being written to disk.

The actual pixels readout is 58 seconds long but the pre-exposure sequence adds several seconds to configure properly the acquisition system: grab an unique file odometer number, check if the camera voltages are all right, clean the detectors, etc...

The filter selection (CFH12K is equipped with a four position filter wheel) takes about 20 seconds but this task is put in the background and the final position will be checked for successful reach when the exposure will be launched. This is why selecting a new filter should always be the first command issued when changing the observing configuration.

Commands can also be piped into DetCom and they will be quickly executed when the prompt is returned at the end of the current action (usually an exposure). This allows saving the time typing and validating the commands (a few seconds!).

Focusing CFH12K required a mixing of camera and telescope actions. Its total duration is determined by the number of exposures (from 3 to 9) and the individual exposure times (from 10 to 20 seconds typically). Guided offsets take about 5 seconds. The number given in the following list is valid for a seven 15 seconds exposures sequence.

An automatic offset is applied when a new filter is selected. Moving the vertical stage of the bonnette is faster than getting to the new filter position.

The following list contains the exact timing of these actions:


Example of observing overheads

The following list contains several examples of observing scenarios timing:

  • Going from T1=[RA=18:00/DEC=-20.0] to t2=[RA=21:00/DEC=+17.0]

    1: Start slewing as soon as the exposure is completed
    2: Select new filter and exposure parameters when slewing
    3: Be prepared to launch the exposure as soon as the Observing Assistant has selected a guide star

    This sequence will be dominated by the dome rotation and the guide star selection, say 45 + 20 = 65 seconds. Since the camera readout time is less than a minute, by the time the new guide star is selected the camera has been "waiting" for only a few seconds.

  • Taking a set of exposures using a dithering pattern

    1: Focus if necessary in the current filter
    2: Launch the "dp12k" script from the graphical interface

    This sequencing is simple because all the commands have been integrated in the dithering pattern script. Based on the optimal breaking of the total exposure time into individual exposures times that leads to the total number of exposures (depend on the filter used, see the "Detector constraints" section in the "Optics and optical performance" page). The exposure time should be long enough to be photon noise dominated and short enough so not too many objects are saturated. Also fragmenting into several dithered exposures, while increasing the overhead, however helps recovering the signal lost in the gaps between the CCDs and the bad columns, as well as eliminating the cosmic ray hits.
    Let's consider that we are observing in the R band, using 600 seconds exposures and a 8 positions dithering pattern.
    The total time is (600*8)+65*8+5*8 = 5360 seconds. The overhead in that case is low: (600*8)/307 => 11%. This is a very typical number for CFH12K which overall efficiency ranges usually from 80% to 90% for such sequences.

  • Taking photometric standards through 3 filters

    1: Focus if necessary in the current filter
    2: Have a script ready to launch (see observing manual):

    clicmd etype o
    clicmd fits object SA 113
    clicmd fits comment Photometric standard - half night
    clicmd etime 10
    clicmd filter 2 R
    clicmd go
    clicmd etime 7
    clicmd filter 3 I
    clicmd go
    clicmd etime 15
    clicmd filter 0 B
    clicmd go

    This sequencing is simple because all the commands have been integrated in a script. The total time is (10+7+15)+3*65+2*20 = 307 seconds. The overhead in that case is large: (10+7+15)/307 => 89%! But these are photometric standards and there is no choice but taking very short exposures.


Typical observing overheads numbers
For example, using the 6 positions dithering pattern (dp6) for a total exposure time of 3600 seconds (1 hour), when one includes the pointing, focusing, and pattern's overheads, the total telescope time on this field ends up being 1 hour and 15 mn (20% overhead).

Comments on the CFH12K pages to J.-C. Cuillandre: