Subject: drift scanning with SPICAM


Submitted: Tue, 05 May 98 22:06:30 -0600

Message number: 54 (previous: 53, next: 55 up: Index)

To: SPICAM observers
From: Jon Holtzman, Marla Geha (NMSU)
Re: drift scanning with SPICAM

Over the past several weeks, we have used SPICAM a few times to do
moderately short drift scans to observe spiral galaxies and background
fields. We decided to try this in an attempt to improve the flat
fielding and photometric performance of SPICAM, even though we take
a performance hit on overhead for the deep exposures we require because
of the limited exposure time allowed on a drift scan (although the
overhead hit becomes smaller as one takes longer scans).

In general, things performed well and as advertised in the SPICAM manual and
separate notes on drift-scanning. We discovered one important feature,
however. Unless one takes special measures, the beginning of the
drift scan will NOT necessarily be where the telescope is pointing
before the scan is commanded.  When SPICAM tells the telescope to start 
drift-scanning, the telescope will proceed to move to _intersect_ the
scan path. It will _not_ necessarily move to the _beginning_ of the scan
path. The difference depends on the amount the rotator needs to move
to align the columns with the commanded drift direction. If the amount
of rotation is small, then the actual path starts at the prior location,
but if the amount of rotation is large, then the actual path will start
downstream of the location which was slewed to.

Consequently, if you want to start scanning at a location which you slewed
to, you must set the proper rotation before starting the scan.
Of course, even if you start the scan at the location you slewed to, you will
not get good data at that location because it will fall in the initial
ramp frame.

As a related issue, we found that the rotation as commanded
from the SPICAM software commands the instrument to an absolute _physical_
rotator angle, i.e., not relative to N-S. There does not appear to be a
way to query the telescope for the current rotator angle (contrary to what
is written in the manual), so this does not allow the rotator angle to
be set from a SPICAM script without a full calculation of the absolute
rotator angle as a function of position in the sky. Consequently, we had
to command the rotation from REMARK (or the TCC). It would be very nice
to be able to command rotations relative to N-S from the SPICAM software, 
and to update the manual with the details of what actually happens when
rotations are commanded.

We also determined that the telescope is returned to the start position after
each scan automatically.

Given this behavior, we generated a simple procedure/script which will
give you n scans of a desired length with desired start position. The
procedure is as follows:

      1. Slew to the position of your object (e.g.,
         the spiral galaxy). When commanding the slew, set the desired
         PA:  if you want a scan from S to N, this will be 0, if you want
         it E to W, this will be 90, if you want it N to S, then use 180,

      2. Determine the desired length for fully exposed data in arcsec.
         Determine the desired offset for where you want the good (not the
         ramp) data to start. For example, if you want the first good
         data to be the center of the frame which you slewed to, the offset is 
         0, if you want the full frame where you slewed to to be good data,
         the offset is 1/2 frame width (~150 arcsec), etc...

      3. run the following script (driftgal) as follows
           driftgal pa length offset n

         This will back the galaxy off by the desired offset plus a half-chip,
         and start the scan. It will perform the desired scan n times without
         any intervention from you.
         Note that, although the pa is specified as a parameter (to get the
         drift direction right), the telescope must have already been commanded
         to go to the correct pa in order to get the drift started at the
         proper location. The sign conventions for the pa as needed by the
         script and as specified to REMARK are the same.

The driftgal script is enclosed below. The current script has a
built-in scan rate which is the slowest (longest exposure) allowed.

We did some brief tests of photometry using drift scanning by looking at
a field of standard stars, and drifting across the field at two perpendicular
directions; this gave us stars observed at both similar column positions 
(for which flat fielding should be irrelevant) and similar row positions 
(for which flat fielding is needed). We generated row flats from the data
themselves. For observations of just a few stars, we found the worst case
relative photometry to be 0.02 mag between the two drifts. Further tests
are clearly required to quantify performance better, but we were encouraged
by this result.

If any of the drift scanning software is modified in the future, we would
appreciate it if we could be informed so we can incorporate any changes
in our observing procedure.

# Usage: driftgal PA len offset nexp
#   PA: galaxy PA, we will drift along major axis
#   len: length of image we want with full full exposure time, in arcsec
#   offset: amount (of fully exposed frame) to offset before starting scan, 
#           in arcsec
#   nexp: number of exposures
# Current version has built-in scan rate at slowest possible speed
macro driftgal

  if ($0 < 5)
    echo "Usage: driftgal PA len offset nexp"
  end if
  tcc hi
  echo "offset -a 0 {($offset+145)/3600.}"
  offset -a 0 {($offset+145)/3600.}
#  time 25
#  object
  for var 1 $nexp 1
    echo "drift $nrows {$rate*dsin($pa)} {$rate*dcos($pa)}"
    drift $nrows {$rate*dsin($pa)} {$rate*dcos($pa)}

APO APO APO APO APO  Apache Point Observatory 3.5m  APO APO APO
APO  This is message 54 in the apo35-dsc archive. You can find
APO  the archive on
APO  To join/leave the list, send mail to
APO  To post a message, mail it to