Useful tips on writing TTF proposals
The signal-noise performance calculator
is a good place to start, not least because it provides a summary of most
of the free parameters which go into a proposal.
Detectors. The CCD detectors available
at the AAT are detailed here.
The field of view and plate scales are shown below:
f/8 (9' max field of view)
f/15 (5' max field of view)
1Kx1K (24um pix)
MITLL3 2Kx4K (15um pix)
2Kx4K (13.5um pix)
The key TTF parameter is simply the resolving power
you wish to use at a given wavelength. This figure illustrates the range
of "safe" resolving powers across the optical window.
Choosing a bandpass.
If you are unsure what bandpass to use, here are some useful tips.
The TTFs are periodic in their instrumental profile. It is
important to consider which blocking filter best suits your particular
need. At small plates spacings, where the resolving powers are lowest,
the orders are well separated in wavelength (i.e. the free spectral range
is large). It is possible to block light pollution from the neighbouring
orders with conventional broadband filters. At the largest spacings, where
the resolving powers are highest, neighbouring orders are closely spaced
in wavelength. This requires much narrower filters to block the neighbouring
orders. As a general rule, the blocking filter should be comparable to
or smaller than the free spectral range.
Are there other emission
lines that you need to avoid, i.e. night sky lines, astrophysical lines
? If so, select your bandpass to be less than the separation of your chosen
line from the neighbouring line. If the other line is a strong atmospheric
feature, you should make the bandpass as narrow as possible. Play with
the TTF calculator to see the effect of this.
Are there atmospheric
absorption troughs that you need to avoid (see below)?
If so, select your band accordingly. Again, the TTF calculator will
Does your program have
internal kinematic structure? If so, choose your band to encompass the
internal dispersion. For extragalactic programmes, don't forget the (1+z)
Alternatively, you may
want to see kinematic structure, in which case, make the band as
narrow as possible and step the TTF through the velocity range with
maybe a few extra steps to see the continuum level. But of course, this
is now a bigger TTF proposal.
is an instrumental effect to be aware of: the TTF field of
view is far from monochromatic at the highest resolving powers. From centre
to edge, there is a parabolic phase shift of 18A. Thus, at f/8, if the
central wavelength is 6563A; at a radius of 2.5', thewavelength is 6557A;
at a radius of 5', the wavelength is 6545A. Bland & Tully (1989,
AJ) show that the phase effect is (a) wavelength independent, and (b) a
fundamental consequence of any form of constructive interference,
monolithic filters. To underline this further, a splendid comparison
of the phase effect of the TTF vs. the phase effect of a monolithic
26A filter is given here.
The crucial point is that you can compensate completely with the former,
and not at all with the latter. If your objects extend over large
fractions of the field, and your bandpass is 6A say, you will need three
steps to compensate for the phase effect. Obviously, the effect is much
less important for programmes which exploit broad bandpasses.
This is what the distribution
in FSR looks like:
= 40 * lambda / R
If you do not block
the neighbouring orders, you are susceptible to (a) spurious emission lines
appearing within your bands, (b) an increased sky background by a factor
of the blocking filter width divided by the FSR. In this event, you will
need to degrade the SNR result derived from the calculator by the square
root of this factor (assuming, as in most cases, that you are background
The TTF produces
40 independent spectral bandpasses between neighbouring orders, i.e., FSR
= 40 * lambda / R =
40 * bandpass.
Therefore, try to ensure that your blocking filter is at least comparable
to this value, if not narrower.
The available broadband
blockers are illustrated here. The available
intermediate band blockers are shown here.
The RTTF blockers are shown superimposed on the atmospheric airglow
lines here. Note that the
intermediate blockers give you 20% more coverage to the blue with filter
tilts, as shown here. Text files for all
of the transmission profiles can be retrieved from here.
Surveyors of arbitrary emission lines at arbitrary redshift routinely
forget to check what the atmosphere is doing in the chosen bands.
This is particularly true of those important off band exposures.
The TTF calculator incorporates a detailed model of the atmosphere
based on echelle observations at the AAT and at CTIO. Spend some
time trolling through these plots. Note that
if you are working in narrow bands, you could end up observing in an atmospheric
trough with near zero transmission!
The web tool for making arbitrary shuffle files is not quite ready. There
are different modes of shuffling one might consider. For now, since the
shuffle overhead is essentially negligible in all modes, proceed with your
SNR calculations for each of the individual bandpasses required.
The MITLL chips are sufficiently large to allow the full TAURUS field to
be shuffled between two discrete frequencies.
You can shuffle any number of frequencies (maximum 250) although the available
CCD field begins to decrease like N/(2N-1). For example, a 3-shuffle partitions
the 4K rows into 5 parts, and therefore the available field of view for
each frequency is 5' x 10' (f/8).
line imaging is almost always performed in comparison to an off band image.
Rather than observing a single off band in shuffle mode, it is possible
to use any number of off bands around the on band, although the off bands
are all imaged onto the same detector region. This method is well suited
to removing slopes in the underlying continuum imposed by the sky, object
or instrumental response.
to be commissioned.
series shuffles. Yet
to be commissioned.
series read out. This method bypasses the N/(2N-1) limit and
has been used to good effect by Margon & Deutsch (1998; PASP,
Aug.) and Tinney & Tolley (1998; MNRAS, in press).
line shuffles. The
instrument sequencer allows up to 250 shuffle steps; this is the current
limit on the detail of an arc shuffle image. Since the TTF finesse
is 40, Nyquist sampling means that we can map the spectral response fully
through three consecutive orders of interference.
is possible to open & close the shutter between each shuffle step,
or to keep the shutter open for the entire exposure.