Like any other observing mode, frequency switching has its own set of special
problems. Most importantly, this mode is sensitive to atmospheric line
features which are efficiently canceled in other observing modes. The
principal trouble maker for many observations is mesospheric CO,
but other minor atmospheric
constituents, like ozone, may also play a role. At 2.6 (1.3) mm, these CO
lines were found to be about 0.5 (4) K strong and about 0.7 kms
wide, not
unlike emission from a typical galactic dark cloud. These features can
interfere
either directly with the target line if the Doppler shift of the target is near
zero, or interfere in other less obvious ways (see Fig. 3
for a simple example)
We strongly recommend that the
prospective observer checks such interference which can usually be avoided
by chosing a suitable observing season.
The Technical Report gives recipes on how to do this. In addition, it is
planned to include in ASTRO a command which provides further help.
Another important boundary condition is that the frequency throw must be
(considerably) larger than the velocity width of the target line
if line cancellation is to be
avoided. Since the maximum practical throw is currently about 45 kms
,
many potentially interesting FSw targets remain out of reach, like CO sources
in the galactic plane, many circumstellar envelopes, or extragalactic lines.
A different problem arises if sources with a strong continuum are observed.
On such sources, the spectroscopic baseline ripple is of more complex nature,
since more than one scattering path tends to be involved. The simple trick
of canceling the ripple by setting 
equal to the ripple period does
therefore not work as efficiently. But not much experience could be acquired
yet on this matter.