At the 30m telescope frequency switching is done with the reference frequency (normally near 100 MHz) of the phase lock unit, thus switching the first local oscillator. Hardware and frontend software is in place to switch all 4 standard SIS receivers. The present limitation of the OBS frequency switching command to a maximum of 2 receivers and other considerations make it, though, that during the trial period only the 3mm SIS and/or the 1mm SIS receiver No.1 (commonly called ``G1'') can be used.
The quality of the spectroscopic baselines in FSw spectra was found to depend
on a number of factors (see the Technical Report), most notably
(i) the frequency throw 
,
(ii) the number and location of receivers used, and
(iii) the stability of a receiver at a particular frequency.
In frequency switching on sources with little or no continuum, the baseline
ripple is found to be strongly dominated by the path involving the
subreflector, giving a dominant ripple period of 7.9 MHz. We found that this
ripple can be suppressed to a large degree by making the frequency throw 
equal to (or, to a lesser degree, twice) this period. Efficient simultaneous
1 and 3 mm observations are thus possible (Fig. 3).
Thanks to the flexibility of the autocorrelator backend whose bandwidth should
ideally be twice the frequency throw, it is even possible
to have the same velocity scale at the two receivers.
Very good baselines can be obtained if only one receiver is used, the 3mm SIS
or G1, by setting the Nasmyth beam distribution optics to a
special mode (main polarization grid removed, dichroic set to 
)
where the scattered wave is better rejected than with the standard optics.
Usually still acceptable baselines are obtained when these two receivers are
used in parallel as long as the dichroic is optimized for the 1 mm receiver.
All receivers have acceptable stability at most frequencies
investigated (for details see the report) if SSB tuned with the recommended
amount of USB rejection is used. We cannot exclude, however,
that there might be a few bad
frequencies where receiver stability is so poor that FSw observations are
impractical. Best stability (often 
on a 1 sec time scale)
and best baselines were obtained at 3mm.
Figure: Frequency switched observation of CO transitions at a
position in a dark cloud near the northern celestial pole.
The spectra were observed with the 3 and 1 mm receivers in parallel, with
frequency throws set to 7.9 and 15.8 MHz, respectively. The
autocorrelator was used with 20 (40) MHz bandwidth and 20(40) kHz
resolution for the 3 (1) mm transitions. Integration time is 15 min, the
weather was poor (1mm opacity 
).
Special optics (see text) was used to reduce
the baseline ripple, but no baselines are removed. Note the presence of
the mesospheric CO line (arrow) which is partially blended with the
negative line from the source. Such interference can usually be avoided
by changing the frequency throw.
