M.A. Holdaway and F.N. Owen
National Radio Astronomy Observatory
Tucson, AZ 85721
February 20, 1996
20 antennas need to be moved, a reconfiguration
might proceed smoothly with only two transporters. Observations to
determine the pointing parameters and the baselines should take on the
order of an hour each. High winds will hinder reconfiguration.
During the winter, daytime winds exceed 30 mph about 25% of the time,
and exceed 40 mph about 8% of the time, so we must either design the
transporter/antenna system to be able to withstand fairly high winds,
or we must be prepared for many reconfiguration delays.
With four planned MMA array configurations, it is important to understand how quickly the 40 element array can be reconfigured. We will estimate the antenna move time for the MMA using the experience of the VLA's antenna group, modifying the times of various operations which will differ between the VLA and the MMA.
According to Jim Ruff and others at the VLA site, a typical VLA antenna move consists of the following operations:
Hence, for a short VLA move of 1500 m, an individual antenna move consists of about 95 minutes of disconnection/reconnection overhead, 30 minutes of move time, and perhaps 30 minutes of miscellaneous overhead. At this rate, 6 antennas can be moved per day with two transporters. Obviously, with 40 antennas, the MMA will need to move the antennas in less time.
A newly moved antenna cannot participate in any observations until its
pointing parameters have been determined. During reconfiguration at
the VLA, pointing solutions are determined each night for the subset
of antennas which were moved earlier that day. Interferometric
determination of the pointing parameters is much faster than single
dish pointing determination. Gross pointing offsets are often several
beam widths, making standard ``five point'' pointing observations
impossible. A simple pointing procedure makes a rough solution for
the dominant pointing errors with about 30 minutes of observing. In
order to solve for the full seven parameter pointing model, the VLA
covers its AZ-EL space by observing 
80 bright calibrator sources, one
ever 3 minutes, over 4 hours.
Accurate baseline solutions are usually made after all antennas have been reconfigured, using all antennas for the solution. Observations performed with bad baselines can be corrected for the new accurate baseline solutions retroactively. The baselines are determined from observations of about 80 sources taken over 4 hours, covering AZ-EL space.
The VLA proceeds with useful astronomical observations during the few days required for each reconfiguration, often performing flux monitoring of unresolved sources which does not require a good synthesized beam. However, it would be advantageous to reconfigure the MMA antennas more quickly, making the array more flexible. We may want to cycle through the entire ensemble of configurations during the winter months so the best phase stability benefits every kind of observations. Significant reductions in the time required for a single antenna move may be achieved for the MMA:
The transportation The actual two way transportation time would be about 2 minutes in moving from a D to a C array station, about 7 minutes in moving from a C to a B array station, and about 25 minutes in moving from a B to an A array station. Hence, the move time for a single MMA antenna will range from 45 to 70 minutes if a separate crew reattaches the cables, or 60 to 85 minutes if one crew performs all move operations.
The work shifts at 5000 m in Chile will probably be less than 8 hours
to prevent physical exertion, and we assume 6 hours. The ``Sistema de
Turno'' as envisioned by Mark Gordon for the MMA operation entails two
staffs of 
30 workers changing about once a week.
Reconfiguration could be performed when the two staffs overlap,
enabling about 12 hours of reconfiguration work each day. If a single
antenna transporter is used, reconfiguration of all 40 antennas would
take about four days. If three antenna transporters were used,
a reconfiguration could take just over one day. If the ``Sistema de
Turno'' is not used, we may need to make special arrangements for two
shifts during reconfigurations in order to keep the move time down.
We are investigating the utility of hybrid configurations to aid in low declination observations, and possibly to provide different Fourier plane distributions. Reconfiguring between a regular array and a hybrid array requires moving about 20 antennas, which could be performed by two transporters in one day.
The reconfiguration will be slowed down by bad weather, which will
come mainly in the form of high winds on the Chilean site. The VLA
does not move antennas when the winds are above 20 mph. The daytime
wind speed during the months June through October on Chajnantor is
higher than 30 mph about 25% of the time, and higher than 40 mph
about 8% of the time. We would want to design the antenna transport
vehicle to be able to function, loaded with an antenna, in winds
at least as high as 30 mph. Since the atmospheric pressure is half
of the sea level pressure and the force on the antennas goes like
, the 30 mph specification might not be unreasonable. Even so,
we will need to expect significant delays in the reconfiguration.
Hence, we boost the full reconfiguration time estimate from a bit over
a day to two days.
The nighttime winds on the Chilean site are fairly benign, typically half as strong as the daytime winds. If we were able to reconfigure antennas at night, it would eliminate numerous delays and relax the specifications on the transporters. However, night antenna moves may not be possible.
The MMA will rely heavily upon the 30 GHz system for pointing and
baseline determination. The 30 GHz system will be very sensitive due
to the low system temperature and the high fluxes of the calibrator
sources, and the long wavelength will largely eliminate wavelength
ambiguities in the baselines. The raw sensitivity of the MMA will far
exceed that of the VLA due to its large bandwidth (8 GHz as compared
to 50 MHz), and the MMA's agile antennas and operating system will
enable the array to leap around the sky quickly. We assume 80 sources
can adequately sample AZ-EL space (as at the VLA), and that each of
the 80 sources will be brighter than 1 Jy at 30 GHz (there are about
200 sources over the entire sky which are brighter than 1 Jy at
90 GHz). We will have adequate SNR on a 1 Jy source for pointing in
just a few seconds, or 15-30 s to perform a ``five-point''
observation, including move time. The next source will typically be
about 
distant, requiring a slew of 7-15 s. Hence, the entire
80 source pointing run will take under one hour. Atmospheric phase
fluctuations and opacity will almost never limit pointing observations
at 30 GHz.
The requirements of the global baseline determination are similar to
the pointing, so they can also be completed in under one hour. The
requirement of no coherence loss across a 8 GHz bandwidth implies that
the baselines need to be known to only about 4 mm. With observation
of 
80 calibrator sources on 780 baselines, the atmospheric phase
fluctuations will permit this level of accuracy on even the longest
baselines (3 km) essentially all the time. While the atmosphere will
usually limit the accuracy of the baseline determinations, accuracy of
a few tens of microns will often be achievable. With phase
calibration being performed on minute time scales with calibrators a
few degrees distant, phase errors from baselines determined with
millimeter accuracy will yield acceptably small phase errors.
If the MMA is designed with fast reconfigurability in mind, we should be able to reconfigure the entire array in under two days with three transporters. If hybrid configurations are used, as at the VLA, and only some of the antennas need to be reconfigured, each partial reconfiguration can be done in under one day, and only two transporters may be required. The transporters must be designed to operate in at least 30 mph winds. Even so, delays due to high wind speeds will preclude antenna moves about a quarter of the daytime during the winter months at the Chilean site.
Reconfiguration would probably proceed during the day, when the opacity, phase noise, and wind velocity are all highest. Hence, we will not usually lose quality observing time to moving antennas. Shortly after sunset when the phase stability and winds settle down, pointing and baselines will be determined. The 30 GHz system will be important for pointing and baselines due to the brightness of quasars at this frequency, the sensitivity of the low frequency system, the large beam size, and the reduction of wavelength ambiguities in the baseline determination.
Features of the transporter, antenna, pad, array, and operations design which will speed up array reconfiguration include: