The AT Antenna Panel Upgrade
Tests on antenna CA03

M.Kesteven & R.Subrahmanyan

ATNF doc: at/39/090/

Introduction

The perforated panels on antenna CA03 were replaced by new, solid surface panels in September 1998. The period since then has seen a number of investigations designed to determine whether the upgrade should continue. Two questions, in particular, were posed:

• Will wind-loading make mm operation doubtful? In other words, on average, during the mm season, is the dish likely to be deformed or mis-pointed to the extent that operations are compromised ?
  
  
• Will wind-loading compromise lower frequency (currently successful) operations ? Will the pointing be so affected during windy conditions as to degrade a significant fraction of the low frequency observations?

Deformations will lower the gain. Mis-pointing enters at two levels - the main beam gain will drop (A shift of 14% of FWHP results in a 5% loss), and the mosaic imaging quality. (The MMA criterion is 2% of FWHP) These criteria translate to 5" and 1" respectively at 90 GHz.

The two main lines of attack were :

• Experimental - we compared the tracking stabilities of the resurfaced antenna and the un-modified antennas on a number of occasions. Very little difference was found.
  
  
• The observations were hampered by the difficulty in finding windy observing periods - there are not many at Narrabri.
  
  
• Theoretical - we revisited the wind-loading calculations for the AT antennas, with data relevant to a solid surface.

These investigations indicate that the increased wind-loading on the outer panels will not have a significant effect on observations.

The investigations do reveal some low-level problems in the tracking stability; we believe that these arise in the control system and should be addressed in the new generation control computer.

Wind-loading Calculations

Mr. J. Schafer of Connell Wagner (our consulting engineers) recomputed the surface deformations for a wind-speed of 8 m/sec, at a number of antenna-wind orientations. The critical wind-loading data were derived from the JPL wind-tunnel experiments with a solid surface model.

These calculations (summarised in table 1) show that at these wind speeds there should be significant deformations. A detailed analysis of the deformed surface shows that the effect amounts to a pointing shift, a change in focal length along with an overall drop in gain due to an increased RMS about the best-fit paraboloid. These three effects appear in different proportions, depending on the wind-antenna direction. The calculations summarised in Table 1 do not include the focus shift, so that the half-path error becomes a realistic measure of the degradation in the gain (eg, with the Ruze formula). For reference, the largest focus shift is 0.8mm.




Observations at mm wavelengths under these conditions would seem to be affected. However, we note from figure 1 which summarises the Narrabri wind statistics of the past three years, that the wind speeds during the mm observing season are substantially lower than 8 m/s - the median during the winter nights is less than 1 m/s.



We have made a number of direct measurements of the antenna deformations -vs- elevation which indicate that the theory is pretty good for the gravitational loads. This gives us confidence that the wind-loading calculations are unlikely to be very wrong.

Experimental

We have tested the tracking stability of the Compact Array antennas in the traditional manner:

a. The array is setup to observe a strong continuum source at 8.6 GHz.

b. We set one antenna as reference, keeping it pointing directly at the source.

c. The remaining antennas of the array are mispointed by a half-beamwidth, first with an offset in azimuth, then with an offset in elevation.

d. A calibration run with no offsets is included.



Figure 2 shows a typical run. Each row relates to a different offset condition : on-axis; offset in azimuth; offset in elevation. Each column relates to a different antenna.

The observations contain two separate bands, centred on 8.640 GHz and 8.768 GHz, with two polarisations/band. We expect that pointing will affect all four signals (band and polarisation) in the same way, so the average, shown as the upper trace in each panel is our best estimate of the pointing irregularities; the lower set of traces shows the difference between the mean and the individual signals.

The quality of the data is good - the signal to noise is such that fluctuations at the 1 to 2 arcsec level are visible. We note that antenna CA03 looks remarkably like its neighbors. We also note the suspicious cycling which is most evident in the elevation offset run, and is most pronounced in antenna CA05.

Almost all the observations, scheduled at random times, have failed to find any differences between antenna CA03 and the other antennas. (In effect, this is simply confirming the weather statistics: Narrabri is not a windy site). So far the only occasion where a difference is detectable occurred at a wind speed (mean) of 3.5 m/s; the results, shown in table 2, are encouraging, and suggest that the theoretical calculations are possibly a bit pessimistic.

Observations under seriously windy conditions (V > 6 m/s) do appear to show an increased tracking error, but antenna CA03 is no worse than the others.

Table 2 attributes all the error to tracking stability, which makes for a worst-case picture: even the amplitude stability in the calibration ("no-offset") data is affected. We believe that this is due to the increased phase variations that we see in the visibility data. In effect, we have some low-level decorrelation within the 10 second integration cycles.

Control System Problem

The tracking stability tests have shown a problem - a 5 arcsec cycling, on all antennas, which could affect mm mosaicing.

The cycling is most evident in the elevation-offset data.

Detailed studies have so far failed to clarify the problem. We have uncovered cycling at a number of frequencies. All are modest in amplitude; most (eg, the sub 0.1 Hz components) will average out. The cycling on the several minute time-scale (such as seen in figure 2) is a problem, as we have been unable to find any trace of this cycling in the encoder data.

More investigations are planned.

Holography

Holography at 22 GHz works well at Narrabri. The panel installation technique gets the panels in place to within 0.5 mm. Adjustments, not surprisingly, are needed.

Holography at mopra is another matter - we can use 12 GHz for a first approximation, using Optus. This should get us to 0.2 mm (RMS), which is not good enough (50% loss in gain at 90 GHz). Since we have plenty of signal, we might be able to get below 3 degrees phase rms in the holography image, corresponding to a surface error of 0.1mm, but we'd need to work at it.

22 GHz VLB holography may be possible, but we can give it a try in the winter.

Conclusions

• Wind is unlikely to make matters worse.
  
  
• Mosaicing at mm wavelengths will be difficult.
  
  
• The cycling evident in the tracking tests needs to be understood.

------------------------

file: /corfu0/mnrf/panels/ca03_report

---

Last modified by Michelle Storey 26/2/99