ca-forum #20 minutes - 4/12/1996

Acc pointing problem

The symptoms

  1. Every so often (perhaps once/week) an antenna develops a large pointing error (1-2 degrees).
  2. The observer has few indicators of the problem an apparent gain change is seen in an antenna; and the ACC reports "encoder jumps". Most observers will therefore only become aware of the problem when their calibrator flux is low, and assistance starts to complain. All the data back to the previous "good" calibrator most likely must be deemed suspect.
  3. The only known cure is to reboot the ACC.
  4. It is likely that the ACCs have suffered this problem at least since oct 1994.
  5. The problem is with the ACC, and not the encoder. The ACC was programmed to operate in a "pseudo-incremental" mode in order to cope with the 1.4 degree jumps to which the NPL encoders are vulnerable. The algorithm which implements this mode appears to be failing in such a way that the ACC's version of the antenna's azimuth differs from the NPL's version; unhappily, the ACC is wrong - the algorithm has been mislead in its increments.
  6. The problem is only seen in azimuth.
  7. The problem seems only to occur during a slew.
  8. Every antenna has suffered at some time or other; on one or two occasions two antennas simultaneously have been affected, but generally it is just one antenna.
  9. There is some suggestion that the time frame may be involved - time frame errors were seen in the last two events, but in no other.

Diagnosis

The ACC operates on a 10 hz heartbeat. The "pseudo-increment" algorithm in question reads the encoders every heartbeat, and determines the increment from the previous heartbeat. If the change is above a threshold an error is declared and a reduced increment is computed. The problem is that the threshold is predicated on an accurate heartbeat - fluctuations exceeding 25% , if they occur when the antenna is slewing at full speed, will trigger an error when in fact no error obtains.

The threshold is based on the azimuth slew rate; since the elevation rate is half the azimuth rate, it is reasonable that the events will be confined to the azimuth reading.

The hypothesis is that we have a problem with the regularity of the heartbeat - one suggestion is that occasionally one beat is lost.

Current plans

Dave McConnell fears that there may be some underlying weakness that has been revealed by this draconian side-effect; thus the plan is to pursue the heartbeat regularity matter, before raising the error threshold (which would eliminate the pointing error).

Advice to observers

The error can only occur during a high-speed slew - thus reducing the exposure to long slews will help. The antenna will not reach full speed if the calibrator is within two degrees of the target field.

Observers processing data from previous observations should check carefully the calibrator amplitudes; a decrease could signal a pointing error. If so, then you'll probably have to disacrd all the data associated with that antenna, from the onset of the pointing error up to the time of the next ACC reboot.

Bear in mind that the error is an error in azimuth - its impact is thus reduced at high elevation.

ACC Hardware problems

The ACC hardware is dated - and no longer available. The electronics group at narrabri has developed expertise in rejuvenating the components, but the effort is becoming ever more desperate.

The floppy disk drives are being replaced; several back-planes have been rebuilt.

We have 7 ACCs in use, one hot spare, and one tepid spare. On the most recent replacement exercise the hot spare failed in transit from the laboratory to the antenna.

It is difficult to maintain the ACCs in the face of their inevitable aging and decay.

It is difficult to replace components which fail from acts of God (eg, lightning).

Ron Beresford estimates the current load at two weeks every three months, with the additional caveat that a life expectancy beyond 1998 is improbable.

The ATOMS group propose to prototype a replacement ACC at mopra in late 1997.

North-South track for ATCA. J.Caswell with R.Sault and L.Staveley-Smith

Benefits and cost of adding a north-south track to our existing east-west compact array at Narrabri (especially valuable for observations at mm wavelengths - frequencies of nearly 100 GHz).

Firstly a reminder of the features of the existing EW array. This has a maximum baseline of 6 km, but there are many situations where fairly compact configurations of less than 1 km are especially useful, eg;

  1. Where objects are diffuse and of low surface brightness so that sensitivity is more important than high resolution.
  2. Where objects are very extended and require multiple field observations in mosaic mode. In the mosaicing mode, we may rapidly cycle round many fields in the mosaic, and if the array is fairly compact this is done without loss of uv coverage.

We have a fairly compact 750 m configuration which in 4 x 12 h sessions almost fully fills the uv plane.

But in more extreme cases of weak diffuse objects, we would like a 375 m array which can be filled in just 2 x 12 h sessions; and also a 180 m array needing only a single 12 h session.

We are adding 3 stations to the EW track to allow these options, as already specified in the MNRF proposal. (see also AT 31.6.7/012).

This enhancement will be valuable at both the lower frequencies covered by existing receivers, and even more so in the future, at higher frequencies where atmospheric instabilities may often render longer baselines unusable because of rapidly varying phases.

New proposal - Addition of a North-South track

Even when the improved short baseline array capabilities on the EW track have been completed, there will remain two additional problems at high frequencies:

  1. In order to achieve a 12 h observation at declinations other than in the extreme south, the beginning and end necessarily extend to low elevations. For example, at declination -30, the start and end of the track occurs at elevation 15 degrees. At such low elevations the high frequencies are heavily attenuated.
  2. Since good observing weather is less common at the high frequencies, even the requirement of a continuous observation lasting 12 h is often dificult to satisfy. Shortening the observation leaves an unfilled sector in the uv coverage which then requires increasing reliance on image restoration and the consequent reduced confidence in the results.

These problems can both be alleviated by adding a north-south track to the array. We first consider the simplest solution where, for some observations, the 5 antennas currently located on the EW track are all rearranged to lie on a NS baseline.

It is estimated that a quite sharply curving track (minimum radius of ~ 21 m) could be negotiated by the existing wheels and bogies. This fact suggests that a turntable is not needed and it would be cheaper to make the transfer from EW to NS track of each antenna by a curved track and 'points'.

A quick assessment shows that using both EW and NS arrays, there are considerable benefits at declination -30 degrees (near the galactic centre at -29 degrees) and worthwhile ones even at -60 degrees, compared with the EW array alone.

For example, at dec -30 degrees, to avoid any missing sectors in uv coverage, an EW array alone needs +/- 6h and the elevation is only 15 degrees at the start and end. Alternatively, if the observation is restricted to elevations above 35 degrees, (HA +/- 4.3h) the gap in the uv coverage is a sector of angle greater than 90 degrees. Adding the +/- 4.3 h of NS data reduces the gap to 17 degrees, a relatively small defect that can be recovered with image restoration. As a bonus, the NS resolution is improved by a factor of 2 and matches the EW resolution.

As a less dramatic example we consider a field at -60 dec. EW array requires +/- 6h and elevation extends down to 30 degrees. With NS as well, full coverage can be obtained all at elevations above 40 degrees using +/- 4h on both the EW and NS arrays.

Shadowing problems are much reduced on the EW array if coverage ceases near +/- 4h. However they still need consideration. One argument might suggest that the shortest spacing be varied to smaller than 31 m to yield better short spacing coverage at southerly declinations. Another would argue that it be increased, in order to reduce shadowing at more northerly declinations. A further unknown is whether the data obtained when shadowing first occurs can be used with very little ill effect, or whether one should be conservative.

Shadowing is insignificant for the ns array; for 22 m dishes spaced 31 m, geometrical shadowing does not even begin until we go to declinations south of -75 (where the NS track will probably not be needed).

I should forestall 2 potential misconceptions.

  1. The purpose of this NS spur is not primarily to allow observations near dec 0, or even to generally improve the capability of the array at declinations north of -30. We are specifically attempting to improve the high frequency performance.
  2. The usefulness of a north south array rapidly diminishes as one considers observatory sites closer to the equator. In fact the ATCA at latitude -30 is nearer the equator than other existing arrays that make use of NS spacings (except for planned arrays in Hawaii). Despite this, the NS arm option remains an effective way of enhancing the existing EW ATCA.

To summarise the current suggestion:

(measurements are in units of 15.308 m; EW locations are measured from the east end of the track).

  • EW additional pads should be built at 104 124 125 (near control room) these give very-compact array spacings:

  • ewvc1 6 3 11 4 giving 3 4 6 9 11 14 15 18 20 24 and

    ewvc2 2 5 3 13 giving 2 3 5 7 8 10 13 16 21 23

    combined, they omit 12 17 19 22 and repeat 3

    This solution allows ultracompact of 4 2 5 3 using 98 102 104 109 112 pads

  • A new NS arm is needed. Preferably it should be 365m long and located near the control room, near the region where the new EW stations are to be built. It should possess 8 pads, located as follows:
  • NS pads (all new) at positions in units north of EW track:

    3 6 8 10 14 19 23 24

    nsuc uses 3 8 10 14; (ie with 1 antenna on EW track, adjacent antennas are separated 3 5 2 4, to give the same array spacings as EW ultracompact).

    nsvc1 uses 3 10 19 24

    nsvc2 uses 6 8 19 23

    This may well be too costly at present and a modest version with 153m of track and its 4 pads would offer a good interim solution. We would have the antenna separations of 3 3 2 2 yielding a uc array with spacings

    2 3 4 5 6 7 8 10 (ie 9 missing and repeat of 2 and 3).

    Not only is it straightforward to extend it to the recommended array, but it is potentially extendable in simple fashion to larger baselines and/or more antennas in much the same way as the BIMA array which extends to 1000m baselines.

Other alternative or additional proposals that are receiving consideration are:

  1. Seek funding for another antenna, since a 6-antenna very compact array is much more effective than a 5-antenna array.
  2. Make it possible to move the '6km antenna' to join the other antennas for some observations.
  3. Move the Mopra antenna to Narrabri.
  4. Consider a 'north spur' which is not perpendicular to the ew track.
  5. Consider in detail hybrid NS/EW instantaneous 2-D arrays. Some would be possible within the scheme outlined above. They may have calibration advantages.

Comments are sought urgently, since an assessment of NS options is needed before the EW plans can be implemented. Especially welcome will be any alternative suggestions for which you have assembled enough information to demonstrate that it is indeed attractive, and with details that allow it to be compared with the scheme suggested here (the default option).

Odd notes

Any suggested enhancements must not add significantly to manpower and effort needed for operating the array.

The track gauge is 9.6 m, and antenna 6 sits on an 80 m length of track.

Note that the Az limits will rotate through 90 for antennas on the NS track. The preferred direction for this rotation is so that the symmetry point changes from 60 to -30. Note of course that some software changes will be needed.

Cost estimates of the scheme outlined are not yet available. The source of funding for such an enhancement would not be MNRF funds (which are fully spoken for).

next meeting: Thursday, February 13.

Staff space
Public