CA-forum minutes - 8 May 2001

Phase rotator 2^nd derivative problem (Ravi)

(see Ravi's presentation which covers this and the next two sections)

Ravi notes that there is a parabolic phase change during a (10 sec) cycle which is pretty close to twice the second derivative term expected from the motion of the array. That is, the phase rotators are not correctly correcting the second order term, as advertised.

John Reynolds noted this problem in June 1998, and there was much discussion about it at the time (see the old mail exchanges). However the problem was never resolved (investigation was abandoned when Ravi failed to find the problem in an experiment where it should have been apparent). No one claims to have fixed or changed it then. The most likely culprit appears to be an error in the sign bit that send out to the array (in 1998 it was verified that the problem was not in the ephemeris calculation).

Ravi and Mark Wieringa will investigate future, when the high frequency receivers are put back on the array.

Gain/elevation at 3mm (Ravi)

Ravi has made some measurements of the primary beam pattern at 3mm, and its elevation dependence. Ravi notes in azimuth, the gain patterns are asymmetric. However their asymmetry does not vary with azimuth or elevation. It corresponds to a subreflector movement or tilt of 1.5mm in CA03 and 0.7mm in CA04. The beam pattern does vary with elevation. The gain drops off by 25-35% at high elevations! A movement of the subreflector of order 2-3mm would produce the observed effects. This could be either a result of the subreflector moving relative to the quadrupod, or deformation of the quadrupod.

Gross errors resulting from deformation of the the primary surface were ruled out: The primary surface was originally set using surveying when the antenna was parked towards the zenith. Panels have subsequently been reset holographically from observations of the Optus beacon at 60^o elevation. There is no gross difference between these two.

Actions: Photogrammetry is to be performed on the dish in June or July.

Clive is working on rigidly fixing the subreflector to the quadrupod to see if subreflector movement relative to the quadrupod it the culprit.

If it turns out that the effect is caused by deformation of the quadrupod, to recover the gain we will need to continuously adjust the subreflector position to compensate.

Mike Kesteven has offered some comments from Bonn.

3mm sensitivity and Round trip phase system (Ravi and Daron)

The interferometric sensitivity at 3mm has been seen to be down by about a factor of two on the single dish sensitivity (i.e. the interferometric Jy/K was twice the single dish Jy/K).

Ravi notes that disabling the round trip phase system eliminates the problem. He infers an rms phase error in the LO phase of 40^o rms. This problem has been noted at lower frequencies (see ATCA Bug #24). The LO distribution to the antennas is interrupted once every 125 ms during which the phase is returned back down the path to allow the length of the LO path to be measured. During this short period ever 125ms, the phases in the antennas are nominally locked. However there appears to be some deviation from this ideal state. Daron has produced a document describing the problem and its measurement in some detail.

Because the period of between measurements is 125ms, the effect is likely to be significantly washed out in the ATCA's high time resolution mode. When looking in a 32 bin period, there are 5 round-trip-phase cycles for in two bins. So the bins should show a "up/down" pattern, but with an amplitude much smaller than 40^o.

Dave Rayner tried to observe the effect in pulsar bin mode, using a period of 125ms. Although he saw a periodicity, its amplitude was a much smaller effect than had been expected, and the inferred periodicity somewhat different from 125ms. The data suggested a period of 124.48 +/- 1.56*n ms ("n" being some integer). The periodicity of the phase system should be closer to 125 than this apparent 0.4 ms error. If the period was different to 125 ms, then the amplitude of any effect seen in pulsar bin mode will be significantly washed out, as it is binning up about 80 periods. For example, if the period is wrong by 1 part in 80, the effect will be completely washed out.

Discussion: The excursions that Daron measures do not seem consistent with the 50% decorrelation being seen at 3mm. 50% decorrelation would require "random" phase 30% of the time, whereas the effect seen seemed smaller. There still seems to be something missing in our understanding of the problem. However George does note that the excursions that Daron was measuring were low down in the LO chain, and that the phase locked loops at the top of the chain might be being thrown out significantly by the jitter.

If the problem is as supposed, then the new round trip phase system will eliminate it. The new LO system will be used at the higher frequencies after the June shutdown. It will become apparent then if we still have a problem.

No Action: The current round trip phase system is being upgraded as part of MNRF. If the hypothesis is correct, then the new system

Encoder cycling problem (Dave Rayner)

Dave Rayner reports that the sin/cos gains on CA04 were adjusted to help balance them. The result is a significant improvement in the "encoder cycling" problem. See Dave's illustation. Work on balancing the sin/cos gains is progressing.

[Subsequently CA02 was adjusted and Dave notes on 25 May that
"The periodic variations in the corrected azimuth encoder are now less than one arcsec rms. I think you got it right...
The elevation encoder has also been improved, in that the 0.7deg period has been mostly removed. The longer period variations remain, obviously."
See the before and after plots for azimuth and elevation.]

Measurements of system temperatures (Bob Sault)

See the presentation here. There are two web pages describing much of this (on estimating the sky brightness from met data and measurments of the system temperatures).

Bob notes a significant increase in the T_dish component when it starts raining. The increase is significantly different for CA03 and CA04. Graeme doubts this is the goretex soaking up water. The angle of the goretex covers are such that water should never pool. Ravi notes that effects similar to this have been seen at 3cm. CA03 is reputed to have a different sort of paint to the other antennas.

The current noise diode system at 12mm has significant problems - the response is extremely frequency dependent, being far too strong at the bottom end of the band, and far too weak at the top end. George understands these, and envisages that they will be fixed in the next incarnation.

The next incarnation of the 3mm receivers will have a paddle under computer control. This will not have a temperature sensor on the paddle, but this was not seen as a major concern. There was some discussion about whether a paddle at 3mm is needed for the long term or not. The aim was to provide a check on the contribution of the noise diode.

Without a noise diode on the next incarnation of the 3mm receivers, we will need to calibrate non-linearities in the current total power measurement systems. The analogue GTP measurements are reputed to have offsets of order 250 mV. Ravi has measured the offsets for the DSDs. Bob measures a 4% non-linearity in the DSDs even after these offsets have been removed. It is hoped that these are calibratable.

Water vapour radiometer (Graeme Carrad)

See the separate report here.