Justin Greenhalgh with IMC, RMP, IAS, DDC, others.

Most recent update: 15 Dec 1998

Assumptions
the function of the current track model element of the pointing system assumes that

1. The telescope’s geometry can be reduced to two rigid, independant triangles (the A frames), whose tilts can be measured by inclinometers on the bases of the A frames.
2. The geometry of the track, and the way in which it affects the bottom corners of the A frames, can be adequately captured in that way.
3. The geometry of the track and the way in interacts with the antenna will be constant in the medium (1-2 month) term.
4. The current inclinometry system (mounts, inclinometers, wiring, electronics, data acquisition, processing software) is sufficiently accurate and stable to make the required measurements.
5. Data taken with the telescope stationary (during inclinometry) are applicable when the telescope is moving (during astronomy/pointing).

   The use of the symmetry measure to assess variability of the inclinometry data further assumes that

6. The data from one A-frame can be flipped and inverted and should reproduce the data from the other A-frame.

Observations
we observe that

1. When the new pointing system was installed, it gave pointing on the sky of around 1.5 arcsec. This supported assumptions A1 to A4, at least at the level necessary to give such good pointing. We infer (I1) that A1 to A5 held at the level of 1.5 arcsec on the sky during most of 1996-1997.
2. We made some early checks of cross-coupling between the A frames and found that they are not independent. (I can't find this in the early reports - help, anyone?) We infer (I2) that assumption A1 is not true at the 1 arcsecond level.
3. During 1995-97, the "symmetry" measure had a value which was not all that low. We infer (I3) that one of assumptions A1 thru A6 was not true, but only at a level of 4 to 5 arcsec on the sky.
4. There have been times when the symmetry measure was very low. (need to fill in details from IMC’s web plots) This implies (I4) that it is measuring something real, and that A6 is true at least some of the time.
5. Clockwise inclinometry gives results different to CCW inclinometry, at about the 2-3 arcsec level. However, the difference is fairly consistent. (Need some data to support this.)
6. The pointing of the telescope seems to have become worse at some time since the current pointing system was installed. This may be solely due to the next item. However, recently made plots suggest that, at least since SCUBA started to take regular pointing data, the SCUBA rms pointing has not deteriorated.
7. The inclinometers have recently been giving results which vary quickly with time, which implies (I5) that one or more of A2, A3 or A4 has become untrue. However, the time history of the variation, now we have it, is hard to understand.
8. During some of 1998, the symmetry measure has been high, suggesting that one of A1 thru A5 was not true. But only some of the time.
9. During the variable inclinometry phase, the results for RY and TY are consistent. This implies (I6) that both of those inclinometers, and all their associated processing, are working well. Suspicion focuses on the LY inclinometer.
10. Our attempts to measure pointing repeatability astronomically (the five night experiment) were inconclusive. This test was intended to distinguish between flaws in assumptions A3 and A4.
11. A further assault on A3/A4 was to double up the left-hand inclinometer with a spare. S and L differ at the 2 arcsecond level, but (at least on the basis of two similar inc runs) consistently so at the .2 arcsecond level. Furthermore, the relationship between SY and LY makes clear that this is not a calibration issue and hence (I7) there is a real difference between what happens at the two points on the beam.
12. Further tests with the spare inclinometer show that when the inclinometry varies by up to 2 arcseconds, SY and LY agree about the variation. We infer (I8) that, at least for small changes, the inclinometry system is functioning well. It may still be that for the larger changes some bizarre effect in the inclinometry manifests itself – we have not yet seen a large inclinometry change with the spare in place. Furthermore (I9) spikes that we see reflect real motions and, because the track model works at all, they must be affecting the telescope beam on the sky.
13. The history plots show that all of the inclinometry results are equally susceptible to variability – evidence that no single inclinometer channel is to blame and thus, rather more tentatively, that none of them are?
14. The conclusions from time-history plots of the differences are that the changes in inclinometry seem to be getting no worse, but the spikes are an increasing component of those differences. No abrupt start to the problem.

Theories
- ideas as to what might be going on :

1. A whole class of theories here to do with problems in the inclinometry. These seem to be pretty thoroughly ruled out by observations 11 and 12.
2. Movement of the concrete. Hard to see how this could cause problems just at the track joints.
3. Movement of the track. This should be measurable
4. A change in the way the wheels follow the track: to do with weight
5. A change in the way the wheels follow the track: to do with radial location
6. Something to do with the central bearing glitches. This seems pretty thoroughly ruled out by the known pitch of the c brg and the complete lack of any coincidence between that and the track joints. Having said which, I wonder if the two are beating to produce the clusters of spikes that we see?
7. Something to do with the c cbrg adjustments. Clearly adjustment earlier this year had a big effect - see the time history plots. But you can also see from those plots that not every adjustment has an effect.
8. Thermal problems in the structure - perhaps related to the radius of the wheels (radius changes about 0.5mm in 10 degrees)
9. Thermal problems in the plinth
10. Humidity problems in the plinth
11. Something to do with the turner

Questions
which we cannot answer and whose answers may have a bearing on the problems :

1. Is there any correlation with the onset of the problem and
   1. Arrival of SCUBA/rebalancing of dish
   2. RxW compressor (probably not - too recent. See time history plots.)
   3. Track welding (probably not, see time history plots)
   4. Az roller swaps?
2. Why is LY worse than SY in terms of symmetry?
3. Can we deduce anything about timescales by looking at two successive inclinometry runs and looking at symmetry and Q1/Q5 data (eg see 981027 writeup)?
4. Why is CW data consistently different from CCW data? Why is it different at all?