Before reading this page you may want to read this overview first which gives a summary and describes all the relevant terms.

Q. How often do I need to point?
A. It is recommended that you do pointing observation approximately every hour on a source close to your science target. This local pointing eliminates many of the sources of error either from the model or from changing conditions.
L. Observing Techniques

Q. Do I need to include anything in my MSBs for pointing?
A. Pointing is performed automatically whilst observing. Typically every hour, or either side of an average length MSB. In this case the TSS will select a pointing sources from our standard catalogue that is near to your science target and as bright as possible. If you require anything other than this you will need to include an option for it in your MSB or write your special request in the notes section.
L. Heterodyne Observations

Q. How is performance measured?
A. Performance is measured by analysis with a programme called TPOINT. TPOINT calculates the differences between the demanded position and the final (corrected) position for each pointing observation. When the data is fir to the model, collimation terms are allowed to vary for each dataset to account for nightly differences. Once the collimation terms have been adjusted TPOINT calculates the rms's of the residuals from this position. They measure the scatter or stability of the pointing over the night.
L. TPOINT Data Format Plotting Data

Q. How accurate are these residuals calculated by TPOINT?
A. TPOINT is extremely useful as a guide about the performance of the existing model and the pointing over a given time period but it is by no means exact. One issue arises due to TPOINT only reading in a single humidity value for a given dataset. This can prove problematic if the data has been collected over many night or even a single night which is particularly unstable. for this reason, the TPOINT calculated performance for 2 nights of data is different to the sum of the two analysed individually.

Q. What is the pointing performance of the JCMT?
The nominal pointing performance of the JCMT is 1.5"-2" in azimuth and elevation.

Q. How is a pointing observation carried out?
A. A typical heterodyne pointing observation is done in a five-point pattern. This is where the telescope slews to the requested position then takes spectra at 5 positions. Above, below, left, right and centre. If the target is perfectly positioned the 3 points connected in elevation and azimuth should trace a Gaussian curve where the two edge positions are perfectly symmetric about the central one. If this is not the case the software adjusts the position of the dish to centralise the source and the correction it is required to make gives the offsets from the pointing model (uaz and uel).
L. Data Acquisition

Q. How often is a new pointing model required?
A. Typically the pointing model remains stable over many months, however any major engineering work requires a new pointing model. Examples of this are a central bearing adjustment or the addition or removal of SCUBA2 from the antenna. If pointing performance seems to be deteriorating it may also be time for a new model.

Q. What data is needed to create a new model?
A. We use RxA to determine our pointing model. A new model need as many data points as possible (though typically this means ~30) covering as much of az/el space as possible. Quadrants 1 and 5 cover the same physical position of the dish but at either end of the potential rotation of the dish, it is important to get data in both of these quadrants. also a good signal to noise ratio is required and very stable weather conditions. they also must be close together in time to minimise any effects of changing conditions, for this reason we do dedicated pointing runs where consecutive pointing observations are carried out until enough data has been collected. This is preferable to using the normal pointing data collected over a night (or two).
L. Updating the TPOINT model

Q. How is the model calculated?
A. Once the data points are collected we take the results logged by the TCS and run them through TPOINT. For normal analysis we import the current model in use and fix all the parameters except the collimation values to assess the performance. However when we want to generate a new model we let all 7 parameters be fit then write out the resulting values as the new model. These parameters are then copied into the TCS to be read for the next observation.
L. Updating the TPOINT model

Q. What about HARP?
A. HARP does not have its own pointing model. Instead it uses the RxA model with a constant offset applied which account for the different positioning of the instrument on the Nasmyth platform rather than in the receiver cabin. the constant offsets are known as collimations and are written into the TCS. The data is also analysed with TPOINT and the collimation values which we allow to change should reflect those encoded int the TCS.
L. Collimations

Q. I heard there were complications regarding HARP pointing?
A. True. Complications have arisen due to the presence of the k-mirror. This mirror is in place to keep the beam oriented on the array. This mirror requires its own pointing terms which have not yet been incorporated into the pointing model leading to slightly poorer performance with HARP than with the other two receivers. Work is continuing to solve this problem so watch this space!
L. HARP Pointing Terms

Q. So how bad is HARP pointing?
A. The rms of the residuals given for HARP are generally poor at the moment (see monthly pointing reports). However this is not the full picture. HARP experiences different pointing offsets depending on the position of the K-mirror itself. The caluclated rms includes the offsets for all these different mirror angles combined giving a poor overall rms. However, as long as the K-mirror does not rotate between your pointing source and your science target then the pointing accuarcy is once again ~2". The operators are aware of this and will ensure that a flip does not take place.

Q. What about RxW?
A. RxW has a couple of pointing options. Direct pointing is rare but possible under the right conditions.the are few sources at these frequencies which are bright enough or close by enough to point on. Like HARP it has its own set of offsets from the RxA model. More commonly for RxW however we perform 'differential pointing.' this is where we point with RxA then apply an offset and switch to RxW for the science observation. We have done tests to determine this offset and we find them to be independent of elevation and free of systematics.
L. Collimations RxW Differential Pointing

Q. How can I find out the pointing performance during my observations?
A. It is usually sufficient in publications to quote our typical performance mentioned above. We also produce reports analysing the pointing performance for a given month. Remember, the exact offsets from the model (uaz & uel) are not what are important; these are simply collimation terms. What is important is the scatter of these offsets which indicate the stability (and applicability) of the model across az/el space. However if you do regular local pointings you correct for all local variations for the subsequent observation. The stability of the pointing model primarily becomes an issue if a large slew is required between the pointing source and the science target. If you have specific queries however you may contact me (follow the link at the bottom of the page).
L. Reports 2009

Q. How much does it change during a single observation?
A. I think you mean tracking. Once the telescope is positioned on a source the tracking errors have been confirmed to be extremely small. Once on source the antenna is known to track to an accuracy of ~0.5" in az and el (away from transit). The dominant factor is then how accurately the beam is positioned on the source in the first place.
L. 20040119 Report

Q. How can I minimise the pointing errors for my observations?
A. This is a simple one. Regular local pointings. These will effectively remove all errors inherent in the model or the nightly collimation differences. You want to aim to complete pointing observations approximately every hour and on a source as near to your science target as possible.