Pointing Tests Page 1 MT/UN005.1 SCIENCE AND ENGINEERING RESEARCH COUNCIL MT/UN005.1 CAVENDISH LABORATORY MULLARD RADIO ASTRONOMY OBSERVATORY MT Project Issued by MT Note UN005.1 _ S.Kenderdine J.Fairclough 20 December 1988 ________ _____ Pointing Tests 1 Introduction This note describes how pointing observations are made and reduced to derive axis misalignment angles, encoder zeros, etc. 2 Observing 2.1 Procedure This is as follows: 1. Log in to OPERATOR and do JCMTC to setup the system 2. Type ADAMCL 3. Type OPT, short for OPTICAL_LOAD, to load the SCREEN_TEL, TEL, NEARSTAR, PLANET and LOGGER tasks. 4. If you wish to acquire a star, decide which catalog you wish to use (see MTUN003). The default is normally YALE. You change it by the command SEND NEARSTAR SET CATALOG 5. If you wish to acquire a star, type STAR. You will then be prompted for the star field centre and asked to select a star to move to. If you do execute a move, then when you have acquired the star on the TV screen, and centred with the tracker ball, answer yes to log the data. - 1 - Pointing Tests Page 2 MT/UN005.1 6. If you wish to acquire a planet, type PLANET. The sequence is then just as for STAR. If you wish to acquire the moon, type PLANET_MOON. You should repeat this command every MOON_INTERVAL hours (normally 0.25 hours) to ensure accurate tracking. 7. You can close the logger file at any time by the command LOGGER_CLOSE. If you want to open a new version then you must do SEND LOGGER OBEY OPEN. 8. When you have finished, just type OPTICAL_KILL. This kills the screen, nearstar and planet tasks but leaves the TEL software running so that you can then perform radio observations. If you do this then you should reset the refraction definition. If you want to end the observing session, just do TEL_KILL. Pointing data is stored in POINTING.LOG in the default directory. 3 Fitting Of Logged Data To Axis-alignment Constants This takes place in two stages: (a) extracting relevant data from the (binary) logger file and converting it to an ASCII file, which may be edited if necessary; and (b) running the least squares fitting program with data from the ASCII file. The directory pointingdir contains the programs and examples of data files. 3.1 Conversion Of Logger Data To ASCII Type the DCL command POINT_EXTRACT to convert the binrary logger data to ASCII. Examine, and if necessary edit, the output. The format should be * (date and time of start of run) One line for each pointing observation, consisting of the following items: az el (degrees) az el errors (0.0001 degrees) other data (not used in fitting) / (to end the list) - 2 - Pointing Tests Page 3 MT/UN005.1 Lines of comment can be added ad lib; they should begin with * (as does the first line of the unedited text). Comment lines will appear transcribed on the output from the fitting program. 3.2 Least-squares Fitting Type the DCL command POINT_FIT to run the program Fit7; the program will prompt for the ASCII logger file, the intended output file, and whether 'pre-whitening' is required (normally not, but see below). The output gives the sums of squares for the best parameter fits, mean mod and rms residuals in azimuth and elevation, the 'best' values for the 7 coefficients being fitted, the correlation matrix, and the individual residuals. The program determines the values of the 7 parameters which minimise the quantity Q = sum over the sources oberved of (actual error in azimuth minus predicted error in azimuth)**2 plus (actual error in elevation minus predicted error in elevation)**2; errors are in seconds of arc on the sky (i.e. azimuth errors are multiplied by cos(elevation) to convert from encoder angles). The algorithm is the standard linear least squares procedure, obtained by putting dQ/d(parameter(i))=0 for each of the i's. The 7 parameters (theta(i),i=1,7) of the current (August 1986) version of the program are as given in the table below; the table also gives the effect that a particular parameter has on azimuth and elevation (in the sense that e.g. the azimuth axis being N of vertical by theta(1) gives errors in azimuth and elevation of sin(A)*cos(E)*theta(1) and cos(A)*theta(1) respectively. Param Origin Effect on azimuth elevation 1 azimuth axis N of vertical sinA.sinE cosA 2 azimuth axis E of vertical -cosA.sinE.cosL sinA.cosL 3 elevation axis not perpendicular to azimuth axis sinE 0 4 telescope beam not perpendicular to elevation axis -1 0 5 azimuth encoder zero cosE 0 - 3 - Pointing Tests Page 4 MT/UN005.1 6 elevation encoder zero 0 1 7 refraction too much/little 0 cotE (approx) A=azimuth, E=elevation, L=latitude Notes: 1. Items 1 to 4 of these correspond (in order) to the four 'axisalignments' angles listed in the telescope.dat file; items 5 and 6 are explicitly listed in telescope.dat; item 7 refers to the first of the three numbers given in telescope.dat under 'refraction' (it is the main tan(elevation) term). The signs are such that in all cases a positive output for a fitted coefficient from fit7 requires an increase of the relevant quantity in the telescope.dat file to correct for the effect. 2. The refraction 'contribution' is currently cotE+0.00125*(cotE)**3 (for elevations greater than 8 degrees); i.e. it is assumed that the functional dependence of refraction on elevation is as in Allen but with a scale factor. At present to correct for a change in refraction deduced from fit7, the first of the three terms in telescope.dat should be increased (algebraically) by the amount given by fit7, and the other two terms changed in the same proportion. 3.3 Additional Features 3.3.1 Fitting with a constraint on the magnitude of the coefficients - In addition to computing the values which minimise Q, the program also computes those which minimise Q' = Q + constant x (sum of (coefficients)**2 ); the constant is set empirically. This provides a check on how well the parameters are determined by the data, since well-determined values are unlikely to be affected whereas ill-determined values are likely to be reduced. - 4 - Pointing Tests Page 5 MT/UN005.1 3.3.2 Pre-whitening - The program allows the user to provide initial values of parameters when these are approximately known. If this 'pre-whitening' option is taken, the program will prompt the user for 7 preliminary values for the parameters; the program will then allow for these known values before doing the fitting. Prewhitening cannot affect the final answers, but it does affect the 'raw' deviations, since these are computed after pre-whitening and before fitting; it allows the user to see how well a given set of parameters fits the data. 3.3.3 Extensions to the program - The Fit7 program is designed to be easily extendable. The number of parameters to be fitted is given by the fortran parameter NPAR. For each new parameter introduced, the contribution it makes to azimuth and elevation deviations as functions of azimuth and elevation (and possibly other variables) must be set up in the arrays (ca(i),i=1,npar) and (ce(i),i=1,npar). If deviations are believed to depend on other variables (e.g. sidereal time), the input routine POINT_INPUT will need to be extended to get the information in (conceivably the Extract and logging process may also need to be extended). - 5 -