Directions for Taking Pointing Data Bill Peters 11 Oct 96 Pointing measurements are used to create an improved pointing model for the telescope. Since there is error in any measurement and some pairs of model parameters contribute in similar ways in small regions of the sky, it is essential that many measurements be done and that they be as widely spaced and uniformly spaced as possible. Redundant measurements help cancel out the measurement errors and allow one to throw out spurious measurements caused by sky variations that happened at a critical time in the measurement. Widely spaced data allows the fitting process to determine more accurately the value of pairs of parameters whose functions only differ from each other significantly over large angles. And if large areas of sky aren't measured at all, the pointing model in those areas is poorly constrained and may deviate from the true model by large amounts in order to fit too accurately, data in an area a large distance away. Finally, if many data points are clustered in the same area of the sky, that portion is weighted too highly and partly cancels out the benefit of widely spaced data. There are not many objects strong enough to use as pointing calibrators. The strongest objects are the planets Venus, Mars, Jupiter, Saturn, and Uranus. (The Moon is the strongest of all, but it is so large that it is usually more trouble than it is worth, since a different observing procedure from all the rest would have to be used to measure it. Mercury is never far enough away from the Sun, 30 degrees, to observe safely.) A few galaxies and quasars are strong enough to use. A list of those which have been successfully been decected can be found in the Observer Documentation binder. Additional objects may be added in time as there is now a second observing procedure to use which ought to be better for measuring fainter objects. A third procedure that will work by measuring a small object with a strong spectral line will be added soon. Before beginning observations, check that the PC that reads out the weather station is still doing so. Occasionally it halts with an error message. Continuing the BASIC program (strike F5) usually cures this, but sometimes a reboot is necessary. If so, login as VISITOR when it asks for a username after the boot. When the boot is complete, type weather to start the program. The weather data is used to compute the index of refraction needed to correct the object apparent positions for refraction. Next start the xephem program: type XEPHEM in one of the windows on an X-terminal. After about 30 seconds a window will pop up. Pull down the View menu and select Sky View. Another window will pop up showing a map of the sky with the pointing calibrators above 10 degrees elevation. If this window overlaps the Update bar on the first window, reposition one of them so that you can press Update periodically. Each time this is pressed, the object positions on the sky are recalculated for the present time. Pull down the View menu on the Sky View window and select Grid. An Azimuth-Elevation grid will appear. The circles are lines of constant elevation spaced 5 degrees apart from 10 to 85 degrees. The lines are drawn every 30 degrees of azimuth with south at the bottom and east at the left. The scroll bars can be used to change the view, but this is not recommended. The scroll bar at the left sets the size of the field of view. It should be 160 degrees so that the last elevation circle is for 10 degrees elevation. The scroll bar at the right selects which elevation should be at the center of the circle (90 degrees). The scroll bar at the bottom selects the azimuth at the bottom of the circle (180 degrees). Find a blank Azimuth-Elevation grid in the Control Room or type PLOT TEL$POINTING:SKY_TEMPLATE to print another one. You should write the date and time on this sheet and mark the positions of the objects as they are observed in order to document the sky distribution of the (successful) measurements. The names of the objects are written on the map alongside the dots marking their positions. The size of each dot is an indication of the radio flux from the object at 1.3 mm wavelength. (The dots for the planets also indicate their relative brightness, but these are on a different scale.) If one moves the mouse pointer onto the map and presses the left button, the Elevation ("Alt") and Azimuth of the pointer are shown on the top corners and the Right Ascension and Declination in the bottom corners. If one drags the mouse button, the SEP value in the upper right corner shows how far you have moved. (This can be used to checking how far an object is from the sun.) If one points at an object and presses the right button, the name and coordinates of the object are printed. (This is particularly useful when the names overlap each other.) Some other options are also available with the right mouse button. Create Trail... lets you see where an object will move as time passes. The first time it pops up the following selections are recommended: Path-Right orientation, 1 Hour interval, time stamp None, and start Now. Press OK and the pop-up will disappear and the trail is drawn. If a different object is right-clicked, one can select Create Trail... then press OK in the pop-up and a trail with the previous selections will be drawn. Right-clicking an object that already has or once had a trail brings up an additional option, Trail, which can be used to re-draw or erase the original trail. The Point and Point & Zoom options should probably avoided as they change the center of the map and the magnification which then must be restored in order to get back to the view corresponding to the Azimuth- Elevation grid used to record the object positions. The OBST window on Frankenheim is used to select the objects and start the observations. Currently, the observations are done via two observing procedures in OBST, POINT and GPOINT. One first selects the object with the command, SOURCE (Note, only enough letters of any OBST command to make it unique need be typed: usually 3.) Then, if the command defaults are set properly, one simply types either POINT or GPOINT. These commands both make observations in the shape of a cross aligned in azimuth and elevation and centered on the object. The POINT command causes the antenna to drift along the arms of the cross at a constant rate and the GPOINT command causes observations to be made while the antenna is held at evenly spaced points along the arms (usually 3 on each arm). During the observation, plots are made periodically displaying the data already accumulated. At the end of the observation, a curve is fitted to the full set of data and the position where the object would have the peak brightness is estimated. This data is automatically recorded on disk for later input to the pointing model fitting program. The final plot shows the pointing error at this position in the sky and the recommended values to two of the pointing constants, COL* and NULE, which could zero that error, at least at this position in the sky. These are simply constant offsets to be applied in azimuth and elevation, respectively. The recommended values are automatically transmitted to the OBST program. (Temporarily, the REFRESH command must be typed in order to see the fitted values on the OBST window before the CORRECTIONS command is executed.) In order to adopt both of them, type the command CORRECTIONS F F To accept only the azimuth correction, COL*, type CORRECTIONS F and to accept only the elevation correction, NULE, type CORRECTIONS * F Optionally, one may explicitly type number(s) for the COL* and NULE values on the CORRECTIONS command: CORRECTIONS or CORRECTIONS or CORRECTIONS * If the corrections are small (a few arcsec), the nominal position of the object measured nearly the peak intensity and the CORRECTIONS command may be skipped. If the corrections are half the size of the antenna beam (the beam is about 32 and 22 arcsec wide at 1.3 mm and 0.9 mm wavelength, respectively), the peak intensity measured was less than half the true peak intensity and the position error estimate may be inaccurate. In this case, the CORRECTIONS command should be executed and the pointing measurement repeated with the cross better centered on the peak. When making measurements with the POINTING command, the most accurate results are obtained when each arm of the cross is measured in both the forward and reverse directions. The Drumbeat backend makes a slight error in assigning a time to the observation which then results in a slightly wrong position of the antenna being attached to the datum. This systematic error is exactly canceled by combining forward and reverse scans. Since the antenna is not scanning across the object during a GPOINT command, this error does not apply. However, at low elevation, the attenuation of the atmosphere may be changing significantly with time due to the change in the amount of atmosphere between telescope and the object with elevation. A forward and reverse scan may also be beneficial in these circumstances, especially along the azimuth arm. When a successful pointing measurement is made, the name of the object, the azimuth and elevation (as marked on the pointing fit plot), and the COL* and NULE recommended by the fit should be marked on the log sheet (regardless of whether the CORRECTIONS command was executed after the fit). One should also press the Update bar on the xephem window and mark the object's position on the Azimuth-Elevation grid. The name of the object is not needed, but it is sometimes helpful to note down the recommended COL* and NULE for that position in the sky. Sometimes, the current pointing model is so poor that one must continually change the COL* and NULE. After a number of measurements, one might recognize a pattern and anticipate the correction when moving near an area of the sky already measured. It is a good idea to make more than one successful pointing measurement on an object, especially when first starting a pointing session. This rule may be relaxed later if one is finding that the measurements always repeat to a couple of arcsec. When a pointing session is begun, it is recommended to first point on a planet (preferably above 25 degrees elevation). Some planets are so bright that the signal may saturate near the peak with the bolometer gain normally used for faint objects. The Drumbeat data is saturated if the numbers are near +-250000 and the peak of the drift scan is flat-topped. For the brightest planets set the Drumbeat gain to 4 (or 5) with the command DRUMBEAT /BOLOGAIN 4 When fainter objects are observed, set the gain back to the normal 1 to get the best signal-to-noise ratio. (The "gain" is mis-named. It would more logically be called attenuation.) DRUMBEAT /BOLOGAIN 1 After making a successful pointing measurement and applying the CORRECTIONS, check the axial focus with the ZFOCUS command. And the ZFOCUS measurement should also be done several times to be sure that the measurement is consistent to 0.1 mm. If there's a choice, the focus procedure is more sensitive if the angular diameter of the planet is small. However, even focusing on Jupiter is better than nothing. The CORRECTIONS command will adopt the last focus fit if one types CORRECTIONS * * F Or one can substitute a number for F as with COL* and NULE. If the above ZFOCUS was successful, a few measurements of the vertical focus with the XFOCUS command is wise (after checking and correcting any pointing errors with another POINTING observation and a CORRECTIONS). The fourth parameter for CORRECTIONS is the X focus value. So to accept the XFOCUS fit, type CORRECTIONS * * * F Again, one can substitute a number for F as with COL* and NULE. When choosing the next object for pointing, one should select an object based upon its being positioned in an area of the sky with few or no other measurements and which will not require a long slew of the telescope. Ideally, one should systematically work one's way from one sky area to an adjacent area until the whole sky has been measured. But the few objects available for pointing rarely are positioned to make this possible. The minimum number of objects required for a good pointing model fit is about 30 if they are well spaced over the sky. The more objects, the better the fit and the less suseptible the fit is to being spoiled by a spurious measurement. Although the signal to noise ratio goes down with elevation, it is important to make measurements at as low an elevation as is feasible. But note that the "horizon" as seen by the HHT is not at zero elevation, but ranges from 15 degrees in the south to 30 degrees in the north. A plot of this horizon vs azimuth is posted on the bulletin board. Syntax of OBST commands ("verbs"): Verbs begin with the verb name followed by zero or more parameters followed by zero or more adverbs (they begin with a /) each adverb followed by zero or more parameters. All items are separated by blanks (equal signs are not allowed). Verb and adverb names may be abbreviated to as few letters as will make them unique. Usually, only the first letter of non-numeric parameters is checked. The adverbs may be given in any order but the order of the parameters of verbs and adverbs is significant. Every verb will accept the adverb /DEFAULT. This means that the parameters specified become the defaults for future executions of the command. The "current" and default values of the parameters of any verb and its adverbs may be displayed by typing SHOW (The "current" values are the values used in the most recent execution of the command.) Like in VMS command syntax, an up arrow will recall the previous command which may then be modified before is struck to execute the command. To recall earlier commands, press PF1 then up arrow as many times as desired. The next previous command will be displayed with each strike of the up arrow. One may recall the previous command which began with by typing REC . Details on the POINTING command: The full syntax of the POINTING command is POINTING /LENGTH_PT /TIME_PT /SUB_PT [BY_2] /ORIENTATION [ONLY] Where is the length of each arm of the cross in arcsec, is the length of time to drift from one end of the arm to the other in seconds, is the sum of the number of drift scans done on both arms of the cross (should be an even number), BY_2 if present means that drifts along the arms of the cross should be done alternately, two at a time (forward and reverse; and ought to be a multiple of 4), otherwise /2 scans are done on one arm (alternating forward and reverse) followed by the remainder on the other arm, is the direction of the first arm of the cross: L=LONGITUDE/R.A., B=LATITUDE/DEC., A=AZIMUTH, E=ELEVATION, and ONLY if present means do NOT do the perpendicular arm of the cross. Usual values for the pointing command are: Planet: /LEN 180 /SUB 4 /TIME 30 Bright radio source: /LEN 180 /SUB 4 /TIME 60 Fainter radio source: /LEN 180 /SUB 8 BY_2 /TIME 60 Really faint: /LEN 180 /SUB 12 BY_2 /TIME 60 If the CORRECTIONS are likely to be large, /SUB 2 may be used (without the BY_2) for the first POINTING to get a quick estimate of the CORRECTIONS needed to roughly center the object. Details of the GPOINT command: The GPOINT syntax is similar to the POINTING command. The full syntax of the GPOINT command is GPOINT /LENGTH_GPT /TIME_GPT /NUMBER /SUB_PT [BY_2] /ORIENTATION [ONLY] Where is the length of each arm of the cross in arcsec, is the length of time in seconds spent averaging the signal at each point on the cross, is the number of points on each arm of the cross, is the sum of the number times observations are made on both arms of the cross (should be an even number), BY_2 if present means that measurements along the arms of the cross should be done alternately, two at a time (forward and reverse; and ought to be a multiple of 4), otherwise /2 scans are done on one arm (alternating forward and reverse) followed by the remainder on the other arm, is the direction of the first arm of the cross: L=LONGITUDE/R.A., B=LATITUDE/DEC., A=AZIMUTH, E=ELEVATION, and ONLY if present means do NOT do the perpendicular arm of the cross. The GPOINT command is too new to have "usual" values. We expect that ought to be 32 arcsec at 1.3 mm wavelength, 22 arcsec at 0.9 mm if is 3. Alternatively, if better results are obtained with = 5, ought to be increased by about 30%: 42 and 29, respectively. /TIM should be 20-30, and /SUB between 4-10 depending on the strength of the source and the repeatability. BY_2 should only be specified if /SUB is 8. GPOINT should not be used on planets or strong radio sources, as the dead time between subscans will make POINTING more efficient. There are several options for the fit Chef performs on the GPOINT data. The fitted curve can be a Gaussian or a Parabola. The default is parabola. MBC SUPPLY GRIDPOINT GAUSSIAN or MBC SUPPLY GRIDPOINT PARABOLA changes Chef's action. The parabolic fit takes much less time and is the better choice if is only 3. The center points of one arm may be used in the fit of the opposite arm (the default). This may not be a good idea if the atmospheric emission is changing significantly with time. To prevent this, and accept only center points bracketted in time by the two end points, type MBC SUPPLY GRIDPOINT BRACKET or MBC SUPPLY GRIDPOINT NOBRACKET to restore the default.