|
Chopping with NIRC
|
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
The f/25 chopping secondary can be triggered by a signal from NIRC, synchronizing the chop to the frame pulses, and allowing the two chop positions to be gathered into two slices of a "data cube." This allows the removal of rapidly varying atmospheric and instrumental effects that otherwise would drive the noise of some types of observation to unacceptably high levels. In particular, 3-5 micron observing with its high and variable backgrounds can benefit from chopping.
Note that the max12ur timing pattern does not work well
with chopping. In general, the OA will only be responsible for initializing the chopper at the start of the night. Control of the chopper during observing is up to the observer. Quick ReferenceA typical observing scenario:
The general outline of chopping is as follows:
Some details on each step: Set up NIRC to use chopping triggerThe command "chop" will set five NIRC keywords to appropriate values, to tell NIRC to generate the chopper signal, save two slices, etc. The five parameters are CHPCOADD.P, SECMODE.P, CHPMODE.P, CHOPBEAMS.A, and FRMSETTL.P. They are all set to one except CHOPBEAMS, which is set to 2. Although observers will generally operate at a higher level, we describe below the operation of each keyword:
The chopper will be told to use the NIRC trigger signal, and to trigger off changes in the level of the signal by the chop script. This is the type of trigger NIRC outputs. This command can also be accomplished with the command "chptrig level". Set NIRC parametersThe observing parameters fall into two categories: NIRC parameters and chopper parameters. To set the NIRC parameters is for the most part like normal observing. Data taking will proceed as follows: (1) the chopper will start its move to its first dwell position, triggered by the change in level of the NIRC trigger, (2) FRMSETTL NIRC frames will be discarded while the chopper is allowed to settle, (3) FRMCOADD (set using "coadd") frames will be coadded into the first slice of the final file, (4) the chopper will start its move to its second dwell position, (5) FRMSETTL frames will again be discarded while the chopper settles, (6) FRMCOADD frames will be coadded into the second slice, (7) steps 1-6 will be repeated "CHPCOADD-1" more times. Note that to set CHPCOADD, use the command "nchop n". In general you will want to choose NIRC parameters using the following rules of thumb:
Set chopper parametersTo set up the chopper itself you need to define a chop throw (amplitude), a chop direction, a "fiducial" for that direction, and a chop frequency. The relevant NIRC scripts are chpamp, chpang, chprelto, and chpfreq. All four parameters can be set with the chopmaster command (in the order discussed; the command will prompt for input if no arguments are given). The chop throw and direction are fairly self-explanatory. chprelto can take as an argument either "posang", which defines North as the angle zero-point (East being 90°), "vertical", which defines the elevation axis as the angle zero-point, or "rotator", which defines up on the detector as the zero-point. There are other options, but these three should suffice for all observations. The chop frequency, when the chopper is used in trigger mode, is determined by the integration time, number of coadds, etc. The act of turning on chopping with the "chpon" command will automatically calculate the necessary frequency. To preview the value which will be loaded, use chpfreq 0. (One can also type in a different frequency, but there is generally no reason to override the automatically calculated value.) chpfreq with no arguments will display the current value. Note that chopmaster automatically does a chpfreq 0. The Keck chopping secondaries operate in a somewhat different manner than many other choppers. One manifestation of this is that one cannot change chop parameters it is necessary to turn off chopping and then turn it back on once the new parameters are loaded. Simply reissuing a chpon command (even though the chopper is already running) will update the parameters. (Type in new values using chpamp, chpang, chprelto, etc., then type chpon.) Turn chopper onAs mentioned above, chpon will turn chopping on. Begin taking dataOnce NIRC has been configured for chopping, the normal goi command will take two-slice chopped images. A new display command, cdiff n, will display the difference between the two slices of image n.. If cdiff is not followed by a frame number, it will display the difference from the last image taken. Note that some scripts, such as "bxy5", are set to automatically display the difference between two images. With chopped data this may not behave as you might expect; it will subtract the same chop beam from the last two images. You might want to use "quiet" versions of the scripts, such as "bxy5q", and display cdiff's of each chopped frame from the "Auxiliary" window. Change observing parametersAlso mentioned above, changes to the chopping parameters, such as changes in the throw, chop direction, or direction "fiducial," will not be applied unless and until chopping is turned off. One can accomplish this by setting the new parameters, then reissuing the chpon command. Turn chopper offchpoff will turn chopping off. You probably will want to do this every time you move to a new object, so that the field can be more easily identified, and guiding initiated. You also will want to do this, of course, when you reconfigure NIRC for normal images; as a safety, the unchop command which reconfigures NIRC for normal imaging also turns the chopper off. Chopping and Nodding (Offsetting)One specific type of chopping is called "chop/nod." Generally a chopped pair is not sufficient to completely remove the background in an image. The reason is that the thermal background is different in the two dwell positions; as the chopper moves, it sees the telescope structure in a different manner. This so-called "radiative offset" can be removed by combining one chopped pair with another chopped pair, but with the telescope offset in the meantime so that the target is placed on a different part of the detector. The radiative offset under the target in the first chopped pair can be removed by subtracting the second pair, which will have the same radiative offset background since it uses the same chopping parameters. Two techniques are commonly used to take the second chopped pair. In the first, called "chop/nod", the telescope is offset the distance of the chopper throw, so that the "sky" end of the chopper "dogbone" now becomes the "object" end, and vice versa. When the appropriate math is done, the target will be coadded onto the same pixels in the image, while the radiative offset is subtracted and hence cancels out. The second technique nods (offsets) the telescope in a different direction, usually perpendicular to the chop direction. For small, "on-chip" chopping, where the object is actually on the detector in both chop beams, one can nod perpendicular to the chop throw by the same amount, hence capturing all four images of the object on the detector. Hence the term "quad chopping" for this technique. Although at first glance quad chopping might appear to lead to higher signal-to-noise than the chop/nod technique, a careful calculation in the background-limited case shows that the two produce similar S/N values. The basic reason is that in the chop/nod calculation there are two background images underneath the object's PSF, whereas in quad chopping there are three background images underneath each of the four object PSFs. Quad chopping does, however, have the advantage that there are four "dither" positions, and hence detector blemishes can be removed more readily. The choice of technique often depends on whether the target is expected to be significantly extended or not. Chop/NodThe chop/nod technique requires that the target be accurately placed on the same pixels in both chop pairs. Because of various inaccuracies in the telescope and chopper control software, this unfortunately may require a small extra offset during the second, "nodded" chopped pair. In order to check and set this extra offset, use chopnod to take a set of data on a star which is bright enough to yield a good centroid in a single chopped pair, and reasonably close to your target. Use cdiff to display each chopped pair, and calculate the star's centroid on both sets of data. Input the two coordinate pairs, (x1,y1) and (x2,y2) into the script "nodoff x1 y1 x2 y2" and new offset values will be calculated and saved. The next time chopnod is run, the images from both sets of chopped pairs should overlap. Note, however, that the offset may be a function of guide star brightness, chop frequency, etc., and may need to be recalculated at each field. This requires some investigation, and the observer should be wary in the meantime. Quad ChoppingIn quad chopping the nod position can be anywhere relative to the chop positions. In this case you can interleave goi commands with whatever telescope move commands you wish. To simplify the common case in which the nod is perpendicular to the chop (and the same distance), use "quad [n]". This script automatically calculates the appropriate commands to send to get to and from the nod position. The optional argument n is the number of quad chop sets to take. More helpAn "index" of commands in the chopping subdirectory can be seen by typing "help chopping" or "help chp". Below is a table and description of the current state of the scripts:
|
|