21 May 2008

The SKY software used to predict the PA necessary to place a guide star on the guider FOV is not predicting the PA correctly. There is an approximate 5 degree difference between the SKY predicted guide star PA and the actual PA. As an example, if SKY is reporting a PA of 100 for a guide star, tell the OA to move to a PA of 95.

16 May 2008

NIRC software is now opperated under VNC. This change in the opperation of NIRC is necessary to migrate the HQ machines to Solaris 10. As a result, FVWM is now the window mangager. On this window manager, observers will use the desktop menu to start VNC sessions on the VNC server. The NIRC software is then launched in the VNC windows as normal. The NIRC startup checklist contains the appropriate steps to start the NIRC software in the VNC session.

19 Dec 2005

During the K1 shutter upgrade, the K1 computer room was reorganized. Part of the reorganization effort was to connect the NIRC host computer MAILI with the electronics computer RABBIT via a lantronics terminal server. As a result of this change, the "tip" session with rabbit no longer monitors rabbit output and may not be used to reboot rabbit..

The new method for rebooting rabbit is to use the "Reboot Rabbit" desktop pulldown menu option, which runs a script located in /home/maili/nirc/p3/bin called "reboot_rabbit". Optionally, observers may run the reboot script from a maili xterm. .

The "tip" line is now viewed via a telnet session to k1consoles 3016. The "tip" window now has the title "rabbit tip via telnet."

1/24/05-New Look

There is a new look to the Web pages. The Observing Support group has designed a new look and organization for all of the instrument Web pages. We hope that you bear with us during the transition, and provide feedback on things you like or don’t like about the design.

12/6/03-Guide Stars

Many observers use the SKY program at Keck to find guide stars for their NIRC targets. The SKY program represents the field available to the guider as an annulus, with a superposed rectangle representing the guider field.

Due to recent realignment of the guider, the annulus that you see on the SKY planning tool is being plotted too far inward. In otehr words, you should look for guide stars near the outer edge of the annulus. These stars should then show up in the center of teh guider field of view.

7/31/02-Guider Focus

Observers and OAs have been complaining for several months now about the poor focus of the NIRC guide/acquisition camera. As background, the guide camera is fixed on the Forward Cass. Module, rotating with the entire module. It does not have a separate remote focus control. To refocus the camera requires someone to ride the telescope at night and turn the focusing ring on the commercial lens in the camera.

After some engineering tests it was determined that the guider focus was somewhere past the LWS focus, on the far side from the NIRC focus. (LWS and NIRC differe in focus by about 150 microns of secondary movement.) On a subsequent night, Randy Campbell and Bill Mason adjusted the focus to be parfocal with NIRC. So when you focus NIRC, you should now have good image quality on the guider.

4/11/01-Sensitivity Loss in Part of the Detector

During the January 2001 observing run we discovered a large area of the detector that suffered a large decrease in sensitivity (to 10% of what it should have been). This looked like another ``flashlight effect'', as described below from 10/99. Observers during that run limped through with the problem, and after the run we warmed the NIRC dewar to ambient dome temperatures.

There was no NIRC run in February 2001, but when we cooled NIRC again for the March observing run we found that most of the effect had gone away, except for a more modest drop (to ~1/3) in a smaller part of the detector.

We are hoping to warm the dewar a second time once an appropriate chunk of time can be identified, and we are debating whether to ``bake'' the detector. There is ambiguity as to whether or not baking the detector has ever helped in the past.

We apologize for the inconvenience.

Update: The second warm-up seems to have cured this problem.

2/17/00-3-5 Micron Cheat Sheet

There is a new page on 3-5 micron observing. Included will be suggested timing patterns and subarrays for NIRC broadband 3-5 micron filters under good to excellent sky conditions.

10/15/99-More on Spectroscopy

Another new command has been added to accomodate the most common forms of spectroscopy. The command "spec [jh,hk,kl,lm] [slit]" will insert the most common grism for the specified pair of passbands, the blocking filter, and if specified a slit. This is all done as parallel motor moves to maximize setup speed.

If nothing is specified, the user is prompted for the passband. If no slit is specified, the program checks to see whether a slit is installed, and if not prompts the user. The slit is always sent to column 180.

10/15/99-Non-functioning Outer Filter Wheel and Decrease in Sensitivity over Part of the Detector

During the September run we found that we had inadvertently caused an intermittent problem with the outer filter wheel. This occurred during reassembly of the instrument after the repair of the broken slit mask (see below). At certain orientations of the instrument the outer filter wheel was not making contact with the drive gear. If a filter move was requested in this orientation, the outer wheel would not move, causing all levels of the software to lose track of where the wheel was physically. (The only fiducial is a home switch up very near the drive motor in the gear path.)

As a temporary fix we placed the outer wheel in one of two positions which would allow broad-band filters in the inner wheel (which was functioning normally) to be used.

During diagnosis of this problem, unfortunately a flashlight was shined into the instrument while the inner filter wheel was not in the "cold plug" position. When this detector is cold, optical light falling directly on the detector will cause a decrease in sensitivity of the exposed part of the detector. We accidentally exposed about 1/3 of the chip to the flashlight beam.

During warmup of NIRC for the repair of the outer filter wheel, this effect has mostly gone away, although there is still a small (~30%) decrease over a much smaller (maybe 10%) part of the detector, in the upper left corner.

8/24/99-Slit Mask Mechanism Broken/Repaired

In mid-July observers noticed that something was blocking half of the focal plane of NIRC. Investigation showed that the slit mask slide in the dewar was not functioning properly, so we took it out of commission for the remainder of the July observing run. This meant that no spectroscopy could be done for observers in that run.

At the beginning of August, during a long HIRES run, Bill Mason opened the dewar and repaired the mechanism, with help from Jim Bell and Pat Crivello. A shaft was found to be bent, and a preload washer found on the wrong side of a shaft.

The mechanism is back in service now, and may be used by observers in the upcoming September run. Unfortunately, we did not remove the dirt on the 1.5-pixel slit during this surgery (see below).

8/24/99-Setting up for Spectroscopy

A new "pointing origin" has been defined for easing spectroscopy setup. When you are first acquiring a target, ask the OA to send it to the "SLIT" pointing origin rather than the usual "NIRC" pointing origin. This should place the target close to the nominal slit center at (x,y) = (180,128) in NIRC pixel coordinates.

Following the NIRC slit mask repair described above, the zero points of the slits were redefined, hopefully making them go to the requested column more accurately. This should also help in setting up for spectroscopy.

Finally, for those wanting higher efficiency in setting up for spectroscopy, note that you can effectively run more than one command at the same time, by using the ampersand (&) character at the end of each command. The & character spawns off a subprocess to run the requested command, giving you back the command prompt immediately so you can type in another command.

Since the commands are no longer under your direct control, you will likely want a way of waiting until the motors are at their requested positions before continuing. For this use the motwait command, with no ampersand. This will wait until all instrument motors are idle before returning control of the command line.

Hence, to put in the grism, slit, and blocker at the same time, you might type:

     slt4 180 &
     gr120 &
     blkhk &
     motwait

These commands can all be placed into a script, of course, if you find yourself using the same sequence multiple times.

Note that to be conservative you might want to first center the target on (180,128), then put only the slit in and take further images to assure that the slit is centered at column 180, and the target image is well-centered in the slit. In general, however, putting the grism and the order blocking filter in at the same time makes sense.

6/28/99-New Commands and Functionality

Some new commands have been added, and some new functionality added to old commands.

• For some observers the most useful command may be angles. If typed with no arguments, it returns the current astronomical p.a. of three vectors: "up" on the detector, "up" along the slit, and the NIRC-to-guide vector. The latter has long been the canonical reference vector for NIRC, leading to confusion over the values of the other vectors.

  If angles is used with a parameter, the SKYPA is printed for the three vectors. This is the angle which must be typed into the SKY program and the angle which shows up on FACSUM. Hence if you want to place the slit at astronomical p.a. 33.4 degrees, you could type angles 33.4 to find out what to tell the OA to put into SKY.

• Another interesting script is the dither routine. This takes as an argument a file name. The file must contain two header lines, the first either "absolute" or "relative", describing whether the dither moves contained in the following are absolute to the initial telescope position, or relative to the previous position. The second line must contain either "pixel" or "arcsec" to define the units of the moves.

  Subsequent lines contain the dither positions, given as (dX, dY) pairs. Note that the dither command has not been fully shaken down, so consult with your IS if you plan on using it.

• A set of three new commands, px, py, and pxy, allow moves in instrument pixels, similar to the existing commands x, y, and xy. The latter commands move in units of arcsec.
• Finally, some functionality has been added. If the mov command is not given any arguments, it prompts for them. If dp is not given an argument, the last frame is displayed (equivalent to typing dlp).
• The buffer math commands buf1, buf2, buf3, and buf4, are now complete. They support addition, subtraction, multiplication, and division of buffers. Hence a command like buf4 = buf1 / buf3 will divide buffer 1 by buffer 3 and put the result in buffer 4. (Note the spaces around the operator, in this case "/". These spaces are NECESSARY!)

4/21/99- NIRC Field Distortion

A new Web page discussing the field distortion of NIRC in normal imaging mode has been produced. The maximum deviation from a completely Euclidean transformation from position on the sky to position on the detector is less than a pixel.

An astrometric field provided by George Gatewood of Allegheny Observatory was imaged in five different field positions, and the distortion map was recreated from those five images. See the Web page for more details.

4/21/99-Dirt on 1.5-pixel Slit

During the 1998 September surgery to repair NIRC's broken slit mechanism, apparently some dirt was left on the narrowest (slt1; 1.5-pixel) slit, somewhat below center. None of the other slits seem to have been affected.

Observers should take this into account when selecting a slit and when choosing dither positions for the 1.5-pixel slit.

4/1/99, 4/21/99-Updated HST Standards List

With the publication of the "Persson standards," a list of 65 highly accurate faint IR standards, I have revised the listing of these photometric standards on the Web site, and in the starlist. The Web page should look similar, only with more stars. All of the finding charts but one (at Dec. -84°) have been produced, and guide stars suggested.

2/16/99-Revised Manual

The manual has been revised, with a new section on overheads, information on background rates and sensitivities, and various other updated information. There has always been a PostScript version available, but it was not linked into the other Web pages. You will now see a link to it after the link to the HTML version.

1/29/99-3-5 micron Photometry

Analysis of a set of data taken on the Elias standard G77-31 and the IRTF bright standard 16 Tau has been completed, and converted into sky brightness and a magnitude zero-point. The numbers can be found in the list of filters on the Web page. The 1-2.5 micron numbers already on that Web page have been converted into the units of electron/sec, to more quickly allow signal-to-noise calculations.

1/18/99-Slit Names in Header, Status Display

There is a new keyword giving information on what slit is in the beam. The keyword, "SLTNAME", will show up in both the image headers and the "xshow" status display (labeled "NIRC_Status" on the title bar). This should allow observers a faster check to make sure that the correct slit is in place.

10/2/98-Faster Motor Moves

Many NIRC commands which involve moving motors now perform faster. Any commands which performed more than one motor move used to do the moves in series. Now they are done in parallel, saving some time. In the most extreme filter moves (both wheels traveling 180 degrees), the move time was cut from 21 sec to 13 sec in tests. Shorter moves will show less improvement. Slit moves and slit clearing has also benefited.

For those observers using home-brewed scripts, the parallel moves can be accomplished by spawning off the inidvidual moves using an ampersand at the end of each command, and then using the new command "mtrwait" to wait until all motor moves are complete.

9/98-Broken Slit Mechanism

The discussion of the broken slit mechanism has been moved to a separate file.

8/30/98-Filter Commands No Longer Clear Slits

In the previous incarnation of the NIRC top-level software, when a filter command was typed the procedure also (silently) cleared the slit and its mask from the beam. This surprised a number of observers, who after carefully setting up so that the finger (which is on the slit slide) was accurately occulting a bright star, found that changing filters made the finger disappear! The solution was to create new procs which were characterized as "blocking filters," because the blocking filter scripts (blkjh, blkhk, etc.) were originally written to not move the slits. The current software was originally written to maintain this behavior, for the sake of backwards compatibility. However, when a recent observer noted that scripts could be appreciably faster if the slit clear were removed, we sprang into action and removed all slit clearing from the filter scripts. Please note that if you expect a filter command to automatically clear the slits from the light path, you will now have to do so by hand, by typing sltclr.

6/29/98-Chopping with NIRC

To aid in 3-5 micron observations, chopping with NIRC has now been commissioned. When in use, NIRC will send a trigger signal to the f/25 chopper, synchronized with the NIRC frames. The result is a two-slice file, one taken at each end of the chop. There are a whole slew of scripts written to make access to the chopper parameters as well as the relevant NIRC parameters. Note that the old list contained in the manual is not up to date. On the short term a document describing the chopping technique and the various chopper scripts will be created and posted on the Web. After some experience and user feedback, a more refined collection of scripts and capabilities will be created and documented in the manual.

6/29/98-Lower Image Overheads

Now that the new NIRC software is fairly stable, we have undertaken a concerted effort to reduce frame overheads, in particular in the normal imaging mode. Tests done when the new software was young showed that our overheads were comparable to the overheads with the new software, during the daytime! For some reason at night the overhead was 6 second larger per frame. This extra overhead was seen in collecting the DCS header keywords. Recently, however, we created a new way to retrieve the DCS keywords, one which has a much smaller overhead, and which is the same whether it is daytime or nighttime.

We have also tried to streamline the scripts in various parts, and have managed to shave some more time off the overhead. We are now at a roughly 8 sec overhead per frame, down from 21 sec during nighttime tests in March. (Note that the older overhead estimate of 11 sec was from daytime testing, before we realized that the overhead was larger at night.)

6/29/98-New Functionality

Various small improvements have been made to the software over the past couple of months. These include:

• Image converter and tertiary mirror cover status, shown in the "NIRC_Status" window.
• Startup script now checks to see whether motors have been homed, and if not, allows the observer to home them at startup time. A test for the state of the tertiary mirror cover is also made, as is one for the DCS simulator.
• MAlign now checks to make sure that (a) the tertiary mirror cover is open, and (b) NIRC is talking to the real DCS, not the simulator.
• The wd command used to write buffers as data files (i.e. temporary buffers such as buf1 will be written to your data directory as the next "s*.fits" file in sequence) now updates the frame number to the appropriate value.
• The ibufs command used to extract slices from a speckle data cube has now been made more powerful. It can take an arbitrary slice number, and an arbitrary buffer, or it can take "-999" as an argument for the slice number, signaling it to take the last slice in the cube, whatever that may be.
• In the "tail log" the full file name is now printed out when an image is written to disk.

4/27/98-tint Found to be Wrong!

Randy Campbell recently noticed that the time between NIRC frames did not correspond to the tint values. As background, when you request an integration time by typing "tint" the NIRC software actually is calculating another parameter, called "FSTRETCH," which is the length of time the timing pattern is paused during the exposure. Because the timing pattern itself has some overhead, the true tint value is this overhead plus FSTRETCH. This is why you cannot request an exposure time less than some minimum value (the overhead, which is defined by the keyword TINT_MIN). Randy had discovered that the overhead calculation is wrong, hence the effective integration time was wrong.

How does this affect your data? First of all, if your tint values are typically measured in seconds, the overhead error is unlikely to make a significant difference. A table comparing old and new formulae is available; in it you will see that the error in the effective integration time is 13 msec for patslow in full-frame mode. Hence a requested tint of 1 sec actually had an effective exposure time of only 1 - 0.013 = 0.987 sec.

Secondly, if your flux standards and science targets used the same integration times, they will be affected the same, and you will still have perfectly accurate photometric numbers.

Where the errors are largest (in terms of % error) is in centered subarrays of fast timing patterns, such as max12ur. Again, this type of data is usually taken at the highest frame rate (tint = tint_min), so the only potential problem would be comparing data taken with different subarrays. This may explain some of our problems reducing 3-5 micron photometry.