We have made the first initial release of the NIRSPEC data reduction pipeline (NSDRP) available for observers to use for quick-look assessment of the data at the telescope. Currently designed for high-resolution mode only, and best suited for Nirspec-1 through Nirspec-7 filters, the DRP runs completely automatically without any user interaction, producing sky-subtracted, wavelength calibrated, one-dimensional extracted spectra for each order. It requires only a flat frame and an object frame. To use, type "nsdrp flat object" at the waimea prompt in the data directory, where flat and object are the filenames of the respective frames (e.g., nsdrp may29s0054.fits may29s0138.fits). For further details and instructions, simply type "nsdrp". Give NSDRP a try next time you're out observing with NIRSPEC!
The Keck Observatory is phasing out the auto DVD writer sometimes used by observers for data backups. Options to launch the DVD autobackup gui are no longer available from the background menu. Because the Keck Observatory Archive (KOA) now ingests all raw data from every instrument at the observatory, observers are encouraged to use KOA to retrieve a copy of the data following your observing sessions. Instructions and links to possible methods of data backup including KOA, scp, and rsync are found at: BackingUp.
To remove accumulated water ice inside NIRSPEC, we have successfully serviced the instrument by warming and pumping for several weeks. This is routine maintenance performed annually. The dewar was not opened during this service.
The Keck Observatory is phasing out the auto DVD writer sometimes used by observers for data backups. Although there is no time line for the removal of the DVD writer, we will no longer troubleshoot or maintain the DVD writer. Because the Keck Observatory Archive (KOA) now ingests all raw data from every instrument at the observatory, observers are encouraged to use KOA to retrieve a copy of the data following your observing sessions. Instructions and links to possible methods of data backup including KOA, scp, and rsync are found at: BackingUp.
To remove accumulated water ice inside NIRSPEC, we have successfully serviced the instrument by warming and pumping for several weeks. This is routine maintenance performed annually. The dewar was not opened during this service.
As an alternate option for acquiring objects in the NIRSPEC slit, one of the slits was removed to permit a direct imaging mode onto the science detector (SPEC). As long as SCAM remains functional, this alternative will not need to be used.
Severe readout banding with SCAM this past spring required that we warm up NIRSPEC to investigate the SCAM wiring inside the dewar. One wire bundle that was touching the inside of the dewar wall was re-positioned further back inside the dewar. Two other wire bundles that looked tightly clustered were "fanned" out to reduce the chance of noise pickup.
We changed out both the cold heads on NIRSPEC after 7 years of service.
Not to jinx it, but SCAM has been reading out on all 4 quadrants since October. The entire length of all slits is visible
We have found that the long slits are under-filled by the calibration beam train. In low res mode, the slit covers about 290 pixels. Before an adjustment, the cals were covering only 230 pixels. After an adjustment, the cals will cover 255 pixels. We are investigating ways to expand the beam coming from the calibration unit.
The upper right-hand quadrant of SCAM started reading all zeros during observing last night. This is the same quadrant as in the past. Observers may define a custom nod pattern so that the object of interest falls only left-hand half of the slit. Please note that if you are using the 12 arcsecond slits, you will need better than 1 arcsecond seeing to avoid beam contamination.
The NIRSPEC calibration pinhole, calp, has been re-enabled to assist in the development of a data reduction pipeline for the Keck Observatory Archive (KOA). If you suspect the pinhole is in beam during observing, please see this entry on the troubleshooting page.
NIRSPEC Arcplots will plot neon, argon, or OH lines for a given wavelength range. One can use Arcplots to match arcs or sky frames to get a rough wavelength solution and identify lines. Arcplots assumes a linear dispersion.
The OH line list is from Rousselot et al., 2000, A&A, 354, 1134 and goes out to 2.62 microns.
The NIRSPEC server has been very reliable of late. When one sees 1-2 crashes per night, a thorough cleaning of the fibers has been very successful is reducing on-sky crashes. Fiber cleaning is generally a daytime task.
Recently, server crashes have increased to 1-2 per night. We think this is caused by marginal fiber optics that carry signals to and from the instrument. We are looking at replacing the fiber cable. For reference, this cable is a mil-spec fiber cable longer than 200 meters and travels through the telescope's azimuth wrap.
NIRSPEC was serviced to extend AO capability to the filters that reside in Filter Wheel 1. This includes the KL and M-Wide filters. Details on operation will follow. The filter complement has been updated.
Especially, DO NOT select the Br-Gamma filter.
Our master machinist has improved the performance of the NIRSPEC hatch so that it no longer binds or sticks. Observers should not see any issue with the hatch, other than needing the "hatch" command to open/close the hatch.
After much effort and a long wait, the NIRSPEC webpages have been revised and improved. We have given the pages a new look and feel. Among the changes:
The NIRSPEC warm and pump service successfully removed the observable water ice from NIRSPEC. Please see the icemonitor webpage for more details. The SCAM dectector is still reading out on 3 of 4 quadrants. There do not appear to be any mechanism issues after the thermal cycle.
Due to increasing water ice inside NIRSPEC, we are servicing NIRSPEC between August 14 and September 9. Please see the icemonitor webpage for more details.
We have found that killing XNIRSPEC via the command line during what appears to be a server crash can "uncrash" the NIRSPEC server. We are watching this.
Since our noise investigation of April 7-10 2008, we have seen an increase in the night-time NIRSPEC server crashes. We are investigating.
Icemonitor data shows the start of ice formation within the dewar. The dip is maybe 10% +/- 5%. Please see the icemonitor webpage for more details. At 12 months after the service, the dewar ice is about the same as 3 months after a dewar window replacement. We hope to service NIRSPEC to remove the ice in fall 2008.
We spent most of 1 week investigating noise issues on NIRSPEC. We replace da pre-amp board with a bad A/D that was causing digitization noise. We are continuing our investigation of analog noise on the chip.
The calibration pinhole stage has been moved out of the beam and its power disconnected. It got stuck in beam for over an hour last night. The pinhole is never used on-sky, but its initialization sequence moves it in, then out of the beam. The start up script has been changed so that the pinhole believes it is initialized. Do not init the calp stage via keywords or XNIRSPEC.
If the calp status reads "UNKNOWN", you may clear this condition with the following command: m calpinitloc=0
The MAGIQ guider camera has been installed in NIRSPEC and tested on-sky. The new camera boasts higher sensitivity, faster readout, and should be more reliable than the PXL series. The wavefront sensing capability is still in development. The photometric filter wheel contains V, R, I, RG780, and open positions. The neutral density wheel contains open, ND1, ND2, ND3, and ND4 filters. Each night after the first MIRA, we will run a guider auto focus routine to focus the guide camera to the instrument.
For all you IDL ATV fans, asinh stretch has been added to the Quicklooks.
The flaky quadrant is the upper right (as before). This blocks half of all slits. For SCAM guiding, please see the User-defined nod entry below. Please note that if you are using the 12 arcsecond slits, you will need better than 1 arcsecond seeing to avoid beam contamination.
During engineering, the MAGIQ software has proven itself ready to go using the "legacy" cameras.
We performed an installation pathfinder on NIRSPEC. This involved removing the PXL system (camera, collimator, fold) and installing the MAGIQ system (camera, collimator, fold, alignment targets, filter wheels, image quality mechanism, and camera focus stage). MAGIQ does fit in the cal unit, although additional modifications will be necessary. On sky, we found focus was flat across the field, we adjusted the focus remotely, and measured FWHM of 0.56 arcseconds.
NIRSPEC is the first instrument to receive the guider upgrade (known as MAGIQ). Work began in September to prepare for the December installation. Prep work has included re-routing the glycol plumbing, adding additional glycol cooling, and fitting of a new electronics enclosure onto the instrument. NIRSPEC's on-sky performance has not been affected.
7 months since the NIRSPEC service and the dewar window is still clear. Please see the icemonitor webpage for more details.
2 months since the NIRSPEC service and the dewar window is still clear. Please see the icemonitor webpage for more details.
During a pre-run checkout today, SCAM readout all 4 quadrants 100+ times. Normal nod sequences will work with SCAM guiding. If the upper right quadrant goes away again, users can define their own nod pattern. The method is described below in the "User-defined Nod Patterns Available (16 April 2007)" entry.
During the last 2 weeks, the SLIT wheel has been initializing on the primary init switch. The problem seems to have fixed itself. Observers should be able initialize all the motors via the startup script.
After the recent thermal cycle, the SLIT mechanism's primary initialization switch is faulty. I am working on an automated workaround. Until then, please follow the new procedure to initialize the SLIT mechanism. Note to observers: this is most likely a support astronomer task.
The NIRSPEC Echelle Format Simulator (EFS) now handles user-defined nod patterns. When the EFS loads, select "USER" from the "Nod Pattern" pull-down menu. The EFS will prompt you to load a nod pattern. Example nod pattern files are stored in /home/waimea/nirspec/setups/nirspec. Currently, only moves along the slit are supported and the EFS does not check input. All moves are relative. An example of an ABBA nod pattern that only uses one half of the slit is: "-2.0 -4.0 0.0 4.0 2.0".
The thermal cycle and dewar pump successfully removed the water ice from the dewar window. The icemonitor webpage shows the results. This test showed that we can remove the ice from the window without breaking vacuum on the NIRSPEC dewar.
NIRSPEC will be out of service from 7-28 March. During this time, we will warm the dewar and pump on it. This service will test whether the ice that forms on the dewar window can be removed or whether we have to replace the dewar window.
In late January, we found the upper right quadrant of SCAM to either read all 0s or to read mostly 0s with bands of other values. It has since started working correctly again. We currently have no plan to replace SCAM.
I have added the first high-resolution sensitivities to the NIRSPEC Sensitivities webpage. They are only for NIRSPEC-7, but it's a start. Source count rates for some very bright "standard" stars are given as well as an estimated SNR/pixel. Essentially, exposure to about 3000 counts in high-res N-7 will yield SNR=100.
The NIRSPEC service went as planned. The window was replaced on 21 March 2006 and NIRSPEC was returned to the vacuum pump. We baked the getters for several hours at 1 A and 17 V, then overnight at 0.75 A and 12 V. We stopped pumping on the dewar on 27 March and turned on the CCRs. Temperatures are going down. Interested parties may follow the NIRSPEC temperatures via this webpage.
NIRSPEC started warming to ambient temperature to replace the Calcium Fluoride dewar window at 12 pm on Tuesday 14 March 2006. NIRSPEC is in its park position and its CCR compressor is off. Interested parties may follow the NIRSPEC temperatures via this webpage.
The Calibration Lamp Exposure Webpage has been updated to reflect the itimes and coadds that the Echelle Format Simulator (EFS) uses to take a "Lamps Only" script.
After too much time, there is a new NIRSPEC troubleshooting page available. The new page occupies the same link as the old so no other pages need changing. The new format is similar to that for DEIMOS. It consists of clearly marked sections of "Symptom", "Problem", and "Solution". Feedback is encouraged!
NIRSPEC will be warmed to ambient temperature to replace the Calcium Fluoride dewar window. The window is contaminated with water ice that shows distinct aborption at 3.08 microns over the lower half of the window (see flatfields and cuts here). The NIRSPEC CCRs will be turned off on March 14 and the service will take place on Wednesday March 22. NIRSPEC will be cold and ready for operations on April 4.
There are now two scripts available from any headquarters machine that read the fits headers of NIRSPEC data and create a basic log. Output can be redirected to a file. The two scripts are called nspeclog and nscamlog. Type nspeclog -h to get examples of their use. These are not replacements for hand-written logs, but meant as a supplement.
The SLIT pointing origin was redefined during engineering on 26 Jul 2005 UT. In addition the slit center in PXL coordinates has been measured to be (544, 508) vs. the previous position of (540, 508). Both of these changes should be transparent to observers.
Data taken by the Keck Observatory to support the NASA Deep Impact mission has been released to the public. Please see our Deep Impact page.
There are 3 new scripts that are very useful for offset (PXL) guiding on NIRSPEC. They allow the observer to easily see objects on the slit that are reasonably bright, but are too dim for SCAM guiding.
The 3 scripts are:
Both bgsub2 and bgsave2 exit gracefully upon a CTRL-C. If a CTRL-C gives you an IDL prompt, you can simply type "exit" to return to a unix prompt.
All 3 scripts must be run from a waimea prompt.
I'm hoping this is a more visible page for the new rotator procedures!
If the rotator becomes unresponsive, you may not have to perform a complete shutdown of the NIRSPEC software, nor re-initialize the rotator. Please follow these steps:
For extremely detailed help, please see this page.
During the recent NIRSPEC service to replace the cold heads, we also replaced the dewar window. The window was showing signs of water ice contamination in L-band flats; however, when viewed by eye, there is no "cloudiness" to the window. The first mirror of the internal rotator mechanism does look discolored and we thought that this mirror maybe causing the absorption. New L-band flats show virtually no absorption across the beam.
There seems to be a procedural problem for guiding on the PXL annular guider in PA mode. The procedure calls for "diff mode" guiding to be used. However, we found this to not work for blind offsets. Normal mode guiding seems to work fine.
The NIRSPEC service mission to replace the cold heads was successful. NIRSPEC is colder than it has been in some time. The Aladdin-3 detector is kept at the proper temperature by a heater circuit.
NIRSPEC had another warming incident on Wednesday 23 Feb. We were able to move NIRSPEC behind AO and get it cool for the weekend's observing. Early Monday 28 Feb, we turned off the compressor to the cold heads and began to warm the instrument for service.
NIRSPEC cryogenic temperatures are near, or very close to, normal operating range. Observers may initialize motors via the startup script and may use the EFS to configure NIRSPEC. Observers may monitor the temperatures via this webpage.
NIRSPEC will be warmed following observations on Sunday 27 February 2005. When warm, the CTI cold heads (models 350 and 1050) will be replaced during a repair mission. It is hoped that this will cure NIRSPEC of its unplanned warmups. Also during this time, the NIRSPEC entrance window will be replaced as the transmission has dropped. The current plan has cooling beginning on Monday 7 March with NIRSPEC ready for use by Wednesday 16 March.
NIRSPEC is back on-sky. On Wednesday afternoon, the Aladdin-3 detector was at 35 K (30.5 K is nominal) and the measured dark current was 1.5 e-/s (0.2 e-/s is nominal). We are observing with minimal motor moves. The detector looks more noisy than normal, but we are getting spectra.
Update: 6 p.m., Wed., 16 Feb. 2005
After consultation with James Larkin and Ian McLean, minimal motor moves are possible, but with severe limitations such as only moving one motor at a time and waiting 5 minutes between motor moves. (Jim Lyke)
Update: 8 a.m., Wed., 16 Feb. 2005
NIRSPEC continues to cool down. The SPEC detector is getting closer to its operable range, butf watch for the other temperatures. Motors cannot be moved, even to initialize them, until they reach thermal equilibrium. Otherwise we risk breaking a mechanism. (RWG)
Monday afternoon, 14 Feb 2005, NIRSPEC's 1050 cold head again "failed" and the instrument began to warm. This has been an intermittent problem for some months and we had already planned on replacing the cold head after the February run.
In the past we have recovered from these cold head glitches by cycling power on the He compressor. While the compressor is almost surely not the problem, the power cycle seems to do something useful at the cold head (put a pulse into the flow rate, drop the pressure temporarily, ...?). On Monday, this did not work, and the instrument continued to warm.
Tuseday morning, we again cycled power on the compressor, this time a little more deliberately. We allowed 15 minutes between turning the compressor off and back on again.
As of 5 p.m. Tuesday, 15 Feb, NIRSPEC has started to cool again. Based on past cooling curves we expect it to take 48 hours or more before it reaches operating temperature. However, there is clearly something wrong with the system, and we might reasonably expect a slower cooling curve (lower cold head efficiency), or even another failure, and a return to the warming trend.
Observers for the late February run should consider alternate programs with other Keck II instruments: preferably NIRC-2, but it may be possible to switch to DEIMOS or ESI. There is some suspicion that the vibration involved in swapping NIRSPEC and DEIMOS may trigger an event, so we may not allow such a switch. NIRC-2 is only a twist of the tertiary mirror away, and is the preferred alternate. ESI is a major reconfiguration, and cannot be done on short notice.
We will attempt to keep this Web page up to date with the latest information. Importantly, you can see plots of the cryogenic temperatures on the Web. A description of some features of the temperature plots is also available.
On November 26, One of NIRSPEC's cold heads "failed" and the instrument began to warm. Observers caught the error and the problem was "fixed" by switching to a backup compressor. Since then, we've seen this "failure" twice: once each when moving to and from the AO enclosure. In both cases, the problem was "fixed" by power cycling the CCR compressor. We are monitoring the cryo-temperatures via this webpage.
We are planning to service NIRSPEC during the time it is off-sky from 28 February through 16 March. This service will include a thermal cycle of the instrument.
The computer control of the K2 air conditioners failed this week. They are now being controlled manually. Before the failure was noticed, the K2 dome warmed by 5 degrees Celsius. Warm dome temperatures can allow water to condense on the NIRSPEC window. Recent L-band flats show increased absorption near 3.1 microns.
We have completed a rewrite of the keyword-- and rotator--servers for NIRSPEC. Essentially, these rewrites prevent the overload of commands to the transputers that were causing the bulk of of the server crashes. These new codes are transparent to the end users.
NIRSPEC is now cold and operational. The repair mission has pushed back the release of the new keyword and rotator servers until early August. With the window replaced, the ice absorption near 3.1 microns is gone. We have increased the airflow across the window in the hopes this will slow the ice formation. Click Here to see a comparison of before and after the window replacement.
NIRSPEC began its weeklong journey from room temperature to approximately 30 K at the ALADDIN detector when Bill Mason turned on the CCRs.
A team from UCLA (John Canfield, George Brims, and Nick Magnone) arrived on 23 Jun 2004 to begin the NIRSPEC service mission. We lifted the cover of NIRSPEC on 24 Jun and removed the slit wheel module, filter wheels module, and image rotator motor. We found the bearings inside the stepper motor for filter wheel 2 to be seized. This caused the observed failure of filter wheel 2. Also of interest, we found the motor for the image rotator to be excessively difficult to turn by hand. John Canfield replaced the stepper motors for the slit wheel, both filter wheels, and the image rotator.
During this mission, Hector Rodriguez replaced the shutter on the PXL optical guide camera, and replaced the 2 getter resistors as well as the getter material.
The cover was placed on NIRSPEC on 27 Jun. We then replaced the window. The old window was not blurry from the ice on its surface.
The CCR compressors have been shutdown on NIRSPEC. The warmup should take about 6-7 days to complete.
A previously scheduled service mission to replace the window has been expanded into a full service mission to fix the stuck filter wheel. The NIRSPEC closed-cycle refrigerators (CCRs) will be shut off on the morning of 16 June. Based upon past warmups, NIRSPEC will warm to room temperature in about 6-7 days. While warm, we hope to perform the following maintenance:
The NIRSPEC filter wheel is stuck in the K-prime filter. Efforts to move the wheel have failed.
Those with questions about NIRSPEC are advised that Jim Lyke is now the primary contact for the instrument. As part of our system in which each instrument has primary and secondary masters, Grant Hill will continue to serve when Jim is unavailable.
Two new GUI's are available to observers that will provide greater flexibility in how NIRSPEC is used. Both can be brought up from the background menu via the NIRSPEC tools link. The control options widget allows a number of operational options such as readout notification sounds to be selected. The calibration acquisition tool allows observers to tailor their calibrations as desired rather than accept the options hardwired into the the EFS.
Brownouts and power outages often cause the iBoots connected to the Computer Room Black Box (crbb), On-Instrument Black Box (oibb), and the Matchbox (mbox) to enter a non-responsive state. Inline power switches have been installed for both the oibb and the mbox to allow easier resets. Please see the iBoot detail page for more details on troubleshooting the iBoots.
Absorption by water ice on the NIRSPEC window has worsened slightly near slit center and significantly near one end. At the "good" end of the long slit, there appears to be virtually no loss at 3.1 microns. At the "bad" end of the long slit, it appears about half the light is being absorbed. We will continue to monitor the situation and start planning to replace the window.
There is now a crash recovery script available to diagnose and recover from NIRSPEC server crashes. Details can be found on the server troubleshooting page.
Historically, a significant fraction of time lost due to server crashes has been the time spend realizing a crash has occurred. NIRSPEC now has an automatic popup warning that informs observers a crash has occurred. If desired, during afternoon setup, a demonstration of what the popup would look like like can be arranged.
Observers should realize that the machine waimea is where NIRSPEC operational software actually runs. There have been instances where operational problems appear to have arisen due to the machine being badly loaded down at night with things like ssh, ftp, file editing, etc. It is best to avoid using instrument host machines for anything other than running the instrument.
Evidence now seems fairly conclusive that ice is reforming on the NIRSPEC window. Near 3.1 microns, at slit center the throughput is lowered by about 20 percent or 30 percent. For the low res slits there is a very noticable variation along the slit. For the high res slits (which are oriented roughly perpendicularly) any variation is not obvious.
In the coming months, observers will hopefully see the time lost due to server crashes decrease as we improve recovery procedures. Ultimately as a result of a major effort just now getting underway, we hope to better understand the source of server crashes, and sharply reduce their frequency. The troubleshooting page that advises how to recover from a server crash is the place to look for the latest info.
The 20 second overhead hit when running scripts from the EFS (see below) has been reduced by a factor of two. The overhead reduction was achieved by fixing some time-wasting errors in "waitfor" scripts, scripts that wait for motors to get into position.
Command-line versions of the ABBA, 2-position nod, and 3-position nod patterns have been written. The commands are "ABBA", "dither2", and "dither3". They can accept a step size parameter (in arcsec) and in fact should be used this way for now. They do not have very sophisticated default step sizes. The ABBA script has been tested on sky, but the dither 2 and dither3 scripts have not.
Use of these scripts from the command line avoids the EFS overhead mentioned above, although you are required to set the object name, integration time, and readout mode before starting the scripts.
There is some indication that ice may be reforming on the NIRSPEC window but the evidence is not yet conclusive. The transparency of the window at 3.1 microns will continue to be monitored and should it need replacing we will do so at the first opportunity.
It appears that when used in PA mode, normal tracking of the rotator does not induce wavelength drifts.
Four sets of comparison spectra were acquired while the telescope tracked across the meridian with the rotator maintaining a constant slit PA on the sky. The sets were obtained at declinations of -30, -15, 0 and +15 degrees. These correspond to an increasing burden on the rotator to maintain a constant PA.
Each set was acquired from roughly 10 minutes east to 10 minutes west. During the course of each set, numerous, brief reports appeared on XNIRSPEC that the rotator was slewing. This would happen whenever the accumulated error between desired and actual position (in motor steps) exceeded a threshold. In the worst case (at dec = +15) the box on the rotator gui which is white when the rotator is successfully tracking occasionally turned red.
Within the measurement errors, there is no evidence from the comparison spectra that wavelength drifts are induced by normal tracking motion of the rotator, even when XNIRSPEC reports brief slew episodes. Observers still need to be aware though that stopping and starting the rotator software may induce drifts as may changing PA and/or resetting the rotator going from one object to another.
Traditionally, upon hitting the go button on the EFS gui to start a nod sequence or to get a set of cals, NIRSPEC spends some time checking the configuration of the spectrograph. However we are not aware of any observers who simultaneously configure the spectrograph and launch a set of observations. Everyone launches a nod sequence or a set of cals after having previously configured the spectrograph via the setup-only option. The time between hitting the go button, and the start of the integration is 10's of seconds. Some fraction of this is thus time wasted checking that the configuration of the spectrograph matches that of the EFS gui.
A command (issued from a waimea prompt) is now available which enables or disables this pause for configuration checking upon hitting the go button. The command is wait4config and takes as arguments on or off. We have not yet gone very deep into the degree to which configuration checking is invoked and as of this writing, only a second or two of time is saved. As we proceed though we hope that more time will be saved.
Traditionally, in between nods one gets a pop-up requesting the observer to confirm that the object is still on the slit. Integrations don't start until the observer clicks okay on the pop-up. Many observers don't want, or don't need to check.
A command (issued from a waimea prompt) is now available which enables or disables this pause for centering confirmation between nods. The command is wait4center and takes as arguments on or off.
Evidence has been accumulating that NIRSPEC image rotator motion is causing motion of one or both of the gratings in NIRSPEC. Presently the problem is not well quantified but we are planning to improve this. Here is what we know at the present:
Until we better quantify (and hopefully fix) this problem we suggest that observers do the following:
In late April and early May, NIRSPEC was successfully warmed up, the dewar window replaced, and the instrument cooled back down.
A plot of the warming curve is available here.
A plot of the cooling curve is available here.
The ratio of flats taken after and before the window replacement indicate that throughput has improved by about a factor of 20 near 3.1 microns. Here is a plot showing the ratio.
Recent spectra near 3 microns indicate the absorption due to water ice has worsened. We are hoping to replace the window during a 3 week period in late April and early May when NIRSPEC is not in use.
Here is a low resolution spectrum showing the ice absorption. Very little light is getting through on the left edge of the spectrum around 3 microns. The flux here is down by about a factor of 20. Near the right edge at the same wavelength, the absorption is not nearly as bad but is now roughly (a non-negligible) factor of 2.
Here is a high resolution spectrum showing the ice absorption. The absorption is more uniform along the slit and is something like a factor of 10.
Prato, Kim and McLean at the UCLA IR Lab have provided Keck with REDSPEC, an IDL based data reduction package for NIRSPEC. The manual and code are available via the NIRSPEC home page through the data reduction link. At this point, we have not installed and tested it here at Keck but plan to do so soon. We make the code available now though, so that observers can have access to it as quickly as possible.
The web page describing how to recover from a server crash has been completely re-written. Observers are strongly urged to read it over. Familiarity with this page may save you lost time on the sky. Most of the extra information has been moved to separate pages via links so that the page is more streamlined and easier to follow. Step by step instructions are supplied along with pictures to help summit staff find any components that may need power cycling.
The SPEC detector was switched to constant current mode on Dec 19 2001 and monitoring of the noise characteristics was undertaken until Jan 10 2002. Comparison of the plots shown in this news item with those in the item dated Dec 17, shows that the variability of the 8-row pattern in quadrants 3 and 4 of the SPEC detector is not dependant upon whether the detector is in voltage mode or constant current mode.
Here is a histogram showing statistics of the variability of the amplitude of the 8 row pattern.
Here is a histogram showing the distribution of readout noise.
Here is a histogram showing the distribution of dark current.
Monitoring of the noise characteristics was begun on Nov 27 with the SPEC detector in voltage mode and continued until Dec 17 2001.
Here is a histogram showing statistics of the variability of the amplitude of the 8 row pattern. This is a zero point (additive) component measured from 0.25 sec darks in CDS mode. The quantity measured is the peak-to-peak amplitude, defined within each quadrant as the highest row minus the lowest row.
Here is a histogram showing the distribution of readout noise. The quantity measured is the standard deviation for all the pixels belonging to a given output channel. There are 8 such channels for each quadrant. Not only is the noise in some channels higher than in others but it is variable in some channels as well. The data measured are the same 0.25 sec darks used to study the variability in the 8 row pattern.
Here is a histogram showing the distribution of dark current.
Each of the 4 quadrants of the Aladdin detector used for NIRSPEC spectroscopy has 8 output channels, each with similar but not identical gains and zeropoints. It is thus known that each quadrant will have its own 8 row pattern. More disturbing, are reports from some observers that the 8 row pattern in one or more quadrants has been seen to vary with time and/or the noise in one or more quadrant is variable.
We have begun monitoring and characterizing the noise in the SPEC detector and observers are urged to maintain an awareness of these efforts and plan their observations accordingly. Links to the efforts will be maintained on this page. We have made one change to the detector electronics already which may have resulted in a reduction of the noise. This means though that from this point on, observers may notice minor changes in things like bias level.
One figure of merit with which to study the 8 row pattern is the peak-to-peak amplitude of the pattern. All the rows belonging to each output channel are averaged and the difference between the highest and lowest is monitored versus time. Here is such a plot of 8 row peak-to-peak variations. The pattern was monitored from powerup to investigate if heating of any of the electronics boards may be responsible for variability. Here is a plot of cabinet temperatures over the same time period. A figure of merit with which to study noise quadrant by quadrant is the average noise per row. Simply taking statistics of an entire quadrant could be misleading if the noise within each row is constant but the 8 row pattern is varying due to zero point drifts. For each quadrant, 8 different sigmas are thus calculated and averaged. Here is a plot of how the average sigma per row varies over time for each quadrant.
Spectra near 3 microns in both high and low resolution modes have been taken to better characterize the absorption due to ice on the dewar window. The absorption varies from a factor of about 7 to near zero along the length of the the 42 arcsecond low resolution long slit. Along the length of the 24 arcsecond high resolution slit, the absorption is more uniform. The variation in this case is about a factor of 3 to a factor of 4.5.
Here is a low resolution spectrum showing the ice absorption.
Here is a high resolution spectrum showing the ice absorption.
Recent observations near 3 microns indicate that ice has reformed on the NIRSPEC dewar window. At the moment, the problem is still being characterized but more than ever, observers are urged to observe with the rotator in stationary mode so they retain some ability to remove the effects of the absorption in their processing.
As noted below, the cal unit circuit which controls the flat field lamp was modified in July. The high setting for the lamp, intended for use when NIRSPEC is behind AO was made dimmer. Previously it was too bright, and for certain instrumental setups the high setting was too high and the low setting was too low. The situation is now improved but is still far from perfect. When behind AO use the following guidelines when running the lamp script from the EFS gui:
Currently, switching between the low and high settings can only be done by someone on the summit. If your observing assistant does not know how to do this contact your instrument scientist.
A link to a page maintained by CSO is now present on the NIRSPEC web pages. The CSO page has tau plots and a button to push which gives the most recent measurement. Click on Telescope and facility information to find the link.
Testing has indicated that offset guiding on non-sidereal objects is possible in rotator stationary mode. However, observers should be aware that the process is labour intensive and thus inefficient and error-prone. Nodding is accomplished by having the OA stop guiding, change the pointing origin by entering (by hand) new PXL coordinates, and restarting guiding for each nod.
Preliminary indications are that the NIRSPEC window replacement has been largely successful:
Rough measurements of the telescope emissivity have been made. Spectra were taken of the night sky and with the cal unit dust cover closed. Assuming that emissivity of the cover is 1.0, the emissivity of the telescope was found to be about 0.15 at 3.4 microns and about 0.18 at 5.05 microns.
NIRSPEC is currently cooling back down after being warmed up to allow various work to be done. The main accomplishments were:
Revised procedures have been released to the community for non-AO NIRSPEC observing with the image rotator stationary. A full description is available and can be found on the "Observing Procedures" page under the heading, "Stationary Mode Guiding, or Observing With A Fixed Image Rotator." Stationary mode observing can be useful for a number of reasons, including:
A limited capability has been implemented to move the NIRSPEC rotator during AO observations, to allow a means of setting the slit PA that is independent of the AO system rotator. This capability should prove useful for observers who have an AO guide star relatively far from their spectroscopic target, such that the AO rotator setting is dictated by the need to pick up the off-axis guide star. Using this new mode, such observers can set the AO rotator as needed to lock on the guide star and still select an arbitrary slit PA using the NIRSPEC rotator. There is a modest efficiency penalty in this mode, in that all moves, no matter how small, require opening and reclosing the AO loops (see 26 June 2001 item below). Instructions on the use of this mode can be found on the page of notes about NIRSPEC AO observing, under the heading "Using the NIRSPEC Rotator to set PA." Observers should note, however, that this new capability has received very limited testing, and problems may still be found.
In AO observing, small moves (less than 0.2 arcsec) made with the "NIRSPEC/AO Slit Nod" widget no longer require opening and reclosing the AO loops. This change should reduce the time required to nudge an object a short distance into the slit. Please note, however, that this change is only effective when the NIRSPEC rotator is in its default position (see item above).
Recent observers have noted more frequent occurences of small drifts in the angle of the echelle grating between frames. In the worst cases, spectral lines are seen to move as much as 20 pixels in the X-direction. The drifts are believed to be associated with slews of the image rotator, which is known to vibrate more than the other mechanisms. Observers are cautioned to take the following steps:
We have added a page of general notes and advice about startup and observing procedures specifically for use with NIRSPEC in combination with the Adaptive Optics (AO) system. There is also a brief discussion about the most important differences between AO and non-AO (or normal) use of NIRSPEC on the NIRSPEC Observing Procedures page. These new pages will (we hope) be the first steps towards better user documentation for the AO-NIRSPEC combination. Watch this space for more announcements.
We have recently noticed a strong absorption in NIRSPEC spectra, in the range 3.0-3.2 microns. This is probably due to a thin film of water ice, most likely on the outside of the dewar window. It is not uniform spatially, and the spatial effect varies with rotator physical angle. In low-resolution mode, at rotator physical angle 0.0 degrees, the absorption is much stronger at one end of the slit than at the other. At rotator physical angles of +90 or -90, it is nearly uniform over the length of the low-res slit. In high-resolution mode, the relationship with rotator physical angle is reversed, because the high-res slits are effectively perpendicular to the low-res slits. At its worst, the absorption seems to reduce the throughput in the affected region by a factor of about 3.
We expect that this absorption will go away when the dewar window is replaced. This replacement is currently scheduled for late July 2001. Observers who will be using the KL filter between now and mid-July should be aware that throughput in the 3.1 micron region is substantially reduced.
A new tool for calculating Sky PA's with any WMKO instrument is now available on our Web site. It's a Java-based tool that runs in any web browser. All you need to do is enter the three header keywords it needs as data: PARANG (the parallactic angle of the observation), EL (the telescope elevation), and ROTPPOSN (the rotator physical position). Give it a try here.In the particular case of NIRSPEC, the slit angles are not computed because the offset between SCAM columns and the slit changes from slit to slit. The calculator does not include the information about those offsets, nor about what slit was in use. It still is useful for getting the angle between North and SCAM columns, and from that and your SCAM images of the slit you can determine the slit PA.
On 2 March 2001, the original PXL guider camera was re-installed, eliminating the annoying y-flip on images from the spare camera. Objects can now be acquired on the PXL annular guider in any rotator mode, and at any rotator angle or sky PA.
With the return of the original guider, the full set of pointing origins needed for stationary mode guiding have been re-calibrated. It is now once again possible to use stationary mode guiding with either the echelle mode 0.43x24 arcsec slit, or with the low resolution mode 42x0.57 arcsec slit. See this page for an updated description of the procedure and its limitations. Other slits may be tried but may not work. See this note for details about other slits.
In order to reduce the risk of poor rotator performance, it is highly recommended to set the rotator so that it preserves its physical angle during a telescope slew. A new command fixpresrot has been implemented to make this chore a little easier. See this note for more information.
Because of the small field of view and very fine plate scales associated with AO observing, effects of differential refraction can be significant for AO+NIRSPEC observing. Henry Roe of UC Berkeley has written an excellent discussion of the problems and ways of dealing with them. You can find his discussion here.
As mentioned below, the current condition of the dewar window affects the sensitivity of NIRSPEC in the thermal IR. The most recent observers to work in L and M bands using stationary mode guiding report that they were able to obtain the following S/N values:
In both cases, the spectra were taken in high-resolution (echelle) mode, with the 0.42 x 24.0 arcsecond slit.
Many thanks to John Carr, Joan Najita, and Bob Mathieu for this information.
To enable better subtraction of the spurious point sources, we have re-implemented stationary mode guiding with NIRSPEC. In this mode, the rotator remains stationary to keep the telescope pupil (hence, dewar window) fixed with respect to the slit, while the position angle of the slit on the sky changes as the telescope tracks. Currently, due to problems with the PXL guide camera (see below), stationary mode guiding works only with the rotator at physical angle = 0.0 degrees, and only with the 0.43 x 24 arcsecond echelle-mode slit. Instructions for using stationary mode guiding can be found here. PLEASE BE AWARE that observing overheads are somewhat higher in stationary mode guiding. PLEASE NOTE ALSO that it is NOT possible to use the "Object" mode observing sequences from the Echelle Format Simulator in this mode.
The condition of the NIRSPEC dewar window has degraded to the point where it is significantly interfering with observing in the thermal IR. The window is only 1 inch away from the telescope Cassegrain focus, so dust and blemishes on the window show up as nearly-in-focus point sources on thermal IR spectra. In "position angle" rotator mode, the telescope pupil and hence the dewar window rotate with respect to the slit, so the spurious sources come and go, making it difficult or impossible to subtract them off. See the item just above for a workaround. The condition of the window also raises the overall background somewhat in the thermal IR. There is unfortunately no workaround for this problem.
Revised procedures for beginning NIRSPEC observing at the start of the night are now available at this page in the NIRSPEC web documentation. The revised procedures include step-by-step instructions for checking pointing and focussing the telescope.
The PXL guide camera (the "annular" CCD guider) failed recently and was replaced with a spare. Unfortunately the detector in the spare camera is read out from a different corner, resulting in a vertical "flip" in the guider images. The guiding software now handles this flip correctly but there is still a problem with setting up on new objects using the PXL annular guider.
Most significantly, during this downtime the Aladdin-2 InSb spectrographic detector was replaced with a new Aladdin-3 device. The new detector has vastly improved cosmetics and seems to be free of the "floating" hot pixels that troubled the earlier device. In addition, early measurements suggest that the readout noise is somewhat improved. Like the previous array, the new detector suffers from a "charge persistence" problem. Exposure to bright flux sources will leave a "memory" of charge behind. The length of the persistence seems to vary with fraction of full-well acheived, but can last as long as 30 minutes after a severe overexposure. Please see our new Frequently Asked Questions list for more information about the persistence.
Because the AO pupil stop is located in the same filter wheel that holds the long-wavelength filters, and because the AO dichroic is opaque beyond 2.7 microns, use of NIRSPEC behind the AO system is limited to wavelengths shorter than 2.6 microns. Filters available are all those in NIRSPEC's Filter Wheel 2: Nirspec-1 through Nirspec-7, K, K', CO, and BrGamma. Nirspec-3 and Nirspec-5 are very close to J and H, respectively. In addition to these cold NIRSPEC filters, the warm KCAM external filter wheel is installed in front of NIRSPEC, so all the filters that can be used in that wheel are available as well.
Users can now select the nod pattern ABBA for object exposures in EFSgui, which will take 4 exposures (1-2-1) with the object in two slit positions (A-B-A). The telescope offsets applied are the same as those for the nod2 pattern, namely, to place the object at the 1/4 and 3/4 slit length positions (high-resolution slits), or approximately at the 1/3 and 2/3 slit length positions (low-resolution slits).