LWS
Frequently Asked Questions

Contents

Comparison to other Keck instruments

How does LWS compare to other mid-IR instruments used on Keck?

Although the full answer depends on your specific needs, in general LWS is quite competitive with OSCIR and MIRLIN. The relative capabilities of the three instruments are shown in this table.

For imaging and spectroscopy in the near-IR, should I use LWS or NIRC?

For most objects you will find NIRC to be more efficient, mostly because the InSb detector used by NIRC has much greater quantum efficiency at 3-5 µm than does the Si:As array used in LWS. However, there remain two situations where LWS would be preferable:
  1. Bright source. If the source has extremely high flux it may saturate NIRC, yet be on scale with LWS due to the low LWS QE.
  2. Extended source. With NIRC, one must read out a subarray in order to prevent saturating the detector in the mid-IR; depending on the filter, you may be limited to reading out a small image section with very little resultant sky coverage. Under these circumstances, LWS may allow you to obtain wider field-of-view than NIRC.

Detector

Is the LWS detector sensitivity altitude dependent?

No. The original LWS lacked a thermal control system, thus allowing the detector temperature to fluctuate by up to 3°K depending on the dewar attitude (elevation). Because the quantum efficiency and gain (etaG) of BIB detectors is strongly temperature dependent, the response of the original LWS thus varied with elevation. The new LWS detector no longer has an attitude (elevation) dependence due to a precision closed-loop thermal control system which maintains the detector at its optimum operating temperature of 8.50±0.02 K.

Imaging

What filters are available for imaging in the 20 µm wavelength regime?

There are 4 filters currently installed in LWS for use in the 20 µm wavelength regime. The broadband 17.9 µm ( 2.0 µm wide) filter is recommended for imaging in the 20 µm region. This filter has good transmission properties and is the most sensitive in this region. Another choice is the narrow 18.7 µm (  0.4 µm wide) filter. The 18.7 has poor transmission and is not very sensitive, however, it provides an alternate choice during marginal weather conditions.   There are two other filters that do not have the proper diameter for use in LWS, the 20 and 22   µm ( 2.0 µm wide) . These filters are undersized and only transmit about 50% of the LWS beam, which effects the sensitivity and point spread function (PSF). They have been temporarily installed for testing and help extend the wavelength coverage for users that need measurements in this region. The 20 and 22 µm filters are NOT recommended for typical LWS programs.

Spectroscopy

What's the LWS sensitivity in spectroscopic observing modes?

A flux of 150 mJy in LRES (R=100) mode, or 500-1000 mJy in HRES (R=1200) mode, yields S/N=1 per spectroscopic resolution element in 1 second on source. The LRES estimate assumes you are using: Quoted flux is light incident on telescope. Also see the next item...

What's the typical observing efficiency in spectroscopic observing modes?

About 25%. A peculiar software feature requires us to wait an integral number of frames for the chop motion to settle. Thus, if your frame time is 0.5 s you must wait 0.5 s after the chop, (tossing this frame) then take a single 0.5 s frame, cutting effficiency. The alternate which may be acceptable sometimes is to wait zero frames after the chop and take the hit in the psf in return for higher efficiency.

What's the procedure for setting up and guiding for spectroscopy?

Because LWS does not have a slit-viewing guider, LWS must be placed into imaging mode in order to position the target on the slit. In a nutshell, the procedure is:
  1. Reconfigure LWS for imaging (mirror in, imaging filter in, slit out)
  2. Image the target and measure centroid
  3. Insert the desired longslit
  4. Image the slit and measure centroid
  5. Move the target to the location of the slit center
  6. Image the target in the slit and manually tweak centering using handpaddle
  7. Reconfigure LWS for spectroscopy (grating in, blocking filter in)
More detailed instructions are available on the Target Acquisition: Spectroscopy Mode checklist.

Guiding, Chopping, and Seeing

What sort of image quality does LWS deliver?

In the near-IR (3-5 µm) LWS and NIRC will give roughly the same image quality, typically 0.3-0.5 arcsec FWHM. In the mid-IR (10-20 µm) LWS generally provides diffraction-limited images at FWHM=0.28 arcsec.

Does image quality vary with observing conditions?

Seeing can degrade a little when guiding/tracking isn't so good, but is usually 0.3 arcsec FWHM or so.

Can the guider keep objects in those narrow LWS slits?

Thus far in the commissioning of LWS spectroscopy modes, keeping an object in the 0.25 arcsec (~3 pixels) slit has not been a problem. In other words, over periods of ~30 min the guide performance has met the requirements for spectroscopy. Another test, designed to look for rotation-dependent flexure, showed the long-term guide performance to be better than 0.3 arcsec. The test duration was 3 hours while crossing the meridian within 5° of zenith. Since this is such a critical parameter, more tests will be performed to measure differential flexure with different gravity vectors acting on the instrument and guider (i.e., lower elevation). Reasons for improvement from the old system is attributable to the LWS mounting hardware being redesigned to make the instrument more rigid in the Forward Cassegrain Module (FCM). Also, there is a new guide algorithm in use at Keck that may be less sensitive to chopping effects on the guide star PSF.

What are the advantages and disadvantages of the different types of chop-nod operation?

At Keck, "chop-nod" mode is defined as using a secondary chop throw and telescope nod throw of equal amplitudes but of opposite directions, 180° apart. The result is that for each nod position one of the other chop beams is in the same xy location (within the accuracy of the telescope guiding/tracking system, typically 0.06" radial RMS) on the detector. For LWS, if the chop-nod throw amplitude is greater that 5.2" then the opposite chop beam will be off the detector (assuming your chop direction is aligned with detector rows; if you chop at a 45° angle relative to the detector "up" direction, you can stretch this to about 7 arcsec).

It might seem that nodding 90° from the chop and keeping all four beams on the detector (defined as "quad-chopping") would increase the signal-to-noise ratio (S/N). However, analysis shows that there is no gain in S/N (for the same elapsed time) from using "quad-chopping" versus using "chop-nod". See proof.  This is counter-intuitive since quad-chopping increases the exposure of the object on the detector by factor 2 for an equivalent amount of time. However, quad-chopped data requires 4 additional shift and add steps in the image processing. This increases the background noise by sqrt(4) resulting in the same signal to noise ratio as equivalent chop-nod data.
 

How are guide stars selected?

The guide camera and LWS are fixed in different locations of the forward Cassegrain module, FCM. The rotation of the FCM causes circular motion of the guide camera relative to LWS. The center of the guide camera is 432 arcsec from the center of the LWS field of view. The guide camera is about 60 arcsec wide in this dimension so that is the width of the annulus for available guide stars. The sky position angle between the science object and the guide star is the angle the telescope control software uses as an input for rotation ( note: this is different from the resulting PA for LWS images, in other words, this is the angle you tell the OA. The limiting magnitude, based on almost any DSS object being suitable, is about 18v under typical seeing conditions. A tool called sky is available for making guide star selections relatively easy and can be used during the afternoon preparations or the OA can use it in real time without much of an overhead penalty.

What is chop-nod mode?

Chop-nod is the standard mode of data acquisition for LWS.  Chopping is performed to cancel out sky radiation and nodding of the telescope is performed to help cancel out the radiation contribution from the telescope. The LWS chop-nod style uses a nod throw equivalent to the chop throw but in opposite directions. See figure. The resulting image, after double difference process, requires no further shift and add processing. The chop-nod technique is very effective at canceling out background radiation.
 

What is the LWS efficiency in chop-nod mode?

In imaging mode the efficiency can approach 50%; in spectroscopy mode it is closer to 25%. The reason for this is that the data acquisition system forces us to discard at least one frame during each chopper transition. For imaging, the frame times are short (10 millisec or more) and the losses are small. For spectroscopy, frame times must be increased in order to reach the background-noise-limited regime (100-400 millisec) and hence the losses are much more significant. We are working on a fix, but it involves very low-level reprogramming of the electronics and will take some time to complete.

Software and User Interface

Is the LWS datataking software robust?

Yes, the problems previously suffered by the LWS software during its first use in 1996 have been corrected. Of course, no software is perfect and although the LWS software still has some "features" that need improvement, its reliability has improved vastly from earlier versions. Software crashes have not been a significant problem during the commissioning of the new LWS. The few crashes we have observed occurred when LWS was operating in a "free running" video mode, as opposed to the data-saving mode employed during observing. After several hours in this video mode the acquisition software has a tendency to crash, but the recovery time is typically less than 2 min.

It should be noted that the old LWS system also had a hardware problem which caused spontaneous rebooting, a factor that helped lead to its de-commissioning. This problem is now corrected and not one occurrence of the "spontaneous reboot" has occurred during the LWS re-commissioning.

Data and Analysis

LWS saves 6 dimensional FITS files. How are the dimensions defined?

The LWS data acquisition system allows the user a choice in setting up the various parameters that determine the FITS file dimensions. In particular the user can select a save-to-disk frequency (maximum 5 Hz) . The save frequency and the total integration time determine the chop sets (sometimes referred to as savesets) and nod sets (see below). A higher save frequency may allow the user to apply "shift and add" or other techniques in order to improve the data quality. The penalties for high save frequencies are a drop in efficiency (the acquisition must stop during disk writes ) and very large resulting FITS files. For a further explanation of terms and timing see this figure.

The 6 dimensions of the output FITS file are:

  1. X (image columns)
  2. Y (image rows)
  3. chop beams (2 when chopping, otherwise 1)
  4. chop sets (number of frame pairs saved per nod)
  5. nod beams (2 when chop-nodding, 1 when only chopping)
  6. nod sets (number of nodsets)

What software is available for processing of LWS images?

The trickiest part of the LWS data reduction effort is to combine properly the six dimensions of the FITS image into a co-added chop-nod image. Once that is done, standard optical/IR processing packages such as IRAF are well suited to further processing.

The WMKOLWS package is an IRAF package written by CARA for LWS which features a task called lwscoadd to transform the 6-D images into 2-D arrays. It is available on-line here at Keck and can be downloaded for installation at your home institution; see the package link above to retrieve the code from our FTP site.

IDL users can retrieve IDL software which contains an LWS version of lwscoadd.
 

How is OBJTIME (integration time on source) computed and why doesn't it agree with my intended value?

The OBJTIME keyword value in LWS FITS headers is the correct value for on-source integration time. OBJTIME can be a source of confusion  since you don't always get what you ask for.  The reason it doesn't agree with the intended value (the value entered in XPOSE-LWS ) is that LWS software computes a "ceiling" value based on other parameters. The determination of OBJTIME in chop-nod mode is based on the number of telescope nod cycles. Since the minimum number of nods is 1, the minimum OBJECT time is based on a 30 second (*2) nod dwell. When you enter an OBJTIME in the exposure tool, the system calculates the nearest ceiling value to make sure you get at least that much time on your object. The default 30 second nod dwell time is based on telescope performance and is tailored for longer integrations.

Go to: LWS Home Page - Instruments Home Page - Keck Home Page

Last modified: Tue Mar 14 15:53:18 HST 2000