(Webpage Last Updated: March 5, 2008 : This page is under construction and will change often)
TripleSpec Cookbook
Terry Herter 21-Feb-08
Version 0.2
These are a few notes to help with taking science data and calibrating it. Hopefully this will help until we get the users manual written. For those of you using TripleSpec during this at risk period, send me comments. See Gustavo's memo for technical details on using the software interface. Please keep in mind that this interface is still evolving and not fully debugged. We are working on the most efficient way to operate TripleSpec.
Specs:
Slit 1 x 30 arcsecond slit
Wavelength coverage 1.0 - 2.4 micron
Spectral resolution 2500-2700
Sampling ~ 2.7 pixels per resolution element
Minimum integration time 4 seconds
Gain ~ 3.8 e-/DN
Saturation level ~ 28000 DN (conservatively)
Data sampling CDS, Fowler, or sample-up-the ramp.
Read Noise 10 e- (CDS), 5 e- (8 samples)
3.5 e- (16 samples)
Dark Current ~ 0.085 e-/sec
Time to background limit ~ 300 seconds
Spectrograph Notes
In the near-infrared the OH lines from the sky are strong and variable, and in the K-band thermal emission is seen from the telescope and sky. There are also a significant number of bad pixels in the array. As such it would be a good idea to take spectra at different positions along the slit.
There can also be some flexure (~0.5"-1 pix) when going from vertical to horizontal with the spectrograph, so even if the OH lines didn't vary you may see poor subtraction of the lines.
You can also get a quick look at your data by taking the difference of two different positions along the slit (with the provisos above). For short integrations (~30 seconds) a simple difference is likely to work well.
Spectrograph Setup:
Setup directory for saving data
Select starting file numbers for the night (e.g. 1000 for night 1, 2000 for night 2, etc).
Select number of Fowler samples. At present 8 or 16 seems fine.
The integration time will be set appropriately for each object.
Calibration Data:
Dome flats:
Take with dome dark, mirror cover open.
High lamp: t = 30 seconds, 10 images to average
Lamp off: t = 30 seconds, 10 images to average
Calibrators:
Choose an Elias standard or one from the supplied list of A0V stars
For future reference place a marker at the location of the star before moving it onto the slit (this will help with acquiring sources later)
For Elias stars, t = 30 works for most
Take spectra at 5 positions along the slit (you can use the "Take Seq" button in the Guider window.
Object Data:
Set up dither positions along the slit
It is recommended that you take data in at least two different positions along the slit. For point sources you may want to do more (e.g. 5 positions or so) for good removal of bad pixels and sky - we're still exploring the best way to do this.
There is a check box below the slit zoom window which causes the markers to be confined along the slit. It is suggested you use this feature unless you have some special needs.
Acquire object in guide field
Note: the default position will be off the slit
Place a marker box on source and use "cm" command in a zoom box for the window to center the box on the source.
Place the guide box over a star that will be in the field when the source is moved onto the slit. Use "cm" command in the guide box to center the guide star.
Turn the guider on
Note: Autoguiding has been a bit problematic, but does seem to work when the settings are adjusted properly. We're still working on what works best see Gustavo's memo.
Select the marker with the object (e.g. green) and use the "bring to 1" command to place it on the slit in position 1.
Check to make sure your source is on slit position 1 and the guide box is center okay.
Select the integration time you want for the spectrograph
Note: the value in the Guider window will override the spectrograph value if you use a "Take Seq" or "Take Spec" command.
Use "Take Spec" or "Take Seq" command to start-up taking data
Sometimes you want to take a spectrum first before starting a sequence to make sure everything is okay. You can then go to the next position by using the "Goto" button.
Note: if you have the "gdr img" box checked, each time you take a spectrum from the guider window, a guider image will also be taken.
Exposure times:
The shortest exposure time is about 4 seconds. There are three possible limits to the maximum exposure time: source saturation, saturation in the K-band and saturation of OH lines. The plot below shows the background measured by TripleSpec near zenith on 16-Feb-2008 (ambient temperature ~ 0° C). Using a saturation level of 28,000 DN, the maximum integration time is approximately 2400 or 1200 seconds for saturation to start occurring in the H and K-band respectively.
The longest integration times we have done on sky thus far are 600 seconds. With a dark current of 0.085 e-/sec and a read noise of 3.5 e- (w/ Fowler sampling), then the noise due to the dark current will equal the read noise in ~ 144 seconds. The inter-OH continuum is difficult to measure but appears to be ~0.1-0.8 e-/s over the J and H-band. This implies that 300 seconds should be sufficient to overcome the read noise. (Note: this result is preliminary and will require a bit more analysis.)
Figure 1: Plot of sky + telescope emission for TripleSpec in units of DN/sec taken on February 16, 2008.
System Response:
The response (integrated across a column) to at 10th magnitude star is given in Figure 2. This can be used for scaling to other magnitudes.
Figure 2: Plot of sky DN/sec per column for a 10th mag A0V star.
Standard Stars
These are a subset of the Elias standards (A-stars). Typical integration times will be 30 seconds. They will be saturated in the guider field. A K-star (or model profile) will be needed to remove the H recombination lines. Additional calibrators can be found via the usual searches on-line.
Name RA (1950) Dec (1950) J H K
HD 225023 00 00 11.8 35 32 14 7.065 6.985 6.96
HD 1160 00 13 23.1 03 58 24 7.055 7.045 7.04
HD 3029 00 31 02.3 20 09 30 7.25 7.12 7.09
HD 18881 03 00 20.5 38 12 53 7.125 7.13 7.14
HD 22686 03 36 18.7 02 36 07 7.195 7.19 7.185
HD 40335 05 55 37.6 01 51 09 6.54 6.47 6.45
HD 44612 06 21 09.7 43 34 35 7.06 7.035 7.04
HD 77281 08 59 05.4 -01 16 45 7.105 7.05 7.03
HD 84800 09 45 35.9 43 53 56 7.56 7.53 7.53
HD 105601 12 06 56.1 38 54 39 6.81 6.715 6.685
HD 106965 12 15 24.0 01 51 10 7.375 7.335 7.315
HD 129653 14 40 38.2 36 58 07 6.98 6.94 6.92
HD 129655 14 41 11.0 -02 17 38 6.815 6.72 6.69
HD 136754 15 19 24.3 24 31 19 7.15 7.13 7.135
HD 161903 17 45 43.3 -01 47 34 7.17 7.055 7.02
HD 162208 17 46 20.7 39 59 40 7.215 7.145 7.11
HD 201941 21 10 13.6 02 26 12 6.7 6.64 6.625
HD 203856 21 21 37.1 39 48 12 6.925 6.88 6.86
Continuum Sensitivity Plots
The following are plot of estimated S/N ratio for an "A0V" star at various magnitudes. These are computed per column so that averaging across a resolution element improves the signal-to-noise ratio by a factor of sqrt(3). The estimates are based on calibrator and sky measurements taken in February 2008. The ambient air temperature was ~ 0°C.
Note: An increase in noise due to subtraction of two spectra is NOT included. Whether this is done or not depends on the final reduction technique. If a simple difference of spectra taken at two positions in the slit to remove background and bad pixels then the signal-to-noise ratio will decrease by a factor of sqrt(2).
Assumptions:
Read Noise 3.5 e- (16 samples)
No. Pixels in extraction: 4
Fraction of flux in extraction: 0.7
Spectra not subtracted: no sqrt(2) loss due to differencing
Spectra are not averaged no sqrt(3) improvement in S/N or
over a resolution element factor of 3 reduction in time (or coadds)
The plot label gives the magnitude, exposure time for each image and the number of coadds (ncoadds). Obviously, the S/N will improve by the sqrt(ncoadds).
Figure 3: S/N ratio per column for an m = 15 (A0V type) source. Coadding across a resolution element (3 pixels) will decrease the number of coadds by a factor of 3 or (for fixed number of coadds increase the S/N by a factor of sqrt(3). No differencing noise is included.
Figure 4: Same as figure 3 but for m = 17.
Figure 5: Same as figure 3 but for m = 18.
Figure 6: Same as figure 3 but for m = 19