User's Guide to the Kast Double Spectrograph

Table of Contents

Quick Reference
Hardware Overview
Common Path
Blue Side
Red Side
Detector Characteristics
Kast Controller
Data Taking System
Position Angle
Arc and Flat-field Lamps
Diagonal Mirror
Kast Focus
Telescope Offset
Setup and Observing Hints
Setup Procedures
Observing Hints
Calibration Lamp Spectra
Exposure Time Calculator

Data Archive
Mt. Hamilton Homepage

Detector Characteristics

CCDs | System Response and Throughput | Flexure and Fringing | Shutters | CCD Controllers | Field Size and Orientation


The red side uses a Hamamatsu 1024x4096 device with 15 micron pixels, which corresponds to 0.43 arcsec per pixel. The dispersion direction is along columns (and the detector is tilted slightly, about 1 degree, with respect to the slit). The blue side uses a Fairchild 2k x 2k device with 15 micron pixels, which corresponds to 0.43 arcsec per pixel. The dispersion direction is along rows on the blue side. The blue device has excellent UV response (>70% at 3200A) and the red device has excellent red response (>94% at 6500A) and both are relatively free of blemishes.

The CCDs should be protected from unnecessary exposure to bright lights. The telescope technicians are responsible for keeping the dewars cold.

The full well depth for the Hamamatsu chip is 150,000 e-, but starts being non-linear in response at about 32,000 DN in slow readout mode. In fast readout mode, the detector is linear until the ADC saturates at 65,000 DN. The gain is about 1.9 e-/DN (0.55 e-/DN if fast readout mode). Two amplifiers are used to read out the CCD. The overscan regions occupies the far right (highest numbered) columns. Readnoise depends on the readspeed for the chip, as listed below. The Hamamatsu chip is thick and hence is very sensitive to cosmic rays. Taking multiple shorter exposures to median filter out the cosmic rays may be desired to deal with the large number cosmic rays. The Hamamatsu detector is much larger than the region which the spectrum covers and there is a mask in front of the detector to prevent scattered light from illuminating unused areas of the detector. The spectrum covers rows 760 to 3310 on the detector, though more spectrum may be visible it may be vignetted beyond these rows.

The full well depth for the Fairchild chip is 110,000 e-, but the ADC saturates at about 65k minus the baseline, or usually about 64,000 DN. The chip is linear up to the A/D saturation point. The CCD is read out using two of the four amplifiers. The gain is about 1.2 e-/DN for both amplifiers. The overscan regions occupy the far right (highest numbered) columns. Readnoise depends on the read speed for the chip, as listed below. [Note that since the installation of the Fairchild CCD in Sept 2008 until Apr 11, 2011 there was a readout programming error where the central 4 columns (two on each amplifier) were not read out. This leads to larger than expected errors near the center of the spectrum when doing wavelength calibration. The ad hoc solution for analyzing older Kast Blue data is to add 4 fake columns to the center of the data or some similar compensation when doing the wavelength calibrations and analyzing the data.]

CCDRead SpeedGain1 Read Noise1 Approx. Readout Time2
Blueslow 1.2 e-/DN 3.7 e- 7 sec
Bluefast 1.3 e-/DN 6.5 e- 2 sec
Redslow 1.9 e-/DN 3.8 e- 20 sec
Redfast 0.55 e-/DN 4.3 e- 14 sec
1Read noise and gain measured 2016 Sept.
2Readout time is for typical sub-region of CCD. Blue: 325 2048 900 0 (nr nc sr sc). Red: 2725 525 675 409 (nr nc sr sc).
Full CCD readout time (in slow mode): Blue (2110x2048) 43 sec, Red (4096x1024) 34 sec.

Suggested operating temperature for the red and blue side detectors is -105 . A warmer temperature will increase the dark current but diminish charge transfer inefficiencies, and vice versa.

In general, if you don't have any prior knowledge of the expected exposure time, a good practice is to take a one second exposure (not recorded), and then scale that to the desired count level to determine the exposure time. It's best not ot overexpose the CCDs. Although no permanent harm will be done, it may take some time to completely flush the extra charge.

This is a small point, but perhaps worth mentioning. The saturation level is determined not necessarily by the well depth, but by the limitation imposed by the 16 bit A-D converter. Thus, if you bin pixels, you may need to reduce the exposure time correspondingly to stay within the 65k capacity of the A-D. Remember too that the actual dynamic range available is not 65k, but 65k minus the baseline. The baseline should be set to 1000 DN (as measured in the overscan regions). If the baseline is vastly different from this figure, contact a support astronomer.

System Response and Throughput

The cameras on both sides are all-refractive. The camera lenses are temperature controlled; focus is a function of lens temperature. The focal planes on both sides appear to be quite flat. You may observe some small-scale variations in the focus due to irregularities of the chip.

Blue: The design range of the camera is 3000-7000 A. It looks good to atmospheric cutoff. Peak efficiency of the entire system including the telescope is between 5 and 20%, depending on setup. The blue side operates in first order, and due to the wavelength coverage, red leak should not be a problem. If you're imaging on the blue side, remember that the lens performance deteriorates past 7000 A.

Red: The design range of the camera is 4000-11000 A. Peak system efficiency is in the vicinity of 30-40% (with the old Reticon detector, figures have not been calculated for the Hamamatsu chip), depending on configuration. Response is decreasing rapidly by 10,500 A, but successful observations have been made out to 10,830 A. You will need to suppress second order if you go beyond twice the effective cut-on point of the dichroic you use. Remember that the user, lower, and upper filter wheels are in common for both beams, so use the 5.5" round filters in the red camera filter wheel. The Hamamatsu detector shows little fringing until the very reddest wavelengths. For most objects, red exposures will probably go faster. Do multiple reds if necessary to avoid red saturation during one blue exposure.

NB: Additional, more current, throughput data has been provided by Dr. Prochaska (UCSC):

Shane 3 meter Kast Throughput

Flexure and Fringing

As the position of the telescope changes, the Kast spectrograph flexes. There is up to nine pixels of total shift on each side (depending on CCD), moving between extreme positions on the sky.

The blue side may shift as much as nine pixels parallel to dispersion, but fringing will not generally be a concern, and the shift may ordinarily be accounted for by reference to skylines.

The red side may also shift as much as nine pixels in the dispersion direction, and because of fringing this may be a more serious problem than on the blue side. The Hamamatsu chip is thick and thus fringing has been minimized, though may still be seen to a small extent at wavelengths greater that 9500 A. This may become a particular problem due to the flexure described just above, because if observations at large zenith distances are flattened with straight up flats, the object and flat fringes may not match. If this is a concern, you may with to take "local" flats.


The Ilex shutter technically has no minimum exposure time. However, exposures of less than 1 sec are not recommended (a 1 sec exposure will have a timing errors of a few milliseconds).

CCD Controllers

There is a separate controller for each side. They are in close proximity to the dewars. The power supplies for the controllers and CCDs are mounted above the instrument in the 19" electronics racks. These contain most of the temperature and readout electronics for the CCDs, and will be set up by the dome crew. The only things the observer need to be concerned about is the temperature readout on the power supply for each controller. It reads in degrees Celsius to the nearest 1/10th degree at a location in the dewar near the chip. It should be fairly stable, and very close to -105 C for the CCDs.

Field Size and Orientation

The long slit capability of the spectrograph is very useful for extended objects as well as collinear ones. For these sorts of observations, one nearly always wants to rotate the instrument TUB to some predetemined (or in some cases, determined on the spot from the guide camera or CCD direct frames) position angle. It's useful to know which diretions are which, for various position angles, in order to plan setups and verify that they are correct.

On the guide camera, the scale and orientation varies with diagonal mirror position. If the TUB is at the standard position angle of 90 degrees, then in mirror position 2 one sees a field about 2 arcmin across on the guider with north up and east to the left. When offset guiding in position 3, this remains the same. The slit runs left-right on the guide camera. At a position angle of 90 degrees, this means the slit runs E-W. In position 4, the field size is roughly halved to about 1 arcmin, so about half the total slit length is seen on the guide camera at a given guide camera position. The guide camera may be moved by the telescope operator on its stage to view locations farther along the slit if necessary.

On the CCDs, with the TUB at position angle 90 degrees, north is right and east is at the top for the blue side (note the dispersion direction is along rows). The red side dispersion direction is along columns, so it is rotated 90 degrees with respect to the blue side CCD, so that north is down and east is right. It could be worse; at least these are simple rotations of most charts. The slit is of course still E-W, so on the blue CCD the slit appears vertical with east at the top. On the red CCD it appears horizontal with east at the right. Note, these orientations are for when the data are displayed with the XVideo software used by the Kast data-taking system. Images are flipped top for bottom in DS9, IDL, and most other image display software (meaning that for PA=90 deg, east would be down on the blue side and north up on the red side).

For different position angles of the TUB, the relative orientations of the guide camera, slit and CCDs are unchanged, since they all move together. The effect of going to a higher position angle (that is, in the usual sense of north through east) is to rotate images on both the guider and CCDs clockwise.

Support Astronomers (
Last modified: Mon Oct 2 20:09:19 PDT 2017