Overview

MIKE (the Magellan Inamori Kyocera Echelle) is a double echelle spectrograph.  The red side covers 440-1000nm.  The blue side covers 320-480nm.   The first optical element in the spectrograph is a dichroic which reflects (transmits) light into the blue (red) arms of the spectrograph, so the both sides can be used simultaneously with independent exposure times.

As of May 2004, a new dichroic has been installed.  The new crossover wavelength is  ~490 nm.  (The old dichroic produced a crossover wavelength of ~455nm.)

Each side has a 2k x 4k CCD with 15micron pixels.   The blue ccd was replaced  in May 2004 with a Lincoln Labs 2k x 4k CCD (also 15micron pixels).

The CCD electronics are a Thompson & Burley production.   Christoph Birk did the control software for the detectors,  so that data acquisition window will look very familiar to LCO observers.  Each side has a shutter which is controlled by the CCD electronics so that exposures can be taken independently.

For a more complete description of the instrument see Bernstein, Shectman, Gunnels, Mochnacki, Athey 2002, Proc SPIE 4841.
 

Efficiency:

Typical efficiencies with realistic slit losses (e.g. 1" slit in 1" seeing) are such that we get 1 e-/sec/A  for an object with 19.2 AB mag (4000A) on the blue side, and 18.5 AB mag (6000A) on the red side.

The plot below shows the source flux (in AB mag) from which we detect 1 e-/sec/A.  The slit-to-detector efficiency is roughly 20% (~6500A) on the red side and 42% (~4500) on the blue side.  (Measurement and calibration courtesy of Scott Burles and Kristin Burgess, MIT)

Taking data.

A.  Atmospheric Dispersion

MIKE was designed to be mounted at the Nasmyth or auxiliary port.   We expect that the preferred observing strategy will be to use MIKE in a gravity invariant mode in which it is mounted on the East Nasmyth platform, not attached to the instrument rotator.  This is the only mode available now.  The virtue of this mode is obviously the stability of the instrument.  The down side is that the slit orientation on the sky cannot be changed.  For that reason, we have positioned MIKE at a 30 angle to the platform.  At this angle atmospheric dispersion lies exactly along the slit when observing at a zenith angle of 30 deg.  Closer to the zenith atmospheric dispersion is fairly small and the component in the direction across the slit is negligible. At zenith angles greater than 40 deg the magnitude of the atmospheric dispersion increases rapidly and the component across the slit can become a significant problem when the goal is to achieve broad spectral coverage.

The figure below shows the atmospheric dispersion (3500-9000A) across and along the slit as a function of zenith angle for MIKE in its standard position (tilted by 30 degrees) on the Nasmyth platform.   At a zenith angle of 30 degrees,  the dispersion is entirely along the slit with a magnitude of about 1.1 arcsec.    The dispersion  across the slit is small (<0.3 arcsec) for all zenith angles less than 40 deg.  At higher zenith angles (>50 deg) atmospheric dispersion across the slit is significant.  With the 5 arcsec long slits,  most of the light should get into the slit at zenith angles less than 60 deg.  If you are using one of the aperture pairs and are interested in the full wavelength range, then it is probably better to avoid zenith angles greater than about 40 deg for the 2" apertures.

B. Selecting a slit, binning and sky subtraction.

MIKE has a polished reflecting slit plate with a variety of slits machined into it.  Behind the slit plate is a fixed plate which blocks the light from all of the slits except the one which is positioned in front of a single hole at the center of the field. The slit plate is mounted on a motorized stage so that any of the slits can be positioned in front of the single hole in the blocking plate.

To select a slit, make sure the box on the guider camera is set to the appropriate screen location corresponding to the center of the field (see the accompanying table). Open the mirror cover and put some light on the slit so that you can see the image of the slit plate clearly.   Make sure the control switch on the spectrograph control box is set to REMOTE. Turn on the slit move enable switch on the remote control box in the data room. Then push the slit position momentary switch either to the left or right and watch the slits move on the guider camera TV screen.  Position the desired slit in the center of the expanded guider box. There is a handy knob for controlling the speed of the slit motion.

The scale of the CCD detectors is 7 pixels per arcsecond on the red side and 7.7 pixels per arcsecond on the blue side.  So if you are using a slit which is wider than 0.5 arcseconds there is not much point in taking unbinned data.  The loss of resolution due to binning will be negligible, but there are significant gains in readout time and especially amplifier noise when the CCD image is binned.

Most of the available slits are in fact pairs of apertures which can be used to observe in A-B mode (one exposure with the star in the top aperture followed by a second exposure with the star in the bottom aperture).  In this mode it is possible to bin the data very aggressively in the cross-dispersion direction (a value of 4 can be used with the 2 arcsecond apertures and higher values are probably OK with the 1.5 arcsecond apertures).  This is especially useful when observing very faint objects.   A picture of the slit plate as it is seen in the slit-viewer is shown below.

Available slits, from left to right on the guider screen:

Aperture Pairs (separation 3"):

 1  0.35 x 0.35 (for focusing)

 2  1.00 x 0.35
 3  1.00 x 0.50
 4  1.00 x 0.70
 5  1.00 x 1.00

 6  1.50 x 0.35
 7  1.50 x 0.50
 8  1.50 x 0.70
 9  1.50 x 1.00
10  1.50 x 1.50

11  2.00 x 0.35
12  2.00 x 0.50
13  2.00 x 0.70
14  2.00 x 1.00
15  2.00 x 1.50
16  2.00 x 2.00

Single Slits:

17  0.35 x 5.00
18  0.50 x 5.00
19  0.70 x 5.00
20  1.00 x 5.00
21  1.50 x 5.00
22  2.00 x 5.00

To select a slit, set the guider box to X = 512, Y =524  and center the slit in the box.  This value will change if the guider camera is removed from the spectrograph for service.  If this happens, contact RAB or SAS for details on how to re-locate the center.  The pixel scale on the slit viewing camera is 0.67" per pixel.

C. Comparison Lamps

The internal comparison lamp is a Thorium Argon source.   It can be turned on and off from the remote control box in the data room. The switch labeled "comparison lamp" will both turn on the lamp and move a flipper-mounted mirror into place to direct the light onto the slit.  A 1 second exposure is more than enough.  It is very important to turn the lamp off because it has a lifetime of only a few hundred hours and external sources will not be bright enough for blue-side calibration.  The required exposure time is nice and short, so please get into the habit of leaving it on for only a few seconds.  While there is no warning on the remote electronics box that the comparison source is on, the flipper mirror will block the field of view of the slit viewer.

The control box on the side of the instrument has a knob which controls the brightness of the comparison lamp.  Turning the knob clockwise makes the lamp brighter.  At any setting, the lamp is already too bright, but the power supply doesn't regulate properly at the lowest setting.  There is a mark on the box at or above which the power supply works properly.  Leave the knob set to the mark.

A 1 second exposure provides a good calibration spectrum for the blue side and most of the red side.  However, the very bright Ar lines in the redder orders will be totally saturated.  We may put in a filter in the future, but for now you will need to use the dome lights to get a calibration spectrum at wavelengths greater than ~7500A.    We have left the red grating positioned such that these reddest orders do not fall on the CCD, however that can be changed.

To wavelength calibrate the red side above you need to use the external calibration lamps. To do this you need to move the flat field screen into place using the GUI shown below. To start the GUI, type "ffs" in a xterm on the data computer.

The row of green boxes represent individual arc lamps mounted near the screen.  The available lamps are Neon, Argon, and Helium.  Ar + Ne together work well for the red orders.  Click on one of the green buttons to turn on the lamp.  The button will turn red when the lamp is on.  Click it again to turn it off.  Below the row of arc lamps is a row of boxes that move the flat field screen in and out of the telescope field of view.  When the screen is in, the button in the lower right corner will be red and read "In."    A few seconds is adequate for the red side.  Use the internal comparison source for the blue side.

D.  Flat fields:

A diffusing glass slide can be positioned in the optical path just downstream of the slit for taking "milky flats." This must be done manually (see below).   The slide will blur the slit  image into a  roughly 40-50 pixel smudge.   This is a good size because it means that there is not much light shared between orders in the flat.  However, you will want to use a light source which does not have many lines.   We found that a 100 second exposure of a bright (2-3 mag) O star will provide a very good flat field image during twilight.    It is not too hard to position it in the slit using the slit viewer, although it is impressively bright.

There is also an internal quartz lamp which can be used for flats with the milky flat diffuser.   It can provide enough photons at redder wavelengths, although  bluer orders will probably require an O star milky-flat.   The daytime crew can show you how to use this internal lamp.   To summarize the important steps,  the arc lamp must be disabled by unplugging it on the electronics box mounted on MIKE;  the flat field lamp must be turned on manually from its power source on the platform;  and the comparison lamp switch on the MIKE control box in the observing room must be "on" (to move the flipper-mounted mirror into place).

E. Aborting an Exposure:

The easiest way to end an exposure quickly is to change the exposure time to a few seconds in the future.  You can do this by typing a new exposure time into the appropriate control box and hitting return.

F.  Orientation of spectra:

Blue side:
Reddest orders are at x=small (on the left).
Red end of each order is at y=large (at the top)

Red side:
Reddest orders are at x=large (on the right).
Red end of each order is at y= small (on the bottom).

G. Recommendations:

Bias:  Occasional jumps in the bias level occur but can be subtracted out nicely by using an average of the overscan region at the end of each row (x=2048:2176,y=*, unbinned).  The highly-repeatable bias ramp which occurs at the beginning of each row can be easily subtracted using the overscan at the end of each column (x=*, y=4096:4224, unbinned).  Using these overscan regions is more effective than an using a combined bias frame.

Dark:  The new Lincoln Labs blue ccd (installed May 2004)  have some dark current (~10 DN/1800 sec, depending on running temperature) which will add noise to the exposure but does not appear to have significant 2-d structure over the CCD.   Check the dark level during your run to estimate and keep this in mind when estimating the signal to noise ratio of your final exposure.

Arcs:  It is a good idea to take arcs every between long exposures (every 15-60 minutes).  Although there is no motion in the image due to flexure (the instrument doesn't move), temperature changes of the air, glass, and metal can cause small (~ 1pixel) motions of the dispersed image over the course of the night.  So while the telescope pointing doesn't affect the wavelength calibration, time does.

Flats:  It is a good idea to take flats with the diffuser for standard pixel-to-pixel flat fielding calibration.  You may also want to take some flats (either internal lamp or sky flats just as the sun is setting) without the diffuser also. These can be used for an illumination correction and to trace the edges of the orders.  Take an arc at the same time to provide an easy way of identifying the offset between the order position in the flat  and the order position in each program exposure.  (The orders can move by fractions of a pixel in both the dispersion direction and the cross-dispersion direction due to temperature changes in the spectrograph.)
 

Moving parts

A. Electronically controlled

A control box is mounted on the side of MIKE for "local" control.  A second, remote box is located in the telescope control room.   All of the controls are manual switches or knobs and are labeled.   Be sure that the local control box switches are set to "REMOTE" so that the slit, comparison, and focus can be controlled from the data room.

1. Slit plate

The position of the slit is controlled by a motorized stage and positioned using the internal slit-viewing camera as described above (see Taking Data).

2.  Internal Comparison Source

The comparison source switch also controls the flipper mirror as described above (see Comparison Lamps) .

3.  Camera Focus

The blue camera focuses around 1.150.   The red camera focuses around 1.050.

If you refocus, try moving in steps of 0.010.  We do not yet have an empirical temperature correction.  The optical benches under the cameras each contain an invar plate which will move the optics as the aluminum expands or contracts with temperature.  This passive thermal control should keep the cameras in focus for changes in temperature up to about 5 degrees C.

B. Manually controlled:

1.  Diffusing slide for "milky flats"

A diffusing slide is mounted on a sliding post and can be positioned in or out of the optical beam just downstream of the slit.  The post sticks through the blue adapter plate below and left of the blue dewar (between the blue and red dewars) and has a brass knob on the end.   When the knob is pushed in (flush with the adapter plate),  the diffuser is out of the beam.   When the slide is pulled out,  the diffuser is in position for milky flats. There is a detent at each position.
 

2.  Grating position

The gratings are positioned manually against preloaded, threaded rods.  Dial guages underneath the red and blue boxes indicate the linear position of the rods, and therefore the position of the grating. There should be no need to move the blue grating.   The azimuthal position of the red grating can be adjusted to select the orders which fall on the chip.  Each of the guages and threaded rods are labeled on the spectrograph.

As of May 2004, the standard grating settings and corresponding wavelength coverage are as follows:

    Blue Azimuth:    0.346 (was 0.374)
    Blue Elevation:  0.540 (same)
    Orders 71-106
    5000 - 3350 A

    Red Azimuth:     0.474 (was 0.535)
    Red Elevation:   0.380 (same)
    Orders 37 - 70
    9300-4900 A

These elevation settings center the free spectral range well, and probably should not be changed.  The azimuthal settings can be adjusted to change the order coverage.
 

MIKE data acquisition GUI.

The data control window is shown below. To start the window, type "mike" in an xterm on the data control computer. An instrument configuration window will appear, also shown below. The observer can enter his/her name and then select three camera configurations, blue camera only, read camera only, or both blue and red camera. There is an entry for offline (simulator) operation of the GUI, this is for testing purposes only. Next the observer should select the number of pixels of overscan at the end of each row and the number of bias lines at the end of a readout that should be appended to a frame. Finally the observer can select the telescope, and again an online or offline operation. If the telescope is online then telescope coordinates and various rotator angles are read from the telescope TCS.

The very top portion of the acquisition window has two pull down menus, the first controls the window status (select Exit GUI to exit the window) the second has an entry for user name and another to select the path to the data. This path can be changed without having to restart the window. These are followed by a bar graph showing the current data disk capacity, and a display of the current UT.

The top half of the window shows information relevant to BOTH the red and blue side, like telescope coordinates, dome temperatures, airmass, etc. Exposures for simultaneous operation of both cameras with all exposure parameters identical are controlled by the Exptime, Start, Pause, Abort, and Loop buttons in the upper right portion of the top half of the GUI. The temperatures of the CCD's are given in the top half of the window. The current set points are -125 C, and these numbers will turn red if the temperature rises above the set point by more than 5 C.

Entries of numerical values for image number and exposure time will cause the background in these windows to turn red, the values are entered into the program when you hit the return key. Other character fields (white boxes for comments, settings for camera focus values, grating angles, slit-size) can be simply edited. Entries for X binning, Y binning, full or subrastered readout, and speed are controlled with pop up menus, indicated by the presence of a small circle in the right side of a menu. Two readout speeds are available, a fast, high-noise mode (full frame readout 95 seconds, readnoise approximately 4.7 electrons), and a slow, low-noise mode (full frame readout 160 seconds, readnoise approximately 3.7 electrons). Note that the read times in all cases will be significantly shorter when a camera readout is binned. When exposures are running, the start buttons turn yellow in the exposing cameras. Start/Pause/Abort buttons are beige when those functions are available.

Cameras can also be operated completely independently with the control buttons in the lower left (blue) and lower right (red) portion of the window. Exposure times can be different, exposures can be started asynchronously, and the number of exposures in a loop can be different.

There are two message windows at the bottom of the GUI. These give the status of an individual readout, both the elapsed read time and the number of lines read and the total number of lines to be read.
 

rabernst@umich.edu
Last updated: July 2004