We are trying here to present a general description of the hardware of the BME. If you want to learn how to use the BME to take data, be sure and read the BME User's Manual, in addition to the present document. Once you have some data, you will find the BME Data Reduction Manual useful.
The BME contains fibers 100 and 200 microns (1.8 and 3.6 arcsec)
in diameter. There is a slit assembly which allows the fibers to be masked
in the spectrograph to discrete widths of 200, 140, 100, 70, 45 and 30
microns to improve resolution at the expense of efficiency. The 200 micron
fiber is almost always used as it is more efficient except when extremely
high resolution is required and seeing is excellent.
The slits are selected with a small micrometer in front of the fiber.
Nominal settings of the micrometer for the 200u fiber are:
|
(200u fiber) |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Setting of the slit width is done by Observer Support and not the observer. Slit location should always be verified by illuminating the fibers from the telescope using the quartz source, putting a Questar focussed at infinity in the beam, pointed at the collimator looking back at the fiber. With this arrangement, it is easy to see when the slit is properly centered on the fiber.
The fibers presently used are 15M long and are of the so-called "wet" variety, doped with OH- to improve their UV transmission at the expense of reduced throughput in the near IR. These fibers have excellent transmission of approximately 80% from 4000A to 8500A, except for a dip between 7100A and 7300A, in which the transmission falls to a minimum of about 60% at 7200A. This is a relatively broad spectral feature which is eliminated by the flat fielding process and after the data has been reduced can only be detected through careful measurements of absolute efficiency on standard stars.
Below 4000A the fiber transmission falls gradually to about 30%
at the atmospheric cutoff. Above 8500A absorption bands begin to appear,
decreasing the efficiency by about 1/3 between 8500A and 9000A. The fiber
throughput is low beyond 9000A.
Light comes from the telescope to a fiber head which holds the fibers at the telescope focus. The fibers are mounted in a stainless steel post which is polished to mirror smoothness along with the fiber tips. Light from the telescope passes through a pellicle mirror before forming a focus on the tips of the fibers. Although the pelicle wastes a small fraction of the light (about 7%), this system permits continuous viewing of a target object while and allows very accurate centering of the target on the fiber.
Reflected light returning from the stainless steel mirror surface is reflected off the underside of the pellicle and reimaged onto a television camera. A clear tv image is seen of the tips of the fibers and a small area around them. The telescope is then pointed at a target object. The TCS is precise enough so that the target will almost always fall automatically into the field of the television camera. Using the hand paddle, the telescope can be focussed and then moved until the brightness decreases dramatically as the image "drops" into the fiber. Brighter objects can still be seen by the light reflected off the tip of the fiber.
It is relatively easy to guide manually since whenever the image begins to move out of the fiber, the brightness increases rapidly and corrections are easily made. Normally, the "GAM" Autoguider is used. The GAM moves a guide probe into the edge of the telescope field without blocking the fiber. A guide star is found and the telescope is set to guide automatically.
There are currently two options for the television camera: An RCA ISIT camera and a cooled CCD camera. The RCA ISIT is the best camera to use under most circumstances. When stars are dimmer than about 11th magnitude, they become difficult to see with the RCA camera and the CCD camera should be installed. The CCD camera should not be used on brighter targets. It is significantly more difficult to use because it saturates on bright objects and does not have the dynamic range nor fast refresh time of the RCA camera. Changing the camera cannot be done during the night.
The head contains quartz and Th-Ar reference lamps whose light can be directed into the fibers by means of a small motorized mirror in the fiber head.
In the optical bench, the light emerges from the fiber and is collimated by an off-axis collimator of e.f.l. 900 mm and sent to a 200 x 375 mm 31.6 l/mm echelle grating rotated about a vertical axis by approximately 5.8 degrees. This so-called "quasi-Littrow" mode causes the dispersed light to emerge in the same plane as the incident light. It then hits a cross dispersing grating and is sent to a camera for detection. A small amount of light is diverted from the center of the beam to a photometer. No light is lost, since the light diverted would otherwise have fallen in the central hole of the camera mirror. The photometer can sometimes be helpful in centering objects on the fiber, though it is only sensitive enough to be useful for objects of at brighter than 7th or 8th magnitude and is not heavily used.
The CCD covers the free spectral range of the echelle with overlapping orders at almost all wavelengths, so it is almost never necessary to tilt the echelle grating. The only adjustments required in the spectrograph are focus and the tilt of the cross disperser.
The 4M R/C gratings are normally used as cross dispersers, though the 4M user has first right of use so the 1.5M observer's first choice grating may not be available. As a rule, one uses the minimum cross dispersion necessary to separate orders. The preferred gratings are KPGL2 in the UV or Blue and #400 in the red.
The BME uses a 750 mm folded Schmidt camera. This camera has excellent
throughput and resolution from the atmospheric cutoff to beyond one micron.
The TEK2048 CCD with 24 micron pixels is normally used. This setup covers
the full free spectral range out to past 9000A. The camera covers the entire
area of the CCD without vignetting. The wings of adjacent orders
begin to contaminate each other when orders are less than about 20 pixels
apart on the Tek Minimum noise is 3 e- or less. Readout time
is under a minute. The spectrograph has an optical reduction factor of
approximately 1.2:1. This means that the 200 micron fiber will project
to a width of approximately 7 pixels on the CCD, giving a resolution of
about 15,000, with maximum throughput. As the slit width is decreased,
the resolution increases as the throughput declines, according to the following
approximate table
.
|
|
||||||
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
The maximum resolution of the system with the Tek2048 is thus about 60,000. There is no benefit from using the 30 micron slit with this chip. If the image quality is extremely good, throughput with a small slit should be better with the 100 micron fiber, though this sitiuation is uncommon. Results are almost always better with the 200u fiber and a slit.
Normal grating setups with Tek 2048:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
The on-blaze tilt is theta(blaze) + phi/2 where theta(blaze) is the blaze angle, phi/2 =20 is half the collimator-camera angle, and the tilt is measured from collimator-camera bisector (zero order- approximately 188.8 degrees on the scale on the cross-disperser rotator). Gr400 will center at approximately 6300A with a tilt of 192.1 degrees. KPGL2 will center at approximately 4000A with a tilt of about 192.7 degrees. The scale on the tilt is fairly crude so these numbers are only approximate. There is a micrometer which can be used for fine tilt adjustments, but the scale on it is relative. Starting from these numbers it is not difficult to find the desired grating tilt since the spectral range is quite broad.
The instrument has been used at wavelengths as short as 3150A, though
the efficiency is several times lower than it is at 3700A, primarily because
of increasing losses in the fiber.
Grating KPGL2 in first order:
3800A: 0.8%
4000A: 1.6%
4500A: 2.3%
5000A: 2.8%
5500A: 2.3%
6000A: 1.3% (way off blaze)
This diagram is a plot of BME efficiency measured on January 17, 1997 using the 200u fiber and the 750mm camera. The open circles are measured with KPGL2, while the dark circles are with grating #400. The efficiencies of many lines at various places on the orders have been measured. The envelope gives a good estimate of the peak efficiencies. Peak efficiencies are somewhat higher than those measured with the 590mm camera (e.g. ~3.1% vs 2.3% at 4500A) as is to be expected because the 750mm camera has less vignetting. The graph shows clearly that KPGL2 should be used at wavelengths below about 5600A and #400 is best at longer wavelengths. The efficiency of the instrument aproaches 5% at 6200A.
In September 1997, the folding flat mirror on the 750mm camera was changed for one with a larger hole in the center. This increased spectral coverage by a factor of 4-8, but decreased efficiency by about 12%.
A realistic representative example of the current efficiency of the system can be seen here. This is a plot of BME efficiency measured on 18 May 1998 using grating #400, the 750mm camera, the 200u fiber but now with a 70u slit. The efficiency would be expected to be about 40% of the efficiency in the previous graph because of the presence of a slit and the larger hole in the folding flat mirror. (.44 x .88=.39) It is actually about 30% of that shown in the previous diagram, the difference presumably coming from variations in seeing or guiding between the two runs.
Note: The "wet" fibers used have a broad absorption dip caused by OH- from 7200A-7800A. The transmission dips by as much as 50% in this range. This attenuation band can be clearly seen in the graph. The fiber transmission is good from 8000 to approximately 8700A. Beyond this wavelength, fiber absorption is very large and use of this spectrograph is not recommended. Both of these regions of high attenuation can be clearly seen in the graph.
Efficiency measurements must always be taken with a grain of salt since
they are strongly seeing and guiding dependent. However, these numbers
are believed to be representative of the efficiency which would be obtained
by a observer of normal competence on a photometric night.
Sample exposure time calculation:
At an efficiency of 2.8% at 5000A, the raw flux (no slit) from a 0 magnitude star would be about 19,000 detected photons per second per pixel (0.04A) along the spectrum reduced to one dimension. This implies that an 90 minute exposure will give a S/N of 100 on a star of 10th magnitude.
For the maximum resolution of 60,000, the 45 micron slit must be used, which will increase the exposure time by about a factor of 3.6.
As a practical limit, it is quite difficult to locate objects dimmer
than 12th magnitude with this instrument.
Return
to CTIO Home Page
