Introduction
This post describes some construction details for an improved version of the SGS spectrometer in what I called the MFSS or Multiple Fixed Slit Spectrograph. The design was intended to overcome several drawbacks of the original, in particular its weight and size: at over 3 kg and a length of over 40cm, this placed the centre of mass of the spectrograph some distance away from the main telescope and gave rise to considerable differential flexure when the telescope was pointed to different parts of the sky.
This new design brings the centre of mass of the spectrograph much closer to the telescope, resulting in a more compact unit. Its reduced weight of less than 2kg including the camera helped to reduced differential flexure when moving the telescope across the sky.
The design is significantly different in that the input slit and CCD detector are at 90 degrees to each other. A schematic diagram of the basic components of the MFSS can be seen below and the various stages of construction are outlined after:
Construction
The beginnings of the spectrograph, showing the frame, partially completed. Since I do not have access to a machine shop but use only hand tools, my first choice for materials for the main body was wood or wood composite (MDF). A block of MDF of dimensions 60x40x20 mm, with a 30 mm hole, serves as the support for the collimator lens and lens barrel. Nothing could be simpler!
The collimator lens in its holder. The lens is an achromatic doublet with focal length of 90mm. A helical focusing tube, with a very fine thread, allows for small adjustments to be made for final collimation and is then locked in position with a small set screw. Available from Edmund Optics.
I chose to use commercially available supports for the front surface mirror and the grating mount. Since a wooden body lacks the rigidity of metal, good quality positioning mounts were used so that more precise adjustments could be made, if needed, and any flexure compensated for. Commercial mounts like these can be obtained from companies such as Newport or Edmund Optics.
This is essentially the same coupling ring that is used on the larger SGS shown earlier, but without the variable slit and attachment plate. The large central hole is tapped to receive a CS-mount threaded video camera lens for times when I want to record spectra of non-astronomical light sources without the telescope.
Above are 2 images of the slit holder with one of the interchangeable slits inserted on the side. This holder was temporarily borrowed from an Oriel MS125 1/8 metre focal length spectrograph. The slit holder fits snugly into a hole in the side of the FSCS. Recall that in this spectrometer design the slit and detector are at 90° to each other. Using a plane front surface mirror just after light passes through the slit does result in some light loss (the mirror’s reflectivity is about 95%) but this was considered acceptable. The main advantage comes from the reduction in size of the whole instrument.
The family of 25, 50, 100 and 200 micron slits used with the spectrograph. Again, these originate from the MS125 and are available from Oriel-Newport Corp. or found on eBay. The “slit” on the right is simply a square hole that can be used for wide slit or slitless spectroscopy. The small square cut-out at the bottom right-hand corner of each slit enables the aperture to be positioned precisely and reproducibly at the centre of the slit holder (previous photo).
A preliminary alignment test at this stage. I am observing from the position of the grating to check for correct mirror orientation. Fine adjustment with the helical focusing ring produced a sharp clear image of the slit which is now collimated. Thus light rays focused on the slit will be parallel when incident on the grating.
The grating (this one from the MS125) with a “rotation arm” bolted to the mount. This is simply a piece of 10 mm angled aluminium and is used to provide controlled rotation of the grating when coupled to a micrometer screw. Just visible in this picture, behind the angle bracket at the red marking, is a knurled nut that has been glued to the arm at a precise position. Seated in the nut will be a small steel ball which will maintain contact with a micrometer screw via a suitable spring attached to the arm and the rear wall of the spectrograph.
The mechanism used to provide rotation of the grating on its mount, to allow different wavelength ranges to be detected by the CCD, is shown here. A micrometer screw pushes against a hardened steel ball in a small socket. Firm (but not too firm!) contact between the micrometer and ball is maintained with a tensioning spring of the right force constant that I retrieved from an old ink-jet printer. When rotated, the grating arm moves through an arc of radius about 39mm.
I am using a 1800 lpm grating with this spectrograph, and I want to have access to the whole visible spectrum from about 380 nm to the near IR at about 900 nm. When turned throughout the full wavelength range, this angular rotation corresponds to a lateral displacement of just over 6 mm between the point of contact of micrometer screw and the steel ball.
Since the diameter of the micrometer screw is exactly 6.35 mm, some careful experimentation is necessary to ensure the extreme wavelength limits in the violet and the near IR are still accessible without the screw and ball losing contact. This procedure was done by trial and error.
The micrometer screw chosen must also have enough linear displacement to cover this wavelength region. The model I found from a local supplier has a total linear translation of 25mm which is acceptable for a 1,800 lpm grating. The total distance covered from blue to near IR is about 21mm.
Two views of the completed spectrograph in its final version. The original Olympus f/1.8 lens has now been replaced by a Pentax f/1.4 50mm lens and the camera and lens brought as close as possible to the grating without cutting off the 18mm f/5 parallel beam from the collimator lens. An earlier configuration that had the CCD camera and lens about 60mm further away from the grating suffered from severe vignetting at the camera lens, causing a loss in signal intensity at the edges of the CCD sensor.
The spectrograph measures 27cm x 15cm x 10cm. Compared to the larger and longer SGS, it is a significantly more compact unit and weighs much less, even with the CCD camera and large telescope adapter needed to hold it in place. The theoretical resolving power R = 2000 with the 1800 lpm grating.
Evaluation and Testing
The Solar Spectrum
To test this improved version of the spectrograph the solar spectrum (from reflected blue sky) was captured. Several specific regions of the visible solar spectrum are presented below. This was mainly to check the resolving power of the spectrograph and to test its sensitivity at the extreme ends of its range. None of the spectra have been calibrated in intensity (vertical axis in the line profiles). I was more interested in verifying the practical resolution of the instrument at this stage, which from the results below appears to be close to the theoretical value.
A group of solar absorption lines known as the Magnesium Triplet around 5180 Angstrom units (Å) is a common test of the practical resolving power of any spectrograph. Additionally, this region falls near to the maximum sensitivity of many CCD detectors (65% QE in my case). The sensor in this detector is a Sony ICX285AL sensor and the camera model is a Starlight Xpress SXVH-9.
An exposure of only 1 second, while pointing the spectrograph at sunlight reflected from a bright blue sky, was enough to record the spectrum given below. At these short exposures the subtraction of a dark frame is not necessary, since the dark current of the Sony sensor is exceedingly low. In fact, the camera manufacturers do not recommend dark frame subtraction (normal practice with deep sky imaging) for exposures less than 10 minutes.
Subtracting a dark frame from the raw image at short exposures can do more harm than good since the statistical noise of the dark frame may exceed the pattern noise from hot pixels and consequently add to the result of the dark subtraction.
Any “hot” pixels, which are principally caused by cosmic rays passing through the sensor chip, can simply be eliminated using image processing software. The images below were pre-processed using either IRIS or ImagePro Plus software. Since perfect alignment of the grating grooves and vertical pixel columns of the CCD chip is never or rarely achieved the first time, controlled rotation of the image is usually required prior to binning to give the spectral profile.
The two images below show the Balmer H-alpha region first as the raw image and then as the spectral profile. Atmospheric O2 absorption lines around 670nm are clearly evident. The raw spectrum image exhibits some slight curvature (a very common effect, especially for home-made spectrographs) which has not been corrected for. This has probably degraded the resolution slightly during the binning process to obtain the spectral line profile. Any corrections for line curvature would slightly increase the resolving power.
This next spectrum is a closeup of one of the several atmospheric H2O absorption bands in the solar spectrum, this one around 670 nm. The x-axis here is in pixel number and not wavelength:
The spectrum below shows the Calcium H and K region in the violet. The labels H and K have no direct scientific relevance to atomic structure; they refer to the original Fraunhofer lines, a series of dark prominent absorption features in the Sun’s spectrum.
They were labelled arbitrarily in the early 19th century when astronomers first investigated these features and tried to understand what they meant.
The apparent absence of absorption lines below 3700 Å is actually due to a combination of vignetting and the drop in sensitivity of the CCD sensor used. All low cost consumer CCD sensors exhibit a fall off in sensitivity (measured as quantum efficiency or QE) at wavelengths below 400 nm (4000Å).
Nevertheless, the spectrograph has no problem detecting the longer wavelength lines in the hydrogen Balmer series up to about Hθ (3797.9Å) and can readily detect diatomic species in molecular band heads such as the cyanogen (CN) free radical at 3883Å and diatomic carbon at 4393Å that are found in comets and cool M- and C-type stars.
Conclusions
Overall, this MFSS has fulfilled its design objectives. Future posts may well record spectra with this instrument from a variety of light sources
