wpo - low-res prism spectrograph - used in afocal mode
astro-optics page text & images [c] Maurice
Gavin 1997/2002 -
BAAJournal -1998
A
home-made low resolution slitless stellar spectrograph for coupling to
a Meade LX200 Schmidt Cassegrain telescope is described and some sample
spectrograms are shown. There is no universal spectroscope
for astronomical application - each must be tailored for a particular
need. This instrument was designed to record faint stars in
short exposures with sufficient resolution to determine their spectral
class. The spectroscope is a slitless design and when located at
the focal plane of a telescope, it performs a similar function to a full
aperture objective prism placed before a telescope producing spectra
of all the objects in the field. There is no option for visual
use - the CCD detector far outperforms the eye or film in sensitivity and
results are displayed immediately the exposure is completed.
A proto-type spectroscope was used to record the spectrum of SN1998bu in M96 and refined here. The complete instrument (excluding camera) weighs 500g and built from accumulated material.
Optics:
The spectrograph comprises three optical components
of similar aperture to maximize through-put - a Barlow
lens to collimate (make parallel) the light from the telescope,
a prism to dispersion the light into a spectrum
and a camera lens to focus the spectrum onto
a CCD detector. Their separation is not critical. The small direct-vision
(dv) prism gives a straight-through view i.e. spectral dispersion
with zero deviation for the mid spectral region. [A
single 30o or 45o prism could serve equally well
but requires offsetting the lens/CCD combo to catch the spectrum deviating
sideways from a prism].
Pentax screw threads couple the brass tube housing the dv-prism to a standard 1-1/4inch push-fit eyepiece adapter and the Starlight Xpress MX5 CCD camera respectively. Card is used to space the prism from the tube walls in a good fit. A Barlow lens conveniently fitted the eyepiece adapter barrel [or may project into the prism hosing]. The diminutive camera lens of 19mm clear aperture is a 32mm fl.. f/1.7 Yashica lens cannibalized from a half-frame camera i.e. stripped of shutter/ iris and epoxied in its mount to a Pentax threaded ring. [A regular 35mm lens of compact design could equally serve]. The lens is prefocused at infinity* when the CCD camera is attached. When the spectrograph is fully assembled and attached to the telescope, the telescope focusing knob is adjusted until the spectra (displayed on PC monitor) appear as sharp streaks. This occurs when the Barlow lens is inside the telescope focus by an amount equal to the Barlow’s (negative) focal length [see figure adjacent]. The prism is rotated until spectra align parallel to the long edge of the CCD (again viewed on monitor). By accurately recording the spectrum onto one or two rows of pixels across the width of the CCD the minimum exposure may be used. Spectra can be post stretched into rectangles [top image] without wasting photons at the telescope.
Other optical effects:
The combo of Barlow lens (collimator) and camera lens, if of dissimilar
focal length (-/+ values ignored in this application), will alter the final
focal length, f/ratio and imaging scale of the telescope. The
Barlow [107mm fl] plus camera lens [32mm fl] acts as an 0.33 focal reducer,
[32/105] converting my 30cm aperture f/10 SCT to f/3.3 and 1000mm fl.
The shorter efl covers a larger field aids locating the target and produces
pin-sharp images by reducing the effects of seeing and minor telescope
guidance errors with the downside that faster f/ratio increases vignetting
around the image and sky-fog.
Note - the length of the spectrum is a product
of the prism’s dispersion and the focal length of the camera lens imaging
the spectrum onto the CCD. It is not effected by the telescope
fl as modified by the spectrograph.
Spectral
dispersion and resolution: A prism gives a non-linear
dispersion i.e. progressively compressed towards the red end of the spectrum
- a minor inconvenience. A complete spectrum from 390nm (violet)
to 900nm (near infra-red) can be contained within the width of the Starlight
Xpress CCD which peaks in sensitivity about 550nm in green light.
The spectrograph's spectral resolution ~1.6nm/pixel [@ 550nm] and dispersion
of 130nm/mm. Classified as very low spectral dispersion it is however
adequate for general stellar classification by the amateur. Resolution
and dispersion improves marginally at 400nm (violet) where CCD sensitivity
drops to near zero. The
spectra were rarely found to overlap because of the low dispersion of the
prism and small field covered by the CCD detector. Exposures are
limited by sky-fog and stellar penetration by dilution of points of starlight
stretched at least 100 fold into spectra.
A low resolution spectrograph has pros and cons. This instrument is maximized to record spectra of mag 12 stars in 5 minutes exposure or less. By comparison 1st mag stars are fully exposed in a remarkable brief 0.1s. On the downside, the spectra are not very detailed. Only A-type stars like Altair show the prominent Balmer series of hydrogen lines with clarity (useful for calibration of other spectra) whilst in cooler K and M-type stars record a fluted spectrum. In other stars the absorption lines are poorly recorded except the Telluric absorption lines (from Earth’s atmosphere - again useful for calibration ) in red which record on most spectra especially with a large zenith distance. For M-type stars there is a loss of data in the blue spectrum and IR begins to dominate. CCDs with their extended IR sensitivity will sharply record IR if the focus is adjusted. This is done by tweaking the telescope focus in preference to disturbing the camera lens prefocused for infinity in visible light*.
Bright emission lines are virtually uninhibited by the low spectral resolution and a few faint Be-type emission line stars were ‘discovered’ whilst trawling the Milky Way. P Cygni (B1 IA) and Beta Lyrae are bright examples of these stars and their emission lines are easily recorded. By comparison Wolf-Rayet (WR) stars and planetary nebula (PN), both dominated by very bright emission lines, prove spectacular and amongst the easiest to record. They also appear to break the rules for the faintest spectra recorded via the spectroscope. Spectra of WR star #144 at mv 15.5 proved relatively easy because most of its light is concentrated into a single blue emission line about 460nm (He II+N V).
Further experiments: To test if an even lower dispersion (for a brighter /shorter spectrum) were viable, experiments were conducted in a similar set-up as above using single 45o [disp= 200nm/mm] and 30o [disp=480nm/mm] prisms respectively or extremely low spectral dispersion. However it can be demonstrated that even this dispersion can resolve, under bench test, the two mercury light emission lines at 546nm (green) and 579nm (yellow) with a separation of about 30nm - the same gross width of the Si II band at 615nm as recorded in type 1a - SN 1998bu.
Image
capture and processing: My goto Meade LX200 simplifies
target acquisition. Exposures are made at sidereal drive rate and without
manual or CCD guidance. Although there are common types each spectrum
will have a different appearance both in brightness/ lines structure.
Exposures are increased to give a good density image for acceptable results
are titled, autodated and saves to disk.
Calibration via darkframes and flatfields is applied and image processing keep to a minimum. A linear stretch (preserving the relative contrast) or an unsharp mask filter (to enhance line structure) are applied. The most revealing is a line profile through a spectrum from within SX Pixwin [or AstroArt] software. This permits any row of pixels across the CCD, and containing a spectrum, to be converted into a graph. The most subtle changes of brightness are apparent. Like an objective prism, this afocal spectrograph has no option for a comparison source to be added. An approximation of wavelength can be made by overlaying spectra from say an A-type star like Altair or PN emission lines which are well documented.
Conclusions: Building a spectrograph
is not difficult. Results are both rewarding and enlightening as
the sample Wolf-Rayet and regular
stars' spectra indicate.
Worcester Park Observatory - UK
* the camera [lens + CCD but less prism] is pointed at the night sky and adjusted until stars are sharply recorded onto the CCD.