Cameras for Spectrographs
Which camera to use with the spectrograph??
Which is the best camera to use with my spectrograph? Can I use my (insert your favourite camera) with my spectrograph?
These are typical questions, frequently asked. The basic answer is - you can use whatever camera you have available.....BUT to obtain the BEST results (in terms of resolution and recording acceptable SNR and faint objects) you really need to invest in a cooled, mono CCD/ CMOS.
Cambridge has a good page on cameras and sensors here. The explanation of the Bayer matrix and the De-Bayer process as well the current use of microlens on sensors is well presented.
Buil has done much work over the years analysing and comparing various cameras. His Camera Page should be consulted for the latest information.
Cooled -why?
Signal to Noise Ratio (SNR). The complex measurement of SNR (See "Handbook of CCD Astronomy, 2nd Ed, by Steve Howell , section 4.4 for all the details) is based on the "CCD Equation" this includes factors for dark noise, background noise, read noise, sensor A-D noise etc. The basic SNR approximated by the Poisson (shot noise) distribution, based on a 1 sigma error, and gives = N / (N)^-2 = (N)^-2 where N is the signal strength of the source. An example: If the total signal (intensity) of the source (measured in ADU for simplicity) is 40,000 ADU, then the SNR will be (40,000)^-2 = 200.
Using cooled cameras reduces the dark current noise and hence improves SNR.
Most serious ProAm type work will be looking for a SNR>200. This in turn determines the exposure needed to achieve this result. When you stack your images, say as an average stack, we will not see these high ADU values.... i.e if you have a source signal = 10,000 ADU and you stack 4 images "Sum" the result would be 40,000 ADU, but as an average stack this would still only give 10,000 ADU - BUT, with a much smoother profile. The SNR calculations are therefore based on the 1 sigma error limit. The SNR in our 1D spectrum are usually measured in a reasonably flat area of the profile, close to the target wavelength. All the spectral processing software has the capability to performing this SNR calculation.
Mono -why?
Response curves. The response of the sensor chip to the incoming photons varies with wavelength, normally peaking around 5500A. The shape of this curve (response v's wavelength) can peak at up to 70% absolute. I.e 70% of the incoming photons are recorded by the chip. The sensitivity at Ha wavelength can vary, around 40%. With a mono sensor this is a smooth curve which can easily be replicated when carrying out an Instrument Response curve for your set-up.
Colour sensors, have the inherent issue of the Bayer matrix. This produces three QE curves - one for each colour RGB which are then superimposed. Messy!
Pixel size - this variable becomes important when considering slit spectroscopy. Ideally the slit gap should cover at least 2-3 pixel (Nyquist sampling ), there are arguments for an even higher number of pixels. (See: "Spectroscopic Instrumention", Eversberg & Vollmann, Section 2.6.6 - Shannon's Theorem or the Nyqist Criterion). For a typical slit gap of 20 micron this would suggest a pixel size of 6 - 8 micron would be acceptable. CCD cameras allow pixel binning which presents the possiblity of using smaller pixel sized camera and x2 binning.
CCD -why?
CCD cameras have been used for spectroscopy for many years. They are well studied and the QE curves well established. The dark current and noise of most CCD cameras have also been investigated. For slit spectroscopes like the Spectra-L200, LhiresIII or Dados the older ATiK 314L+ works well. The later ATiK 428/ 460 is also widely used.
Colour camera - DSLR/ OSC
Beginners will obviously use the camera they have available. This means in most cases a DSLR. Used with low resultion grating set-ups they can indeed present workable results. The larger DSLR chip size can record longer spectral images and wider field of view (when used with gratings). The limitations:
Response curves - see above.
Bayer matrix - see above.
Modded DSLR cameras - With some DSLR cameras (notably the Canon series) the internal colour correction filters can be removed (sometimes replaced by clear or UV-IR cut filters) to provide an enhanced response to the red Ha region. Improvements of up to 40% have been achieved. In a FULL mod, both internal filters removed, we can record spectral features further into the NIR, and down in the UV.
CMOS cameras
These cameras are becoming more and more main stream in astronomy. There are many suppliers now selling cooled mono CMOS cameras which many be of interest to the amateur.
The recent advice that Sony are no longer going to produce CCD chips has been followed by a press release saying ON Semiconductors will cease manufacture of CCD chips.
The writing seems to be on the wall....we will need to find acceptable CMOS alternatives.
Buil has already carried out some comparisons with cooled CMOS cameras. The ASI 1600MM cooled seems to be gaining acceptance.
Guide cameras
Many amateurs use the Starlight Lodestar cameras for guiding. It is a well supported camera with high sensitvity and few faults. The CMOS cameras, especially the ZWO range of ASI Guide cameras are widely used in AP applications and can equally perform with the spectrograph.