Digital single lens reflex cameras are now commonplace with a number of manufacturers offering a variety models and many options for lenses. They range from relatively inexpensive consumer level bodies, often packaged with average quality zoom lens, up to very expensive professional bodies and lenses. My experience is definitely at the cheaper end of this range.
I will offer suggestions based on my limited experience using such a camera for photometry. These are just suggestions and I encourage everyone to experiment and find their own preferred mode of operation – then let the rest of us know what works for you.
CMOS or CCD sensors in DSLR cameras are very similar to the sensors in astronomical CCD cameras. Typically they have 12 bit or, more commonly now, 14 bit analogue to digital converters. More bits equates to potentially greater dynamic range.
DSLR cameras can save images in a variety of formats, however to preserve the integrity of recorded data always use the native RAW format. Other formats compress information to reduce files size and/or alter colour balance and gamma.
The ability to change lenses is one advantage of these cameras, allowing different configurations depending on the needs of the individual observing program. Fixed focal length lenses are preferred as they tend to have fewer air to glass surfaces and lower distortions compared with zoom lenses. High dispersion glass elements, fast optics and anti-reflection coatings are desirable but can add considerably to the cost.
Excellent quality lenses dating back to the age of film are readily available second hand for very reasonable prices. With the addition of an appropriate adapter they can be attached to a DSLR body. Be aware, however, that not all lenses can be used on all DSLR bodies. Some lenses can cause damage because they protrude too far into the camera body. So research thoroughly before spending any money.
I use a 1990’s vintage Nikkor 180mm f2.8 lens (stopped down to f4 for better images) on my Canon 450d DSLR. Focus and aperture setting must be manually adjusted so the camera must have a manual setting available.
The simplest approach is to use a sturdy camera tripod and wired remote control (to avoid shaking the camera when starting an exposure). Exposures will be limited to less than 10 seconds by star trailing since the camera is not tracking. Longer focal length lenses equate to shorter exposures. This will limit you to bright targets and preclude time series acquisitions.
A better option is to piggyback the camera on an equatorially mounted telescope which can track the target as the earth turns. There is no need for guiding (auto or manual) as long as polar alignment is reasonable.
Bayer matrix and focus
An advantage of DSLR photometry over PEP or CCD photometry is that R, G and B images are recorded simultaneously thereby reducing acquisition time. However there are a few issues with one shot colour imaging that have significant implications for photometry.
In the Bayer filter array used in DSLR cameras half the pixels record green light and one quarter of the pixels record red or blue light (see Figure 1).
Software that interprets and displays these images has to estimate how much red, green and blue light would have fallen on each pixel in the image. It does this by looking at, for instance, the surrounding green pixels and interpolating how much green light should have fall on the red and blue pixels.
Figure 1a. Schematic of a focused star image superimposed on a Bayer filter matrix. Only a small number of individual pixels of each colour are illuminated leading to poor photometry in each colour. Also, each pixel reaches saturation relatively quickly despite total photon flux (and therefore starADU) being relatively low.
Figure 1b. Schematic of a defocused star image superimposed on a Bayer filter matrix. More individual pixels of each colour are illuminated allowing better photometry. Furthermore, longer exposures are possible before saturation occurs so total flux (starADU) is increased.
As shown schematically in Figure 1a, a tightly focused star image extends over only a few pixels of each colour. This leads to imprecise interpolation. Furthermore, pixels reach full well capacity in relatively short exposures.
Defocusing extends the star image over many more pixels which leads to better colour interpolation. It also allows longer exposures before pixel saturation, hence better counting statistics.
Achieving correct focus is greatly simplified when the camera has a “live view” mode where a live image on the viewing screen is updated several times per second. Canon introduced this feature with the 450d and subsequent models.
Linearity and ISO
Use the lowest ISO compatible with reasonable exposure times. Higher ISO only scales the gain so ADU per incident photon is increased. If a 1 second exposure at ISO 100 gives 5,000 ADU then the same exposure will give 10,000 ADU at ISO 200. There is no more information because the noise is also scaled by the same factor.
Camera response will deviate from linear as the well capacity of pixels is approached. The Canon 450d showed linear response up to at least 14,600 ADU at ISO 100. Higher ISO settings cause ADU values to exceed the analogue to digital converter maximum (14 bit = 16,384 ADU) before well capacity is reached. Hence higher ISO settings should have linear response throughout the ADU range, but at the cost of reduced dynamic range.
Choice of f-ratio
Fast camera lenses allow shorter exposure times; however, it may be necessary to stop down the lens aperture one or two stops. Coma, spherical and chromatic aberrations can cause unacceptable distortions at the edges of images, especially since we intentionally defocus to spread star images over more pixels.
FWHM of R, G and B channels
Chromatic aberration causes FWHM to be different for each colour, especially when the lens is out of focus. With my Nikkor 180mm lens the Red and Blue channel star images have larger FWHM than the G channel, so appropriate aperture radii for aperture photometry analysis must be determined for each channel.
Untracked cameras will be limited to less than 10 second exposures. Tracking allows much longer acquisitions and therefore fainter targets can be measured. Maximum exposure time on my Canon 450d camera is 30 seconds unless a wired/wireless remote control is used and the camera set to BULB mode.
Choice of exposure time will be determined by the brightness of the target and whether the camera is tracking or not. We must avoid overexposure of any pixels in the target and comparison star images.
This feature is available through the special functions menu on Canon DSLR cameras and is very useful when imaging through a telescope or long focal length lens. Framing and focusing is carried out first then the mirror is locked in the up position before the shutter is opened to start an exposure. This reduces camera shake which is more noticeable at longer focal lengths.
I have not found it necessary to use mirror lockup with my 180mm lens.
Long exposure and high ISO noise reduction
These features are also available through the special functions menu on Canon DSLR cameras. I recommend you do not use these functions. High ISO should be avoided because dynamic range is reduced.
When long exposure noise reduction is used the camera records an image normally, then records a second image of the same duration with the shutter closed (a dark frame). The second image is subtracted from the first to remove dark current noise. Therefore a 10 second exposure takes 20 seconds. This isn’t much of a time cost if only a few images are to be recorded but can be significant when high cadence time series datasets are required.
It is far better to record a set of dark frames which can be averaged to make a master dark frame with reduced random noise. The master dark frame can then be used in the image calibration process later, along with flat field correction.
DSLR battery capacity is usually sufficient for several hundred images. If possible turn off the display screen to conserve power. Recording images directly to a computer hard drive instead of the camera memory card may also help extend battery life. I have recorded over 500 images in one night on a single fully charged battery.
Cold conditions will lead shorter battery life. A second battery would be a good investment. Alternatively you could purchase a power supply and adapter to power your camera from the mains.
Computer control of camera
Software supplied with Canon and, presumably, other DSLR cameras can control many aspects of image acquisition via a USB cable connected to a laptop computer. BULB exposure mode is available without the need for a remote control.
A number of third party software packages are available (at a price) specifically for astronomical imaging. These provide even more options for image automation and analysis.
Computer control of the camera greatly facilitates focusing, framing and exposure checking. Images can be saved directly to the computer hard drive instead of the camera memory card. Images can be examined with your photometry software to monitor focus, tracking, etc which can’t easily be done with a standalone camera.
Accurate time stamping of images is easily achieved if the computer clock is synchronised with a time server. See Tom Richards’ article “How Good Is Your Timing?” in the August 2010 VSS Newsletter.
Multiple images can be acquired automatically at predetermined intervals via software control thereby allowing time series datasets spanning many hours to be easily acquired.
No matter which camera you use, dew on the front of the lens can be a real problem in cooler months. I use a commercial dew heater system on my lens, but instructions for DIY versions are available on the web.