Sun, 10/27/2013 - 13:25
I am interested in knowing how to do photometry with a DSLR body at prime focus. The cameras have noise reduction features that can be switched on or off, save raw image files and have pixels which subtend about 1 arc second at the prime focus of the f/6 MAK-Newt. I find the Citizen Sky stuff somewhat irrelevant (FOV, exposure times, air mass adjustment). Could anyone point me to a reference? (and would the moderator kindly allow this to be posted so I might receive some feedback?)
Hi SFS,
I can't help you with references specific to DSLR at prime focus, but there are people doing this (not me though). However, with respect to FOV, exposure times and airmass just do the same as you would for a CCD imager. See the CCD Observing Manual (http://www.aavso.org/ccd-observing-manual).
You should not use the DSLR's built-in noise subtraction features as it doubles your total exposure time (say 30 sec for the science frame then another 30 sec for the dark frame). So when recording a time series the cadence is halved. Also, with only one dark frame there is extra noise added to the image compared with a good master dark frame. Much better to record a number of dark frames (I use at least 16) of the same duration as your science frames and make a master dark frame from these.
You'll still need to defocus the stars a little so the light is spread over several of each of the r, g and b pixels in the Bayer array. And do not use photometric filters. Cheers,
Mark
Thanks, Mark!
I found some more information at www.variablestarssouth.org/techniques/dslr-photometry-guide that explained some of the whys in your post. One of the interesting claims is that one should set the ISO (sensitivity) at a low level, because increasing it increases the noise by the same amount as the signal; i.e., it behaves as a low noise amplifier and thereby at very best does not change the SNR. (In reality, though, amplifiers add noise so there should actually be an SNR degradation when cranking up the ISO level.)
The other thing I got from the material cited above is that it's problematical to measure really red stars because although the green filter is close to the Johnson V, the red and blue are about useless. They didn't say it in so many words, but that's the impression I got. This, if true, casts a real damper on my enthusiasm, because my favorite targets are CVs. Any comments?
Clear skies,
Stephen
Hi Stephen,
Stan Walker and I wrote the DSLR photometry guide on the Variable stars South website two or three years ago so they probably should be updated. I've been meaning to do that for a while but keep finding other things to occupy my time.
If I remember correctly, not only does high ISO add noise it also reduces dynamic range. I usually use ISO 200 but there may be situations where higher values are justified, say for faint targets that would otherwise require long exposures.
The spectral response of the Bayer red filter is quite different from the standard R photometric filter, but for many stars it is possible to transform DLSR magnitudes to the standard system. It depends on how much molecular absortion or emission line intensity is present. Very red stars and emission line object should probably be avoided. It would be a good project to try to determine which spectral types are suitable targets for DSLR photometry.
I know very little about CV's but if you were only interested in seeing when they are in outburst or measuring the period then transformation isn't really necessary. But I might be wrong there, happy to be corrected by anyone who knows better. Cheers,
Mark
Hi Mark, Stephen,
It's clear it's better to use the lowest ISO when possible, this is where the highest number of electrons can be stored and by the way the shot noise SNR be the best (it's sqrt(number of electrons)) and then the dynamic range is the largest, fully agree.
But if the star is really faint and you can't push the exposure time or the aperture it's right to use higher ISO until there is no saturation, either on the target or the comparisons... That doesn't affect the SNR at all, it's even somewhat improved. In fact the electronics amplifiers have their best noise factor at the highest gain, the highest ISO (here it's no more the shot noise that is dominant but the Johnson-Nyquist noise of the amplifier)
The color issue is limited to the M6III to M10III spectral types, in particular when using my VSF technique, there are more problems with the standard transformation. The problem is due to a specific blue excess in the very red spectra of those stars. As the green is very low that generates a wrong B-V, said sometimes "degenerated". The B-V doesn't reflect the true effective temperature of the star. For those stars there is a large difference between a standard Bessell B pass-band and the DSLR B. In fact the DSLR is less affected than the Bessell B-V ! For all other stars good B-V and V-R can be extracted from the DSLR log of the B/G and G/R ratios. As the channels measurments are simultaneous into the same silicon chip those ratios are very stable. Results within a few % of the catalog values are usual. This is also a good indicator of the extrinction reddening.
Clear Skies !
Roger