Reducing scatter in DSLR images

Affiliation
American Association of Variable Star Observers (AAVSO)
Fri, 07/29/2022 - 15:29

Hello all,

I have been trying to get good data on the low-amplitude Delta Scuti variable Epsilon Cephei for several weeks now with a Canon EOS T6i DSLR camera. So far I have gotten convincing evidence of periodicity from only 1 run out of 7. Admittedly, in a couple of these runs the target star counts indicated they were outside the camera's linearity range, or even saturated in one or two pixels. But in the 5 or so good runs, only once was the variability clearly present, with a period consistent with the accepted one. The scatter in the intensity data is usually comparable to the amplitude of the variability, which of course makes detecting variability impossible.

Are there any techniques to reduce this scatter? Or is it inherent in DSLR cameras (or perhaps my model)?

Thanks,

Liz

Precision expepctations for DSLR photometry

Liz,

It may be that you are near the limit of your precision for the task with your equipment. From your previous posts, I recollect that you are using a tele zoom lens with your camera. Usually, such lenses are not 'fast' (i.e., they do not have wide apertures). In a previous post Mark Blackford, I think, recommended a prime lens for this reason, but I understand that you have some limitations on your ability to acquire equipment.

Eps Cep according to VSX has an amplitude of 4.15 - 4.21 in V. That's 0.06 Mag, which in my opinion is asking a lot of a setup like yours.

It would be important to determine what precision you can actually achieve by taking multiple images of comparison stars of a similar magnitude to Eps Cep and plotting the scatter. Standard deviations will of course put a measure on the scatter.

In general, the following may help. Use a degree of defocussing that will allow you to lengthen your exposure without reaching saturation. The more photons you capture for your target stars, the lower your S/N ratio, other things being equal. My personal view is that if your comp stars are close in magnitude to the target it is better. If they are significantly brighter, your exposure is limited to avoiding saturation of comp stars, rather than maximising precision of the measurement of the target.

On an optimistic note, Eps Cep at mag 4 is bright, and in general precision of DSLR photometry can be very good with bright stars.

Final thought. At mag 4, you do not need a long focal length lens (in fact, a shorter focal length for that mag enables easier inclusion of comp stars in the field). Could you get access to a prime lens with a shorter focal length and wider aperture?

Roy

Affiliation
American Association of Variable Star Observers (AAVSO)
Re: Precision expepctations for DSLR photometry

Roy,

Did you perhaps mean that the more photons gathered, the higher (i.e. better) the S/N ratio? Is the scatter due largely to scintillation?

With Eps Cep there is a limit on how much I can defocus because there are faint stars very close to it in the FOV. I have been forced to compromise, not defocus too much, and reduce exposure time to 20 s. It is possible that I could defocus a little more and use a 25 s exposure, but I know that 30 s is not doable - limiting the defocus to leave enough space around the target star for sampling dark sky background results in saturation.

As far as doing photometry on my comp stars, I have done that already and the precision so far is clearly not good enough with my current exposure parameters. That's why I posted this question here, hoping there might be some other option. Unfortunately I do not have any fixed focal length lenses and no way to easily acquire any (I would need to apply for a faculty development grant). I do have a shorter focal length zoom lens, 18mm to 135mm, but that lens's aperture is even smaller. If I needed a wider FOV to get more options for comp stars, the other end of my 250mm lens's range is 55mm and would probably be a better choice. So far, for Eps Cep, most of my comp stars have actually been fainter - the only (essentially) fixed magnitude star of similar brightness in the FOV is Zeta Cep. Typically I use Zeta along with a handful of fainter stars as comp stars.

What I have not done, but should, is to forget about Eps Cep and just do a longer exposure (maybe a minute) on one of the more isolated comp stars with sufficient defocusing to avoid saturation - since they're fainter than Eps Cep, that should be easily doable. It sounds like that's the only avenue I have open to me to improve the S/N ratio anyway, so I will see what I find and take it from there.

Thanks,

Liz

Affiliation
American Association of Variable Star Observers (AAVSO)
No worries, I thought so…

No worries, I thought so anyway. I'm still concerned that some of the scatter might be due to factors other than scintillation, such as differential response of different sensor pixels (or even the Bayer pattern itself) as the star image drifts slightly from exposure to exposure. This seems to be unavoidable with the SkyTracker, as its tracking is good but not quite spot on, which could possibly be due to imperfect polar alignment. The reticle has significant parallax with even tiny motions of the observer's head. If that's what is limiting my precision, then longer exposure times won't help, though greater defocus should help to a degree. Anyway, hope to try some longer exposures tonight to see what my precision limit really is.

Liz

Star tracker polar alignment

I use a Star Adventurer and polar align by star drift. Blinking pairs of exposures with the central part of the image displayed at full resolution easily reveals drift. No detectable drift between images 3 minutes apart in my setup (200mm focal length lens on a camera with a sensor close to APS-C size) allows tracking throughout the night with minimal drift. I don't autoguide, so periodic error is evident in images.

Polar aligning by star drift takes time, but as I observe only one star per night (short period EBs), it doesn't bother me.

Roy

Bayer pattern affecting photometry

I meant to include the following:

Mark Blackford has demonstrated that a DSLR image not defocussed enough will display detectable shifts in ADU counts from a colour channel as the image moves across pixel groups due to periodic error (I hope I'm describing his work accurately here).

Roy

Affiliation
American Association of Variable Star Observers (AAVSO)
I've read the same thing,…

I've read the same thing, not sure where (it might have even been in the AAVSO DSLR Observers' Manual), and it is what gave me the idea to suggest that possibility.

It makes sense that the greater the defocus, the more the star image is spread out over a range of pixels, and the less the effect of the shift would be on the *total* ADU count from the pixels of that colour channel that the image occupies.