Adaptive optics vs. being slightly out of focus

Affiliation
American Association of Variable Star Observers (AAVSO)
Thu, 10/17/2019 - 18:59

At least for DSLRs with a beyer filter, the image is suppose to be slightly out of focus so the light will spread over more pixels.  Yet, I've also been told adaptive optics is a useful tool to have for photometry.  But adaptive optics are used to compensate for atmospheric scattering in stars.  This would "tighten" the focus of the image and reduce the size of the airy disk.  Now, wouldn't this be counterproductive for someone who set their telescope to be a bit out of focus to spread the light around more?

Affiliation
Variable Stars South (VSS)
Adaptive optics

I have an AO unit on my SBIG camera which is a mono camera. I'm not sure how it could be connected to a DSLR.

 The need for defocusing depends on the image scale of your system. If you are using a long focal length scope, usually the star image spreads over plenty of pixels already so there is no need to defocus.AO units that are available to amateurs are really just fast accurate autoguiders that only have to move a small mirror rather than the entire scope so often result in tighter stars. This is more useful for longer focal length setups where defocussing is less needed.

Cheers

Terry

Affiliation
None
I've found that the best way

I've found that the best way to look for insights in answer to questions is to consult the literature, or, in the case of established science, textbooks.  I'm sure adaptive optics is addressed in modern optics books, but my own reference is the excellent review article by Roggemann, Welsh and Fugate in Rev Mod Phys (Vol 69 Nr 2, Apr 97).  As the title, "Improving the resolution of telescopes" implies, the objective is to compensate for atmospheric effects which degrade resolution. 

In adaptive optics systems, an auxilliary imager together with one or more active probes, typically green lasers, are employed to measure the wavefront curvature and tilt, which are used to derive commands to actuators that change the shape of the primary mirror in real time.  TCB168 is correct that the so-called AO components of imaging sensors amount to nothing more than fast autoguiders.

AO is potentially useful in photometry because when successful it narrows the PSF, thereby increasing the SNR.  However, as no professioal grade imaging system used in conjunction with large apertures employs filter arrays such as are used in consumer cameras, any discussion of the pros and cons of using AO with a DSLR is uninteresting.

Unless the pixels of the sensor array are huge - much larger than the PSF - it is necessary to aggregate pixels to avoid measurement errors, especially if simultaneous measurement of a target, one or more comparison stars, and a check star is to be accomplished.  The reason for this is that even though one could in principle center the PSF of a single star on a pixel, doing it simultaneously for multiple stars is practically impossible.  To compensate for the random placement of the stars of interest with respect to the focal plane array, some defocusing is necessary.  When the sensor is a multi-color consumer grade camera, further defocusing is required to ensure that the number of green, or blue, or red, as the case may be, pixels sampling the PSF is approximately the same for all the stars being measured.  This is an impetus for defocusing to achieve a radius of 9 or 10 pixels. 

It is a simple exercise to write a Monte Carlo simulation that generates a three color array and random star positions together with some nominal PSF, to see how the error resulting from spillover of the signal varies with defocus.  Running such a simulation should be sufficient to convince one of the need for significant defocus. 

Affiliation
American Association of Variable Star Observers (AAVSO)
Follow-up question about

Follow-up question about focal length and defocusing.  Would I need to defocus if my scope as a focal length of 730mm and my camera was a monochrome camera with a pixel size of 9µm, which would result in a resolution of 2.5 arc-seconds per pixel?