Nice little Airy Disk..

[NickK]

The idea was to take some star twinkles yesterday evening but it was getting late so I decided to create my own using my little SRV-1 robot. Continuing my idea of using deconvolution of guide stars to map to the final image deconvolution.

The setup - EQ6 pointing the Vixen A80mf at the fence at the bottom of the garden with the Titan taking fast exposures (1000 x 0.001 exposures). The natural movements of air from the hot day and wind will add some reality to it. Placing a piece of paper on the fence and then pointing the SRV-1's ranging lasers at it should create a basic artificial star.. all because there's no stars about! It also makes it a little easier to work with but with the same issues (i.e. non-continuation, movement, aberration, airy disc rings, non-uniformity etc).

In the first b/w shot you can see a stack of 100 images using PI. You can see the laser dot and the respective airy disc.

The second colour shot you can see 150 images represented as voxels to show the wobbles and movements in 3D over time (top to bottom)

So why is is this so interesting? Well normal deconvolution works on the final image.. that's 10-30minute of wobbles from the seeing that make it very difficult to undo without loosing information. Now if you use the guider images then you have a history of the deviation of the point spread function over time. This means you can sum up the deviations and the required corrective actions and then apply that sum to the final 10-30 minute main image. (that's the simple description)

[Merlin66]

Interesting stuff there...

I'm trying to get my head around it - see if it would help in spectroscopy....what about the guide corrections - on a timing/ duration basis??

[NickK]

Interesting stuff there...

I'm trying to get my head around it - see if it would help in spectroscopy....what about the guide corrections - on a timing/ duration basis??

The PSF is a function of wavelength so the narrower the band, the more accurate it could be. Also the atmosphere impacts different wavelengths differently.

If you're guiding on the same star you're testing, it should be possible to approximate, however to get a more accurate mechanism narrowing the wavelengths on the guider inline with the area of spectra you're interested in would help. For example if you were sampling spectra for components that have lines only in the red band it may be worth putting a red broadband (i.e. a normal AP red filter) on the guider - better still narrowband filter but it's unlikely you'll be able to guide robustly with that!

[Merlin66]

Nick,

The actual spectrum is generated by the target star image in the slit gap and the collimator/ grating in the spectroscope.

I was really thinking about being able to more accurately "lock" a target star image into/onto a 20 micron entrance slit for a few hours.......

[NickK]

Nick,

The actual spectrum is generated by the target star image in the slit gap and the collimator/ grating in the spectroscope.

I was really thinking about being able to more accurately "lock" a target star image into/onto a 20 micron entrance slit for a few hours.......

I see accuracy in guiding rather than deconvolution.

The guiding would be more accurate if you could guide using the specific wavelengths you're attempting to sample.

Sub 5 second images means that the image really is impacted by seeing. The faster the framerate, the more the image will change in position. The key then is to average over time so that the slower mechanical mount can keep up a smooth movement. I think most guiding mechanisms do this already so they don't attempt to move the mount every frame but then attempt to move the mount according to the predicted location (based on the previous average) for the next mechanical movement.

One option is - could you shift the position of the slot in the X/Y plane 20 times a second? The spectra would move over the camera image for each frame but the lock would be faster.. It's a bit like Starlight's Active Optics, rather than move a block of glass to correct using refraction 20 times a second, the slot could move 20 times a second. The problem is that the x/y movement would have to be very accurate and fast.. to the point that having zero friction table and voice coils to move it would work.. now there's an idea..

The resulting moving slit could then be guided using the guider. Post processing could then assist if there's any spectra movement on the final image to clear up the frames.. just some initial thinking on the idea..

[NickK]

I took the mathematics and wrote a program to show the theoretical Airy Disk for two "scopes". To make the disk a nice size to plot I've used a very small aperture.

Light wavelength = 650nm

Aperture = 0.000001 m

Obstruction = 0.0000001 (for reflector disc, representing a 10% aperture obstruction).

Interestingly the obstruction means the like-sized aperture reflector will hit refraction limits first (i.e. the blurring on the CCD).

The last image (titled as wideband) is the distribution caused as if an image had a red filter then wavelengths from 575nm to 695nm are used on the same image.

[Merlin66]

Hmmm - very interesting.

Unfortunately I think the smearing of the spectrum due to the movement of the slit could be a major problem. The calibration accuracy and the resolution are dependent on the effective slit width.

[NickK]

Hmmm - very interesting.

Unfortunately I think the smearing of the spectrum due to the movement of the slit could be a major problem. The calibration accuracy and the resolution are dependent on the effective slit width.

If you know the position of the guider slit then you know how to post process the image to sharpen the blurred spectrum. So that's very similar to the guider deconvolution.

[Merlin66]

The total spectrum is made up from a series of "images" of the target star in the slit gap, usually 20micron wide (3 pixels with my camera - Nyquist OK)

The length of the spectrum recorded is usually around 10mm long....

I think to shuffle sub-pixels across the length of the spectrum would be a daunting task!!!