Optical shadow- can you help determing the cause

[MartinB]

The standard advice for flat fields is  30-50% the value of fullwell/adu gain.  This ensures that your flats stay in the linear range of the chip.  Despite having a NABG chip with a very linear response my flats are normally between 13000 and 16000 ADUs and I take 100 for each filter channel.  I'm not saying this is a panacea but I do think it is sensible to keep within the linear range of the chip.

 The only time I have really had a problem has been due to image tilt with exactly the problem you have shown here.  I replicated the tilt for the light and for the flats and the problem recurred.  It disappeared when I collected a few flats with an EL panel with he image train at the angle it was when collecting the lights, problem disappeared.  Have you tried this?

[MartinB]

Ah, just realised, you've done some cleaning so I guess you're not going to be able to replicate the situation.

[ollypenrice]

7 hours ago, pyrasanth said:

Does this work for the Atik 11000?- I know that if I take increasingly longer darks the hot pixels become quite apparent & I would image the same would hold true for the light frames- let me know your mileage with this.

It works fine for the 11000. I never take darks for flats, just use a master bias as a dark for flats. The camera with which I struggle is an SXVH36 capturing in Nebulosity on a Mac. Parallel with this is an identical scope and Atik 11000 on which flats work perfectly using the same panel etc.

William Optics paint the inside of dewheilds white?  Are they mad? I just checked my ZS66 and it's blackened inside. You'll probably know this already, but to blacken anything properly you need either flocking or hight temp (barbecue/stove) paint which is pigment rather than dye based. Dye based paints are reflective in the IR.

Olly

[andrew s]

1 hour ago, ollypenrice said:

The camera with which I struggle is an SXVH36 capturing in Nebulosity on a Mac. Parallel with this is an identical scope and Atik 11000 on which flats work perfectly using the same panel etc.

Olly

I am sure you have tried many things but have you plotted a Photon Transfer Curve for the SXVH36 ? It could show up problems with the camera that would not otherwise be obvious.

This describes the method http://www.narrowbandimaging.com/images/ptc_talk_wsp_2009_crisp_final_comments_web.pdf.

I am not sure what telescope it is on but I would check for light leaks before doing the curve. I know two ways to do this. One, put a pin hole close to the camera window and take a flat. The pin hole acts as a pin whole camera and takes an image of the inside of the scope. I discovered it this method by accident when a shutter failed just open. The other way is to have a blank in the filter wheel and compare a light and dark of the same exposure with the "lens" covered in aluminium foil - they should be identical.

Doing a PTC is a bit of a black art but it does reward the effort in my opinion.

Regards Andrew

Note in the presentation Richard measures the offset (bias) using the overscan region but I don't think SX cameras support this so start with measuring a bias frame.

[andrew s]

9 hours ago, MartinB said:

The standard advice for flat fields is  30-50% the value of fullwell/adu gain.  This ensures that your flats stay in the linear range of the chip.  Despite having a NABG chip with a very linear response my flats are normally between 13000 and 16000 ADUs and I take 100 for each filter channel.  I'm not saying this is a panacea but I do think it is sensible to keep within the linear range of the chip.

 The only time I have really had a problem has been due to image tilt with exactly the problem you have shown here.  I replicated the tilt for the light and for the flats and the problem recurred.  It disappeared when I collected a few flats with an EL panel with he image train at the angle it was when collecting the lights, problem disappeared.  Have you tried this?

Here are some experimental data on flats of different lengths http://www.narrowbandimaging.com/images/ML8300_Flat_Fielding_Experimental_Data.pdf you might find the conclusions of interest.

Regards Andrew

[pyrasanth]

9 hours ago, andrew s said:

Here are some experimental data on flats of different lengths http://www.narrowbandimaging.com/images/ML8300_Flat_Fielding_Experimental_Data.pdf you might find the conclusions of interest.

Regards Andrew

I have never really tried to emulate the angle with the flat panel at which the lights were taken-guess I should really. I normally point the telescope up 180 & take the flat with the panel on top of the dew shield. I guess this is just lazy. I will try an angle closer to the target flat in future & find a way to anchor the panel to the top of the dew shield- it will slide off otherwise- gravity is a bitch!

I've ordered some flock paper from FLO to flock the inside of the dew shield- it is black but I reckon this wont hurt. I concur with the 30-50% of the maximum ADU- this appears to be borne out with my tests on the Atik 460- the flats appear to better & more accurate with a lower ADU especially when fully processed from start to finish in PixInsight.

I have not had a look at your experimental data link yet but I will take a look. Thanks for the info.

 

[pyrasanth]

11 hours ago, ollypenrice said:

It works fine for the 11000. I never take darks for flats, just use a master bias as a dark for flats. The camera with which I struggle is an SXVH36 capturing in Nebulosity on a Mac. Parallel with this is an identical scope and Atik 11000 on which flats work perfectly using the same panel etc.

William Optics paint the inside of dewheilds white?  Are they mad? I just checked my ZS66 and it's blackened inside. You'll probably know this already, but to blacken anything properly you need either flocking or hight temp (barbecue/stove) paint which is pigment rather than dye based. Dye based paints are reflective in the IR.

Olly

 

Digression warning.........

I'm temporarily forced to use darks on the Atik 11000 as it is currently being pressed into service as a guide camera so the use of a dark is somewhat essential as I end up trying to guide on a hot pixel otherwise. When I get the ONAG XM (the big brother of the SM) the cameras will swap roles. The Atik 11000 will be the imaging camera & the 460 the guide camera. I was tempted to buy another CCD camera like the Atik infinity for guiding  but the 460 is so sensitive it will make a good guide camera as I can't use the 460 whilst using the 11000 so all options are kept open.

The Lodestar X2 which I have struggles with the focus lock software as I think the centroid shape is a bit too distorted because the camera resolution is quite low but again I need to do more testing- I seem to get better results with a higher resolution. Focus lock software relies on the shape of the star changing to measure how far the focus has drifted in order to compensate so chunky pixels might confuse the software.

I'm learning to use the Optec fast focus SMFS & it is very good. The C11 mirror is now locked down & all focusing is done using the SMFS which has completely eliminated the mirror flop. I generated a new PEC curve for the Paramount & get an RMS of under 3 arc seconds which is very good. My guiding does not reflect this yet as I'm still trying to find the optimum guide settings. I will start a new discussion with regard to guiding as I need to get some help in this direction.

[andrew s]

As pyrasanth said above often an FL panel is often placed directly on the dew shield unfortunately this can violate one of the key assumptions of the theory of calibration using flats.

This is not uncommon as often several of the assumptions are not met and my hypothesis would be as to why, as Olly often points out, theory does not match practice.

Some of the assumptions are:

The Flat source must illuminate the entrance pupil identically to the sky background.
There must be no additional sources of illumination (no light leaks)
The optical system must not change between the taking of the Flats and the images to be flattened.
The Flat source and sky background should have the same spectral response.
The CCD system should be in its linear response range.
The Flat illumination should be in the Fixed Pattern Noise range of its response . (The four regions are - read
noise, shot noise, fixed pattern noise and saturation)

Putting the FL panel on the dew shield violates the first as it puts off-axis light into the entrance pupil beyond that of the sky background which will not be removed by other baffles or the field stop. I think this is why lower ADU flats are preferred as multiple reflections have less impact. It would be better to move the panel further from the entrance pupil. 

Clearly this list is difficult if not impossible to meet however, they can be much better approximated if they are understood and mitigation steps taken.

One point on the last one.  As well as removing vignetting and dust bunnies a flat can reduce the impact of Fixed Pattern Noise if suitably exposed.

For the real nerds on SGL you might like to read this http://www.narrowbandimaging.com/incoming/flats_part1_part2_part3_expanded.pdf this http://www.narrowbandimaging.com/images/daytime_sky_flats.pdf and this http://www.narrowbandimaging.com/incoming/Flat_Field_Anomalies.pdf . Well it is snowing here!

Regards Andrew

 

PS for super nerds I can loan copies of "Scientific Charge-Coupled Devices" by James R Janesick and also his "Photon Transfer DN to Lamda"

[sharkmelley]

On ‎01‎/‎03‎/‎2016 at 20:20, ollypenrice said:

William Optics paint the inside of dewheilds white?  Are they mad?

Olly

WHAT!?!?!