Time series photometry problem

[andrew s]

32 minutes ago, james_screech said:

The images don't need rotating as CMuniWin matches stars without it. So the only thing must be something is moving. As I never had this problem with my refractor I think I will just use the newtonian for visual work and go back to the refractor for photometry.

That's the point if it matches without rotating then the flat could be being applied in a different orientation with the before and after flip images.

Regards  Andrew 

[james_screech]

Just now, andrew s said:

That's the point if it matches without rotating then the flat could be being applied in a different orientation with the before and after flip images.

Regards  Andrew 

The flat is applied to the image files without any rotation as the image train has not changed, just the orientation of the train with respect to the stars. The star matching is done in software which can compensate for the flip, the flats are applied before this so should work, unless the image train changes (movement somewhere).

[andrew s]

26 minutes ago, james_screech said:

The flat is applied to the image files without any rotation as the image train has not changed, just the orientation of the train with respect to the stars. The star matching is done in software which can compensate for the flip, the flats are applied before this so should work, unless the image train changes (movement somewhere).

Exactly, so if there were a gradient on the flat then as the stars are in different positions relative to the gradient they will have different "flat" values subtracted before and after the flip and before any photometey processing.

It could in principle explain what you are finding.

Regards Andrew 

[andrew s]

This is your master flat. It seems to me that stars placed differently on it will have the apparent magnitude changed as it should. So the question is does it accurately  reflect the response of your system both side of the flip. Movement could be shifting the collimation and thus invalidating the flat.  Can you take flats on either side of the meridian and compare them? Regards Andrew

[james_screech]

13 minutes ago, andrew s said:

Exactly, so if there were a gradient on the flat then as the stars are in different positions relative to the gradient they will have different "flat" values subtracted before and after the flip and before any photometey processing.

It could in principle explain what you are finding.

Regards Andrew 

If the image train was stable that would be fine as it would be correct, however if there has been movement the flat will no longer "match" the image and so produce the problem. In fact the flats both side of the meridian will probably be incorrect as movement will have occurred between the flats (tube pointing vertically up) and the first side of the meridian images. 

[james_screech]

Just now, andrew s said:

This is your master flat. It seems to me that stars placed differently on it will have the apparent magnitude changed as it should. So the question is does it accurately  reflect the response of your system both side of the flip. Movement could be shifting the collimation and thus invalidating the flat.  Can you take flats on either side of the meridian and compare them? Regards Andrew

To take flats I need the tube to be vertical, so unfortunately no.

[coatesg]

The flat does have a gradient (well, asymmetry at least, especially towards the bottom right) - how are you making the flat? I mean, this could be fine if this is indeed how the optical train behaves, but, if it's not truly representative, then it may well be the issue (despite my earlier thoughts!). 

Can you take one set of flats on one side of the pier, and then another the other side?

EDIT: Beaten to it 😉

 

Edited September 9, 2019 by coatesg

[james_screech]

I've tried flats with both an LED panel and an EL panel, both show a similar pattern, so it's likely to be real.  

[Darryl]

You could try taking evening (or dawn) sky flats first on one side of the pier, then after a flip and then back on the other side of the pier.

Differencing the resulting flats would give you an indication of their repeatability.

I've had similar problems with flat calibration in the past on the Issac Newton Telescope (in a previous life when I had access to big scopes), we spent a lot of time trying to track down flat field errors and in the end put it down to scattered light upsetting the flats. Flat fields can be a right pain, best to avoid the meridian flip if at all possible.

The only other though that I had is whether the star PSF varies significantly as a function of field position. If the photometry apperture is relatively small then there could be a significant difference in the amount of light outside of the apperture depending on where the star is in the field. An 'apperture correction' is often used to overcome this problem, particularly with systems with strong coma or field curvature.

Cheers,

Darryl