Rotation after meridian flip... why?

[Uranium235]

Theres probably an obvious technical answer to this one, but would anyone be able to enlighten me as to why I get some rotation after a meridian flip? It been bothering me for a little while now, its not serious mind... but just a little annoying that I end up cropping/losing a bit of the frame when putting together mosaics.

[steppenwolf]

There are only two conditions under which there is no rotation after a flip:-

1. when the long side of the sensor runs exactly parallel to the RA axis (90 degrees)      or

2. when the long side of the sensor runs exactly parallel to the DEC axis (0 degrees)

At any other orientation, there will be an error of double the variance from the above.

Edited for 'tosh'!

[ChrisLX200]

1 minute ago, steppenwolf said:

There are only two conditions under which there is no rotation after a flip:-

1. when the long side of the sensor runs exactly parallel to the RA axis (90 degrees)      or

2. when the long side of the sensor runs exactly parallel to the DEC axis (0 degrees)

At any other orientation, there will be an error of double the variance from the above.

? don't understand that - a flip just rotates by 180deg and that's that, doesn't matter what angle the camera is at. If that was not true then I wouldn't be able to stack subs from East and West sides of the mount without rotation showing, but I can.  Maybe cone error will add some rotation though (just a thought of the top of my (shiney) head )

ChrisH

[steppenwolf]

Quote

? don't understand that - a flip just rotates by 180deg and that's that, doesn't matter what angle the camera is at. If that was not true then I wouldn't be able to stack subs from East and West sides of the mount without rotation showing, but I can.  Maybe cone error will add some rotation though (just a thought of the top of my (shiney) head )

Unfortunately, not so! The centre line of the telescope has flipped 180 degrees as has the centre line of the light falling on the sensor but the sensor is now 'upside down'. This means that if it was rotated 1 degree up at the top right hand corner, it will now be rotated 1 degree down at the bottom right hand corner so it will have rotated by 2 degrees.

Yes, of course you can stack such an image because you use a 'rotate and slide' stacking process which allows for lateral movement and rotational movement but if you do not have the perfect 0 degree or 90 degree orientation, close examination of your field edges will show a wedge shaped overlap.

Edited for 'tosh'

[Uranium235]

Ahhh I see, so fairly unavoidable then in most cases. Its not a massive loss mind, but one panel was taken pre-flip - the other post flip and the overlapping edge is the short one.

I ended up with roughly a 5 degree rotation, and managed to reduce the loss by stitching the panels together then rotating the whole image back by 2 degrees. It probably also didnt help that the area im imaging is near the zenith.

[Uranium235]

I was pretty sure in setting up the camera beforehand though.... I usually eye up the edge of the camera body to the flat base of the dovetail to square it up (easy since the 383 is a square camera). I normally know whether ive got it right while calibrating the mount becuase the stars always move parallel to the crosshairs in artemis when Im making adjustments.

[steppenwolf]

To be honest, I am working from what I was told many years ago so before you take it as 'gospel', I'd like to run a test for you - unfortunately, I am clouded out here (minutes into an imaging session grrrr). The theory seemed sound at the time but I've never tested it as I have always found that either 90 or 0 degrees gives me a good framing for just about any object.

I'll report back when I have put it to the test as you have me thinking about it now!

[steppenwolf]

OK, scrub the theory above!! I have tested it in real life with a non-accurate orientation of 103 degrees and the error is not as predicted! Soooo back to the drawing board - but I am now on a mission to find the real cause because I have noticed a small rotation in my own images and previously put this down to not achieving exactly 90 or 0 degrees, now dis-proven.

I am not sure about cone error as suggested by Chris but it is as good a start place as any and I will try to simulate it to see if it has this kind of effect. It can't be an offset telescope centre line or a side by side arrangement wouldn't work so the search begins......

[martin_h]

I get the same effect...... and no I don't know why either. 

[Uranium235]

I couldnt be sky curvature near the zenith could it? I mean, it was a fairly wide FOV as I was shooting with the Star71 + 383.

[alexbb]

How do you realign after meridian flip? Plate solving or just leave it as it is? Because if you try to recenter on the same point and you have a cone error, you have to rotate the axes to point exactly in the same place as before.

Btw, how does the mount compensate for cone error if you do a 3 star alignment? I never needed to do a meridian flip for my balcony imaging and neither when going outside wasn't mandatory as my mount traks past meridian. Maybe you don't even need to recenter after flip, the mount already compensates and that implies some rotation.

[Uranium235]

Thats what I thought.... that proper mount calibration eliminates cone error.

I realign the image by storing the current position in the handset and marking key stars in artemis (no plate solving here!). Flip the mount by doing another goto on the stored position, then line (or attempt to) up the marked stars with a rotated output in artemis (it has a rotate 180 function).... except I found that the stars didnt line up quite right becuase the image was slightly rotated.

In the end I settled for an "average" closeness to the markers so I wasnt losing too much. But the flip needed to be done as I was imaging at the zenith and the camera was quite close to colliding with the mount. If I had being imaging towards the south (ie: Orion) I would have been ok to let it go perhaps four or five hours past the meridian as there is no danger of collision.

Odd thing is, I didnt have anywhere near the same amount of rotation with Cygnus - and that was a 16 panel mosaic!

[ChrisLX200]

I'm not at all sure about cone error either, just something off the top of my head

I do know my camera is not set at exactly 90deg to the scope centre line, it's just approximately so - and I try not to move it between sesssions as I know I won't get it precisely back in place (sheer idleness I suppose!).

Anyway, the 10-Micron is very precise and it will flip and exactly line up the centre of the frame afterwards - no need to plate solve to get that accuracy. When I later come to stack subs from E and W of the pier they often align with no artifacts at all - but sometimes there is a lateral shift of 1 or 2 pixels which leaves parallel dark line(s) on one side or the other. It's no big deal and I know the cause of that (usually not precisely re-balancing at the start of the imaging session, and re-using the old mount model. With a new model generated just prior to starting the session the alignment is perfect).

So what does that tell you? Firstly, under ideal conditions then you won't get rotation. If you ARE getting rotation between E and W sides of the mount then clearly something else is a play - either flexure or some pointing error. How flexure would translate to rotational mis-alignment I can't envisage, but what else is there? I would ask if the rotation is progressive - does it increase the further away from the meridian you get after the flip.

ChrisH

Edited November 8, 2016 by ChrisLX200

[Uranium235]

Cant have been flex mate, that Star71 is rock solid and held in the vice like grip of an ADM saddle.... Hang on... I'll just pop downstairs and see if a cable snag occurred (though highly unlikely as all cables are velcroed up).....

Nope... no detectable rotation in the camera from when I set it up.... hmm Im stumped!

[alexbb]

It depends also on the cone error type. If the dovetail is shifted sideways to the scope, a Dec axis slew will point to the same place. If the front (or back) of the scope is farther away to the dovetail, then you have to slew the RA axis too to recenter. This will also happen if there's something caught between the dovetail and the mount clamp.

I'm trying now to figure in my head how much this matters if you're imaging closer to the pole.

Edited November 8, 2016 by moise212
Misspelling

[Uranium235]

This is the area I was imaging FWIW:

[martin_h]

Here is my overlap...its wedge shaped so would indicate a rotation after the flip, my ccd is 1.5 degrees out of square....according to my last plate solve

Edited November 9, 2016 by martin_h
more info

[Skipper Billy]

When I come to stack images from before and after a Meridian flip there is always a band as shown in the above image from Martin H.

I use Astrotortilla for initial alignment and again for realignment after the flip and PHD2 just sorts itself out and the target is bang on the middle of the frame but I cannot explain the sloping band.

Someone smarter than me (not difficult!) must know the answer!!

 

[Pompey Monkey]

On 11/8/2016 at 23:03, Uranium235 said:

and the camera was quite close to colliding with the mount

Are you sure it did not actually brush against the mount at some point?

The other thing that comes to mind is a less than perfect PA - this is another source of cone error. However, given the quality of your images, I suspect that your alignment is rather better than +- 3/4 of a degree!

[Uranium235]

2 hours ago, Pompey Monkey said:

Are you sure it did not actually brush against the mount at some point?

The other thing that comes to mind is a less than perfect PA - this is another source of cone error. However, given the quality of your images, I suspect that your alignment is rather better than +- 3/4 of a degree!

Nah, it didnt touch the mount Im pretty certain of that ... touch wood that will never happen (well, not unless I fall asleep during an imaging run!). It might be worth me checking if the tripod is completely level, to try and eliminate all possibilities so to speak - though I doubt that would be the case, I triple checked it in setting it up after its trip to SGL XII.

[Physopto]

On ‎09‎/‎11‎/‎2016 at 11:42, martin_h said:

Here is my overlap...its wedge shaped so would indicate a rotation after the flip, my ccd is 1.5 degrees out of square....according to my last plate solve

Looks like a small amount of cone error causing the small strip.  Do you get the same on the opposite  edge ?

Derek

[martin_h]

2 hours ago, Physopto said:

Looks like a small amount of cone error causing the small strip.  Do you get the same on the opposite  edge ?

Derek

Yes I do

 

[Physopto]

3 hours ago, martin_h said:

Yes I do

 

Then I would recon it is Cone error. The slight wedge is probably down to not being entirely orthogonal to the centre of the circle of the mount. it would be awkward to get exactly right, but doable with time and effort I suppose. Is it worth it? I spent ages getting rid of cone error on my EQ8 (well almost). I have yet to try the same on my 10 Micron. Again is it worth it?

 Derek

[martin_h]

It doesn't bother me. Photoshop crop tool soon sorts the problem.

[wxsatuser]

Take a look at the post by Gary, there is not just one source of cone error, I always thought there was just one.

http://www.iceinspace.com.au/forum/showthread.php?t=27686