Rotation after meridian flip... why?

[Stub Mandrel]

I fear you all have bit too much confidence in how well you align your scopes.

See below, if your scope, in its rings, is perfectly aligned with the dec axis (1a), a rotation (1b) followed by a flip (1c) will leave the scope pointing at the same spot.

If it is not fitted exactly parallel to the DEC plane of rotation (2a), the RA rotation (2b) and DEC flip (2c) will leave it pointing at a very different place, at least with my exaggerated five-degree misalignment.

Clearly a 1-minut misalignment, too small to see, would give a 2-minute error in pointing, enough to give two mis-aligned images.

There is no need to invoke an inaccuracy in the mount itself.

 

[martin_h]

I don't align my scope, just slap it in the saddle and fire away, as long as my object of interest in in the centre of fov I'm not to bothered about the frame edges, I have to crop them to get rid of the dither lines.

[ChrisLX200]

3 hours ago, Stub Mandrel said:

I fear you all have bit too much confidence in how well you align your scopes.

See below, if your scope, in its rings, is perfectly aligned with the dec axis (1a), a rotation (1b) followed by a flip (1c) will leave the scope pointing at the same spot.

If it is not fitted exactly parallel to the DEC plane of rotation (2a), the RA rotation (2b) and DEC flip (2c) will leave it pointing at a very different place, at least with my exaggerated five-degree misalignment.

Clearly a 1-minut misalignment, too small to see, would give a 2-minute error in pointing, enough to give two mis-aligned images.

There is no need to invoke an inaccuracy in the mount itself.

 

 

That's pretty much what I said earlier but the point is not the misalignment, that is easily corrected by plate-solving back onto target (i.e., the same centre of frame), but there remains some residual rotation comparing each side of the mount. It's the rotation that is the weird issue.

ChrisH

[Stub Mandrel]

It's fundamentally because. the scope itself is not just moved a foot or two to one side, it's rotated through 180 degrees.

[ChrisLX200]

21 minutes ago, Stub Mandrel said:

It's fundamentally because. the scope itself is not just moved a foot or two to one side, it's rotated through 180 degrees.

I think we have a Missed Point Error in Line.1 here    We know it's been flipped 180deg but frames should still line up (just 180deg out), but folk are reporting residual rotation of a degree or two which cannot be accounted for..

ChrisH

[alexbb]

Imagine you are doing the flip manually and imagine that the situation shown in the drawing describes a scope with a magnification of 1, but the scope is tilted away from the dovetail clamp (as described before by @Stub Mandrel).

First you rotate around the RA axis 180 deg. Then you rotate around the Dec axis as shown. The coloured stars are then away from the center.

What you need to do in the case described to center to the same FOV is to rotate the scope around the RA axis counterclockwise and around the Dec axis ... can't compute at this hour how you have to alter - add or subtract to the degrees number, but there's an arrow for that.

Does this make sense to anyone? Maybe I can explain better when I'm awake.

[Stub Mandrel]

12 hours ago, ChrisLX200 said:

I think we have a Missed Point Error in Line.1 here    We know it's been flipped 180deg but frames should still line up (just 180deg out), but folk are reporting residual rotation of a degree or two which cannot be accounted for..

ChrisH

Fair enough, what I illustrate would only create pointing error, but look again at figs 2a to 2c

The scope is cantilevered out from the mount (unless it is a fork mount, but it isn't it's a GEQ) so inevitably there well be some flexure. In 2a this will cause the body of the scope to droop on the side away from the mount and the image will rotate anti-clockwise compared to the view through a weightless scope. But in 2c it is now balanced the other way and it will rotate the image slightly in a clockwise direction.

This may be a small effect, but acting in concert with any misalignment between RA and DEC (which will cause a vertical error) and DEC and scope alignment (which will cause a horizontal error) we cloud expect the sort of mis-registration see where the main effect is a pointing error, accompanied by a smaller 'twist'.

There may be scepticism the scope will twist enough to cause this, but guide scopes are usually much lighter and their weight can be enough to cause image rotation as the scope tracks. These effects could be compounded or partially compensated by twisting of the OTA as the weight of the imaging train shifts from one side to the other.

I'm confident the full mass of the setup MUST make a measurable twisting movement when it is turned upside down.

This wouldl be a great challenge for someone who has access to FEA software and lots of time!

[vlaiv]

Just repeat diagrams with DEC not being 90degs to RA - it will show rotation.

[vlaiv]

Ok here it is:

1. original FOV orientation

2. After 180 around RA axes

3. After 180 around DEC axes

4. Actual line DEC rotates around - it's angled to what it should rotate about because RA and DEC are not 90degs to each other.

Diagram shows exaggerated angles just to make it obvious. So 1 - initial FOV orientation, 3 - final after flip - clearly FOV has rotated.

[StuartJPP]

If your OTA/sensor had 0.5 degrees of rotation/pointing error in relation to the Dec axis and you then flipped Dec by exactly 180 degrees, would the error now be 1 degree off? I'd suspect so. Now also take into account tilt of the OTA (cone type of misalignment), then it will probably do the same thing when flipped.

In other words the error is 1/2 the value measured. It isn't an error after flipping it is a cumulative error from before and after flipping.

The good thing is that this can be tested during the day quite easily.

My brain doesn't work in 3D so I can be totally wrong.

[MarsG76]

Without putting pen to paper to verify my theory, I think that  even though the telescope flips 180 degrees the scope is on the other side of the RA rotational center and with the DEC having to tilt up to recenter the object, you'll find that the RA bar flip is not exactly 180 degrees with DEC compensating to center the object. 

Think of your finger at arms length and looking at it with alternating eyes, ignoring the fact that one of your eyes would be upside down.. When you alternate your eyes the finger shifts, and the amount you need to pan your head to bring the finger back into the same spot in reference to what's behind it is the angle shift in a meridian flip.

Of course the angle of the camera and a spherical sky creates a tilted angle that's more than a angle tilt that is  left to right.

[MarsG76]

On 19 November 2016 at 13:12, moise212 said:

Imagine you are doing the flip manually and imagine that the situation shown in the drawing describes a scope with a magnification of 1, but the scope is tilted away from the dovetail clamp (as described before by @Stub Mandrel).

First you rotate around the RA axis 180 deg. Then you rotate around the Dec axis as shown. The coloured stars are then away from the center.

What you need to do in the case described to center to the same FOV is to rotate the scope around the RA axis counterclockwise and around the Dec axis ... can't compute at this hour how you have to alter - add or subtract to the degrees number, but there's an arrow for that.

Does this make sense to anyone? Maybe I can explain better when I'm awake.

Yes!!! Exactly my assumption

[Jessun]

In my opinion, for it is nothing more, a flip simply flips and mirrors (turns it upside down if you like) so any camera misalignment is doubled. If you're leaning 5° to one side in the draw tube the resulting stack will have a 10° discrepancy between pre and post flip subs, so the black triangles show up along the edges.

What causes the misalignment in the first place could well be all sources of cone error tricking you to think that the camera is aligned with RA and Dec on that side of the meridian.

PS - I take this all back. It's not the case... Camera retains orientation as good as the mount can do it after the flip. Perhaps it lies with internal bearings or axis alignment allowing a little tilt or something else not squared up internally.

/Jesper

Edited November 15, 2017 by Jessun

[BrendanC]

Hi all,

I know this is an old thread but I'm reviving it!

I have exactly this issue of rotation and whereas it doesn't bother me too much most of the time, if I have a good, full FOV image that has involved a meridian flip, it's frustrating having to crop it.

There are lots of great diagrams and explanations in this thread and honestly, without my brain exploding/imploding, I don't completely understand them.

So, after all the debate here: what's the fix? Is it cone error, because if so I'm considering using ConeSharp to fix it - https://www.sharpcap.co.uk/conesharp

Thanks, Brendan