Quote:

---

No one is claiming anything “new”. This thread is not meant to introduce new concepts or terminology.

---

Post Extras:

Quote:

---

If you follow a sane collimation protocol, lack of squareness of the focuser cannot cause axial miscollimation (or do you mean an optical axis not along the tube axis when you say "axial disparity"? That's certainly a "new" term to me, so I'm unsure as to what exactly it means.)

---

Post Extras:

Quote:

---

Quote:

---

No one is claiming anything “new”. This thread is not meant to introduce new concepts or terminology.

---

I don't know if anyone is claiming anything, but I do know that something that is not new is being called that. If we step back from talking in terms of focuser/primary axis, this collimation mode is essentially identical to the conventional 1-2-3 step procedure, with #1 being centering the diagonal in the focuser's circle by axially shifting it toward primary, #2 centering the primary's reflection in the diagonal by tilting the diagonal so that focuser's axis "reflected" from it hits the center of the primary, and #3, centering reflection of the diagonal in the primary by tilting the primary's axis toward focuser (so that it nearly coincides with focuser's axis "reflected" from the diagonal).

What is new is that I figured out why I couldn't get primary's axis to fall back onto the focuser's axis (didn't take into account that axial shift of the diagonal cancels most of the primary's reflection displacement in the diagonal). The result is that the magnitude of axial disparity is significantly lower than I thought. It is given by 57.3a^2/SH in degrees, with "a" being the diagonal minor semiaxis, S the diagonal-to-primary and H the diagonal-to-focus separation. For a 12" f/5 Newtonian, with a=1.5", H=10" and S=50", that comes to 0.26 degrees (assuming square focuser).

As usual, it's good for me to hang with you guys.

Vla

---

The formula I use is that the displacement of the optical axis is equal to the offset in classical offset positioning of the secondary.
As such, you can then compute the angular error between the optical and mechanical axes as the inverse tangent of the offset divided by the distance from primary to secondary.
For a 12.5" f/5, that was about 1/8" offset and about 8' deviation between the optical and mechanical axes.
With a 2.6' resolution on my encoders, that could cause errors in pointing, but the object is always in the center 50% of the field of a 45' field eyepiece if I took the time to align using a reticle eyepiece. I've achieved 0.0 warp factors.
However, I should point out that my focuser is exactly on the side of the UTA, which means the deviation is only in azimuth, which it compensated for in alignment. If the focuser is at other angles, the axial deviation could lead to more problems in pointing.

Now, if offset is larger and the secondary-to-primary distance is less, the error grows. As I mentioned, it is probably wise to resurrect the classical offset techniques as scopes get shorter in f/ratio.

Post Extras:

Quote:

---

The formula I use is that the displacement of the optical axis is equal to the offset in classical offset positioning of the secondary.

---

Post Extras:

Quote:

---

As I understand it, in classical offset the secondary is moved along the 45 degree plane of the secondary by an offset amount that is required to bring the opto-mechanical axis coincident with the geometric center of the secondary, and that that results in a centered secondary and even illumination.

---

Post Extras:

Quote:

---

and #3, centering reflection of the diagonal in the primary by tilting the primary's axis toward focuser (so that it nearly coincides with focuser's axis "reflected" from the diagonal).

---

Post Extras:

Quote:

---

If these two axes [the mechanical tube axis and the optical axis] are at an angle, it induces pointing and tracking errors

---

Yes, if you care about them (and have a GEM with GoTo or use digital setting circles).

But it doesn't introduce axial miscollimation, i.e. aberrations or defocus in the field.

BTW, it doesn't introduce tracking errors on my scope. I have an equatorial platform and its polar axis is quite independent from the axes of the Dobsonian mount and their squareness.

It also doesn't introduce tracking errors on mounts that don't track at all, obviously, and as I star hop and do not use DSCs, it also doesn't introduce pointing errors on my scope either .

Post Extras:

Quote:

---

You've got it slightly wrong. #3 uses a Cheshire, and *perfectly* overlaps the optical axis and focuser axis and their reflections (even though the intercept angle on the secondary is not 90° and the optical axis is not along the tube axis), because you make the reflected optical axis cross the focuser axis half way between the Cheshire ring and the pupil.

---

Quote:

---

Using the reflected secondary's silhouette is a bad idea, and no sane protocol would use it as a reference.

---

Post Extras:

Quote:

---

It also doesn't introduce tracking errors on mounts that don't track at all, obviously, and as I star hop and do not use DSCs, it also doesn't introduce pointing errors on my scope either.

---

That's nothing. I have a small Tal Newtonian that never needs collimating at all, because I never use it. Isn't that as good as it gets?

Vla

Post Extras:

Quote:

---

Quote:

---

If these two axes [the mechanical tube axis and the optical axis] are at an angle, it induces pointing and tracking errors

---

Yes, if you care about them (and have a GEM with GoTo or use digital setting circles).

---

Not necessarily. If the plane defined by the two lines (optical axis and OTA axis) is perpendicular to the axis of rotation (declination or altitude), then the axial offset will have no impact on DSC performance.

Even when the plane is skewed, the error is often small enough to fall within the resolution of most DSC displays (+/-0.1-degree). The Argo Navis (used with 8000 or 10000 count encoders) pointing accuracy does have a higher resolution, but the AN can also model the pointing error and correct it on the fly...

That said, I've used an f/4.1 with a Sky Commander and an f/4.0 with a Argo Navis, both aligned using the "New Model" (full offset with centered secondary mirror) and, selecting good alignment pairs, have found the pointing accuracy normally places the selected object within 0.1- or 0.2-degrees from the center of the fov. The worst case scenario is when the selected object is within a few degrees of the zenith--the possible azimuth error can be very large, and still fall within the fov of a widefield eyepiece! Remarkably, once the object has been centered, tracking is largely unimpaired.

FWIW--I think the first time I used the phrase, "the new model", was in 1998 (although back then I suggested mechanically tilting the focuser to ensure a 90-degree intercept). I was getting quite a bit of opposition from those who felt collimation was not accurate unless it was either centered or offset. Choosing a model that was both was probably enough to start some debate--but I didn't stop there, going on to assure those who would follow the procedure that the focuser need not be precisely perpendicular to the OTA, the intercept angle need not be exactly 90-degrees, and the primary mirror need not be perfectly centered in the OTA. The goal, of course, was precise axial (focuser and primary mirror) alignment with an optimal secondary mirror alignment (best available offset and minimal skew).

Considering lightweight, open truss OTAs might have small eccentricities or tilt errors from the lower mirror section to the upper focuser section, and the potential for small geometry errors in the upper focuser section that might impact roundness, squareness, and centration, and then adding that the secondary mirror might not be a true sqrt2 major axis ellipse...it seems prudent to focus on the optical alignment first, and then address the mechanical alignment issues as/if they arise.

Post Extras:

Quote:

---

Quote:

---

You've got it slightly wrong. #3 uses a Cheshire, and *perfectly* overlaps the optical axis and focuser axis and their reflections (even though the intercept angle on the secondary is not 90° and the optical axis is not along the tube axis), because you make the reflected optical axis cross the focuser axis half way between the Cheshire ring and the pupil.

---

How do they cross if they perfectly overlap? Nothing gets really perfect, except on our illustrations. One of the reasons is that actual mirror surfaces, when tilted, do not rotate around the point of their intersections with the axis, but around a point effectivelly displaced from it (particularly diagonal). As a result, any tilt introduces decenter, which makes the surfaces and reflections less than perfectly concentric.

Quote:

---

Using the reflected secondary's silhouette is a bad idea, and no sane protocol would use it as a reference.

---

Lots of insanity around: Sidgwick (p204), Thompson (p120), Lecleire (p262), Norton's Star Atlas (p72)...

Vla

---

It should be pointed out that the silhouette (reflected) image of the secondary grows more concentric with all other images as the f/ratio gets longer. The illustration that Jason posted is slightly exaggerated for f/6 and longer scopes. Many of the sources you mention were written back when most reflectors were f/8 to f/12.
It might also be pointed out that standard collimation techniques of the '50s and '60s did not always call for even edge-of-field illumination. As late as 1967-1968, books by popular authors such as Sam Brown of Edmund Scientific called for a centered secondary with all images concentric except for the reflected primary, which appeared dropped toward the lower end of the secondary.
The interesting thing is that there are so many different mechanical alignments that can result in good optical collimation. It is when we aim to optimize field illumination and maintain coincidence between optical and mechanical axes that things get more complicated.

Post Extras:

Quote:

---

Not necessarily. If the plane defined by the two lines (optical axis and OTA axis) is perpendicular to the axis of rotation (declination or altitude), then the axial offset will have no impact on DSC performance.

---

Post Extras:

Quote:

---

It is an unvolved procedure to grasp. The main reason I got into this thread is to make sure the page on Newtonian collimation that I added just recently got it right. Good move for me, since Jason's illustrations and Alexis's comment made me obvious an inconsistency that led me to find and correct an oversight

---

Post Extras:

Quote:

---

Quote:

---

Not necessarily. If the plane defined by the two lines (optical axis and OTA axis) is perpendicular to the axis of rotation (declination or altitude), then the axial offset will have no impact on DSC performance.

---

Hi Vic, I believe we discussed this specific point looooong time ago and the end result of the discussion was that the above error can be corrected but it will have to be by the DSC computer. However, if the DSC computer does not have the proper software to make this kind of correction, the error will impact accuracy.
Jason

---

I just can't think of any DSC computer that doesn't work that way. The ones I've used employ a basic 2-star alignment procedure. If you choose one star at the zenith (90-degrees elevation) and one at the horizon (0-degrees elevation) and use the telescope optics to align to them...

Of course, it's unlikely you would choose two stars with such widely separated elevations, but then whichever two stars you do choose, once their elevation and azimuth has been correlated to their known right ascension and declination, the software will know where the zenith and the horizon are--even if the telescope horizon (defined by the azimuth bearing) differs from the actual horizon.

The Argo Navis manual refers to this alignment error as a "collimation error". In an Alt/Az configuration, it's "...the non-perpendicularity between the pointing axis and the Alt axis." They further note that this error results in a "...left-right shift in the sky that is constant for all Altitudes" and "...there will be an area around the pole (zenith) of the scope that the scope cannot point." This only happens when the optical axis is offset (skewed) away from the perpendicular plane--if it's offset and parallel to the plane, or falls inside the plane, the error will be compensated for when the DSCs are initialized (and the area around the pole will be accessible).


(The Argo Navis also uses an altitude reference (usually 90-degrees), but this is only an estimate and is referenced to the telescope (not the zenith). With the horizon stop turned on, it's possible to have the Argo Navis tell you that the object you have selected is below the horizon, when in fact, you can manually override the GoTo feature and view the object in the telescope (a common occurrence at the WSP when observing Eta Carinae or Alpha Centauri).)

Edited by Vic Menard (10/05/09 02:00 PM)

Post Extras:

Quote:

---

I just can't think of any DSC computer that doesn't work that way. The ones I've used employ a basic 2-star alignment procedure. If you choose one star at the zenith (90-degrees elevation) and one at the horizon (0-degrees elevation) and use the telescope optics to align to them...

---

Post Extras:

Quote:

---

...As long as the “actual” angular distance between the two alignment stars matches the database, then it is a matter of simple math to translate, back-and-forth, between the actual DSC coordinates and the database coordinates.

---

Quote:

---

If the angular distance is not the same then the discrepancy is caused by other reason(s) than optical/mechanical axes misalignment.

---

Post Extras:

Quote:

---

How do they cross if they perfectly overlap?
[µ/quote]
I'm talking about the theory. You make the axes cross at the focal plane and at the primary mirror's centre, and then they overlap everywhere (and "cross" everywhere).

Quote:

---

As a result, any tilt introduces decenter, which makes the surfaces and reflections less than perfectly concentric.

---

In real life, you'll find many procedures are iterative. But of course, once you're within the proper tolerances, you stop, rather than take an infinite number of extra steps .

Quote:

---

Lots of insanity around: Sidgwick (p204), Thompson (p120), Lecleire (p262), Norton's Star Atlas (p72)...

---

No argument from me. I'd bet they didn't try it on a large f/3.66 Newt, though.

Post Extras: