OK. So let me ask this, if I get it all lined up as it is currently, and assuming I am quite gentle with it (as I take pains to be), any likely misalignment is likely to show itself how?
Since you have two tools: the Cheshire will confirm that if you see only one image in the autocollimator, it's a true stack. You can't get one of the "false stacking" scenarios.
If you use the Cheshire, then any remaining misalignment that leads to unstacked images is likely to be in the pointing of the focuser axis.
In other words, if it gets only slightly misaligned, I won't be able to see four images using the Infinity XLK so would it simply be a matter of adjusting the primary to get them all to stack?
No -- what mirror you shoud adjust obviously depends on what error there is. There is only one solution for the tilt of the secondary and the tilt of the primary that avoids both FAE and PAE errors. If you make adjustments to the wrong mirror, then you won't get a reading with no PAE and no FAE, no matter what.
There are tell-tale differences. If there is only one source of error, then the remaining images, if you have three, are always on a line. If you have only PAE (which means that you should only tilt the primary), not only is 3 stacked under P, but the P-1 and P-2 distances are the same. If 2 is further from P than 1, then you know that there is FAE.
And, finally, isn't the #3 image always the faintest one? That's how to identify it correct?
#3 is faint and inverted (and focuses somewhere else than 2, the other inverted reflection -- some people find it hard to focus it). It also does not move relative to P when you tweak the primary. If you have trouble identifying it, don't bother with CDP.
If you know there is an error, you can do two things:
-switch to the Cheshire and remove PAE by making the Cheshire reading correct by tilting the primary; then stack the images as best you can by tilting the secondary (which will reduce FAE), evaluate PAE again with the Cheshire, etc. That's the iterative protocol. The trick is to use each tool for a different mirror, so that you have a reliable protocol that wil converge to a solution without running around in circles (where with one tool you undo what you fixed with the other). It's slow but simple, and using two different tools ensures that you cannot converge to a false solution (although the offset pupil in the newer autocollimator can also help avoid these "false solution" scenarios).
-Use CDP: you can always null FAE by stacking P and 3 (which is the same as making sure that 2 and 1 are at exactly equidistant from P at opposite sides of it, on a line, but that is an indirect reading that is less precise.) Once you've done that, then tilting the primary to stack the images will null PAE without affecting the (now zero) FAE. So that converges to a solution in one go (of course I always check with a Cheshire and make sure no little tweaks are necessary).
The good thing is that you can converge to the solution using only one tool. The danger of using only one tool is that if you think you're done with the secondary but you're not (i.e. 3 is not under P and P, 1 and 2 are either not exactly on a line or P-1 and P-2 distances are not equal), by tilting the primary you can end up in a false solution where some of the reflections disappear --because their light paths cross a pupil where there is no mirror-- before they are truly stacked. But the Cheshire will reveal the remaining PAE, which is why the Cheshire is always used in the final sanity check.
Edited by sixela, 24 September 2020 - 02:45 AM.