Your A should be more like 58 if you use the Camera Plate.
And your B should be around 95-96.
I use 97 on my 5D with 4 around and it has the same Tripod mount length.
There is no "correct" settings for a fisheye but 92 is to far from the medium if you want the zenith to stitch perfect.
How did you get to the A=55. If you looked through your viewfinder and set the centre point to the rotator you should make sure that the bottom plate of you camera is plain first.
I have 4 Canons from 10D to the 5D Mark II and only the Mark II is plain.
You provide no clues as to how you are checking the panohead setup. You should be able to do this accurately in no more than 10mins by checking the position of the entrance pupil directly. See http://www.johnhpanos.com/epcalib.htm.
You provide no clues as to how you are checking the panohead setup.
First, I found the "A" value (horizontal displacement bar) by mounting my camera and looking through the viewfinder to center it with NN logo.
The I shot two overlapping objects (one straight in front of the other) to adjust the "B" value (front/back displacement). However, I'm not satisfied with the alignment I got, something doesn't feel right. I usually do test panos on my living room, where I have some very close objects and some far away ones, my resulting pano had some misalignment issue on not so close objects.
I think my setup is perfectible. I will photograph my camera as in the example you linked to check my alignment.
Pointing the camera down at the rotator is NOT the way to get the lower rail setting. There are way too many variables involved. The pano head and the camera all have to be absolutely square and plumb. You would also have to be looking exactly straight down into the viewfinder. You could be off a couple of millmeters and not even know it. Although it is a good place to start, it is not where you want to finish.
Pointing down on the rotator is actually a very good way as it also takes care of errors in the lens.
There is no other way to figure that out.
In reality there can be as much as 2mm difference between lenses.
You will see that when you optimize for shift and get large differences between 2 lenses on the same camera.
Thats why I suspect that Felipe's camera may have the same error.
The correct A value for his camera on the NN# is 58-59 mm the same as the 5D .
He may also have a vertical arm that is not in 90 degree angle with the bottom rail.
This will produce the same error.
My old NN3 has that and the only way to correct is with some tape on the bottom plate of the vertical arm.
we know there are different ways to go to Rome to take the royal shot with Cesar. Don't we?
As Dennis said, there are a lot more reasons besides non using ducktapes use to find the right LRS. Slight differences in NN production for example.
I never saw Dennis to place "the one and only way to go to Rome" comment. He is experienced, he spend a lot time for developments to hardware and placed his thoughts in several forums. Leading to new NN products with "state of art designs" from Nick. A very good synergy.
What I should have said, which reflects my view in a better way, is that pointing the camera down and centering on the rotator is not the only step in the calibration process, but the first. If, after followng John's and Smooth's tutorial, the camera is not pointing directly into the center of the rotator, further adjustment may not be required. Shimming parts that are not square may bring the rotator into the center of view, but might not be necessary in order to get good results. If the image parameters show a lot of variances, then adding shims to the appropriate area may become necesary for the final "fine tuning".