NavList:

Hi Bill

I've got the two axes of rotation nearly coincident.  They need not be perfectly coincident, as presumably the index mirror is flat enough to not mind the small bit of translation that occurs with Ultradex rotation.

I've also got the parallelism good enough such that the autocollimator can see the reflected beam throughout the travel.  A adjustable kinematic mount (two plates) such that the parallelism could be dialed in would certainly be nice. It just seemed like too much for a wobbly proof of concept wooden structure.  I will include it in the metal sextant mount, complete with rare earth magnets to provide preload between the plates.

There is a kinematic locator for the sextant legs on the existing woobly wood.  I've got the cup under the index mirror leg, a slot radial to that cup for another leg and a flat for the third.  By carefully milling the cup and slot with the same machine set, the parallelism was adjusted by lowering the flat!  The problem I experienced was, as you point out, the overhung sextant load.  Every time the Ultradex was raised and lowered to engage/disengage the teeth, the flimsy wooden plate wiggled and the sextant took a new set in the kinematic locator.  The walls of the cup and slot were not smooth enough to allow the sextant to settle back to true location.  I eventually just provided a preload via the sextant handle being tie-wrapped to the plate.  It was a balance between using a lot of force, and potentially distorting the sextant frame and the wooden plate or too little force, permitting the sextant to move around.  (Your setup does not have this issue, as your ultradex equivalent merely rotates in plane).  With the commitment to metal, I can assure a good enough surface finish in the kinematic locator to assure true location.

As to balancing the overhung load.  The overhung load caused undue friction on the raising and lowering action of the Ultradex.  This caused additional motion in the wobbly wood. I intend to provide a plug of steel directly on top of the Ultradex and under the adjustable kinematic mount.  This will raise the adjustable kinematic plates AND center the mass more generally over the Ultradex.  This will permit me to directly balance the load without interfering with the lever that raises and lowers the Ultradex.  Additionally, with the commitment to metal, the static deflection due to the overhung load will be inconsequential.

Your point vis plasticene is well taken.  The Ultradex is current hard bolted to the existing wooden plate.  When I move to the granite surface plate, there are two options.  I can use either threaded inserts epoxied into the granite or a plate clamped to the granite with threaded holes installed.  As to the Autocollimator, that stand alone is in excess of 50 pounds.  It resists bumping just by mass, however it too can be clamped to the granite, a practice skipped for the woobly wood.

The surface plate will be thick enough to retain the flatness independent of mount.  Too often, I've seen what amount to toy surface plates where the plate just isn't thick.  In that case, the surface plate bends.  I've not yet decided what flatness grade of surface plate I want.

The grade of flatness is going to be a function of how good a sextant collimation has to be.  Starrett offers a croblox reflecting cube, also in custom sizes.  The squareness from one face to the other is on the order of a few seconds.  So if I lay one face against the surface plate, the next face is perpendicular to the plate.  Align the Auto collimator to the vertical perpendicular face.  Install the sextant in that optical path, with the arc of the sextant made parallel to the surface plate.  The autocollimator's optical path will go through the telescope.  If the telescope is then not parallel to to arc of the plane of the sextant, then the beam will be deflected, something the autocollimator will observe.  An error observed can be adjusted out, if the sextant has that feature, or simply measured to be compared against a go/no go figure of merit. 

I very much enjoyed your chasing tenths post.  There are two parts to your post.  Part one described how you certified your reference (the cut down sextant).  Part two described how you used that reference to measure the arc of the sextant under test.  For the external reader, the Ultradex is certified as a reference by the manufacturer and provides an extreme level of accuracy.  The Ultradex substitutes for part one of Bill's post by definition.  In part two, Bill and I are doing essentially the same thing, rotating the reference one way and the sextant under test the other.  I chose to obtain 6 values and average.

As to the length of the test, it takes me just over an hour of determined focus to take the 42 values and record them.  Add in set up and optical alignment for another 1/2 hour with result presentation and breakdown for another half hour.  Call it a two hour event

Brad