Over the past year and a half I have noticed a small amount of tilt in the focal plane of my setup with respect to the CCD sensor's surface, amounting to ~50 +/- 10 microns across the ~25 millimeter sensor diagonal. This tilt leads to uneven focus across the sensor. When I replaced the entire stock Takahashi focuser, camera angle adjuster and adapter train earlier this year with an FLI Atlas focuser and two PrecisePart adapters, the tilt increased to about ~150 +/- 10 microns across the diagonal.
I discovered that three flat head cap screws on the front surface of the Atlas focuser were not countersunk sufficiently. These screws interfered with the proper mating of the flange of the PrecisePart adapter to the surface of the Atlas. The overlap between the flange and two of the screws is visible in the photograph below.
The screws are epoxied in place, FLI recommends that they not be removed to avoid damage to the focuser. I solved this interference problem by machining clearance for the screws into the outer section of the flange. The machined clearance is visible as the non-anodized portion of the flange in the photograph below. The unmodified part of the flange is sufficiently wide for proper mating. This change reduced the amount of tilt but did not completely eliminate it. The residual tilt measures ~50 +/- 10 microns across the diagonal, and is equal to the amount of tilt measured prior to the installation of the Atlas.
The Atlas focuser uses a proprietary Zero Tilt Adapter (ZTA), an evolution of the traditional dovetail mechanism that uses a circular steel spring rod to apply even pressure around the connection when three set screws are tightened. The ZTA rod is visible retracted within its channel in the photograph below. I discovered that the order that the three set screws are tightened is important. The screw that compresses the middle section of the ZTA rod is the first one that should be tightened, followed by the other two screws that compress the sections nearer to the rod's ends. When tightening the middle screw, the section of the adapter's flange located diametrically opposite to the middle screw should be checked carefully to be sure it remains properly mated with no gap.
The impact of the residual tilt on star FWHM measurements is shown below. The contour plot shows star FWHM estimates across the sensor surface using measurements of a typical 40 minute subexposure, binned 2 x 2 with an H-alpha filter. The tilt runs roughly from the upper-left corner to the lower-right corner, with the upper-left corner ~25 +/- 10 microns inside of focus, the center portion approximately in focus and the lower-right corner ~25 +/- 10 microns outside of focus. Both the two other upper-right and the lower-left corners are nearly in focus. The critical focus zone of my setup at its focal ratio of f/5 is approximately +/- 10 microns [astrodonimaging.com]. The +/- 25 micron tilt is significantly larger than the critical focus zone, and it produces a measurable increase of ~10% in star FWHM in two corners. The tilt shown in this plot has stayed roughly the same over all parts of the sky and over all of my subexposures during past year and a half (except for the increase due to the FLI cap screw countersink problem described above).
Because the tilt remained roughly constant across an entire image train change, I believe the tilt is likely due to either sensor non-orthogonality or optical misalignment or possibly a combination of both. According to my camera manufacturer, Kodak does not specify the parallelism of the silicon detector relative to the CCD package, and there can be a small amount of non-parallelism between the sensor and the mating surface.
To solve this tilt problem, I recently purchased a Gerd Neumann Camera Tilting Unit (CTU) XL [gerdneumann.net], shown in the photograph below. The unit uses three cone shaped radial screws to separate the two parts of the unit by a small amount. One whole turn of a screw corresponds to a separation of 0.2mm measured on the outer surface of the 119mm diameter unit. The two parts are held together with packs of disk springs and secured with set screws. I plan on installing the unit between the focuser and the camera using adapters from PreciseParts. I will update this blog once I have experimental results with this unit.