Early pointing runs
With a small camera FOV (0.25 deg), it was going to be important to point accurately. The target stars must fall near the center of the FOV after a 'goto' command if observing was going to be efficient. At first, the pointing was terrible. I found it hard to drop stars into the FOV and often ended up 'hunting' for them. It was clear that I needed a finder scope to learn about the pointing errors.
Here is my finder scope addition to the main telescope. I added a small camera to it so that I could view it remotely.
The finder provided about 8x the FOV. Here the entire Pleiades fits into the finder FOV:
With this larger FOV I was able to steer stars into the main camera FOC fairly easily. Easily enough for testing.
I could point to about 40 different stars in an hour and record their apparent positions for later analysis. It became clear that the az table tilt was minor and correctable, but the az errs signaled a significant problem.
The altitude errors vs. az follow a simple sine wave, as would be expected if the az table was slightly tilted. The half-amplitude of the sine wave is about one arc minute. Very correctable with the tilt adjustment bolts.
However, the az errors as a function of altitude showed that the altitude axis and the az axis were not perpendicular. They were off by about 0.5 degrees. This meant that I could not point to the zenith, and that the closer I got to the zenith, the larger the az errors were going to be. I called this the 'zenith hole' problem. Because a meridian altitude is defined as the altitude when the az = 180 degrees, this meant that meridian altitude measures at high altitudes (stars whose declinations were similar to my latitude) were going to be corrupted.
Because I didn't have a second instrument to check against (as Tycho did), I had to resort to modern catalogs to show me the actual errors. In the graph above, the error values graphed are the difference between what I measured and what the catalogs said I should see. I could use this data immediately to correct my pointing, and, when I did that, I was able to drop stars into my camera FOV easily.
This mechanical error was not correctable. My only recourse was to model the error in software and correct the az pointing on the fly.
I also found, through later tests, that I could not even accurately correct the altitude readings for stars that got closer than about 3 degrees to the zenith. This affected about 30 stars in Tycho's catalog, and several of them were in the list of 100 primaries that would form the basis of my catalog attempt. So it was crucial to find a way to measure the meridian altitudes of these 30 some stars that passed close to my zenith.
A meridian altitude measurement leads to the determination of the star's declination if the observer's latitude is known. I had already determined my latitude via solar observations, so, if I could find an alternate way to determine the declinations of these zenith stars, I could solve the problem.
Via many emails exchanged with Dr. Rosa, he was able to point me to an alternate approach, used by Tycho, to obtain the declination of a star even away from the meridian. (That technique is shown on the next page, the Zenith recovery page)
A more detailed look at the altitude data revealed another problem with the mount. In my pointing runs, I calculated the expected altitude of the target star as if there was no atmosphere. The Earth's atmosphere refracts the starlight slightly and makes the star have a higher altitude than the exo-atmospheric expectation.
I had hoped that my plots of (measured altitude - calculated altitude) would show this refraction effect, but they did not. Instead, I got a behavior almost the opposite, which showed that my mount was not pointing accurately in altitude. The effect was small, several arc minutes, but enough to compromise the final accuracy of the catalog. Here is the key plot:
This plot shows the difference between what I had hoped the data would show (red data points) versus what the data actually did show (blue data points). The red data points were the expected refraction, calculated from a formula, for each of the stars I observed.
The blue points were the difference between the measured declination and the calculated exo-atmospheric declination. The difference should have been due to refraction. It wasn't.
My conclusion was that the error was due to flexure in my altitude optical path. I didn't have any investigative capability (optical models of the optics) so I tried several lab experiments where I attached weights to various parts of the scope. I was able to show deflections, due to the weights, but was not able to definitely explain the flexure.
This meant that I would have to use current knowledge to correct my altitude data. I did this by fitting simple polynomials to each night's data so that I could correct slope and zero-point errors of the data set. These corrections accounted for both the flexure and refraction combined, so my reported altitudes are effectively exo-atmospheric.
I had to keep reminding myself that the goal of this entire exercise was not to produce a catalog equal to or better than Tycho's, but to understand the efforts he must have had to go through to create his catalog. Specifically, how did he plan his nightly observing, how much effort did he put into the actual observing, and how did he go about assembling the catalog.