I just wanted to show the error in the tracking of the Celestron CGEM telescope mount. The mount was polar aligned to within 1.6 arc-minutes of the North Celestial Pole. The data was gathered using PHD software by Stark Labs version 1.13.5 for Mac. It shows the periodic error of just under 20 arc-seconds. The important thing to take away from this is that the waveform is anything BUT periodic. There is NO Periodic Error Correction (PEC) that can be applied to this waveform that could ever effectively correct this error, that is that could be programed into the hand controller. Here is a graph of the periodic error occurring over more than five cycles of the worm drive gear:
The whole point of using a German Equatorial Mount (GEM) is so there will only be a single motor to control the axis (Right Ascension axis) that is, ideally, perfectly aligned with the rotational axis of the Earth. The Celestron CGEM telescope mount DOES NOT AND CAN NOT ACCOMPLISH THIS!!!!! Both axis motors must nearly be continuously driven to follow any object in the sky. Here is a graph showing the cumulative movements of the DEC axis and over the time period was greater than TWO ARC-MINNUTES!!!!! This dwarfs the periodic error in the RA axis so even if the PE error were to be able to be completely removed the drift error in the DEC axis would still have to be dealt with.
So one would have to ask themselves the question: Why would anyone buy this mount when any number of altitude-azimuth fork mounts that are, by design, intended to be tracked using both axis motors and can easily be found much cheaper than the CGEM? Wouldn’t it be nice if there were a web forum who’s members would provide useful non-delusional factual information. Don’t kid yourself, there WILL be image rotation with the CGEM mount due to the DEC movements!!!!
Here is the frequency spectrum graph showing the errors in the RA axis, the highest peak is the worm gear period, the second highest peak is the so-called 8/3 harmonic (with its sidebands) known to exist with the CGEM mount. This is most likely due to the poor quality servo motor and gear reduction system discussed in previous posts. Also note the 6/21 subharmonic at the far right side of the spectrum. If the length of data were to be greater than 24 hours the graphs would also show the non-orthogonality in a combined RA-DEC error spectrum of over 3600 arc-seconds Peak to Peak, that is OVER ONE FULL DEGREE of UNACCEPTABLE ERROR!!!
The above graphs were created using GNU Octave numerical analysis software using the log file created by Stark Labs PHD software. The Octave script and data files can be downloaded by clicking the following link:
phd_pec (corrected 5.2 micron pixel size)
Update, added 2012-10-05:
I decided to combine both the RA and DEC errors into a single graph to give a total error. I’m not sure but I think this is valid since both axis are essentially in motion together. I simply used the Pythagorean theorem; square-root(RA^2 + DEC^2). I then used a linear regression to remove the drift and performed an FFT. Below is the result:
The interesting thing here is the 8/3 harmonic is not the next important harmonic after the worm period. In fact there are at least eight harmonics and subharmonics that show much greater weight than the 8/3 harmonic. I’m not sure how relevant this is but it would not surprise me that this is exactly why Celestron has no plans to address the 8/3 harmonic issue.
When both motors are forced, through the lack of quality control, to be utilized simultaneously the resulting problems unfold exponentially.