Collecting  Photons

Comparing a vintage Byers drive with a circa 1992 G11.

Here are shown two periods from the of a vintage Celestron 8 equipped with a "precision" (at that particular point in space-time)  Byers drive, and from my Losmandy G11 retrofitted with the steel precision worm. Although the Losmandy has a smaller error the trace is much steeper, making it harder to correct. Data shown here have been digitally smoothed with a 2.5 s moving window and so appear cleaner than following traces, that instead show raw data.

The 76 seconds wobble

The Losmandy G11 is a wonderful mount with a major problem:  the incredibly dumb way in which the RA and declination worms are assembled. As every owner knows all too well, the two bearings holding the worm are mounted within two separate blocks, that are screwed directly to a base plate. The adjustment of the two mounting blocks is extremely tricky and the lack of a sure way of ensuring orthogonality is the source of several tracking issues including the horrific non-harmonic error known as the 76 seconds wobble.

The upper graph shows three cycle of my worm. This is a steel "high precision" screw that has been obtained at the end of 2006. The data shows that the curve is quite complex with a peak to peak amplitude of about 15" and with several different overimposed harmonics. The green trace shows two components of the Right Ascension drift: a fundamental component at a period of 240 s and the second harmonic at a 120 s period.  The comparison of the two traces shows that these two components are not sufficient to describe the tracking error. Indeed, the spectral analysis of the data shows that there are at least three components related to the basic periodic error of the worm (240, 120 and 60 s period). Furthermore there is a massive component at about 76 s: cunning investigative work by several users have traced this error to an error of orthogonality of the worm axis compared to the bearings. Notice that the total peak-to-peak amplitude of all the components nearby 76 s is about 8"!Finally, small but numerous components can be seen at higher frequencies, and their presence can be easily detected also on the tracking data in the form of a small but very steep noise.

The 76 s wobble and the high frequency components are very difficult, or totally impossible to guide out, and represent the main obstacle at using this mount for high quality astrophotography.

A new method to mount the screw was required.

Enters the Ovision upgrade kit.

In the upper image the two original mounting blocks can be seen, together with the carter of the kit produced by Ovision, a company in the south of France. They have produced a novel design to mount an high precision worm within a steel monolithic worm block. This design ensure orthogonality between the worm and the bearings and a much easier life when adjusting the system. Furthermore, since worm and block are now made of the same material, it should have less issues with temperature changes.

I received my unit after a few weeks from the order: the worm and the block are beautifully made and they are extremely easy to assemble on the mounting. As a matter of fact the kit is much easier to install than the original blocks. The unit arrived with its own certificate done by testing it under the stars. The certificate described very accurately what I found with my own testing.

Comparing the new and old blocks.

The upper traces show the direct comparison between three periods of the original worm and blocs (red) and the Ovision kit (green). The peak to peak amplitude is decreased to about 7 asec and the trace looks much smoother.

The really good news come with the spectral analysis. Although the first harmonic of the PE is identical the second harmonic is gone. It is also completely gone the 76 s oscillation and the high frequency noise.

The loss of non harmonic components and of the high frequency noise is particularly exciting and suggest that autoguiding will operate much better. Indeed, it is important to stress that the amplitude of the periodic error is not very important per se. The really important thing is the smoothness of the tracking. A large periodic error can be easily guided out, provided it is smooth.  A small periodic error with high frequency components will never be guided out.

Derivative of tracking data

Here I computed the derivative of the tracking data to show the improvement in tracking smoothness. Data (sampling rate 0.25 s) were smoothed by a 5 points parabolic interpolation and were finally derived. What is shown here is the tracking error expressed as angle (asec) per sec. This is what the autoguide has to deal with. Clearly the performance of the new block (green trace) is much better.

The peak-to-peak range of the derivative of the HP worm with the old blocks was 8.1 asec/s. The derivative was pretty well distributed on a Gaussian: 95% of the time, the module of the amplitude of the rate was equal or less than 2"/s. With the new block and worm this limit was lowered to 1.2"/s: this rate is easily handled by autoguiding, provided that the correction period is £ 2 s.

PEC training

One clear consequence of the change to the new block is that now PEC correction is finally working. Infact, even if the periodic error were perfectly regular (which was not, anyway) the presence of a large non harmonic component (about 8 asec) makes PEC useless.

In this test the autoguider was turned on at the beginning of imaging, and after about a minute I started PEC training, which continued for one cycle of the worm (240 s). The resulting trace showed a peak-to-peak tracking of about 3 asec (except for a brief spike). This is a far cry from the 9 asec range I was getting before the change. Indeed, the non-harmonic components were so bad that PEC seemed to make almost no difference. Here the difference is striking as it can be seen in the last period of the worm, when PEC was turned off.

Autoguiding efficiency can be evaluated by the statistics of the tracking error during the autoguided period. The standard deviation of the tracking error was 0.92" in right ascension. Comparing this with the estimate of seeing given by the error in declination (SD=0.71") suggest that the uncompensated tracking error in right ascension was only ±0.59" (seeing error adds up in quadrature to the tracking error.