tPoint, written by Patrick Wallace and sold by Software Bisque, is a software package that works with TheSky6 to augment pointing accuracy all over the sky. It does this by 'modeling' the telescope mount, and predicting the errors anywhere in the sky. The model is built by 'mapping' points in the sky - i.e., telling tPoint where the real stars are compared to where the mount thinks they are (is that clear?). The process of mapping can be done manually or automatically. Manual mapping aims the telescope at an object (e.g., star), then makes an exposure and sees where the star is located in the CCD image. The star is centered in the image, and then "mapped" into tPoint. A model can have as few as 20-30 stars, or as many as hundreds of stars. The more stars, the more accurate the model, and therefore the better the pointing. Automatic mapping can be done with a program such as AAG tPoint Mapper (described separately). Below is an example of a tPoint 'map' containing more than 300 stars. This was done recently using the Paramount in the observatory. Note that most of the stars are within 20 arcseconds of the center of the image, and virtually all of the stars are within 45 arcseconds. |
 |
The above map is then modeled by selecting various parameters in tPoint. This is an art, but basically attempts to reduce the pointing error. Additional terms can be added by tPoint based on harmonics of the various components of the error. The image below shows the modeling window in tPoint. |
 |
Only 4 of the 'conventional' terms are used (including the two polar alignment terms, as these allow measurement of the polar alignment error). Another 12 terms were added to get to this accuracy. The "Fit information" can then be examined to see the standard deviations of each of the terms; these should be minimized to create the most accurate model possible. As you can see in the FIT Information window below, the 16 parameters contribute corrections ranging from 8-120 arcseconds. The righthand column indicates that these corrections are within (i.e., have a standard deviation of less than) about 1.5-3.0 arcseconds of the generated model, based on data from the 300+ stars in the tPoint map. |
 |
Finally, once the model has been built, a natural output is the polar alignment error, seen on the following image. While this can be very accurate with a good tPoint model, I have found that it is very misleading when the tPoint model does not have enough stars (e.g., 30-50 at least, and preferably 100-300 stars, if accuracy below 1 arcminute is desired). Therefore, I do not use tPoint modeling as a polar alignment technique, although it is a good check, once PemPro alignment has already been done. In the Polar Alignment Information window below, the calculated polar alignment accuracy of the mount (as used to generate the tPoint map) is within about 46 arcseconds E-W and 12 arcseconds N-S of the true pole. The Sigma (standard deviation) value indicates that these figures are fairly accurate (i.e., plus/minus about 5-8 arcseconds at the 3-sigma confidence level). |
 |
CAUTION, CAUTION, WILL ROBINSON !!! I initially used tPoint modeling with my LX200 scope, as that could really use some help with pointing accuracy (it wasn't that bad, about 5 arcminutes typically). However, at the time, I didn't know much about tPoint modeling; I would add terms to make the numbers better, but ended-up adding many terms that were not at all appropriate for my type of mount (fork mount). Although this made the model look good ... and the pointing wasn't bad at all (about 2-3 arcminutes) ... it did terrible things to fine pointing with the scope. Unfortunately, I chalked this up to the scope mechanics (backlash, etc.), and didn't really find out why my fine pointing was bad until I bought the Paramount, and found the same thing, until I paid attention to the tPoint model and built one with appropriate terms. If the tPoint model is bad, it can result in trying to make a 1 arcminute move of a star towards the center of the image, and ending up with it moving 2 arcmin AWAY from the center (obviously, this also depends on what side of the meridian the mount is on, and how the centering was calibrated in MaximDL)! This is something that "they don't tell you", and I have never seen published anywhere (except paragraph below, added in January 2011), but something that can be extremely frustrating, if you use tPoint indiscriminantly, and do not focus on building a good model with reasonable terms. |
As of January 23, 2011, I found a post on the SBIG web site that confirmed that these problems can be created using tPoint with incorrect modeling terms. There was a discussions regarding the Paramount ME and MX mounts, and length of time an unguided exposure can be made with these mounts (while retaining round stars). One of the posts, from Stweard Yeung (HERE), described the use of a 250-point tPoint model with incorrectly selected terms. While the pointing accuracy was very good, the tracking accuracy (with ProTrack turned on) was poor. Stewart's web page describes this situation and demonstrates that with proper modeling (correct terms selected) the tracking accuracy is dramatically improved (6 minute unguided exposures look fine). So-far, this has been the only other documented case of someone warning about properl tPoint terms that I have found. |