NavList:
A Community Devoted to the Preservation and Practice of Celestial Navigation and Other Methods of Traditional Wayfinding
Re: Instumental error?
From: Fred Hebard
Date: 2005 Apr 19, 22:46 -0400
From: Fred Hebard
Date: 2005 Apr 19, 22:46 -0400
Alex, I wonder if you could clarify how the standard deviation, average error and maximal span were computed. As best as I can tell, your data appear to fairly precise, based on the standard deviations; personally, I would prefer that the standard deviations be given to an extra digit. In my experience, which is based primarily on using artificial horizon shots to attempt calibration of a Husun sextant, the development of proficiency is an important component in minimizing error. A few times, I was very accurate and precise in altitude shots (with a Cassens + Plath, not the Husun), but only after about a year and a half of practice. However, it may take longer to develop proficiency in altitude shots than in star-to-star shots or lunars since accurate timing is so critical to altitude shots. I have seen a few densely populated instrument correction curves for extreme late-model Husun sextants (post 1955). The curves can take surprising twists. I regret I don't have one to share with the list. If you look at Ebay listings for Husun sextants, you often will see the calibration data on the label. The late model Husuns (post 1940) were wonderfully imprecise, usually with errors up to 40 or 50" of arc; thus one can get an impression for sextant correction data, unlike those horrible Japanese and German instruments, with no detectable error [;)]. (Husuns built prior to 1940 usually were quite precise). Unfortunately, the label for late model Husuns only shows errors for 30, 60, 90 and 120 degrees of arc; it is only the extreme late-model instruments that have the densely populated correction curve prepared by the British military. On the Yahoo sextants list a year or two ago, an optical physicist described calibrating an Astra IIIB using equipment to which he had access. He mentioned that he fit the "observations" using Fourier transforms, I guess since he was fitting on a continuous "circle." That sort of curve fitting/calibration is way beyond my powers, but I would be delighted to see it explained. Fred On Apr 19, 2005, at 8:59 PM, Alexandre Eremenko wrote: > I am still struggling with my SNO-T, trying to determine > any instrumental error from distances. > (Once George asked: "Did anyone succeed in determining > instrumental error of a modern sextant by ordinary observations"? > I cite from my memory: still cannot manage the search engine > on this list:-( > > I think I did. My SNO-T sextant has a certifricate, which is > filled by +10" everywhere on the arc. But even the seller > confirmed that in the last years of Soviet Union, no one > really cared to measure and they filled the certificates with > random numbers (even did not care to vary these numbers:-) > > Anyway, here are the result of my last distance measurements > as they are. I will appreciate any interpretation of these numbers. > > Index correction. Always from 0 to -0.2', since October 2004. > I do it every time when I do observations. From the Sun > or from the stars. The results are consistent. > All distances I mention are > measured under > ideal conditions from my balcony in perfect weather. > Lunars are reduced with Frank's calculator, star distances with > my own computer program. A light filter is frequently used on the > moon at night, an appropriate one on the Sun, > and no filter on the stars. > > April 19 > 2:00 GMT, 8 measurements of Sirius-Procyon, > standard deviation 0.15', maximal span 0.4', > distance approx. 25d40' average error +0.5' > > 2:20 GMT, 7 measurements of Procyon-Capella, > stabdard deviation 0.2' maximal span 0.6', > distance approx. 51d7' average error +0.4' > Same with Galileo scope, average error +0.3' > > 2.37 GMT, 7 measurements of Sirius-Capella, > standard deviation 0.3' maximal span 0.5' > distance approx. 65d48', average error +0.6 > > 22:45 GMT 8 measurements of Jupiter-Saturn, > standard deviation 0.3', maximal span 0.5' > distance approx 81d53', average error 0.0' > > 23:30 GMT 6 measurements of Sun-Moon distance, > standard deviation 0.2' > distance approx 127d10', average error +0.8' > > This is just a sample. My log book is full of similar data, > and they are consistent in the average, > produced since October 2004. Can anyone help me to > interpret this? > (In the beginning I thought that this is my technique > to blame, but now (after having tried once Bill's new > Astra III) I begin to suspect that this is an instrument error. > > a) it is ALWAYS positive. > b) it ranges from +0.2' to almost 1'. > c) Sometimes it is almost zero, but rarely. > > Based on the recent observations I can make the following table > of corrections: > > 25 38 51 65 68 71 82 90 127 > -0.5 -0.4 -0.4 -0.6 -0.2 -0.1 -0.2 -0.2 -0.8 > > Should I attach this table to my sextant box lid as a > certificate, and use it? > > Any comments? > Alex. >