NavList:
A Community Devoted to the Preservation and Practice of Celestial Navigation and Other Methods of Traditional Wayfinding
Sextant calibration in the workshop
From: Bill Morris
Date: 2007 Dec 24, 14:30 -0800
From: Bill Morris
Date: 2007 Dec 24, 14:30 -0800
It may be of passing interest to members to learn how a sextant may be calibrated in practice in the workshop, though not with equipment readily available to the amateur. Two autocollimators and a surface plate are needed. A rotary table is an extra refinement and is not strictly necessary. The rotary table is not used as part of the calibration process, but to rotate the sextant to the desired position easily A collimator is simply a lens with an illuminated graticule at its focus, so that a parallel beam is projected and the image of the graticule appears as if at infinity. If the beam is intercepted by a mirror that is normal(square to) the optical axis of the collimator, it gets reflected right back to where it came from. If the mirror is at some angle to the optical axis, the reflected image is displaced. In the autocollimator there is an optical arrangement called an optical micrometer that allows the reflected image to be viewed and its displacement to be measured directly. The amount of displacement is dependent only on the focal length of the collimating lens and the angular displacement of the mirror, not on the distance of the mirror from the instrument. Unlike length measurements, angular measurements require no official standard to be kept, as there are always exactly 360 degrees in a circle and 180 degrees in a straight line. This latter fact is exploited in the calibration method. The two autocollimators are first set up on a surface plate, a polished slab of granite about 1000 x 600 x 100 mm and flatter to better than 4 microns, so that their vertical cross-wires coincide with each other. One is designated as the "fixed autocollimator" and the other the "moving autocollimator" The sextant is adjusted on a suitable stand so that the plane of its arc is parallel to the surface plate and is set carefully to, say, 15 degrees or some other factor of 180. Its index mirror is then placed in the light path of the moving a/c and the whole sextant rotated until the reflected image is seen in the a/c to coincide with the vertical cross wire. The sextant is then set to 0 degrees, taking great care not to move it on the surface plate, and the moving a/c moved until the reflected image again coincides with the vertical wire. The sextant is reset to 15 degrees and so on. After repeating this 24 times, the sextant will have rotated through a nominal 180 degrees and the axis of the fixed autocollimator will then be normal to the index mirror of the sextant. Any excess or deficit can be measured by the fixed a/c. The excess or deficit represents 24 times the error of the sextant in moving through 0 to 15 degrees; and the axes of the autocollimators in their final positions will be at a known angle of 7 1/2 degrees plus or minus this error(The index mirror rotates through 7 1/2 degrees, though the sextant index reading moves from 0 to 15). This known angle can then be used to calibrate the rest of the scale, from 15 to 30, 30 to 45 degrees and so on. Alternatively, if many sextants are to be calibrated, one of the a/c s can be adjusted to remove the error and bring the axes to exactly 7 1/2 degrees, allowing the sextant error of any step to be read off directly from one of the a/c scales. Needless to say, once the a/c s reach their final positions they must not be disturbed. Errors for values greater than 15 degrees that are factors of 180 can also be estimated ie. 120 degrees(3 steps, see attachment), 90 degrees(4 steps) 60 degrees(6 steps) and 30 degrees(12 steps). I am not brainy enough to be able to state with certainty what effect the number of steps has on the probable error of the result. Perhaps Alex Eremenko can help? Members may wonder about the precision of an autocollimator. A basic Hilger and Watt Microptic autocollimator's least graduation is 0.2 seconds and with a little practice and a mirror of good quality, readings can be repeated to within about 0.3 seconds. Even at first acquaintance, repeatability to within 1 second is easy. Using a photoelectrical readout, precision is about five times better. Its accuracy is of a similar order. When measuring to parts of a second, great care is needed. The instruments need to be given time to reach the ambient temperature, overnight if possible, touching them should be kept to a minimum and accidentally brushing against an autocollimator when stepping out the 7.5 degree increments can mean having to go back to the first of 48 readings again. The setting error of the sextant is greater. Its micrometer is altogether coarser than those of the autocollimators, its thimble is much smaller, making for larger reading errors, the oil film separating the parts will vary in thickness depending on temperature and speed of setting, the whole instrument is liable to move when resetting it unless very great care is cultivated and taken and the error also contains the autocollimator reading errors. I may have left something out! Even so, in a series of 24 repeat readings with a SNO-T sextant, 95 percent of readings can be expected to fall within a range of 3.7 seconds. In setting the sextant micrometer, it must always be rotated in the same direction to avoid backlash errors. In sextants like the SNO-T and Freiberger, it does not matter in which direction, as long as it is always the same, but in the BuShips Mark II and sextants of similar construction including Tamaya and clones, there is only one correct setting direction. As it has only one thrust bearing, preloaded by a rather weak spring, it must be rotated so as to load the thrust bearing, not the spring. In the MkII, this happens to be the direction that decreases the reading of the instrument. The micrometer can be calibrated directly against one of the autocollimators and here the superiority of the SNO-T over the Mk II shows itself. The maximum deviation of the SNO-T from 0 to 60 minutes in 5 minute steps was 1.6 seconds, compared to 6.5 seconds for the Mk II. I am now in a position to answer George Huxtable's question "Has anyone on this list, by measuring star-star distances or by any other method, ever discovered reproducible errors, outside the terms of a calibration certificate or maker's warranty, in a sextant? Has anyone made calibration measurements of his own, in which he has more confidence than in the manufacturer's scale readings, corrected as necessary by the box certificate? And if the answer is yes, what's the magnitude of those errors?" According my SNO-T manual in English translation K52.514.004 TO of 1976, "Instrumental accuracy within angle range 0 to 120 degrees - +/- 6 seconds" and "Accuracy of drum scale reading - +/- 6 seconds". Alex Eremenko's Russian manual gives "Instrumental error within the range of measurement - 12 seconds" I have already dealt with the micrometer. My 1981 instrument meets Alex's specification but not mine: 15d. +1"; 30d +1"; 45d +4"; 60d +7"; 75d +8"; 90d +11"; 105d +9"; 120d +8". My Mk II s/n 14176, 1942 in near-new condition with a certificate from Long Beach Shipyard dated 7 January 1986, does well too, except above 90 degrees. I give the original certificate figures followed in brackets by my own, which are each based on the mean of three careful readings: 15d +3"(+1"); 30d -14"(-12"); 45d -17"(-16'); 60d -25"(-22"); 75d -30"(-31"); 90d -33"(-40"); 105d -12"(-31"); 120d 0"(-30"). For most practical purposes except lunars, the differences are insignificant. If someone can tell me how to add attachments to the list I will post a diagram and a photograph to make the process plainer. Bill Morris --~--~---------~--~----~------------~-------~--~----~ To post to this group, send email to NavList@fer3.com To , send email to NavList-@fer3.com -~----------~----~----~----~------~----~------~--~---