NavList:
A Community Devoted to the Preservation and Practice of Celestial Navigation and Other Methods of Traditional Wayfinding
Re: SNO-T tests
From: Bill B
Date: 2005 Dec 13, 16:37 -0500
From: Bill B
Date: 2005 Dec 13, 16:37 -0500
Frank Trying to get a handle on this and have a few questions and observations. I assume it is important to use a flat-screen CRT or LCD monitor so distance does not change due to curvature of a standard CRT. I also assume you want the screen perpendicular to the line of sight on both the horizontal and vertical to avoid convergence. And old photographer copy-stand trick is placing a small piece of mirror in the center of the base, and aligning an single-lens reflex camera so the lens sees itself in the mirror. An assistant could do that with a monitor in the center of a monitor. Is the line if sight in this case considered to be from the scope, rather than the index mirror? When you wrote of distance from the monitor, what point on the sextant is used as the reference point? Confused on, "I wrote a very simple piece of software that displays two vertical white lines, one above the other, which can be separated at regular pixel intervals (I used 40 pixel jumps)." Does this mean that the ends of the line segments do not overlap, so you are butting one line segment up to the other when you align them? Would the target be similar to below? (Connect the dots vertically.) . . . . . . Referring to, "Then I compared two runs of these measurements with the linear increase that I would predict if the sextant has no micrometer error." How did you predict the values? In my sextant-as-rangefinder experiments, I found the parallax changed (front silvered index mirror) with angle enough to throw things off considerably. I also would expect the distance from the point seen on the index mirror to the monitor to change, but not enough to account for more that a couple of seconds error over 2 degrees. BTW, if I can get a handle on the above, it struck me that instead of using a monitor I could use graphics/CAD software and produce target and have played out at 2400 dpi. Thanks Bill > The setup: > I wrote a very simple piece of software that displays two vertical white > lines, one above the other, which can be separated at regular pixel intervals > (I > used 40 pixel jumps). I set this up and ran it on my laptop which I placed > at the far end of a room about 25 feet away (so that I could focus the > telescope). I placed my sextant on its side and carefully measured the angle > between > the upper and lower line, repeating at each interval. The angles ranged in > roughly 5 minute of arc steps from 0 to 2 degrees (two minutes or so for each > observation and the walk across the room to move the line over one step so > over three-quarters of an hour of work for each run). Then I compared two > runs > of these measurements with the linear increase that I would predict if the > sextant has no micrometer error. Sure enough, there was a nearly sinusoidal, > cyclic difference between the actual observations and the linear prediction. > The amplitude from top to bottom was 0.7 minutes of arc --exactly the same > magnitude as the mysterious errors I had seen previously. I have not yet had > the > opportunity to try correcting real observations, of lunar distances e.g., > with my new micrometer correction table, but I'm optimistic that this will > render this instrument essentially perfect (*after* correcting for arc error > and > micrometer error). Lots of fun! I highly recommend trying this. > > By the way, as a bonus, this is an easy, extremely accurate way to test for > backlash error. With the same sextant, I found zero backlash error every time > I checked for it. It was also interesting how repeatable these measurements > turned out to be. Ninety percent of the time the angles between the lines were > the same to 0.1 minute of arc accuracy when re-measured, and over 95% were > no more than 0.2 minutes different when re-measured.