Sokkisha Subtense Bar

Are you sure you didn't get an extra zero in there? At 100m the angular measurement needs to be good to the order of 0.1 second to achieve 30,000.?ÿ You would be aiming at the targets to within 0.05 mm each. (sorry, I didn't do exact calcs).

I did exact calcs?ÿ

I use a T3, 16-30 sets. I also take 4 pointings of each target per set. Std dev is 0.3ƒ? typ. All sets include 50% increasing micrometer and 50% decreasing micrometer to eliminate micrometer scale error. Micrometrr scale error determined for each set, micrometer proration used for each set. Residuals, tested against the mean, are tested for normal distribution. Rejected obs happen now and again. Std err calculates to 0.08ƒ? - 0.10ƒ?. I use rotating optical plummets to reduce centering error ?ÿ

I have calibrated the bars against a pair of 1 micron m standards, 4x. Starnet reports 10-15 micron 95%. (5 to 10 ppm)

I did a control survey with 10 points with 45 subtense distances; every inter visible line that was max 350. T3 angles and every inter visible point 200-1000 ft. No single angle had angles smaller than 25 degrees, most near 45-70 degrees ?ÿ

A-prior in Starnet I set distances to 1:10k +10mm. Failed Chi. Reduced that to 1:30k +5mm. Failed chi. Reduced that to 1:50k (20 ppm) +2mm. Chi square test passed. 95% confidence regions 2mm typ.?ÿ

Simple loop closure around the perimeter, 4000 ft, using 12 Subtense distances, 250-350 ft, closure was 1:90K. ?ÿRepeated the distances with EDM and differences were very similar to Starnet residuals.?ÿ

Estabished 500 ft baseline with 50 ft stations by Invar, compared with EDM, and further with 2 instrument static phase differential. Subtense distance of 1:30K comparison is routine, but time consuming. Of course 1:20K is observed as well as 1:50K. However, Accuracy of 1:7k+/-, on short, 100 ft or less, is normal, because itƒ??s short.?ÿ

Subtense Bar calibration was critical, and suprisingly large. 3 bars: 382, 270 and 120 microns longer than the assumed 2 m. ?ÿ

Wild documentation cites the accuracy as ƒ??user specifiedƒ??. Do the math, determine std err of the angle, and apply angle the appropriate routine as required, use T3. ?ÿ

An angle of 50 minutes (3000 seconds) is the absolute min. So I hold to 1 degree min, 350 ft. A 10 mm error is a do-over; uncommon.?ÿ

Its not all-weather. Must be overcast, scintillation kept to minimum.

to work out the foibles took hundreds of distances within a fully constrained, over determined network.?ÿ

?ÿ

?ÿ

?ÿ

?ÿ

Wow.?ÿ Time consuming, indeed.

Average Joe wouldn't get that other zero that you earn.

It came up in conversation whether RTK was even as good as Subtense.?ÿ

The white papers cite much you did, 1:5k. Approximate. Just an old down and dirty method. The Wild doc has a very much different take. If calibrated, and turn enough valid angles, normal distribution residual, statistics are the QC.?ÿ

And donƒ??t get seduced by the occasional 2 mm comparison. Evaluate a statistically significant number of data points.?ÿ

So, if it were 1960, and you had to cross a rivine, (bridge control), what are your choices? A chain in catenary or an hour of angles, at night or cloudy conditions, with a right calibrated bar??ÿ

Iƒ??d put Subtense up against any measurement: chain or EDM. 350 lengths for sure, and time, with a one man crew. But a 3 second EDM shot of course is faster cheaper. I tried to imagine in the late fifties just how good Subtense can be, pre-electronic, when the alternative is harder.?ÿ

Of course if your accuracy is limited, use an M16A1. Why there are even tables available...?ÿ

see:?ÿ https://www.globalsecurity.org/military/library/policy/army/fm/6-50/Ch5.htm

?ÿ

BTW, his method is described as a ƒ??hasty survey techniqueƒ?.