I remembered I still have a set of DC files from my days of doing a lot of high accuracy architectural and steel stakeout. On this particular project I had to provide certified accuracy layout and as-built for a robotic surgical suite. I was using a 1" Trimble VX in close quarters (under 200 feet in all directions).
The specs for the instrument said 0.01' plus 2 ppm for the EDM, but bench testing revealed much better than that under real applications (indoors anyway). Here is some pertinent data from the LS Analysis (unconstrained):
Adjustment Settings
Set-Up Errors
Terrestrial
Error in Height of Instrument 0.000 ft
Error in Height of Target 0.000 ft
Covariance Display
Horizontal:
Propagated Linear Error [E]: U.S.
Constant Term [C]: 0.000 ft
Scale on Linear Error : 1.960
Three-Dimensional
Propagated Linear Error [E]: U.S.
Constant Term [C]: 0.000 ft
Scale on Linear Error : 1.960
--------------------------------------------------------------------------------
Adjustment Statistics
Number of Iterations for Successful Adjustment: 2
Network Reference Factor: 0.93
Chi Square Test (95%): Passed
Precision Confidence Level: 95%
Degrees of Freedom: 57
Total Station Observation Statistics
Horizontal Circle Reading:
Reference Factor: 0.79 Redundancy Number: 17.14 A Priori Scalar: 1.34
Vertical Angle:
Reference Factor: 1.06 Redundancy Number: 20.93 A Priori Scalar: 4.72
Slope Distance:
Reference Factor: 0.90 Redundancy Number: 18.93 A Priori Scalar: 0.39
--------------------------------------------------------------------------------
Control Point Constraints
Point ID
Type
North ?
(US survey foot)
East ?
(US survey foot)
Height ?
(US survey foot)
Elevation ?
(US survey foot)
Fixed = 0.000003(US survey foot)
Adjusted Grid Coordinates
Point ID
Northing
(US survey foot)
Northing Error
(US survey foot)
Easting
(US survey foot)
Easting Error
(US survey foot)
Elevation
(US survey foot)
Elevation Error
(US survey foot)
Constraint
10140000 436.878 0.005 1493.253 0.006 0.081 0.001
10140001 398.728 0.002 1515.218 0.003 0.095 0.001
10140003 414.404 0.003 1519.011 0.002 0.000 0.001
10141045 447.382 0.007 1484.763 0.008 0.093 0.003
10141046 406.972 0.008 1479.992 0.003 0.150 0.002
10141047 422.900 0.005 1493.333 0.004 3.714 0.002
10154026 396.879 0.005 1470.019 0.003 0.125 0.002
10154027 447.601 0.003 1512.045 0.006 0.097 0.002
10172000 426.608 0.003 1492.542 0.003 3.567 0.001
10172004 393.445 0.003 1509.554 0.004 9.836 0.002
10172005 393.422 0.005 1475.907 0.004 9.836 0.002
10172006 460.476 0.005 1476.427 0.007 9.833 0.002
10172009 427.134 0.005 1481.582 0.004 -0.125 0.003
10172013 428.597 0.003 1492.590 0.003 3.642 0.001
10173000 393.571 0.003 1491.560 0.003 3.572 0.001
Coordinates from a free adjustment should only be used for analysis of the inner accuracy of the network. They should not be distributed as final results.
Error Ellipse Components
Point ID
Semi-major axis
(US survey foot)
Semi-minor axis
(US survey foot)
Azimuth
10140000 0.009 0.002 50°
10140001 0.004 0.002 56°
10140003 0.004 0.002 29°
10141045 0.013 0.003 51°
10141046 0.009 0.004 178°
10141047 0.008 0.003 39°
10154026 0.007 0.004 163°
10154027 0.007 0.003 97°
10172000 0.005 0.002 54°
10172004 0.005 0.004 65°
10172005 0.007 0.004 151°
10172006 0.011 0.004 58°
10172009 0.007 0.004 40°
10172013 0.005 0.002 54°
10173000 0.005 0.003 132°
The copy/paste loses some formatting, but I think you get the idea. Some of the points that drifted out to near a hundredths worth of error were for elevation only.
I usually don't care about reference factor or Chi Squares all that much, but for some reason the adjustment always looks a little better and makes more sense if I allow a little more weight to the distance component. I know this project was important enough that I performed all of the aiming manually during setting control. This data includes a floor flatness survey that would have been conducted robotically, so you're probably seeing some variance between autolock and eye at close range with the horizontal and vertical component. For some reason I remember that I was never able to get a vertical collimation adjustment on this robot that I was happy with.
Since I did this project all via resection and with a mini-prism I allowed for no error in instrument or target height, no error in instrument setup, and 0.002' in target centering. The mini-prism when tilted to the outer edge of the inner bulls-eye introduces less than 0.005' error.
Not sure what's causing the text strikeout...
