I just did a test of the angular accuracy of my S6, which is spec'd at 1" angular accuracy.?ÿ
Reason: I have a client who is specifying 4 sets of angles in a survey. There are two levels to this survey, a control network of setup points and then a secondary network of monitored points, to be adjusted separately. They want 4 sets for both types of points. We have been doing 4 sets in the control network and 1 set in the monitoring network for 14 years now. I should add that all of the monitored points are observed from a minimum of two different control points. So I proposed 4 sets and 2 sets, but they want 4 sets. In this network distances range from 5 meters to about 150 meters. I am using a Trimble S6 high accuracy total station. We will be doing hundreds of setups to thousands of points.?ÿ
So I did a test. Turned 16 sets between a backsight (130 meters) and a foresight (140 meters). Here are some stats I computed. All values are in gons, (0.0003g=1 arc second).
stdev of 16 sets (32 angles): 0.000418 (1.35"), but that is contaminated by the D+R split. The difference between the direct average and the reverse average was 0.000416 (1.3")
If I average each set (1 set=D+R) and then compute the standard deviation (16 angles), I get 0.000253 (0.8").?ÿ
The foresight is 140 m away. If I take each set's residual and compute the distance subtended at 140 meters, all 16 are under 1 mm. The criteria for the adjusted error ellipses is that they must all be less than 9.5 mm, so I maintain that 1 mm is not significant. In any case I cannot win the argument, just wanted to prove to myself the accuracy and I was a bit surprised. I would say that the spec'd 1" accuracy has been proven, and it is a combination of pointing and reading error. The 4 sets would be more appropriate if I was manually pointing and/or manually reading the circle. Other errors that affect the results like instrument and target centering errors are not improved by multiple sets.?ÿ
I also have some data sets that I did with a T2 and a T3, I need to dig those out and see what difference there might be. I would expect the reading error (manual reading of circle) and pointing error (me pointing to precise traverse target) would make the standard deviations of these sets a bit higher.?ÿ
?ÿ
and that's what makes this site so invaluable to the practicing surveyor.?ÿ Real world testing that checks the 'spin' on the manufacturer marketing sheet
Trimble and Leica have always stated their specifications conservatively.
If I'm remembering correctly Trimble doesn't expressly state an accuracy specification for the robotic pointing on the S6.?ÿ They didn't start stating the pointing accuracy until this latest odd numbered series.?ÿ That's not to say the S6 didn't point accurately, simply that Trimble wasn't quite willing to claim it did just yet - my thinking is they wanted to observe them in the wild for a period.?ÿ I did a ton of monitoring, column line staking, and ALTA surveys with an S6 and wouldn't hesitate to trust the robotic pointing.?ÿ?ÿ
Better than I expected.
The problem with specs like that, is it leads some to believe that the accuracy of their surveys is 1", forgetting that there are sources of error other than the pointing error. Of course, the most of those errors won't be reduced by turning four sets.?ÿ
