I always looks at this part of the listing report first:
Adjustment Statistical Summary
==============================
Iterations = 2
Number of Stations = 15
Number of Observations = 66
Number of Unknowns = 28
Number of Redundant Obs = 38
Observation Count Sum Squares Error of Std Res Factor
Angles?ÿ ?ÿ ?ÿ ?ÿ ?ÿ 16?ÿ ?ÿ25.069?ÿ ?ÿ 1.650
Distances?ÿ ?ÿ ?ÿ49?ÿ ?ÿ 2.425?ÿ ?ÿ ?ÿ0.293
Az/Bearings?ÿ ?ÿ1?ÿ ?ÿ 0.000?ÿ ?ÿ ?ÿ0.000
Total?ÿ ?ÿ ?ÿ ?ÿ ?ÿ ?ÿ ?ÿ66?ÿ ?ÿ27.494?ÿ ?ÿ 0.851
The Chi-Square Test at 5.00% Level Passed
Lower/Upper Bounds (0.776/1.224)
One wants those numbers I've bolded to approach 1. After this I look at the individual distance residuals, then the angle residuals.?ÿ Then the rest.?ÿ
The angles are the weakest part of this traverse, but they aren't blundered. Remember that a 5" instrument spec is for pointing error, and each angle is composed of a minimum of 2 pointings. So a 10" angle error is appropriate for an angle composed of a single BS/FS observation using a 5" instrument that is in perfect adjustment and used under laboratory conditions.?ÿ As Lee suggests, doubling of the angles would be a good thing.?ÿ The mean of multiple sightings will, in theory, be better.?ÿ?ÿ
?ÿ
@rover83?ÿ
I have 10 years (about 3 years now as a residential construction layout party chief - doing the odd boundary locate as well) but I had several years working with people who didn't have enough knowledge to get me where I would like to be.
I became party chief when the previous construction guy quit. I made one mistake on a cut stake (missed a digit on the calculator and didn't catch it). Then I came up with a solution to foolproof the procedure. Now my manual calculation on paper is my final check before I'm done with one cut stake. Before that I create a calc point with the proposed grade, stake it out with prism only....look at the cut value in staking...then sideshot it. Then either do a reflectorless shot or a full 1.5m rod height check shot (in case the robot is locked on something else) then inverse the 2 shots, then inverse to the calc point, make sure it matches what I already saw when I staked it out... write it on the stake, take a picture of it...then record my setup, check, both cut stake shots and do the math on paper and make sure it matches what the inverse was.?ÿ
After doing it that way we had one excavator screw up a basement by 1 foot. I triple checked my work... Took it back to my boss, he looked at what I did and said "Yeah we're not paying for this one - good job, keep doing it that way".
Sometimes there are many right ways to solve a problem.?ÿ
I always looks at this part of the listing report first:
Adjustment Statistical Summary
==============================Iterations = 2
Number of Stations = 15
Number of Observations = 66
Number of Unknowns = 28
Number of Redundant Obs = 38Observation Count Sum Squares Error of Std Res Factor
Angles?ÿ ?ÿ ?ÿ ?ÿ ?ÿ 16?ÿ ?ÿ25.069?ÿ ?ÿ 1.650
Distances?ÿ ?ÿ ?ÿ49?ÿ ?ÿ 2.425?ÿ ?ÿ ?ÿ0.293
Az/Bearings?ÿ ?ÿ1?ÿ ?ÿ 0.000?ÿ ?ÿ ?ÿ0.000Total?ÿ ?ÿ ?ÿ ?ÿ ?ÿ ?ÿ ?ÿ66?ÿ ?ÿ27.494?ÿ ?ÿ 0.851
The Chi-Square Test at 5.00% Level Passed
Lower/Upper Bounds (0.776/1.224)One wants those numbers I've bolded to approach 1. After this I look at the individual distance residuals, then the angle residuals.?ÿ Then the rest.?ÿ
The angles are the weakest part of this traverse, but they aren't blundered. Remember that a 5" instrument spec is for pointing error, and each angle is composed of a minimum of 2 pointings. So a 10" angle error is appropriate for an angle composed of a single BS/FS observation using a 5" instrument that is in perfect adjustment and used under laboratory conditions.?ÿ As Lee suggests, doubling of the angles would be a good thing.?ÿ The mean of multiple sightings will, in theory, be better.?ÿ?ÿ
?ÿ
I'm not fond of this particular set of data, too loose.?ÿ Like you say it is not blundered but it could be much better.
When I do something like this then feed it through StarNet I get really small Error Factors, StarNet really strokes my ego.?ÿ TBC isn't as optimistic as StarNet, it's more like my Wife, well sure that looks okay but you didn't get these other things done.
a 5" instrument spec is for pointing error, and each angle is composed of a minimum of 2 pointings. So a 10" angle error is appropriate for an angle composed of a single BS/FS observation using a 5" instrument
In theory the standard error should grow by the square root of the number of pointings. 5 x sqrt(2)=7.1
Set up at 405 bs 404 it looks like there are some issues.
Line 262 shows an angular error of 8 minutes 56 seconds on a check shot
Line 276 then shows 4 minutes and 06 seconds
?ÿ
Did something happen at that set-up??ÿ Depending on which way it swings, that looks like it would knock out about half the mistake.?ÿ But it still wouldn't be acceptable closure.
Making 476 and 483 match 400 and 401 then working backwards to a different calculated position for 432.
It seems like it would be an odd coincidence that you can work back from the end point through point 432 and then end up with a 7 minute angle between as measured 432-405-backwards calculated 432 as this is exactly where your raw data shows some odd back sight check errors.
While you can massage the data to work, I would highly recommend going back and setting up at 405, back sighting 404 and turning 432 to see if the strange back sight checks that show up were actually carried forward in your work.
Was there something unstable about setting up at point 405?
?ÿ
@lurker I have had that same issue on a TSC3 with older version of Trimble Access. It does that after a reboot. And even if it does not have a reboot. The first time I saw it I was making my cheat sheet for a guy to re observe some GPS observations. All I had was the raw data and I wanted to make sure we had a 4 hour gap when I sent him back out and some with high precisions and poor PDOP. So I made a quick spread sheet with time stamps and saw what you are seeing in the raw data. So I began when i was operating it writing down every time I did a reboot or it did one by itself randomly. With his battery issue I bet he has a board about to go out in that Data collector honestly. I had mine into the shop for a lot of this and they told me the board was about to go and they had no more. I cringe every time I have to grab it off the shelf. 2 hours of work turns into a full day many times because it reboots or locks up. Or will turn off and it takes an hour to get it back on.
I've done a little editing of Dave's data file, mostly formating and removing some duplicate shot records. Also some point renumbering for "closure".?ÿ The angles on shots on 407/480 (presumed to be the same point) are messy enough to blow up the chi-square.?ÿ ?ÿ?ÿ
There is no need to rerun the whole thing, but the top 4 or 5 of these occupations could be re-done, starting with 404 and 475. Plus a closing angle at 407/480.?ÿ This is what LS does for you - tells you were to look for the redo-s?ÿ
This is not to question the least squares or Dave and Norman's solution, but I still have a much more basic concern about your data.?ÿ If you are still checking in on this message you posted, I would love to know what was happening at point 405.
From your raw data when set up at 405, here are time stamps and issues:
12:30 Everything looks good
13:49 angle error of 8'56"
14:15 check shot - delta north 0.58 delta east 0.38, observed angle right 8'56" (should be zeroed on back sight)
?ÿbetween 14:15 & 14:16 angle error 4'06"
14:16 Everything reset back sight check good
17:21 BS check delta north 0.3 delta east 0.15
Everything reset BS check good
time stamp not known when you shoot 432
18:30 occupy 432
To me, it looks like you drifted out of level very badly within an hour of setting up at 405.?ÿ Checked it and reset everything.?ÿ Drifted out of level again within 3 hours.?ÿ Reset.?ÿ Then potentially shot 432 anywhere from just after resetting to an hour after (same time frame of original issue).
@leegreen brother this company is the ones who want the points this way. When i started working for them i would turn in my points with traverse starting at 51, eips at 100s and sideshots at 200's and they didnt like that. Its funny how ppl are so quick to be a smartass but not really have the entire picture. I do agree with repping the angles in. Its something we did when we would turn angles on traverses by hand but never had an issue with the robot turning angles before this situatiion. So not a bad idea to start implementing that again. Thanks
@jon-payne so this setup 405 was beside a creek, 404 up a hill through high trees. 432 was across creek. We had to drive a long way just to get to other side of creek, take down several fences, cut line, etc. So during that time I had the owner who was out there for a while point the instrument at me or something. but hen he did I think he bumped the gun bc i was all out of level before setting 432. So i do remember having to tighten up the level and get a new backsight before setting 432. But when I get out of level the data collector tells me out of tilt and wont let me take a shot like that. So at time stamp 17;21 looks like the last time after re leveling for setting zeros and shooting 432 in. And at 1721 there was no angular error. Also when the robot goes to shoot a backsight and a leaf or something is in the way, it will bounce in and out of a locked state and can probably show craziness in the raw data, i dont know. But before I continue with anything and leave my backsight screen i look through and make sure it indeed is locked onto the prizm and not a rear view mirror and that its not bouncing around due to an object being in the way. Now there was a small hole occupying at 405 and backsighting at 404 through which to see and could be the reason i was checking the backsight so much. if theres anyhing making that crosshair jump around I will go cut it and keep rechecking until i see the angle stabilize and bs check under a hundreth most every time. I got to shoot in pt 405 today before it rained me out so that will be my first setup tomorrow..
@norman-oklahoma so 407 is a power pole shot taken at the top of the pole reflectorless as a sideshot. 480 is that same pole taken from the other side of the property to make sure it was shot in well as the first shot was through some trees so i wanted to make sure the prizmless feature didnt shoot something 10' away. Does this make sense??
@norman-oklahoma 400occ bs at 401 and fs at 404 was first leg of traverse. closing points were 476 at 400 and 483 at 401.?ÿ
So after looking at this starnet and chi squares and all has it shown any busts that can account for a 1.6' closure? Its hard for me to make heads or tails of the data.?ÿ
@ricklife here is the traditional traverse closure:
Traverse Closures of Unadjusted Observations
============================================
(Beginning and Ending on Adjusted Stations)
TRAVERSE 1
Error Angular = 296.46 Sec, 12 Angles, 24.70 Sec/Angle
Error Linear = 0.4455 N, 0.0520 E
Horiz Precision = 0.4485 Error in 3877.3991, 1:8646, 115.67 PPM
From To Unadj Bearing Unadj Dist
400 401 N66-30-18.23W BS
400 404 N20-11-41.91E 233.9714
404 405 N58-02-33.22E 265.8798
405 432 N82-02-56.58E 395.7595
432 434 S52-37-37.43E 460.6778
434 442 S71-04-52.41E 281.1741
442 449 S45-45-15.98E 122.9036
449 453 S25-20-38.04E 346.7069
453 462 S03-00-36.13W 132.6515
462 473 N67-34-22.89W 420.5815
473 475 N76-50-01.28W 621.6645
475 400 N71-49-44.04W 595.4285
400 401 N66-30-18.23W FS
