Maybe there is no answer to this but here goes:
We did a Survey which involved a network of conventional measurements with one triangular GNSS session on a northeast-southwest oriented baseline on the west and a single point on the east side.
So we put the data into TBC which can handle it because we set up the work flow to start on the GNSS baseline. I give it centering errors, etc. TBC almost always fails Chi Square. No matter what I do. So we push all the stars to reset the a priori estimates and run it again, of course it passes Chi Square because it's a self fulfilling prophecy. The error ellipses are a bit rough, some of them are on the 0.05' level.
So I put the exact same data in Star*Net and run it (processed GNSS baselines with conventional run simultaneously) using the same centering errors, etc. Star*Net thinks thinks this is the most wonderful survey it has ever seen, it's going to submit it to the Star*Net hall of fame, it has a network-gasm. It fails Chi Square on the low end which according to the manual is not a problem. You can tighten the standard errors to make it pass but it won't change the answer. The error ellipses are all sub hundredth except for a few that aren't much over 0.01'.
I put both sets of adjusted coordinates into Excel and compare them. They are virtually identical.
Dave - Take a look at the release notes for TBC 3.60, they've enhanced the network adjustment (finally!!). You can now set uncertaincies for individual points and constrain azimuths and distances, among other things. I haven't had a chance to dig into it yet.
http://trl.trimble.com/docushare/dsweb/Get/Document-790348/Release.pdf
From the release notes:
The azimuth thing is probably because I complained to CSDS about that and CSDS turned it around saying it came from Caltrans ;-). Trimble really wants to keep Caltrans happy.
Star*Net and SurvNet are products created for surveyors to greatly assist in their daily undertaking. I can not believe a surveyor had anything to do with TBC. Trimble should be ashamed.
lmbrls, post: 337318, member: 6823 wrote: Star*Net and SurvNet are products created for surveyors to greatly assist in their daily undertaking. I can not believe a surveyor had anything to do with TBC. Trimble should be ashamed.
When was the last time you saw TBC? And did you see Dave's comment that the coordinates were virtually identical? People like what they like, and that's fine... I demoed Star*Net earlier this year; I thought it looked like something straight out of 1991. To each his own...
Could part of the difference be due to the assumptions built into the programs regarding whether it is to accept your estimates of std errors or to come up with its own scaling and use yours only for proportioning the error between measurements?
See my post in this thread:
Relative Positional Precision - Why did I fail?
Also see late in this thread for the discussion that taught me this
Least Squares Ellipse Confidence Calculations?
With TBC you have the option of setting the standard errors to a user setting or an external source. For instance, for GNSS vectors, I can set default standard errors like 0.02'+1ppm HZ and 0.04'+1ppm V; or I can tell it to use the covariance data and precision estimates from the baseline processor, which is the default. As Dave noted, using the estimates from the baseline processor generally causes the chi square test to fail on the initial pass.
Lee D, post: 337328, member: 7971 wrote: When was the last time you saw TBC? And did you see Dave's comment that the coordinates were virtually identical? People like what they like, and that's fine... I demoed Star*Net earlier this year; I thought it looked like something straight out of 1991. To each his own...
I gave up on TBC about a year ago. My biggest complaint is how awkward it is to edit a conventional traverse. We had a traverse that our field crew admittedly screwed royally. We re-observed several setups. We had the Trimble support person come to our office and spend a day. He could not solve the problem. We were working with a large company that uses TBC. They put their expert on the problem and he fixed it on the second day. We had already put everything in SurvNet and fixed everything in two hours. The difference in coordinates between the two softwares was very small. We at that point upgraded our field crew and software.
Lee based on your previous post, you appear to have mastered TBC. You seem to answer 90% of the many TBC questions. If you were to offer support, I would definitely pay you next time we are forced to use TBC.
While TBC gives a large amount of statistical information for each observation, I prefer software that is flexible and intuitive. Being straight out of 1951, I will admit to having a bias for platforms that follow convention. At the end of the day, we want good data for the basis of our work. How we get there, well "to each his own".
I have to agree that the support for conventional data has been weak at best in TBC; most of our work is GPS so I haven't had to deal with that as much. But Trimble must have paid attention to comments and stories like yours above, because in the last couple releases they've taken steps to address the issue. I haven't had an opportunity to test any of the new features, but it seems like they've improved the capability for editing things like setups and prism/atmospheric corrections. I was also interested to note that the release notes for 3.60 say that they'll now support other manufacturer's level data - if this is true, and actually works, that's a definite step in a different direction for Trimble. I like the level editor in TBC and they're doing their customers a service by supporting all levels instead of just theirs.
Okay, I resolved the issue with the help from Lee.
The problem was I had entered the centering errors but it still was set to Imported Files. I changed that to Project Settings and it looks just like Star*Net.
I also upgraded to Release 3.60.
That's a large part of the attraction of StarNet. It worked in 1991 and doesn't need to be fiddled with every year. Actually, 'round about 2000 was a big look and feel change, when it went from DOS to Windows. Recently Microsurvey has attempted to monetize their investment by adding features. While I like some of those features they have also opened some cracks in the rock solid armor of StarNet. I hope they don't break it.
