Yeah that's strange. Why am I sitting there for minutes when I can apparently get the same data in a few seconds?
It's 2024 and there are videos for surveyors on how to do a site calibration?
I started watching and thought it was 1996 again.
As far as the observed control RTK point. If I ever was going to do a something as regressive as a site calibration it would only be done using static data.
Site Calibration, Holy Moly.
How is that still a thing?
How about you explain to all of the plebs how you do things since site calibrations are apparently so bad.
I'm with @MightyMoe on this one. a properly setup site shouldn't require a calibration. We haven't calibrated for a very very long time. I do understand the need for it though, but that seems to mainly be driven by the software, especially for the machine grading gps setups.
The Trimble method you are skeptical of is very nearly the same thing as the Javad process.
A "Kalman Filter" is basically the name of the process that you describe for Javad.
The math of the solution improves as the algorithm iterates, and converges on the "best" solution.
In Trimble, you just don't get to see it or review it.
Javad is still storing a single position of the solution generated by the weighted measurements, just like the Trimble.
In GNSS processing, resolving the fractional part of the pseudorange was quite easy and very precise. Being sure of the whole number of wavelengths was more difficult. That's what "Fixed" used to mean, we know the whole number of wavelengths in the pseudorange, and thus we had the distance to the satellite, for any given number of satellites. The "position accuracy/precision" was not really considered.
Current RTK GNSS positioning solutions have moved away from the "Fixed/Float" nomenclature because those terms were arbitrary with respect to the accuracy/precision of the position, but made the user feel better. Current survey grade tools now reflect the measurement convergence of a solution on the most probable position with the accuracy/precision of some statistical measure with respect to many other variables that affect precision/accuracy, like all the DOPs.
He’s saying the Trimble people told him that despite being stationary for xxx epochs of RTK measurements, only the last (best?) measurement is being stored.
Yeah....that's absolutely incorrect and "Trimble people" probably means "rando dealer person". A glance at the data in TBC will show the number of epochs in the vector solution.
For observed control points, the number of epochs set in the survey style must be met sequentially, and all of them must be within the tolerances set by the user. If it's set to Auto, Access uses the expected precisions from the connected receiver, and if there are user-defined tolerances, those will be overridden.
For topo observations, there are two conditions that must be met for storage, time and number of epochs. The receiver must collect for the time specified, plus the number of sequential epochs that meet expected precisions.
With my Javad gear, all of the epochs are weighted and go into the solution, and you can watch the resulting position change as each epoch is collected.
Same thing with Trimble, although you don't see the final position changing during collection, only the realtime precisions.
@WA-ID Surveyor I was told early on that the systems running machine control needed to calibrate, I've also been told that more recent systems can handle projections and geoid models. I don't know enough about their equipment to say. So apparently machine control may or may not need to calibrate. That doesn't mean the surveyor does.
What we do is give out control for their construction purposes. It seems to work well, as long as my control is tight to their machine data all is good. So far everyone is happy. None of my control is ever calibrated so we don't need to do anything. I've got a dozen subdivisions in various stages. The first job is boundary and everything is tied to our projection data and everything follows from there. If the construction company wants to calibrate, that's on them.
But, any XYZ data can be projected to a geodetic system if the internal geometry and elevations are accurate, if it isn't then it's of no use anyway and needs to be scrapped.
Some years ago Dr. Herb gave part of a seminar about how to do it.
The violence done to the data by a calibration isn't necessary anymore with modern Geoid models.
Another thing...
Since you intend to submit RTK vectors to OPUS Projects, you will probably want to set your survey stlye up to record the QC2 (variance/covariance matrix) associated with the recorded rtk vectors.
QC1: DOP & Time.
QC2: Variance/Covariance Matrix.
QC3: Error ellipse(derived from QC2).
This may help:
https://help.trimblegeospatial.com/TrimbleAccess/latest/en/Home.htm
And back to your original question about precision warnings:
https://help.trimblegeospatial.com/TrimbleAccess/latest/en/GNSS-survey-messages-and-warnings.htm
He indicated he was using a rental.
No telling how that thing is set up.
Turning the screen brightness down solved my power problem -- I'm now getting through even long workdays with battery to spare.
The dealer offered to provide extra batteries -- at $3 per day each. If the screen brightness thing hadn't worked I was going to buy the car charger (not a battery bay, but the thing that plugs into the TSC7) for $255. I figure I could sell if for half that at the end of the project and be better off than renting extra batteries.
You could pick up an inverter much cheaper and just use that and the charger you currently have. I guess dealers do things differently all over. I had to rent gear often and always received a TSC7 with both the DC charger and cradle charger along with a couple extra batteries for TSC7. Now I always rented a base rover so I rented a cowbell extra for base. As most of the time I was renting it was for projects in no mans land so no cell coverage or very poor. You can place the dc in sleep mode with one push of power button it’s not turning it off in between drive times. Simply hit it again once at your point. Glad you are getting it done.
That’s a very common misconception of RTK observations. You are storing essentially a difference in position relative to your base station. It’s not raw data. Once your screen shown precision stop getting lower, continuing the observation is only wasted time and could degrade your position if you lose some satellite signals. If you are only using the collector and not TBC, you are better off taking 3 or 4 observations using the same point name and choose “store another “, then you can use cogo compute average and choose weighted average.
When I observe valuable points with RTK, I will go no more than 20 seconds, again seeing if precision shown has stopped decreasing. Just One RTK observation isn’t advisable. I’ve had points I observed 3-4 times that a point derivation report on that point in TBC showed the first observation as the worst being 0.05-7 off both H and V. Doesn’t happen a lot, but it does. And I reset satellite tracking between observations, fo not just turn your rod upside down. Consider RTK a resection where your control points are continually moving!
If prolonged RTK observations are worse than 180 epochs or 3 minutes and just taking several 20 second observations are better why does NGS recommend 5 minute observations? While I will agree that fewer epochs of data may point to tighter relative differences it does not prove that it is a better solution in the big picture. If I go to a point at 1200 and do a few back to back 20 second observations then got to another point and do the same 15 or 20 minutes later and then observe a third point and do the same. Then set up and turn angles and distances between them with a conventional total station I could see having decent results. Now you have a project where you are doing that around a site and now on opposite side of the site is now 3 to 4 hrs later I have always had better results with a minimum of 180 epochs or 3 minutes with two observations at a gap in time of 4 hrs making those points better relative to each other. I have heard the spill I don’t know how many times from a manufacturer that states 30 seconds is better and all that is needed even saw the math they had in a power point. The question I asked was this. Did you keep the same initialization or break it and were all three of those 30 second observations done back to back. Yes was the answer. I did a 2 1/2 acre Alta where we traversed and I shot all property corners and control with 4 rounds direct and reverse with good traverse kits and bipod on some of the property corners. Way over kill. This was with a 3 second S5. I was lucky as I could locate corners and control from across the site so many good cross ties. I went back did base and rover rtk r10 only. I located all the same points at 180 epochs and moved base and re observed each one. I moved it a third time. And did the same all 4 hr gap in time. Now two observations met the Alta specs but the third made the difference between conventional and rtk almost unbelievable so close it was not even funny. I tested direct distance and interior angles comparing the 2. Both vertical and horizontal positions diffences ended up in the reduced differences deltas in like a .015 ft horizontal and .024 ft vertical. This was a small site but I was trying to see if one could truly get to conventional type relative deltas on a small site. Both were adjusted via least squares. The two observations with rtk were an average.01 higher in both hz and vertical. Now if done like most the main traverse would have been direct and reverse through control only and side shots face 1 only to a poll on corners found. With no cross ties. When I compared those results RTK and conventional difference to corners saw .037 ft hz .05 vt. As the highest on a corner. Angles got a little higher diffence but in such short distance it was still negligible. All the same data set I just disabled observation in between each test. Most of this was done on my own time not companies. We were very close to the meridian and at the lambert parallel between zones all were done on grid.
You can take a 5 second shot and achieve the same results sometimes. Gps or GNSS rtk or static or ppk. Time in collecting and time in between observations are always good. Now to overkill like what I did is an also a trade off between what’s good enough vs what is to costly. For making money. If I were stamping something today I would require two observations minimum at 180 epochs with a minimum of 2 hrs but prefer 3 to 4 hrs. If those results exceeded the minimum standards I would roll with it. Of course what does project requirements need to be met etc. I actually have done rtk with three for vertical with digital levels and have dad tremendous results that far exceeded manufacturer specs in the diffences in heights with those procedures.
"If prolonged RTK observations are worse than 180 epochs or 3 minutes and
just taking several 20 second observations are better why does NGS
recommend 5 minute observations?"
Perhaps they made such a recommendation at one time but I'm not sure that is the current thinking. Check out the recommendations on pgs 42-44 of the following document:
https://www.ngs.noaa.gov/PUBS_LIB/NGSRealTimeUserGuidelines.v2.1.pdf
Redundant 3 minute sessions are their top level recommendations. They put a lot more stress on PDOP management and multipath reduction than they do on time of occupation.
BTW - that whole document is a must read for every RTK user, IMO.
Yes it’s a great document and when it was written it absolutely had focused on just GPS. They are working on the new guidelines I would imagine. Possibly that will change some things for sure. All I know is after years of orbits and years of geodetic work. Things have changed some with how manufactures develop the algorithms to establish the positions in an RTK environment. However no one manufacture follows the guidelines or guidance 100% in what is recommended via the government that established and set up the messaging and why each recommendation should be followed. I for one firmly believe that you can obtain great results in shorter occupation times. But across the board and to achieve the highest possible best solution it makes no sense why shorter is better. It makes no basic sense in GNSS whatsoever. So many things outside of the norms of what most look at like multipath etc that can go wrong with satellites themselves. Things happen all the time and end users never even are aware. The common terms that use to be stated was well it was just an anomaly. Well maybe but When you have watched a satellite for years orbiting and tracking that you get a perspective that’s is much different than a result we see on the user end. I would still say 3 minutes is a drop in the bucket for seeing this on the ground.
"Perhaps they made such a recommendation at one time but I’m not sure that is the current thinking."
Current thinking is reflected in the draft NOS NGS 92, which requires 3 each 5-minute NRTK observations, or 5 each 5-minute SRTK observations, for a 3 cm vertical (EH) project like the one I'm doing. RTK observations must be separated by at least 3 hours and span at least 2 days, and each NRTK group (i.e. emanating from a given PRS base) must have at least one OPUS-S validation session.
There are also post-processed options, but all require longer sessions.
@jhframe since you are doing this to go in OPUS Projects. I know for static it is highly recommended to orient the antenna correctly as opus projects and opus now use the absolute antenna calibration files. Were there any recommendations when doing rtk with the GVS files in doing this.
Trimble HD-GNSS & ProPoint
Windows Screen brightness is number one power draw on TSC7 followed by WIFI. Settings on rental gear may be set to "check for updates", turn that off for sure. There are bad TSC7 batteries circulating out there as well. I can get 8 hours on 2 with suspend mode engaged (screen off) between shots and BT engaged the entire time with R12i's.