Fixed Integers, or Estimated integers? What is the practical explanation, and what is the technical explanation?
I felt that a good discussion of this would be helpful to anybody, who involves themselves in RTK GPS. And, Post processed, Static.
Nate
About 2 to 5 ft.
:bad: KAROLY:bad:
I'm with Dave on this. I was going to say a couple of feet.
There are radio waves sent out from each satellite. Each wave has a 19cm wavelength. The phase center of your receiver antenna is close to one of the many possible intersections of the tops (or bottoms) of all of the satellite's wavelengths.
A fixes solution means that the software has determined that statistical checks on the data are within a certain tolerance that indicate that it is very likely that it has correctly chosen the intersection of wavelengths that your receiver antenna was nearest to.
Float means that the statistical tests did not return that certainty. It is still making a determination, but the tests did not meet a certain tolerance, and it could be wrong.
I used to think that only fixed was good, and float was for pikers, hayseeds, and shavetails. Now I realize that the receiver antenna is almost certainly NOT at the intersection, but merely near it, and given enough time, a float solution should actually be superior to a fixed solution.
But, this only for certain applications. For most GPS surveying, such as collecting topo and planimetric with a roving rod, you want fixed.
Stephen
This image is from Trimble.
Stephen
Stephen,
That is kind of like saying that if you stay put long enough with a code solution you will come up with the correct answer. If you watch the velocities while in a float solution you will notice that it can be quite significant. Until you get the fixed solution, you do have no way of knowing your precision any better than the float values.
Ok foggy, and Karoly
Around here, fixed means that you don't hafta worry about your dog getting pregnant!
🙂
😉
I think a float solution is a differential solution without fixed integer ambiguities. It is good for some things such as GIS mapping.
"Fixed" is short for "fixed integer number of wavelengths", the center of the 3D error ellipse that consists of the distance-distance-distance-distance-distance intersections from all the space vehicles to your rover and your base. It means that the RTK engine has arrived at a solution that is good enough to start playing with the phase data, which is how we get that elusive 1cm from something with a 20cm wavelength. Note that you can have a fixed solution that is not good enough to be a fixed phase solution, and there are different kinds of phase. Someone fresher on that can jump in here.
Float means your RTK has not converged to a fixed integer solution and the RTK engine is still using floating point numbers to try to converge to a fixed integer solution.
it can be more than 5'
if you are using Topsurv, go to Setup, Status, System to see how many Initialized Sats you have.
you need 5, if you do not have 5, you will be in float
it's too bad that the other listings of the sats are for the Common sats, not the Initialized sats
Prior to say, 1991 or so, most GPS solutions were float solutions based on the L1 only hardware of the time. Various ionospheric models were in use as the variable nature of the ionosphere was an important factor in computing the solution:
http://www.navipedia.net/index.php/Klobuchar_Ionospheric_Model
At that time, it was not possible fix the integer ambiguity in a short time. Observations were often in terms of hours and yielded results which are perfectly acceptable today.
FWIW,
See slide 14 at: http://geodesyattamucc.pbworks.com/w/file/fetch/65649128/class25_GPS_bl.pdf
The solution of a baseline using phase data requires the determination of the integer unknown number of wavelengths. These are the wavelengths of data transmitted by the SV prior to the receiver starting to log data.
In vector reduction, this integer number of wavelengths plus the first fractional part are converted to distances. If the correct values are determined a very strong solution results. A worse solution is obtained when the wrong integers are used than when the solution allows these parameters to float (not held to integer values).
While others refer to the single frequency case, frequency combinations are useful not only to determine the ionosphere impact but also to assist in integer fixing. Note that if we combine frequencies we come up with other length combinations that must fit our position. For example, we can better determine our position by comparing our 19cm wavelength solution (L1 only) with a widelane 86 cm solution.
HTH,
DMM
"Float" is what Tony Soprano's enemies did; "Fixed" is what Jimmy Hoffa is. Capuche?
You mean capiche?
I had a Trimble sales guy tell me that the new R10 doesn't do fixed or float, it just gives you its calculated precision.
Correct, it is a "New Realization" of the receiver's true position. It is supposed to be a much more accurate representation of the point position. You are either RTK Initialized or RTK Not Initialized. No more fixed/float. It is a little confusing for a lot of guys. It's good to know the DC software (Access) has set of tolerances that keep you from storing a position that has too high of precision estimates...you can be initialized at 0.50'.
It's a little more complicated than the number of wavelengths being solved between antenna and satellites. Or perhaps it's less complicated. Seldom are the number of cycles resolved. The base and rover are fixed when the assumed cycles are in agreement. At least that's my understanding. However precise point positioning is actually attempting to solve the integers from each satellite to the rover by correcting for all of the known errors inherent in gnss.
> Correct, it is a "New Realization" of the receiver's true position. It is supposed to be a much more accurate representation of the point position. You are either RTK Initialized or RTK Not Initialized. No more fixed/float. It is a little confusing for a lot of guys. It's good to know the DC software (Access) has set of tolerances that keep you from storing a position that has too high of precision estimates...you can be initialized at 0.50'.
Has something significant changed? The rule of thumb is that any RTK accuracy prediction over about 0.3' can actually be off feet.
I have found this to be true.
Perhaps that is why Trimble went this way?
wait...WHAT?
".. and given enough time, a float solution should actually be superior to a fixed solution. "
How did you come to that conclusion?