Inside GNSS Media & Research

JUL-AUG 2019

Issue link:

Contents of this Issue


Page 23 of 67

24 Inside GNSS J U L Y / A U G U S T 2 0 1 9 In nominal operations, our CV so - ware uses the well-known iono-free combination of GPS P1 and P2 pseudor- anges. In the unlikely event of a problem with this combination, for example due to jamming or interference in the GPS L1 or L2 bands, or even due to a total failure of GPS, we need to have alternative CV methods that ensure the continuity of operations. is is achieved by incorpo- rating Galileo, and also by using SF CV in all the individual GNSS bands. An example of the results is shown in Figure 1 which depicts the PHM-A clock model versus UTC(PTB) for April 25, 2019, i.e., MJD 58598 (Modifi ed Julian Day). e accompanying table (Figure 2) helps to understand the GNSS frequencies and pseudoranges involved. Two DF combi- nations are used: P1/P2 for GPS (nomi- nal method), and C1/C5 for Galileo. In SF, the iono models used are Klobuchar for GPS, and NeQuick for Galileo. e numerical values of the adjusted models, corresponding to the plots in Figure 1, are shown in the table below (Figure 3). e reference time for the model is T0 = MJD 58597.5. us the clock value at any time is calculated as: A0+ A1* (MJD-T0) + A2* (MJD-T0) 2 . "Steering" indicates the frequency correction to be applied to the clock; it is obtained extrapolating the model by one day to MJD 58598.5. In a visual way, the steering is the slope of the tangent to the model curves (Figure 1), evaluated at the extrapolated epoch. e steering is conceived to adjust to zero the instan- taneous frequency off set at that epoch. An additional steering correction term is used to adjust the clock phase off set versus UTC to zero in the following 10 days, but this aspect will not be dis- cussed here. "Error" in the A0 term and in the steering is calculated as the diff er- ence versus the nominal DF GPS solu- tion (GPS P1/P2). "RMS of Fit" indicates the RMS of the CV data residuals a er adjusting the quadratic clock model; the RMS value gives an idea of the uncer- tainty of the CV method. As can be seen, SF fi ts are in general nearly twice as noisy as DF ones. What the Results Indicate Several facts can be observed from the results. In the first place you can see a relatively large dispersion in the adjusted A0 values, with maximum errors of more than 3 ns versus the nominal solution. is is explained by the combined calibration errors in the two GNSS receivers. The pseudorange delay in each receiver chain can reach a few hundred ns, considering the accumulated effect of the antenna, the antenna cable, and the receiver itself. e total delay can be calibrated with an uncertainty of 2 or 3 ns for each pseudo- range signal. Since receiver chain delays are fairly stable, the error in the A0 can be considered constant, and thus it does not aff ect frequency transfer, but it does aff ect the time transfer. e second remarkable fact from the results is that GPS and Galileo pro- vide nearly identical steering results in DF, and that in fact the very small fre- quency dri of the clock (A2 term) can only be properly estimated in DF. We can observe a large dispersion if the A2 term from the SF solutions, with in fact an opposite sign with respect to DF. Finally, we can see that the steering error in SF is roughly inversely propor- tional to the value of the carrier fre- quency, with smaller values in L1/E1 and larger values in L5/E5a. is makes sense since the ionospheric error in the pseu- dorange is inversely proportional to the square of the carrier frequency. We can PRECISION TIMING FIGURE 1 This example depcits the PHM-A clock model versus (UTC)PTB for April 25, 2019. FIGURE 2 GNSS frequencies and pseudoranges involved.

Articles in this issue

Links on this page

view archives of Inside GNSS Media & Research - JUL-AUG 2019