Testing of the ICESat BlackJack Engineering Model

1
Testing of the ICESat BlackJack Engineering Model

• Presentation for JPL
• August 22, 2002
• Jacob Williams
• University of Texas at Austin Center for Space
  Research

2
Presentation Overview

• Motivation and Objectives
• Hardware and Testing Setup
• Data Analysis
• Results
• Receiver Performance
• Anomalies
• Conclusions

3
Test Considerations

• Receiver provided by GSFC ICESat project.
• Investigate and analyze receiver performance in
  orbit-like environment.
• Receiver tested in simulated ICESat orbit, using
  antenna gain pattern based on spacecraft antenna.
• In all, over 400 hours of data was collected.

4
Testing Objectives

• Overall receiver performance and characteristics.
• Receiver Observables
• C1, P1, P2 pseudorange
• L1, L2 carrier phase
• Receiver Navigation Solution
• Position, velocity, and time.
• Receiver 0.1 PPS timing signal

5
Receiver and Software

• Engineering Model for ICESat.
• Beep port used.
• Maximum 9 PRNs tracked.
• Goddard software for commands and data collection
• BJInterface. Commanding receiver and collecting
  data stream.
• BJReader. Extracting observation files from raw
  data file.
• BJrnx (JPL) also used to generate RINEX files.

6
GPS Simulator

• GSSI STR-4760
• Dual Frequency Capable
• Pseudo-Y Code
• Can specify receiver trajectory (orbital,
  static), atmospheric properties, antenna gain
  pattern, etc.

7
Detailed Hardware Setup

• This diagram shows the hardware setup used for
  BlackJack testing.

8
Data Analysis

• Receiver, simulator, and timing data collected
  and post processed using MATLAB code.
• Simulated signals are used as truth for
  determination of accuracy of receiver
  observations.

9
Data Analysis

• Double difference used for analysis of raw
  measurement accuracy.
• Interpolation using clock offset for data
  alignment.
• Position and velocity direct difference from
  simulator.

• Timing signals of receiver and simulator
  differenced, and compared to time offset computed
  by receiver.

10
Simulator Gain Calibration

• Simulator gain level calibrated to achieve
  closest match with on-orbit CHAMP observations.
• Calibration curves created by plotting CA PR vs.
  CA SNR.

11
Receiver Performance

• High accuracy of dual frequency GPS observations
  and receiver navigation solutions.
• Receiver able to track simulators pseudo-Y code,
  as well as the unencrypted P-code.
• Orbital and static scenarios.
• High accuracy of receiver clock steering and
  timing pulse.

12
Results PRN Tracking

• Receiver tracks 7 or more satellites 95 of the
  time.

In this plot, ?2 gt 500 have been filtered out.
13
Results Clock Steering

• Sub-microsecond clock steering during regions of
  valid navigation solutions.

14
Results Observable Accuracy

• Double differences between two receiver channels
  and simulator truth.
• Very high quality dual frequency GPS
  observations.
• Error standard deviations for this PRN pair
• C1 123 mm
• P1 247 mm
• P2 285 mm
• L1 0.11 mm
• L2 0.18 mm

15
Results Receiver Navigation Accuracy

• Orbital scenario

Ionosphere On 3-axis s 1.95 m
Ionosphere Off 3-axis s 0.67 m
Due to known 15m bias issue
16
Observed Anomalies

• Corruption of observations from satellites with
  high relative acceleration.
• Affects receiver navigation solution and clock
  steering.
• Linkage between memory usage and receiver resets.
• L2 Ramps.
• 15 meter pseudorange bias.
• Duplicate PRN number and time epochs.

17
SNR Drop During High Acceleration

• In one 18 hr simulation, 22 of data epochs had
  at least one satellite with this anomaly.
• Occurs at high SNR, low pseudorange, i.e., near
  the closest approach of the receiver and GPS
  satellite.

This plot shows the correlation between high
relative acceleration and SNR drops.
18
Effect on Clock Steering

• The observations from satellites with SNR drops
  are corrupted.
• The navigation solution is corrupted, illustrated
  by a large ?2.
• Without a valid navigation solution, clock
  steering cannot be performed.

In this simulation, an 8 hr period existed where
stable clock steering was impossible.
Radial Position Error
19
Effect on Clock Steering

• Some of the anomalous navigation solutions fall
  below 10,000 threshold.
• These solutions are used for clock steering, and
  offset is not properly driven to zero.

?2 gt 500
?2 gt 10,000 No clock steering
?2 lt 10,000 Clock steering
20
Timing Pulse
Receiver Clock Offset
Measured

• Receivers clock offset computation nominally at
  10 ns level accuracy
• Large errors (up to 60 µs) seen during SNR drops.
• This is simply another measurement of the
  receiver navigation solution errors during epochs
  with SNR drops.
• (Possible timing pulse anomaly after receiver
  reset was not tested with this equipment.)

(sec)
Receiver Computed
(Drift over time possibly due to drift
between the two channels of the timing card.)
21
Memory Usage and Resets

• Memory usage in long-duration orbital simulation.
• Fairly consistent 38 hour period for the three
  natural receiver resets.
• Significant change in memory usage was not
  observed in static (ground) simulation.
• This reset rate is within ICESat specs.

These resets caused by the GPS simulator, not by
any receiver anomaly.
22
L2 Ramps

• L1-L2 phase difference plots.
• Previously reported L2 ramps observed, although
  not as frequently as JASON.
• Manifestation of very low (lt 5) P2 SNR during
  initial track.

Here, values of 0 were recorded for P2 SNR
23
Conclusions

• High quality dual frequency measurements,
  navigation solutions and clock steering.
• Anomalies
• In this model/software version, the high
  acceleration anomaly is a severe issue for
  receivers real-time navigation and clock
  steering.
• High acceleration anomaly was present on CHAMP,
  but seems to have been fixed (observed by
  comparing PR vs SNR curves for early and recent
  data).
• Memory reset rate is within ICESat specifications
  (1 per day).
• Other anomalies are relatively minor, and can be
  filtered out in post processing.