Automatic orbit ephemeris update in a low cost ground station

Automatic orbit ephemeris update in a low cost ground station

PII: Acta Astronautica Vol. 41, No. 1, pp. 1±6, 1997 # 1998 Published by Elsevier Science Ltd. All rights reserved Printed in Great Britain 0094-5765...

330KB Sizes 0 Downloads 92 Views

PII:

Acta Astronautica Vol. 41, No. 1, pp. 1±6, 1997 # 1998 Published by Elsevier Science Ltd. All rights reserved Printed in Great Britain 0094-5765/98 $19.00 + 0.00 S0094-5765(97)00211-7

AUTOMATIC ORBIT EPHEMERIS UPDATE IN A LOW COST GROUND STATION A. KIM WARD{ and ROBERT J. ELY Satellites International, Space Innovations Limited, Newbury, Berkshire, U.K.

RICHARD HOLDAWAY Rutherford Appleton Laboratory, Chilton, Oxfordshire, U.K. (Received 25 April 1997) AbstractÐThis paper examines a method of updating the orbit ephemeris of low Earth orbit satellites based on antenna tracking error data from a single ground station. The paper describes Satellites International's low cost S-band ground station which utilizes a 2.4 m parabolic dish antenna with full azimuth and elevation drive control. The antenna incorporates a diplexing feed, with a low noise ampli®er in the receive path, linked to a receiver/demodulator which downconverts and decodes the received telemetry before passing it to a telemetry depacketizer; and thence to the spacecraft control centre. The uplink telecommand data is packetized and passed to a sub-carrier generator, before being upconverted in an S-band sweeper/modulator, followed by a high power ampli®er linked to the antenna feed. The antenna controller drives the antenna in a ®ne resolution program track mode derived from a 10-point orbit ephemeris. Tracking error data is obtained in real-time during a pass by scanning the antenna across-track and along-track with respect to the nominal satellite position. The error measurements are then fed automatically into the orbit generation software to provide an updated orbit ephemeris. The paper demonstrates that the orbit prediction accuracy obtainable using this method can achieve satisfactory mission performance. # 1998 Published by Elsevier Science Ltd. All rights reserved

1. INTRODUCTION

smaller low cost missions. It is also true that if the mission requires the telemetry downlink to be as high as possible, i.e. 1 Mbps or above, then a large antenna will be required [1]. However, for moderately high data rates, up to around 500 kbps, it is entirely feasible to use a small 2.4 m or 3 m dish antenna linked to an appropriate suite of ground communications and antenna pointing equipment. Satellites International (SIL) has developed a low cost ground station of this type and its constituents are described in the following section. As regards antenna tracking, the problem is actually worse with larger dishes because of their narrower antenna beamwidth, typically 0.58. Such antennas require particularly accurate antenna tracking data if they are to track in a program-track mode, or else they may need to operate in an auto-track mode utilizing direct feedback of the received signal strength in the tracking loop. Smaller dishes of the sort described here have a wider beamwidth, typically a few degrees, and tracking is considerably easier, involving a simple program track control algorithm. A tracking error signal can be obtained by scanning the antenna about its nominal position and then using this data to update the ephemeris. This technique is described following the ground station description.

There are a number of pre-conceptions held by many potential users in the Space User Community regarding satellite ground systems which tend to colour their thinking with respect to mission feasibility. Principal among these are: (i) that the ground system should be based on one of the large international tracking networks; (ii) that the ground system is inherently expensive (an inevitable consequence of (i) above); (iii) that moving to Sband and higher frequencies for satellite communications (as opposed to the VHF and UHF amateur wave-bands), inevitably involves complex and expensive equipment, adding still further to the cost; (iv) that a large expensive dish antenna will be required; and (v) that tracking a low Earth orbit satellite is dicult because of perturbations to the orbit caused by atmospheric drag and other e€ects and, therefore, requires support from an external agency such as NORAD, or else an accurate GPS receiver system on the satellite. This paper seeks to dispel some of these myths. It is certainly true that using an Agency tracking network is expensive and should be avoided for {Corresponding author. Tel: 00 44 (0) 1635 46254; Fax: 00 44 (0) 1635 38785; e-mail: [email protected]. 1

2

A. K. Ward et al. 2. GROUND STATION

This section presents a description of the principal elements of the S-band ground station. It has a 2.4 m parabolic dish antenna with full azimuth and elevation drive control linked to pass prediction software. The antenna feed and low noise ampli®er (LNA) are linked via a diplexer to a receiver/demodulator which down-converts and decodes the received telemetry, before passing it to a depacketizer; and then to the spacecraft control system. The uplink telecommands pass from the control system through a packetizer, modulator, up-converter and 10 W high power ampli®er (HPA) before being fed to the dish via the diplexer. These various components are illustrated in Fig. 1 below and described in the following sub-sections.

2.1. SGA-2.4/TR antenna The SGA-2.4/TR antenna is a 2.4 m parabolic dish antenna, with nominal F/D ratio of 0.33, ®tted with a diplexing receive/transmit S-band feed covering the frequency ranges 2025±2120 MHz for transmit and 2200±2290 MHz for receive. The feed incorporates a GaAs FET LNA module mounted in a hermetically sealed housing. The antenna is mounted on a yoke assembly which permits 1808 of elevation movement and 22708 of azimuth movement, thus allowing it to support all types of satellite pass. The azimuth and elevation drives each utilize a high-torque brushless motor, close coupled to a planetary gear box, which in turn is close coupled to a harmonic drive gearbox, permitting the antenna to track within speci®cation at wind speeds up to 120 km/h. The power supplies and servo-drive ampli®ers for these motors are contained within a rack-mounted unit installed in the RF equipment rack. 2.2. TDC-22/70 tunable S-band down-converter The TDC-22/70 down-converter can select as its centre frequency any 1 of the 720 ESA/NASA Category A S-band channels between 2200.000 and 2290.000 MHz, and down-convert it to nominally 70 MHz, suitable for the VRD-300 demodulator. The down-converter utilizes a two-stage down conversion process and has a gain of around 60 dB, split between the two stages. The 70 MHz output still includes all the Doppler frequency o€set which was present at the S-band input; Doppler correction is performed in the VRD-300. 2.3. VRD-300 variable rate demodulator The VRD-300 is a low cost variant of SIL's well established VRD-100 variable rate demodulator as used in ESA's ground station network since the late 1980s. The VRD-300 utilizes the latest technology to reduce the chip-count and physical size of the demodulator, whilst retaining much of the functionality of the original VRD-100. It can support BPSK and QPSK modulation schemes at symbol rates from around 1 ksps to 1 Msps per channel and removes any Doppler frequency o€set present on the 70 MHz input signal. Variations in input signal strength are compensated for by an AGC circuit, whose output is available externally to support antenna tracking error measurements as explained below. The unit also includes an integral Viterbi convolutional decoder to process the output data if required. 2.4. PTC-200 packet standard depacketizer/packetizer

Fig. 1. Ground station block diagram

The PTC-200 packet standard depacketizer/packetizer ful®ls two functions. Firstly, it pre-processes the downlinked telemetry data in order to extract the source data packets before they can be further

Automatic orbit ephemeris in a low cost ground station

processed by the data processing computer in the Operations Control Centre (OCC). Secondly, it accepts telecommand user packets from the OCC and packetizes them into Command Link Transfer Units (CLTUs) together with their associated Idle packets for modulating the S-band uplink. The depacketizing function complies with the ESA/CCSDS packet telemetry standards for nonReed±Solomon encoded packets. It supports any permitted size of source telemetry packets together with any number of di€erent Application Process IDs. The output from the depacketizer is the original user packet source data which can then be distributed to telemetry processing computers via an Ethernet link. Error detection is also provided using the packet standard's cyclic redundancy check (CRC) code. 2.5. DG-100 sub-carrier generator and BPSK modulator The DG-100A Sub-Carrier Generator and BPSK Modulator accepts the telecommand data from the PTC-200 in the form of CLTUs and idle packets, together with an associated data clock. It uses these data to BPSK modulate a 8 or 16 kHz sub-carrier which is generated internally in the unit, synchronously with the telecommand data clock. The output constitutes the BPSK modulated sub-carrier. For ground test purposes, the unit also includes a test pattern generator, which can produce all 1s, all 0s, chequerboard (alternate 0s and 1s), or pseudorandom data, for modulating the sub-carrier. It has an integral bit error rate (BER) counter which can be used in a closed loop mode for testing the complete communications link. 2.6. SG-300A uplink sweep generator and PM modulator The SG-300A S-Band Sweep Generator and PM Modulator accepts the BPSK-modulated sub-carrier from the DG-100A and uses it to phase modulate an S-band carrier. The S-band signal is synthesized using a numerically controlled oscillator (NCO) and is able to be swept either side of the selected centre frequency, thus allowing the spacecraft on-board receiver to acquire. A triangular sweep is used and its width and rate are preset to standard values, typically 260 kHz at a rate of 30 kHz/s. The centre frequency will normally be o€set by a variable amount corresponding to the predicted Doppler frequency o€set computed in the OCC, according to the predicted position of the spacecraft in the orbit. The unit has the facility for an external sweep input to be used if required. The output power level of the SG-300A is nominally 0 dBm, which is directly compatible with the SHPA which follows it, as well as providing a suitable input for spectrum analysers or frequency counters if required.

3

2.7. SHPA-10 S-band 10 Watt ampli®er The SHPA-10 is a solid state S-band ampli®er which accepts a 0 dBm input signal and produces an output in excess of 10 W at S-band which is passed to the diplexing feed on the antenna. 2.8. GPS 1804 universal time standard In order for the tracking software to operate correctly it is important that it should have access to an accurate universal time reference. The Rapco GPS 1804 precision time standard is provided for this purpose. This unit utilizes the GPS satellite transmissions as the basis for generating a high precision 10 MHz frequency standard and a universal time source. 2.9. Antenna controller The antenna controller comprises a 486-PC with a custom antenna control software package which drives the antenna in an open loop program track mode. For each satellite pass, the controller must be provided with a 10-point ephemeris from the OCC which it uses to compute a ®ne resolution program-track. The controller then uses the universal time reference and the program track data to drive the antenna via its drive ampli®er/control unit with updates typically 5 times per second. The antenna control package then superimposes an along-track and cross-track o€set scan with respect to the nominal position of the satellite. The 2.4 m dish antenna has a 3 dB beamwidth of around 48 and an o€set of around 18 from the nominal position produces a usable change in the AGC signal obtained from the VRD-300 demodulator. The AGC data is in fact an estimate of the Eb/ No computed within the demodulator. The tracking error measurements can be made at discrete intervals during favourable passes or else continuously. This data is made available to the OCC's orbit prediction package for updating the orbit ephemeris as described in the following section. 2.10. Ground station price The price of the 2.4 m ground station described above is naturally dependent on the particular con®guration of the antenna and the RF and baseband equipment, which may need to be customized to comply with any special mission requirements. However, for planning purposes, or to compare with other ground system options, it may be assumed that the ROM price of the complete ground station is around $350 K.

3. AUTONOMOUS ORBIT EPHEMERIS GENERATION

This section addresses the problem of accurately determining the orbit of a near-Earth satellite using look-angle tracking data from a single ground station. It then considers the follow-on e€ect of pre-

4

A. K. Ward et al.

dicting the orbit for periods of several months ahead. 3.1. Accuracy requirements The ®rst question to be asked is ``is it possible to determine an orbit accurately from a single ground station, given the small amount of orbit coverage attainable ?'' The answer depends on two further questions: 1. How accurate is accurate? and 2. What is the orbit/ground station geometry? In determining and predicting the orbit of a nearEarth satellite there are usually three factors which de®ne the accuracy requirements. These are: (i) the ability to track the satellite correctly during a ground station pass; (ii) the ability to reconstruct the orbit and hence the position of the satellite at any time during the mission; and (iii) the ability to predict up to many weeks ahead for the purposes of mission planning. It is usually part (ii) which has the least tolerance on allowable errors. In order to track from a ground station, the maximum allowable error in azimuth and/or elevation can be typically 230' for a large antenna. This is interpreted in the worst case as a time error in satellite position of about one second for a 1000 km altitude satellite. Clearly the allowable error reduces slightly with decreasing altitude. For orbit reconstruction, the requirement is based on observations taken at times before and after a reconstructed epoch. For example, in a project which requires mapping features of the Earth, the altitude position reconstruction has to be exceptionally accurate (within metres), and this is de®nitely not possible using a single ground station, even with the aid of advanced techniques such as laser ranging. However, for many satellites an accuracy of 21 km in altitude and 25 km in along-track position is sucient. For orbit prediction, the required accuracy depends not only on the accuracy of orbit determination, but also on the accuracy of the orbit propagator. Long term predictions are usually required for advance planning of pass times, eclipses, attitude manoeuvres or experiment observations. Typical longterm accuracy requirements correspond to a cumulative error of a few seconds per week. This paper shows that these tolerances can be met, dependent on a suitable mix of orbit/ground station geometry. Firstly, the satellite must be visible from the ground station several times per day at elevations not less than 308 above the local horizon. Secondly, good accuracy is only obtainable if a suciently large portion of the orbit is visible from the ground station. For a near-equatorial satellite and a near-equatorial ground station this is not a problem. For a near-critical inclination satellite (i = 638) the ground station latitude should be at least 108 below the orbit inclination, otherwise

the ground station will only see the satellite near the apex (or nadir for a southern hemisphere ground station). In the case of a polar orbit, the ideal latitude for the ground station is somewhere in the region of 508 North (or South), as this will give coverage of the orbit at points approximately 908 apart on the north-bound and south-bound passes. It should be noted, from this latter point, that neither equatorial nor polar ground stations are particularly suitable for high inclination orbits, as their coverages tends to be 08 or 1808 apart, and this does not allow good discrimination of the orbit eccentricity or argument of perigee. In order to illustrate this point, it is useful to consider the case of a polar orbit at an altitude of 900 km (Fig. 2), the geometry of which results in six passes per day over a mid-latitude ground station, such as one located in the U.K. There are typically three usable north-bound passes and three south-bound, as illustrated in Fig. 3. This leads to a typical coverage of two 438 arcs of the orbit with their centres separated by 758 (Fig. 4). Using these data, the orbit can be updated daily and orbit prediction can take place once per week. The method of orbit determination and prediction, and the results obtained during ¯ight operation now follow. 3.2. Orbit determination With large antennas, typically three output signals are provided: one is the channel containing the received signal, whilst the other two comprise error signals for the azimuth and elevation control so that with servo-loop control the antenna can lock on to an incoming transmission using an auto-track technique. In the case of drop-lock, the antenna fol-

Fig. 2. Polar orbit con®guration

Automatic orbit ephemeris in a low cost ground station

Fig. 3. Ground tracks over U.K. ground station

lows an open-loop predicted path across the sky, which is the technique used exclusively by the small ground station described here. Look-angles from the actual satellite track (determined by the scanning process described earlier) are stored at a frequency of one observation point every 30 s at elevations over 108, giving typically 20 points per pass. Data from the ®ve best passes over two days are then accumulated and fed into the orbit determination purposes. Given estimates of the orbital elements of the satellite at some speci®ed epoch, and given observations of the satellite over a period of a few days about the epoch, the program re®nes the estimates of some or all of these elements by an iterative least-squares di€erential-correction procedure. Orbital elements are then generated at a slightly later epoch, but within the observation period. During the orbit re®nement, account is taken of perturbations due to the harmonic coecients of the Earth's gravitational potential ®eld and to atmospheric drag. Formulae for the shortperiodic, secular and long periodic e€ects are included. A typical set of input observations consists of about 100 data points spread in time over two days (Fig. 5), from three north-bound passes and two

Fig. 4. Orbit coverage from U.K. ground station

5

south-bound passes. The data are pre-smoothed, and the chosen epoch times for input and output elements are at midnight, one day apart. The output elements are then used as input for the orbit prediction software described in the next section. The loop is closed when part of the output data from the orbit prediction is then used as input for the next series of tracking. In the real-time process of tracking azimuth and elevation are initially fed to the antenna at a frequency of 10 Hz, in the open loop mode. Coecients of three nth order polynomials describe the motion of the non-inertial X, Y, Z co-ordinates of the satellite during each pass over the ground station. The polynomials are functions of time, hence it is straightforward to reconstruct the instantaneous X, Y, Z of the satellite and, thereby, calculate the azimuth and elevation of the satellite from the tracking station. 3.3. Orbit prediction The output from the orbit determination process is a set of standard mean or osculating orbital elements. These elements are then input to a computer program for the purpose of predicting orbital elements and events well into the future. 3.4. Open loop tracking At altitudes below about 600 km, uncertainties in orbit perturbations are primarily dominated by aerodynamic drag. Above 600 km, the air drag is signi®cantly reduced, and at remote sensing altitudes (say 700±1000) it is quite straightforward to model with high accuracy. As an example, at the altitude of the ERS-1 and ERS-2 satellites (typically 770 km) the satellites are travelling at around 7.5 km/s. For a large S-band antenna such as the 12 m antenna at the Rutherford Appleton Laboratory (RAL), which has a full beam width of 0.58 the satellite will travel through the beam in less than 1 s. With smaller antennas, such the 2.4 m dish described earlier, the beam-width is larger and the time to pass through the beam is around 10 s. It is, therefore, necessary to predict and track the satellite within these error margins, i.e. all pass tracking data stored for use within the period of autonomy must have sucient accuracy such that the predicted satellite position is always within about 0.8 or 8 s, respectively, of the

Fig. 5. Ephemeris update strategy

6

A. K. Ward et al.

lite position, it is straightforward to calibrate the next set of orbit predictions. The set of error measurements derived in the process explained above is fed manually or autonomously into the orbit ephemeris generator to provide an updated ephemeris, time-shifted by a constant o€-set for each up-coming pass. This results in high accuracy orbit data being available for many days in advance, which it is only necessary to update perhaps once per week.

Fig. 6. Tracking calibration

actual position. With current tracking technology and ground data processing, this is possible for periods of 4±5 days in advance of the pass for large antennas and considerably longer than this for the smaller ones. 3.5. Closed loop feedback tracking The main source of error in predicting the position of a satellite is the along-track error. A useful measure of this error can be obtained from the antenna position encoders, which can be calibrated to provide the angular separation between predicted satellite transmitter peak power and actual satellite transmitter peak power, as shown in Fig. 6. This angular separation is a direct function of the error in along-track position of the satellite, and can be fed back into the predicted satellite position as a linear time di€erence. 3.6. Autonomous feedback of error data Because the main source of error is along-track, representing an equivalent timing error in the satel-

4. CONCLUSIONS

This paper has made two main points regarding satellite ground systems which will hopefully be useful to potential users in the mission planning process. Firstly, it has shown that the ground segment need not necessarily be expensive; small low cost high performance ground stations are now available which can support moderately high data rates, reducing the dependence on support from large Agency tracking networks. Secondly it has shown that such small ground stations can produce suciently accurate tracking error data to update the orbit ephemeris autonomously, again reducing dependence on external Agencies or complex orbit determination methods.

REFERENCES

1. A. K. Ward, R. A. Bull, A. J. Barrington-Brown, S. J. Gardner, Ground systems for high data rate small satellites, Proceedings of the 1st International Symposium on Reducing the Cost of Spacecraft Ground Systems and Operations, Rutherford Appleton Laboratory, RAL.GS.63, 27±29 September, p. 63.1 (1995).