Prisma Satellites - CNES Experiment: FFRF Envelope

(Duration of 10 days)

This 10 days experimental slot carries two different purposes:

Pursue FFRF sensor characterization started during the FFRF_Initialize session – this includes sensor calibration, functional tests and performance evaluation.
Perform some preliminary GNC commissioning that consists in activating some guidance and control functionalities in open loop mode.

The experimental slot will imply trajectories similar to FFRF_initialize i.e. slow drift from 100 m to different distances and back that will be repeated several times.

Reports:

FFRF Envelope part 2 takes a good start

The second part of FFRF Envelope experimental slot started Tuesday 8^th evening and will last 4 days. Prior to experiment a new software version was uploaded on both equipments to correct an anomaly and modify some calibration parameters. The purpose of this experimental slot is to pursue the characterization of RF sensor and assess performance improvement brought by a new tuning of parameters. The following activities are scheduled: multipath error calibration assessment, characterization of Integer Ambiguity Resolution (IAR) process for distance and Line of Sight (LOS) in various attitude and distance configurations, test of inter satellite link at high rate (12kb/s instead of 4 kb/s), assessment of performance during antenna handover on Target. Once again IAR and multipath tests have required the elaboration of complex attitude profiles and resulted in heavy TC plans (several thousands of TC!)

Since the beginning of experiment, RF sensor appears to behave very well. The first test was dedicated to assessment of multipath error correction on LOS measurement. We recall that multipath error results from the multiple reflections of incoming RF signal on the satellite structure surrounding FFRF antenna. As this error is spatially dependent and reproducible for a given direction of arrival of the signal, it is then possible to build correction tables (space is sampled vs. azimuth and elevation). Once tables have been uploaded into the equipment, corrections are applied to measurements in real time. The multipath tables that were uploaded for this experiment were obtained using results from Envelope part 1. Errors were computed as the difference between FFRF and GPS POD measurements. Some averaging was necessary to reduce other error sources. Note that corrections have only been computed for LOS (no table for distance) and were restricted to elevations lower than 30°.

To assess LOS performance improvement, the test scenario involved the same spiralling attitude profile that had been performed during Envelope part 1 (figure 1). Below we plot the residual errors obtained with real time correction application (figure 2) and the ones that were obtained in October (figure 3). Improvement brought by the correction (inside the 30°-elevation domain) is obvious. Mean and standard deviation of LOS error are improved by a factor of 2 (mean drops from 0.9° to 0.4° and standard deviation from 0.26° to 0.14°). LOS multipath error calibration appears then to be promising and should result in improved GNC performance.

Figure 1. Attitude profile for multipath test

Figure 2. Line of Sight error without multipath correction (results from October 12^th)

Figure 3. Line of Sight error with real-time multipath correction (February 8^th)

Written by

CNES

2011-02-11 / 22:20:18

FFRF Envelope part 1 successfully achieved

The first part of the FFRF Envelope experimental slot is over. During six days the CNES-FFIORD team has been using PRISMA satellites so as to characterize the Radio Frequency based sensor within an as extensive as possible domain. Various trajectories have been designed to test the FFRF sensor at close (20 meters), medium and far ranges (up to 9 kilometers). The following pictures show the overall trajectory followed during the 6 FFRF Envelope days using the SSC Autonomous Formation Flying (AFF) guidance and control module. The left red dot in the picture gives Tango position, centered on the Local Orbital Frame. Mango is represented by the right red dot.

Complex attitude profiles have been elaborated as well to assess the level of multipath error and to pursue the test of Integer Ambiguity Removal process.

To have a good feeling of the complexity and intensity of the past operations, let’s have a look at the following figures:

- 3757: this is the number of elementary telecommands dedicated to the past 6 days FFIORD experiment, without taking into account the other telecommands included within higher level called procedures and the one required for platform maintenance. Amongst those 3757 TCs are the attitude guidance commands and the FFRF direct commands.

- 2689: this is the number of telecommands sent to the manual attitude guidance module. Indeed the attitude profiles generated on ground consist in a starting date, and ending date, and in series of Chebychev polynomials coefficients used to parameterize the profile along which Mango has to stay aligned. For each segment of attitude profile all those parameters have to be sent to Mango and more precisely to the manual attitude module in charge of following the reference attitude using reaction wheels.

- 450: this is the number of telecommands sent to the FFRF instrument. They were intended to initialize and / or reset the sensor for each of the 105 scheduled elementary test sequences. Amongst those elementary test sequences there had been some quiet performance oriented slots, and some more dynamical test periods for the Integer Ambiguity Removal process and for the multipath level assessment.

No need to say that the validation of scenarios containing thousands of telecommands is a complex process. But efficient tools both on CNES and SSC sides allowed for quick and partly automatic validation using representative simulators of the PRISMA satellites. Only 2 iterations in between SSC and CNES were necessary during the validation process, and finally the go-ahead was given by the PRISMA Mission Control Center 2 days before starting date of the FFRF Envelope.

To illustrate the agility of the Mango satellite in terms of attitude and the way it has been used by CNES, one can have a look at the movement designed for the multipath error assessment on the 12^th of October. The purpose was to evaluate the level of multipath for various angles of azimuth and elevation (multi reflections of the incoming RF signal on the surroundings of the RF antenna shift the signal phase and induce some errors). A spiralling profile was followed, starting from a null elevation and progressively increasing this angle, while spinning around the antenna line of sight axis (one turn ranges from 0 deg azimuth to 360 deg). The spiralling movement was decreasing in speed with the increasing elevation so as to stay within the working domain of the attitude estimation and control subsystems.

After intensive post-processing of all the measurements (9 hours of test produced 32400 snapshot FFRF measurements) and cross comparison with POD (Precise Orbit Determination based on GPS measurements and delivered by DLR) it was possible to assess the mean error of the RF metrology with respect to azimuth and elevation. As an illustration the following pictures shows this error for the Line Of Sight measurement, ranging from -1cm (-0.6 deg) to 2.5 cm (deg). Assuming that the other error contributors are negligible (Tango residual attitude movement inducing more multipath effect, Mango attitude estimation error, POD error, geometrical uncertainties…) then this kind of computation give access to the multipath error on Mango.

Now that the FFRF Envelope part 1 slot is ended and that the operational constraints are behind us (temporarily), it is time to analyze in depth the great amount of data. This will serve for full characterization of the RF sensor performance, as well as for refining the tuning of the sensor (bias and possibly multipath calibration) and navigation function in order to improve even more the performance of the coming closed loop experiment to be performed by CNES-FFIORD end of October.

Written by

Pierre-Yves Guidotti

2011-02-11 / 22:13:45

Handover and orbit control during FFRF Envelope

Short update on how Mango and Tango has been dancing during the last week. AFC 1 ended in a large relative motion, circling Tango in the middle. After updating the software on both Mango and Tango and while Tango was still rotating and transitioning in to a three axis stabilized attitude, pointing one GPS antenna to zenith, Mango was given the command to switch to AFF mode and perform a handover to the starting relative state of FFRF Envelope. This handover involved a few large maneuvers, reducing the cross-track and radial motion to zero and going in to a position only 100 meters behind Tango in along-track. This means Mango and Tango are in exactly the same orbit, the only difference is a small time delay for Mango of 14 ms. The following two figures show the handover relative motion but from two different viewpoints.

The start is indicated by a green star and the finish by a red circle. Also shown is the location of Tango and Mango for one sample time where it is indicated by an arrow the direction of the relative motion. Note that it might look like Mango went very very close to Tango, but at the same time the cross-track component was about 50 m, this means that during the full handover the minimum distance between Mango and Tango was never smaller than 50 m and was perfectly safe.

The handover was performed using the AFF orbit guidance and control software. AFF was also used during the FFRF Envelope scenarios to follow reference trajectories which had been specified by CNES. The following two figures show the complete trajectory executed by AFF during FFRF Envelope from two different viewpoints. The full FFRF Envelope duration was 6 days and required 58 AFF telecommands in total.

Written by

Robin Larsson

2011-02-11 / 22:13:45

FFRF Envelope

FFRF Envelope part 1 the CNES’ second prime experimental slot started on Thursday October 7^th and will last for 6 days. Objective is double: pursue characterization of FFRF subsystem and test some CNES GNC functions in open loop (no application of the computed maneuvers). FFRF tests focus on the characterization of Integer Ambiguity Resolution process for both distance and Line of Sight (LoS), assessment of FFRF final performance (LoS and distance), and calibration of multipath errors. As these errors are dependent on signal direction of arrival, characterization requires the sweeping of a wide range of TARGET and MAIN attitudes. As a result specific trajectories and attitude profiles have been designed for this test so as to cover as much as possible the FFRF working domain envelope. In the meantime assessment of the CNES navigation performance keeps on going.

Prior to the experiment start-up, a new FFRF software was successfully uploaded on-board. FFRF SW update was intended to correct a few anomalies and update some parameters so as to increase performance. This operation was quite intense since it required 5 orbits (resp. 6 orbits) to upload and execute 8651 telecommands to Main FFRF terminal (resp. Target). Thanks to the efficiency of SSC operational team the Target FFRF software upload has been optimized and the activities of the 6^th orbit have been squeezed within the 5^th passage. After a long night of operations, going to bed 1h40 before the expected time is really appreciable…

Following a shaky start on Thursday evening, Day 1 and 2 have run very well since then. FFRF parameters update showed its benefits: the distance bias which is used to be around 1.2m has greatly dropped. Comparison with the DLR onboard navigation (GPS based) let appear a bias of 3.6 cm. Cross validation will likely be improved soon when comparing with POD (Precise Orbit Determination based on GPS). As for LoS, measurements appear to be very stable and accuracy is less than 1°.

Written by

Thomas Grelier and Pierre-Yves Guidotti (CNES)

2011-02-11 / 22:13:45