January 29 th, 2013

P.PRIEUR (CNES, Microscope DFACS architect)

MICROSCOPE Drag-free & Attitude Control System

MICROSCOPE colloquium – 29&30th January 2013 - Palaiseau2/13

Contents

Introduction : orbit, attitude, Fep DFACS driving requirements DFACS hardware equipment Life span evaluation and enhancement DFACS control loop Real time attitude estimation A posteriori attitude knowledge Calibration modes Conclusions

MICROSCOPE colloquium – 29&30th January 2013 - Palaiseau3/13

1) Introduction : Orbit, Attitude, Fep

Orbit : Microscope will be launched into a quasi-circular 700km dusk-dawn

sun-synchronous Earth orbit

Attitudethe instrument main axis is kept into the mean orbital plane

Fep frequency (modulation of ‘g’ w.r.t. s/c frame)

Inertial pointing : Fepi = Forb ~ 1.7 10 -4 Hz (sessions of Ti=120 orbits) Rotating mode : Fepr = Forb + Fspin ~ 1 10 -3 Hz (sessions of Ti=20 orbits)

Calibration mode : based on inertial pointing

- linear stimulations

- attitude oscillations about s/c axes

eq uator

Xsat

N orth gr

MICROSCOPE colloquium – 29&30th January 2013 - Palaiseau4/13

2) DFACS DRIVING REQUIREMENTS

Drag-free

The difference of scale factor between the 2 masses (Kd<1.5 10-4) induces the control of the common acceleration @Fep

This 3 axes linear requirement induces the use of a 3 axes propulsion system The air drag is the main perturbation @Fep : 10 to 30µN depending on the solar

activity => need of rejection factor of 30µN/300Kg/1.10-12m/s2=105 (100dB)

γ1 γ2

21 KK ≠

Hzm/s 3.10 and m/s 1.10 2-10@Fep

2-12@Fep /<Γ<Γ

rr

m/scKd. : that so 21610.2 −≤Γ

MICROSCOPE colloquium – 29&30th January 2013 - Palaiseau5/13

2) DFACS DRIVING REQUIREMENTS

Attitude control

The differential excentring of the masses (∆<20µm) induces the control of the angular acceleration and of the angular velocity @Fep

1.10-9rd/s @Fepr is equivalent to an attitude stability of the instrument better than

0.16µrad at very low frequency

This order of magnitude is far below the thermo-elastic deformations between the instrument base-plate and measurement axes of any kind of attitude sensor

Then attitude estimation is clearly a challenge (see later)

mode) (rotating rd/s 1.10

modes) rotating & (inertial rd/s 1.109-

@Fep

2-11@Fep

<

<

ω

ωr

r&

γ1 γ2

21 OO ≠ m/s and : that so 21610.2)^(^^ −≤∆ΩΩ∆Ω

rrrr&r

MICROSCOPE colloquium – 29&30th January 2013 - Palaiseau6/13

2) DFACS DRIVING REQUIREMENTS

Attitude control (suite) : frequency and magnitude of the disturbance torques

Aerodynamic and solar pressure torques

are negligible due to the shape of the s/c

Gravity gradient produces large torques@2Fepr due to the satellite non-spherical

inertia

Magnetic torques (s/c permanent andinduced moment) produces large

@Fepr+Forb torques

Because of non linear effects in the δextraction method, all the harmonic signals

are taken into account in the @Fep DFACS

budgets via rejection templates

The most DFACS stringent requirementsare Harmonic ones. Stochastic sources

(on sensors, actuators, perturbations) are

quite low

Rotating mode – out of plane axis – FFT of Cpert

Rotating mode – In plane axis – FFT of Cpert

µN.m

µN.m

30

20

15

10

Fepr+Forb

2Fepr

Fepr+Forb

MICROSCOPE colloquium – 29&30th January 2013 - Palaiseau7/13

3) DFACS hardware equipment

In operational mode, the DFACS uses : the scientific instrument as main sensor for linear and angular accelerations

2 star-tracker camera heads (co-pointed but twisted) A set of 8 cold gas proportional thrusters (+cold redundance 1/1)

Range: 1 to 300 µN , τ<250ms, <1µN/rHz

MICROSCOPE colloquium – 29&30th January 2013 - Palaiseau8/13

4) Life Span evaluation and enhancement

Intensive Monte-Carlo simulations have been performed to estimate the number of

operational orbits before completing the propellant gas Some learnings have been taken into account in the design :

The satellite centring tolerance (i.e. the position of the s/c CoM w.r.t. the SUs) is

tighten to reduce the gravity gradient and the inertial forces in rotating mode The orientation of the thrusters has been optimized (trade-off between the force

efficiency of the configuration in case of strong solar activity and the torque efficiency

in case of large magnetic momentum) The gas consumption depends on key parameters :

Solar activity (atmospheric density)

Thrusters minimum setpoint (for quick re-start)

Satellite Magnetic momentum, Thusters ISP

The best effort is done on every key factor

in order to increase the worst cases life span

MICROSCOPE colloquium – 29&30th January 2013 - Palaiseau9/13

5) DFACS control loop

Cold-GasPropSyst

Drag free control laws

Attitude control laws

Thrust Repartitor

Hybridization filter

Star Tracker

6-axis accelerometer

attitude

ωγ&

accelerations

Satellite dynamics

DisturbingForces &Torques

++T

C

F

THRUSTC

F

8:1i

iFc

=

comF

ωQ ,r

Attitude measurement

Angular Acceleration Measurement

mω&

Estimates

MCA software : 4Hz sampling

comT

c

c

Q

ω -

+

cγ -

+

Linear acceleration measurement mγ

MICROSCOPE colloquium – 29&30th January 2013 - Palaiseau10/13

6) Real time attitude estimation

The star-tracker measurements errors @fep are not compatible with the angular velocity

and acceleration requirements (1.10-9rad/s & 1.10-11rad/s2) [see P5]

Fortunately, the scientific instrument complies with such accuracy A first solution consists in using a classic Kalman filter with a forced very low

hybridization frequency, this solution is possible in Inertial mode

In rotating mode, this technique is inefficient because of coupling transfers between in

plane axes. An original method was used to shape the θest/θstr transfer : a deep and wide attenuation around Fep was obtained at the price of very large digital filters

MICROSCOPE colloquium – 29&30th January 2013 - Palaiseau11/13

7) ‘A posteriori’ attitude knowledge

After every session, the attitude, angular velocity

and angular acceleration are calculated at the dates of the instrument measurements

The file is used at the SMC for the extraction of

δ process The restitution methods are still under

investigation : Once a posteriori, the measures from the 6 angular

sources (4 test masses & 2 star-trackers) can be transformed to frequential

A ‘frequency by frequency’ hybridization is then possible, keeping the best of each one of the 6 sources

An ‘expertise file’ giving the DFACS

observations and the evaluation of the DFACS

performances is also produced

Red : real time attitude performanceBlue : a posteriori attitude knowledge

performanceAbove : versus time over 20 orbitsUnder : versus frequency

MICROSCOPE colloquium – 29&30th January 2013 - Palaiseau12/13

8) Calibration modes

Some instrument parameters like ∆, Kd, θc (alignment between common SU sensitive axes and star-tracker ones) have been identified to be calibrated in-flight

The DFACS provides stimulations on the platform to increase the observability : Attitude oscillations Linear oscillations

The DFACS uses specific tunings, for example to estimate the attitude from the STR measurement at the observation frequency

K21xx, K22xxKdx, ηdy-θdy, ηdz+θdz, K2dxx

ηcy-θcy

etηcz+θcz

Kcx·∆x, Kcx·∆zméthod 2

∆yCalibrated param

noAbout Xsat, Ysat or Zsat

NonnonoLinear acceleration

stimulations

nonoabout Zsatat Fcalang

about Xsatat Fcalang

about Ysatat Fcalang

Attitude oscillationsA=0.05rad

inertialInertialInertial basedInertial basedInertial basedAttitude

XisXisXisXis, ZisXisMeasurement axis

2*f’TM2*2 10-2 Hz

fcallin1.3 10-3 Hz

fCor =fTM – fcalang=8.5 10-3 Hz

fcalang1.3 10-3 Hz

Fcalang1.3 10-3 Hz

Frequency of observation (fcalib)

EDCBASession Name

MICROSCOPE colloquium – 29&30th January 2013 - Palaiseau13/13

Conclusions

DFACS performance is credible and compatible with the overall mission budget :

While some requirements were initially set to easy figures, others were very tough. After some iterations in the frame of the performance group, the requirements have been adjusted

DFACS performance is clearly limited by sensors :

The performances of the propulsion system (accurate 6 axes control, quick response time) allow huge strain to the control loops, the resulting control errors are negligible

Looking for accuracy at low frequency (10-4Hz to 10-3Hz), we cope with low frequency errors on the star tracker associated with platform thermo-elasticity

Fortunately, the scientific instrument provides accurate angular measurements. Thanks to a complex hybridization, the angular velocity and acceleration requirements are fulfilled

Microscope, a very challenging mission: The large interweaving of DFACS and Scientific Instrument functions implies that progress in design

and performance budgets can only be achieved by a co-operative work between teams, which is very stimulating