Galileo Rubidium Standard
- Lifetime data and GIOVE-A related telemetries -
Fabien Droz, Gérald Barmaverain, Qinghua Wang,
Pascal Rochat
Temex Neuchâtel Time SA
Neuchâtel, Suisse
mail@temextime.com
Galileo navigation program is in progress under the technical
supervision of the European Space Agency (ESA). The
preliminary activities related to GSTB-V2 experimental
satellites provide the first results and the implementation of the
In Orbit Validation (IOV) phase are in progress. Atomic clocks
represent critical equipment for the satellite navigation system
and clocks development has been continuously supported by
ESA. The Rubidium Atomic Frequency Standard (RAFS) and
the Passive Hydrogen Maser (PHM) are at present the baseline
clock technologies for the Galileo navigation payload. For the
RAFS, initial ground technological project related to lifetime
possible limitation of the clock was already initiated in 2001.
Telemetries of the operating RAFS on board of GIOVE-A
provide highly valuable data to compare with the ground data
and validate the selected method of 2001. This article gives an
overview on the ground lifetime data of the RAFS in comparison
with GIOVE-A data. Extrapolation for the 12 years Galileo
mission duration is also provided.
I. INTRODUCTION
GALILEO is a joint initiative of the European Commission
and the European Space Agency (ESA) for a state-of-the-art
global navigation satellite system, providing a highly
accurate, guaranteed global positioning service under civilian
control. It will probably be inter-operable with GPS and
GLONASS, the two other Global Navigation Satellite
Systems (GNSS) available today.
The fully deployed Galileo system consists of 30 satellites
(27 operational and 3 active spares), stationed on three
circular Medium Earth Orbits (MEO) at an altitude of 23 222
km with an inclination of 56º to the equator.
Atomic clocks represent critical equipment for the satellite
navigation system. The Rubidium Atomic Frequency
Standard (RAFS) and Passive Hydrogen Maser (PHM) are at
present the baseline clock technologies for the Galileo
navigation payload. According to the present baseline, every
satellite will embark two RAFS' and two PHM's. The
adoption of a "dual technology" for the on-board clocks is
dictated by the need to insure a sufficient degree of reliability
(technology diversity) and to comply with the Galileo
Francesco Emma, Pierre Waller
European Space Agency (ESTEC) / Netherlands
Nordwijk, Netherlands
lifetime requirement (12 years). Both developments are based
on early studies performed at the Observatory of Neuchâtel
(ON) from end of 1980s and Temex Neuchâtel Time (TNT)
since 1995. These studies have been continuously supported
by Swiss fundings within ESA technological programs
especially since the set-up of the European GNSS2 program.
The activities related to Galileo System Test Bed (GSTB-
V2) experimental satellite as well as the implementation of the
In Orbit Validation phase are in progress. The first
experimental satellite (GIOVE-A) was launched 28th
December 2005, to secure the Galileo frequency fillings, to
test some of the critical technologies, such as the atomic
clocks, to make experimentation on Galileo signals and to
characterise the MEO environment. There is two RAFS on the
GIOVE-A satellite supplied by Surrey Satellite Technologies
Ltd.
II. DEVELOPMENT & QUALIFICATION ACTIVITIES OF
RUBIDIUM ATOMIC FREQUENCY STANDARD
The RAFS development milestones are chronologically listed
as below:
The first development activity kicked off at TNT in 1997, and
completed in 2000 with one Engineering Model (EM) RAFS
produced.
The updated RAFS development started in June 2000 and
completed at the beginning of 2002. The industrial
consortium is led by TNT with Astrium Germany as the
subcontractor for the electronics package. The results of this
development is applied clock design used for qualification
and lifetime testing.
The last development and qualifications step was initiated at
the end of 2001 and completed at the beginning of 2003 with
the delivery of an EM, which is the baseline unit for the
development of the flight models for GSTB-V2. The main
contribution of this last development step is the Inclusion of a
DC/DC converter and the satellite TT&C interface
compatible with ESA's new requirements. It doesn’t affect the
lifetime sensitive parameters of the clock.
 In the frame of GSTB-V2, one EQM, one Proto-Flight III. LIFETIME EXTRAPOLATION FROM GROUND TESTING
Model (PFM) and five Flight Model (FM) units have been In the frame of the “Lifetime Qualification of Rubidium
delivered. The PFM and FM1 are integrated in GIOVE-B and Clock”, five RAFS EQM’s were manufactured and tested
ready for launch. The FM4 and FM5 are integrated in GIOVE- under vacuum in order to perform the monitoring of the
A and in orbit since 28th December 2005. In addition, the FM2 potential lifetime limitations of the RAFS.
and FM3 are available as FM spare units.
The test bench is composed of five identical units and
RAFS for GSTB-V2 performance achieved are given in the some common elements.
table 1 and the plot in figure 1.
The five identical units are composed of:
Parameter Measurement • the vacuum chamber with pumping system and gauge
Frequency stability < 4*10-14 @ 10’000 sec • the cooled base-plate
Flicker floor < 3*10-14 (drift removed) • the frequency measurement system (including PC)
Thermal sensitivity < 5*10-14 /°C
• the TM output terminals
Magnetic sensitivity < 1*10-13 / Gauss
The common elements are:
Mass and volume 3.3 kg and 2.4 liter
Table 1: RAFS GSTB-V2 general performance achieved • the reference frequency system; H-Maser with GPS
monitoring and the frequency distribution unit (common for
Frequency stability of GSTB-V2 RAFS (Q4 2005) drift removed all the TNT facilities).
Averaging time, T, in seconds
1 10 100 1000 10000 • the frequency measurements instruments; five Picotime
1.0E-11 instruments measure independently each unit.
• the automatic TM measurement system; parameters with
Allan deviation, sigma y (T)

1.0E-12
possible ageing are monitored.
RAFS Spec.

PFM (GIOVE-B)
• the power supply; five RAFS are connected on two 12V
FM1 (GIOVE-B) batteries in serie
1.0E-13 FM2 (Spare)
FM3 (Spare) • The cooling system; the five cooled base-plate are
FM4 (GIOVE-A)
FM5 (GIOVE-A)
connected in serie
1.0E-14
The overall layout of the test bench are illustrated below.
Figure 1. RAFS GSTB-V2 Allan deviation (ground measurements)
Most of the parts are of-the-shelf standard parts.

Figure 2. Picture of the GSTB-V2 RAFS

Figure 3. The five vacuum chambers used for lifetime test.
The detailed description of these units was published in
previous EFTF paper [1]. From the validation of the test bench and the tuning of the 5
RAFS EQM’s, a total period of about 12 months of
testing/adjustment was performed. Then, a first period of one
year of continuous measurement was performed. At the end
of this first year of measurement, the units were dismounted
 and inspected. Then, the five re-assembled RAFS were For the Rb spectral lamp light intensity and the Rb atomic
submitted to the second year of continuous operation with signal level the best fitting corresponds to a formula based on
parameters monitoring. square root of time and is the one used for the prediction table
(see table x). This result was predictable considering that the
In total six parameters were measured: ageing effect on the lamp is related to the Rb diffusion in the
- Output Frequency surrounding glass. This phenomena is well known and was
- Rb spectral lamp light intensity emission measured by several institutes with same results, in relation
- Rb atomic response signal level with square root of time [2]. The prediction demonstrates that
- Crystal control voltage no major impact on performances are foreseen for a lifetime
- Regulated internal reference voltage of 12 years.
- Interrogation power injected in micro-wave cell
An example of Rb spectral lamp light intensity extrapolation
Three typical records are shown in figure x to y. is illustrated in figure 8 (EQM5). The red curve is the
3.100 measurement data during one year, the black one is an
exponential fitting curve, the blue one a linear fitting and the
Square Root Fit:
fit_light= W_coef[0]+W_coef[1]*x^0.5
W_coef={3.097,-9.1714e-05}

3.095
V_chisq= 2.32856e-05; V_npnts= 13433;
W_sigma={4.19e-06,2.77e-08} best fitting is the green curve in square root of time
3.10
Light

3.090

3.08

3.085

3.06
Year exp sqt lin
1 3.081 3.081 3.081
3
0 5 10 15 20 25x10 1.25 3.079 3.079 3.079
Light

x1000 sec 3.04 1.5 3.078 3.077 3.076 light meas

2 3.076 3.074 3.071 plight_exp
Figure 4. One year spectral lamp light intensity monitoring on “lifetime” 2.5 3.075 3.072 3.067
plight_sqt
3 3.074 3.069 3.062
EQM5 3.02 4 3.074 3.065 3.052 plight_lin
5 3.074 3.061 3.042
6 3.074 3.057 3.033
7 3.074 3.054 3.023
3.73
3.00 8 3.074 3.051 3.014
9 3.074 3.048 3.004
10 3.074 3.046 2.994
3.72 11 3.074 3.043 2.985
2.98 12 3.074 3.041 2.975

3.71 3
0 100 200 300x10
x1000 sec
Signal

3.70

3.69
Figure 7. Spectral lamp light intensity TM on “lifetime” EQM5 with
3.68 fit_signal= W_coef[0]+W_coef[1]*x^0.5 prediction on 12 years
W_coef={3.7048,1.353e-05}
V_chisq= 0.716141; V_npnts= 13433;
W_sigma={0.000734,4.87e-06}
3.67
A strong correlation between the spectral lamp ageing and the
frequency drift is also clear. Papers were already published
3
0 5 10 15 20 25x10
Time ( * 1000s )

on this topic [3].

Figure 5. One year Rb atomic response signal monitoring on “lifetime”
EQM5
The table 2 provides the summary of these key parameters
predicted evolution using square root of time for the five
-20
EQM units submitted to life testing.
-40 Square Root Fit:
fit_freq= W_coef[0]+W_coef[1]*x^0.5
Such extrapolation is corresponding to the law of Rb
-60 W_coef={-7.5687e-11,-6.2115e-13}
V_chisq= 1.09772e-21; V_npnts= 13395;
W_sigma={2.88e-14,1.91e-16}
absorption into the lamp walls.
-80
Frequency

-100 Square root of time extrapolation on 12 years

-120 EQM1 EQM2 EQM3 EQM4 EQM5
-140 Freq -9.2E-10 +1.8E-10 -4.2E-10 -6.5E-10 -4.5E-10
-160 Light -15% -18% -20% -11% -6%
-180x10
-12
Signal -11% -1% -7.5% -2% <-1%
3
0 5 10 15
Time ( * 1000s )
20 25x10
Table 2: Extrapolation of ageing data on 12 years

Figure 6. 1 years 10 MHz output frequency measurement on “lifetime”
EQM5
Based on the records, an extrapolation for 12 years was
performed. Different fittings were evaluated to generate the
extrapolation formula.
IV. GIOVE-A TM (JANUARY –OCTOBER 2006)
In order to verify the in-orbit RAFS ageing, as demonstrated
in the lifetime program, the main telemetries are related to the
spectral lamp light intensity level and the atomic signal level.
The 10th January 2006, the RAFS FM4 was switched-on for In the figure below we observe several ON-OFF cycles
the first time in orbit. The establishment of the internal TM including a long OFF period between the day 100 and 140.
was exactly in agreement with the ones recorded during
laboratory test.

on-board data
Warm-up
6.00 period after
switch-off
5.00

4.00

3.00

2.00
Rb LIGHT

1.00 Rb SIGNAL

0.00
08:15:00 08:30:00 08:45:00 09:00:00 09:15:00 09:30:00 09:45:00
Figure 9. Spectral lamp light intensity TM from FM4 (GIOVE-A)

lab data During the life of the clock, the spectral lamp light
intensity will slowly decay due to the Rb diffusion in the
6. 00
surrounding glass. As observed during the lifetime test
program, the decay of light intensity is a function which
5. 00
follow to the square root of time. The TM coming from the in-
orbit FM4 is in line with the expectation coming out of the
4. 00 lifetime program. In addition, when the unit is OFF, the
ageing process stops and the next ON level is equal to the last
3. 00
level in the previous ON period.
It must be noted that the frequency drift of the RAFS is
2. 00
nearly proportional to the decay of the light. So, the frequency
Rb LIGHT drift value decay with time for a better performance.
Rb SIGNAL
As illustrated in figure 9, the FM4 light TM measurement
1. 00

is affected by the digitalisation noise due to the A/D converter

0. 00
low resolution (8 bit resolution).
21:00:00 21:15:00 21:30:00 21:45:00 22:00:00 22:15:00 22:30:00

Figure 8. Comparison of the stabilisation process of the critical TM during

the warm-up in orbit and on ground

Spectral Lamp Light Intensity TM

The spectral lamp light intensity level is collected through the

photo-cell located inside the micro-wave assembly. When the
lamp is switched-off, the reference level is 2.0V. The
saturation is at about 4.8V. In nominal operation, the FM4
level was set at about 3.5V.

After an OFF period, the micro-wave assembly is cold and

the vapor of Rb is at a very low density. By the way, no light
is absorbed and the photo-cell is saturated (4.8V). After about
one hour, the micro-wave assembly reaches its nominal
temperature and the photo-cell indicates the operational level
Figure 10. Zoom on Spectral lamp light intensity TM
(about 3.5V).
Atomic Signal level TM
After the warm-up period (about one hour), the lower is
the TM voltage, the higher is Atomic Signal level. When no
signal exists, the TM provides 5V. The value of 5V is
observed each time after a switch-off of several minutes (or
more) as the density of the Rb vapor in the cell is too low to
collect a signal.
For the FM4, the signal is nominal at 2.5V. The small
ageing variation of TM is due to a decay of the Atomic Signal
related to the decay of the spectral lamp light intensity. This
low decay have no visible effect on the frequency stability.
The numerous switch-on / switch-off sequences operated
on the RAFS FM4 have not affected the overall clock
behaviour.
The real environmental space conditions seem not to affect
the ageing behaviour in comparison with the simulated one
used for lifetime testing.
VI. REFERENCES
[1] F.Droz, P.Mosset, G.Barmaverain, P.Rochat, Q.Wang Temex
Neuchâtel Time SA; M.Belloni, L.Mattioni Galileo Avionica / Italy;
U.Schmidt, T.Pike EADS Astrium / Germany; F. Emma, P.Waller
European Space Agency (ESTEC) / Netherlands "The On-Board
Galileo Clocks: Rubidium Standard and Passive Hydrogen Maser
Current Status and Performance" 20th Europen Frequency and Time
Foroum Braunschweig (D) March 27-30, 2006.

[2] G.Mileti, H.Schweda, P.Thomann, G.Busca, Observatoire cantonal de

Neuchâtel, C. Couplet, P.Rochat, Temex Neuchâtel Time SA;
"Miniaturized Rubidiuim Clocks for Space and Industrial
Applications". International Frequency Control Symposium, IEEE
1995

[3] J. Camparo, Photonics Technology Dept The Aerospace Corporation;

"Frequency Equilibration and the Light-Shift Effect for Block IIR GPS
Rubidium Clock"; 36th Annual Precise Time and Time Interval (PTTI)
Meeting. December 7-9, 2004

Figure 11. Atomic Signal TM from FM4 (GIOVE-A)

The noise measured on the Atomic Signal is related to the

ripple of the analogue TM (+/-150 mV) and is nominal.
Here again, the in-orbit measurement results are similar to
the ground measurement results coming from the lifetime
program.

V. CONCLUSIONS:

The lifetime program running on five reference units has

provided useful results and demonstrated the capability of the
RAFS to operate for 12 years under vacuum without
significant degradation.
Based on the TM from January to October 2006, the RAFS
FM4 operates according to the expectation.
The small ageing trend observed is in line with prediction
of the lifetime program results and is compatible with Galileo
12 years mission.