VERIFICATION AND OPTIMIZATION OF THE PHYSICS PARAMETERS OF THE ONBOARD GALILEO PASSIVE HYDROGEN MASER Qinghua Wang, Pierre Mosset, Fabien Droz, Pasal Rohat Temex Time Vauseyon 9, Neuhâtel, Switzerland E-mail: qinghua@temextime.om Giovanni Busa Kytime Eluse, Bevaix, Switzerland Abstrat Atomi loks represent ritial equipment for the satellite navigation system. The Passive Hydrogen Maser (PHM), with its exellent frequeny stability performane has been hosen as the master lok in the Galileo navigation satellite payload, and will be the first one of its type ever to fly. Temex Neuhâtel Time is responsible for the industrialization of the Physis Pakage (PP) of PHM and has been developing numbers of PP inluding Engineering Qualifiation, Qualifiation, Proto-Flight, and Flight Models in the frame of the Galileo Satellite Test Bed (GSTB-V) and the In Orbit Validation (IOV) phase. This paper provides the verifiation results of PHM physis parameters, aording to the measurement data and the theoretial analysis, for all the PP produed up to now. A theoretial PHM physis model has been developed to extrat inherent physis parameters (suh as osillation parameter, saturation fator, natural line width, various relaxation rates and useful atomi flux) whih annot be measured diretly, but are of great importane in order to evaluate the instrument performane. I. DEVELOPMENT ACTIVITIES OF THE SPACE PASSIVE HYDROGEN MASER Atomi loks represent ritial equipment for the satellite navigation system. The Rubidium Atomi Frequeny Standard (RAFS) and Passive Hydrogen Maser (PHM) are the baseline lok tehnologies for the Galileo navigation payload. The adoption of a "dual tehnology" for the onboard loks is ditated by the need to insure a suffiient degree of reliability (tehnology diversity) and to omply with the Galileo lifetime requirement of years. The PHM with its exellent frequeny stability performane has been hosen as the master lok, and will be the first one of its type ever to fly. Temex Neuhâtel Time (TNT) is not only the supplier of the RAFS, but it is also responsible for the industrialization of the Physis Pakage (PP) of the PHM. Galileo Avionia (GA), who designs the Eletronis Pakage (EP), is responsible of the integration of the PP with the instrument (Figure ).
Figure. Piture of Galileo PHM (8kg). The Physis Pakage (PP) is on the top of the Eletroni Pakage (EP) The industrialization ativity aimed at PHM design onsolidation for flight prodution was started in January 3, based on the Engineering Model (EM) design led by the Observatory of Neuhatel sine. Main efforts for the PP in the industrialization frame foused on the definition of repeatable and reliable manufaturing proesses and assemblies, as well as the parts number and ost redutions, while keeping the ritial performanes unhanged (e.g. Allan / deviation σ y ( τ ) τ for s τ s ). Two tehnologial models (Figure ), a Strutural Model and numbers of PP have been developed at TNT for these objetives and to qualify the new upgraded design. In the frame of the Galileo Satellite Test Bed (GSTB-V) and the In Orbit Validation (IOV) phase, TNT has manufatured two Engineering Qualifiation Models (EQM), a Proto-Flight Model (PFM), and a Flight Model (FM), whih were delivered and tested at payload level. In addition, four Qualifiation Models (QM) are being manufatured and will be submitted to the prolonged testing for the life time demonstration. The first in-orbit PHM will be validated in the Galileo experimental satellite GIOVE-B, whih is now planed for launhing in early 7. Figure. Tehnologial Model with / without over II. MEASUREMENTS OF THE PP PARAMETERS
RF power RF ontrol 38 th Annual Preise Time and Time Interval (PTTI) Meeting Before the integration with EP, the PP is tested and haraterized under the thermal vauum hamber by the Maser Test Benh whih supplies the eletrial ontrols for the PP and monitors all operating parameters (Figure 3 and Figure 4). Thermal Vauum Chamber Physis Pakage MCSA & Low Vauum Hydrogen Assembly Ion Pump Cavity Heater & Temp. sensor C-Field Coil Varator H purifier H supply assembly Ion Pump HDO Interr. signal Atomi signal DC power Var. ontrol DC power High Voltage Vauum eletrial interfaes (Feed through) Atomi Signal Aquisition System MCSA, HSA, Ion pump & HDO Controls AHM referene Final Test Benh Monitoring Computer (PC) Figure 3. Test setup Figure 4. Maser Test Benh with Thermal Vauum Chamber In order to extrat the relevant physis parameters using the theoretial PHM physis model, the maser gain and the operational linewidth as the funtion of the interrogation power for different purifier urrents have to be measured, while keeping other operating parameters at nominal settings. The maser gain and the operational linewidth are measured by the dediated Atomi Signal Aquisition System (ASAS). The system is onstituted by a RF signal proessing unit ontrolled by two synthesized funtion generators in order to san the useful frequeny range step by step with the resolution of.hz. The system gives the output level of the atomi signal vs. the 3
interrogation frequeny from whih the two quantities G (amplitude gain at resonane) and LW (the operational full linewidth measured at the half value of the power gain) are determined. Figure 5 shows the atomi response of the PHM IOV-EQM PP, measured with 5Hz span exhibiting an atomi signal gain of 3.6dB and an operational atomi linewidth of.6 Hz. Figure 5. PHM atomi signal measured by Atomi Signal Aquisition System for IOV EQM III. EXCTRACTION OF THE RELEVANT PP PARAMETERS Based on above measurements, a theoretial PHM physis model has been developed to extrat the relevant physis parameters (the osillation parameter, the saturation fator, the natural linewidth, various relaxation rates, and the useful atomi flux) whih are diffiult to measure diretly, but are of great importane in order to evaluate the instrument performane. 3. THE OSCILLATION PARAMETER AND THE SATURATION FACTOR Unlike the self-osillation of an ative hydrogen maser, the small-size passively operated maser is used as a mirowave amplifier, having a very narrow bandwidth, by injeting in the mirowave avity an interrogation signal. The osillation parameter α is an important parameter determining the osillation properties of the system. The passive behavior is obtained when α <. The amplitude gain at resonane G of the maser amplifier [] is given by: + S G = () + S α where S is the saturation fator at resonane, whih is in proportional to the interrogation power P for a given atomi flux: in S = k s P in () 4
k is inversely proportional to s γ γ aording to the definition of S, with k alulated as the onstant independent of the atomi flux: k k s = (3) γ γ From Eq.(), the value of α is extrated from the unsaturated amplitude gain orresponding to a very low level of the interrogation signal, when P. The value of in S at larger interrogation levels is obtained from the measured (saturated) amplitude gain knowing the value of α. Figure 6 shows the least-squares fit to the IOV-EQM PP measurement data of G vs. P in for three levels of H flux, to find out α and S with k. s 5. 4.5 4. a=.47 a=.395 a=.383 Gain G [db] 3.5 3..5.5 mbar exp.5 mbar fit.9 mbar exp.9 mbar fit.5 mbar exp.5 mbar fit..5. - -5 - -95-9 -85-8 -75-7 Pin [dbm] Figure 6. Atomi gain at resonane vs. interrogation power for three H low pressures, and the respetive osillation parameter by the least-squares fit 3. THE NATURAL LINEWIDTH The transverse relaxation rate γ orresponds to the natural linewidth, distinguished from the measured operational linewidth LW whih is subjeted the power broadening effet. The amplitude gain G of the maser amplifier is []: G = + + x + + x α α (4) S S + + x 5
where x is the normalized frequeny offset ω ω a x =, ω is the interrogation arrier angular frequeny, ω a is the hydrogen resonane angular frequeny and the avity is assumed to be tuned to the hydrogen frequeny. The linewidth broadening fator F an be solved as the value of x at the half maximum of the power gain signal aording to Eq.(4), for known α and S : F = 3 4α 4α + α γ [ S + S ] ( α 3) αs + αs 4( α S ) ( α + S ) + α ( α ) ( α 3) ( α S ) (5) and γ an be determined from the operational linewidth: LW γ = π (6) F 3.3 THE USEFUL ATOMIC FLUX AND VARIOUS RELAXATION RATES These parameters an be solved by the set of equations: α ' µ µ B η Qψ = (7) hvbγ γ 4 γ = γ b + γ w + k eψ (8) 3 γ = γ b + γ w + k eψ (9) where, µ, µ B and h : onstants of the magneti permeability of vauum, Bohr magneton and Plank s onstant divided by π, respetively. Q : Cavity loaded quality fator. V b : Storage bulb volume. ψ : Useful H atomi flux entering the avity in the (,) hyperfine state. γ : Longitudinal relaxation rate. γ : Storage bulb or geometri relaxation rate. This an be alulated by b υπa γ = () b 3L 4V b + 8a where υ is the mean veloity of Hydrogen atoms, L and a are the length and radius of the exit tube of the storage bulb respetively. This value is onstant for all our PP. bulb. γ : Transverse wall relaxation rate, due to atomi ollisions with the wall of the storage w 6
A magneti relaxation rate should be added theoretially to Eq.(8-9). This term is however negligible for our onfiguration and is negleted. The last terms of Eq.(8-9) are beam density-dependent ontributed by the spin-exhange relaxation rate. k is onsidered as a variable for different PP. e In Eq.(7), the filling fator η ' relates to the oupling between the atomi medium and the mirowave field. It is defined by H ' Vb z b η = () V H where V is the overall volume of the avity spae, H is the square of the average amplitude z b of the axial omponent of the magneti field within the bulb volume and H is the quadrati average value of the magneti field amplitude within the overall mirowave avity volume. ' It s hallenging to obtain the value of η by the rigorous theoreti solution for suh a ompliated magnetron avity struture. The numerial simulation of the eletromagneti field for the avitybulb assembly was performed (Figure 7) to solve η ' by integrating the H field distribution ' funtion over the bulb and the avity volumes. The value obtained for η is.4. This is the value we assume for all the PP. Figure 7. TE magnetron mirowave avity and the magneti field distribution simulation For two purifier urrents A and B (orresponding to two H flux setting A and B) the Eq.(7-9), give a the set of 6 equations with known α A, α B, γ A, γ B and γ b. This allows to solve 6 unknown parameters : ψ A, ψ B, γ A, γ B, γ w and k e ( γ w and k e are independent of flux). 3.4 THEORETICAL FREQUENCY STABILITY AND OPTIMIZATION We assume that the noise of the interrogation signal is negligible. In suh a ase the frequeny stability equation for the PHM, has ontributions from the thermal noise of the reeiver and the PP parameters: 7
( + S α ) α S ( + S ) k sktfr σ ( τ ) = τ () y A Q where k is Boltzman's onstant, T is the temperature of the avity, F r is the noise figure of the reeiver, A is the avity power attenuation, and Q is the atomi line quality fator: πf Q = (3) γ For a given PP, being k, s Q and α dependent on the useful atomi flux ψ (Eq.3, 7-9 & 3), σ is now the funtion of y ψ and S, i.e. respetively of the H low pressure and of the interrogation power. σ y s vs. ψ & S,, Figure 9 shows α vs. ψ,and Figure shows H low pressure vs. ψ, for IOV-EQM PP. From these figures the optimum purifier urrent, the optimum interrogation power and the maximum osillation parameter orresponding to the best frequeny stability an be obtained. Figure 8 shows ( ) Figure 8. Theoretial Allan deviation at s vs. saturation fator S and useful atomi flux ψ σ s =6.9e-3 at y S =.37 (orresponding to interrogation power of -8dBm), and ψ =8.3e atoms/s (orresponding to a low pressure of.mbar) (minimum ( ) 8
.45.4.47 8.3E+.35 Osillation Parameter a.3.5..5 theoretial model.5 mbar.9 mbar.5 mbar optimum..5..e+ 5.E+.E+3.5E+3.E+3.5E+3 3.E+3 Useful Atomi flux I [atoms/se] Figure 9. Calulated osillation parameter vs. useful atomi flux for three different H low pressures, and theoretial model with optimum point (α =.47 at ψ =8.3e atoms/s).6.5.4 y =.599E-4x -.7884E-. H Low Pressure [mbar]..8.6.5.9. Optimum.4. 3.E+ 4.E+ 5.E+ 6.E+ 7.E+ 8.E+ 9.E+.E+3.E+3.E+3.3E+3 Usefull flux [atoms/se] Figure. H low pressure vs. alulated useful atomi flux and fitting for optimum value (.mbar at ψ =8.3e atoms/s) IV. VERIFICATION RESULTS OF PHYSICS PACKAGES Table lists the results from measurements of all PP manufatured at TNT up to now. In an attempt to haraterize the PP before delivering to GA the physis parameters were measured using the same eletronis (referene eletronis), with the same interrogation power. 9
The nominal operational H low pressure was set to have the same H onsumption (< bar.liter/year) for eah PP. The values were seleted aording to the ondutane alibration for eah mutli-hole ollimator of the hydrogen dissoiation bulb performed in the Tehnologial Model before the assembly to PP. With this H onsumption, the useful atomi flux for 4 PP is about 5e+/s, varying slightly from unit to unit. The higher value of 7.9e/s for IOV EQM ould ome from the higher dissoiation or state seletion effiieny. Table. Comparison of PP in nominal operational ondition (blue: measurement parameters) PP GSTB-V EQM GSTB-V PFM GSTB-V FM IOV EQM QM Aug.4 Feb.5 Aug.5 Jul.6 Ot.6 Interrogation power [dbm] -8-8 -8-8 -8 H low pressure [mbar]...8.9.9 Cavity quality fator 58 58 95 95 95 Useful flux [atoms/s] 5.5E+ 4.E+ 5.7E+ 7.9e+ 4.9e+ Linewidth [Hz] Geometri relaxation..... Wall relaxation..4.5.9. Spin-exhange relaxation.9.7.9.5.6 Natural (sum of above) 3..3.6 3.6 3. Atomi gain [db]..5 4.8 3.4.8 Osillation parameter.4.35.6.4.34 Theoretial Allan deviation σ y ( τ ) = Aτ Operational (s) Optimum (s).4e- 9.9E-3 3.E-3 7.e-3 7.3e-3 8.E-3 8.9E-3.8E-3 6.9e-3 6.9e-3 @.mbar @.mbar @.mbar @.mbar @.5mbar (.e3/s) (6.8e/s) (6.9e/s) (9.e/s) (3.e/s) -75dBm -8dBm -85dBm -8dBm -8dBm For the flux-independent relaxation effets, the bulb esape relaxation ontributes.hz due to the fat that we assure a onstant geometry for all the storage bulb, but the wall relaxation linewidth appears to be varying. The average of k e for 5 units is 6.e-3, whih is in good agreement with theoretial value of 5.6e-3 aording to []. The first PHM use Al avities oated by Alodine, having a quality fator of 58. The higher atomi gain and the osillation parameter leading to better frequeny stability for PFM are due to the narrower natural linewidth whih benefits mainly from the smaller wall relaxation effet. Sine the FM, the avity quality fator has been inreased from 58 to 95 by a silver oating of the magnetron avity. Signifiant higher atomi gain and higher osillation parameter have been ahieved by the quality fator improvement. Figure shows the frequeny stabilities of PP for GSTB-V EQM, PFM and FM, measured by the referene eletronis. The theoretial Allan deviations are alulated for the operational onditions orresponding to the measurements and assuming the same noise figure ontributed by the reeiver. The evaluation is in good agreement with the measurements of EQM and PFM, but too optimisti for FM. In this partiular ase, the stability of FM is limited by the noise of the
interrogation osillator, whih is limited to 6E-3 to 8E-3 within the maser line bandwidth. For further FM models, the loal osillator ould be seleted with a better short term stability to improve the instrument short term stability performanes if there is a real need at system level..e- Allan Deviation Sigma y (T).E-3.E-4 EQM PP, Aug.4 EQM, theo. PFM PP, Feb.5 PFM, theo. FM PP, Aug.5 FM, theo..e-5 Averaging time, τ, Seonds Figure. Measured and theoretial frequeny stabilities of Physis Pakages for GSTB-V EQM, PFM and FM For reent IOV EQM and QM, a modifiation of the C-field was performed due to weight onstraints. The initial assumption is that the modifiation ould have produed a small inrease of the magneti inhomogeneity. A system is presently under study in order to estimate the assoiated small magneti relaxation. The atomi gain and osillation parameter for IOV EQM are bigger than for QM. This is probably assoiated with a very small bakground vauum obtained for EQM. However the estimated best frequeny stabilities reah to the same value for these two masers. Figure shows the preliminary measurements of PP for IOV EQM and QM. The theoretial prediations are onsistent with the measurements. As shown in the table, the best frequeny stability is evaluated by optimizing the operating parameters. However, if the flux has to be inreased to favor the stability whih follows the penalty of the life time, the trade-off for the life time has to be made arefully.
.E- Allan Deviation Sigma y (T).E-3.E-4 Speifiation IOV EQM PP, Jul.6 IOV EQM, theo. QM PP, Ot.6 QM, theo..e-5 Averaging time, τ, Seonds Figure. Measured and theoretial frequeny stabilities of Physis Pakages for IOV EQM and QM V. CONCLUSIONS An effiient method for extrating the relevant maser parameters has been devised. The maser operational parameters for the 5 units tested are found to be in a reasonable range. The operational flux assures the lifetime speifiation. Exept for the st EQM, the stability of all other units at the operational parameters settings is found to be in the speifiations. The optimum frequeny stability determined by the PP parameters as demonstrated by FM an reah the level of 3E-3 at s, i.e. 3 times better than the speifiation, but in suh ase the frequeny performane is limited by the loal osillator noise. VI. REFERENCES [] P. Rohat, F. Droz, P. Mosset, G. Barmaverain, Q. Wang, D. Boving, U. Shmidt, T. Pike, L. Mattioni, M. Belloni, M. Gioia, and F. Emma, The onboard Galileo rubidium and passive maser, status and performane, in Proeedings of 5 Joint IEEE International Frequeny and Control Symposium and Preise Time and Time Interval (PTTI) Systems and Appliations Meeting, 9-3 August, 5, Vanouver, Canada, pp 6-3. [] J. Vanier and C. Audoin, The Quantum Physis of Atomi Frequeny Standards, (Adam Hilger), Vol., 988.
---- Communiation information: Qinghua Wang Temex Time Vauseyon 9, Neuhâtel, Switzerland E-mail: qinghua@temextime.om Tel: +4 3 73 96 74 Fax: +4 3 73 6 67 3