Z. Phys.

C - Particles and Fields 50, 2 1 ~ 9 (1991)

Zeitschrift for Physik C

9 Springer-Verlag 1991

Partides and Fields

Evaluation and preliminary measurement of the interaction of a dynamical gravitational near field with a cryogenic gravitational wave antenna
P. Astone 2 M. Bassan 4' 5, S. Bates 3 R. Bizzarri 1, 2 p. Bonifazi6' 2 R. Cardarelli 1,2 G. Cavallari 3, E. Coccia 4' s, A. Degasperis 1, 2, D. De Pedis 2, S. Frasca 1,2, E. Majorana 2' 8, L. Merucci 1, 27 I. Modena 4' s, G. Muratori 3, G.V. Pallottino a, 2 C. Patrignani 1, 2, G. Pizzella 1, z, M. Price 3, p. Rapagnani 1, 2, F. Ricci 1,2, M. Visco 2, s
Dipartimento di Fisica, Universit/t La Sapienza, Roma, Italy z Istituto Nazionale di Fisica Nucleate, Sezione di Roma, Roma, Italy CERN, Geneva, Switzerland 4 Dipartimento di Fisica, Universitfi Tor Vergata, Roma, Italy 5 Istituto Nazionale di Fisica Nucleare, Sezione di Roma 2, Roma, Italy 6 Istituto di Fisica dello Spazio Interplanetario, CNR, Frascati, Italy 8 Dipartimento di Fisica, Universitfi di Catania, Catania, Italy Received 30 May 1990, in revised from 24 October 1990

Abstract. We present a detailed analysis of the effect of the gravitational field generated, by a small rotating quadrupole on a gravitational wave antenna and we report on the preliminary measurement of this effect on the Explorer 2270 kg cryogenic gravitational wave antenna of the Rome group. The induced signal had an amplitude twenty times larger than the detector noise when the antenna was equipped with an FET amplifier and was easily detected without requiring integration in time. We remark that with this method we were able to make an absolute calibration of a gravitational wave antenna.

is the description of gravitational forces in the context of the general theory of quantized fields. In this approch Fujii in 1971 proposed a modified theory involving a massive particle as a boson associated to dilatation invariance [-2J. As a consequence of coupling between such a new massive particle and a graviton, Fujii found that the gravitational potential between two point particles should be
U = - G o ( M 1 M2/r)(1 + 7e-r/a).

(1.1)

I Introduction

The inverse square law of gravitation shows that two masses M 1 and M 2 at distance r are attracted by a force F
M F = _G ~ 1Mz r. (1.0)

Other theoretical approaches have led to a similar formula where the non-Newtonian term is in someway related to the unification theories. In the weak field limit, a law of gravitation that differs from the Newtonian law by a small amount can be interpreted just as well in terms of a variation of G with the distance r, to say that, according to (1.1), (1.0) holds with
G = G(r) = Go [1 + e-~/~(1 + r/2)].

(1.2)

Astronomical data have provided confirmation that, in the weak field limit, the newtonian law of gravitation is obeyed to high accuracy [1]. Since publication of Einstein's gravitational theory (Einstein 1916) general relativity has become dominant in the field of gravitation. However, as we know, the description of gravity as a manifestation of curved space has very little connection with the rest of physics, so that a great deal of theoretical effort has been spent on attempting to unify gravity with the other forces of nature. One step in this direction

Among the physical constants G is the worst known: all other constants are known better than 1 x 10 -6, while the latest value [3] of G has an uncertainty that amounts to about 128 x 10 -6. Thus, although the product of G with the mass M of an astronomical body may be determined with a precision of 10 -6 (by observation of the satellite orbits), it is possible that a noticeable difference exists between G measured at astronomical distances and in the laboratory. In 1976 Long 1-4] claimed that the inverse square law of gravitation breaks down at a distance of the order of 0.1 m. In direct contradiction to Long's work, however, Spero et al. [43 performed another laboratory ex-

22 periment which verified the validity of the Newton law at distances of the order of 10 cm. Stacey and Tuck [5] gave a series of measurements of the variation of the free fall acceleration g with depth in mines and claimed that the G value determined from this data is larger than laboratory-determined values by between 0.5% to 1%. In the paper published in 1986 Fischbach et al. [6] suggested that the non-Newtonian term, rather than being part of the gravitational field, may be due to a hitherto-unknown medium-range baryon-baryon interaction. This assumption was based on a reanalysis of the E6tv6s torsion-balance data. The Fischbach hypothesis stimulated new experiments in the field. These experimental efforts were both aimed at testing the 1/r dependence and the existence of composition dependent effects. The tests of the 1/r dependence have been mostly based on the study of the variation of g with altitude, both in deep boreholes and high towers. These experiments led to a much increased understanding of the relevance of the local topography in the estimation of the free air gradient [71 and finally no compelling evidence for deviations from the 1/r dependence of the potential has been found [81. In addition, for the composition dependence experiments, some early results [91 showing indications for such an effect, were not confirmed and there is a large amount of negative results [101. Tests of the 1/r dependence of the gravitational potential have also been performed using generators of timevarying gravitational field and mass quadrupole antennas (such as those developed for the search of gravitational waves) as detector. With this method the ambiguities due to the surrounding mass distribution are completely eliminated, which makes possible to operate over a wider range of distance r (up to 50 m). The first attempt was done in 1968 by Sinsky at the University of Maryland [11]. A dynamic Newtonian field was generated by vibrating a cylinder of around 100 kg at 1600 Hz at distances of the order of two meters from the room temperature gravitational wave antenna of Weber. He observed an increase of the vibration energy of the order of 20% above the Brownian noise level of the detector. Starting from 1980, the gravitational wave group of the University of Tokyo carried on several experiments with room temperature antennas to measure the effect of the dynamic gravitational field generated by a rotating quadrupole source [12]. These experiments were done in a range of frequency below 100 Hz covering a range distance from 0.1 m to 10 m. The accuracy of this kind of experiment is governed by the antenna Brownian motion whose effect was decreased with a long integration time. They assumed that the gravitational force had the form 1/r 2+~ obtaining for the 6 parameter 6=(2.1 _+6.2). 10 .3 in the range 2.6-10.7 m while for the range 0.1-0.3 m they found 6=(-0.7-t-2.9). 10 -3. Since 1985 the gravitational wave group of the Rome University has installed at CERN a cryogenic gravitational wave antenna of 2270 kg and of 916 Hz resonance frequency [-131. This antenna, equipped with a resonant capacitive transducer and a D.C. SQUID amplifier, operated at 2 K reaching in 1989 the goal of a burst noise temperature of 7 mK. With this sophisticated gravitational field detector, the Rome group conceived two possible experiments. One is directly related to the possibility of recognizing a deviation from the 1/r dependence of the Newtonian potential by measuring the signal into the antenna due to a gravitational near field generated by a rotating source similar to those used by the Tokyo group. This approach should be considered as preliminary to performing the second experiment, i.e. a null experiment in which it could be measured the field gradient along the antenna due to a rotating baryonic dipole, a nearly simmetric rotor with suitable materials, which makes the antenna only sensitive to the composition dependent interaction [14]. We report here a preliminary experiment aimed at the detection with the antenna of the field produced by a rotating quadrupole. This measurement, which can also be regarded as an absolute calibration of the gravitational wave detector, represents the first step to define the feasibility of the experiment using the baryonic dipole. In Sects. 2 and 3 we analyse of the source-detector interaction, in Sects. 4 and 5 we give a schematic description of the antenna and the rotor, and in Sect. 6 we report the results obtained with our experimental apparatus.

2 The evaluation of the dynamic gravitational signal

Here and in the following section we compute the oscillation amplitude of the antenna, which is an aluminium cylinder, induced by a gravitational generator, i.e. an object of suitable shape which rotates with angular frequency c3, corresponding to one half of the resonance frequency of the aluminium bar. The motion of the cylinder is analysed in terms of its normal modes, and, in particular, the first longitudinal vibration mode is considered. Moreover, in order to simplify the computation we neglect for the moment the perturbation induced by the transducer on the spatial distribution of the vibration mode. Let us set the origin of the cartesian coordinate system at the center of the bar, a cylinder of radius R and length L; the generator is a smaller object with density

p(x).
The acceleration field a(x, t) produced by the generator at the point x ~ (xl, x2, x3) of the detector is, according to (1.1): a(x, t)= Go Vx S dyfi(y, t) [qb(x- y) + c~~ ( x - Y ) ] (2.1 a)

where ~(x):lx1-1, and ~(x):lx1-1 e -Ixl/z is the nonNewtonian term; moreover fi (y, t ) = p (~) where the integration variable y is related to the new variable ~ by the (time dependent) expression y (t) = 6 fi + R (t) r (2.2)

23 Here 6 i is the position vector (Jill = 1), 6 is the distance of a fixed point of the generator, lying on the rotation axis fi, and R(t) is the rotation matrix. With this change of variable the integral expression (2.1 a) takes the more explicit form a(x, 0 = G 0

7=~FM
F = -1 ~ dxv(x), v { [x2-i(x3-6)]2 ~ 2 detector --X\[x'2-}-X2"l-(X3--)215/21 M=~
1

(2.8) (2.9)

gx

~ d~p(r
generator

(2.1b)

~
generator

dCp(r

+i~2) 2.

(2.10)

Here, and in the following, we set e = 0 because the preliminary experimental data reported below do not justify taking into account non-Newtonian contributions, since is expected to be less than one percent. Now we consider the motion of the elastic cylinder due to the acceleration field. A normal mode, with angular frequency COb and displacement field given by the eigenfunction v(x) has amplitude q(t) which satisfies the following differential equation d2 COb d 2 d t 2 q (t) + ~ ~ / q (t) + cob q (t) = g (t) where dxv(x) a(x, t)
detector

(2.3)

These expressions are obtained by choosing the origin of coordinates (xl, x2, x3) to be at the centre of the cylinder, the axis xa to coincide with the cylinder axis, xl to be along the vertical direction, and, as is the case in the present experimental set up, the generator center to be at distance 6 on the x3 axis and to rotate around the vertical axis, i.e. u 1 = / 2 2 = 0 , //3= 1, n z = n 3 = 0 , n l = 1 (Fig. 1). To proceed further we must introduce the displacement field v(x) corresponding to the first longitudinal vibration mode. An approximate expression is v(x)-(0, 0, sin ~-~3). (2.11)

(2.4) dxv(x) v(x) "

g(t) =

~
detector

In order to calculate this integral, we perform a multipole expansion of the acceleration field given by (2.1 b). This expansion yields a power series representation of the acceleration a with respect to the variable R (t)

This approximation is justified by the low radius - length ratio for our antenna, R/L=O.1. We note also that the corresponding approximate eigen-frequency COb=TCv/L , where v is the velocity of sound, is very close to the observed value of 916 Hz. Under this assumption the normalization factor of the formula (2.7) is v (x). v (x) d x =
detector

rcR2 L 2

(2.12)

g(t) ~ = ( o n + ~ + n+ exp(i05t)+ (_ n_ exp(-i05t)

with
~o=i.~

(2.5)

Assuming R2 ] - 5/2
~ 1 (2.13) 1 -~ (~ _ L / z ) ~

4_+ =~22(~1 q- i~2) the integration in (2.9) yields

fi+_=--~(fil + ii2)

(2.6)
~xI

where fi~, fi2 and the rotation axis i form an orthogonal Cartesian frame. Because of the time dependence (2.5) of the vector R(t) ~, each term of the power expansion is a linear combination of terms oscillating with angular frequency 0, 05, 2 05... Since the frequency of the resonating contribution is 205, the lowest power term comes from the quadrupole moment of the generator, and this yields for the forcing term g(t) in (2.3) (see (2.4)) the expression g(t)= ~
detector

\/V-

Go dxv(x) v(x) [? exp(iZ05t)+?* exp(--i205t)] (2.7)

where

Fig. 1. Model for the rotor-antenna system, c5 is the angular frequency of the rotor and 6 the distance between the rotor and the center of mass of the detector

24 f F_ L2 6~ 2 L2~2 quencies co+, co_ are related to the frequencies of the uncoupled oscillators as: co_+: ]/l[(co2(1 +/~) + 0,2) ___1//(co2(1 +/~) + co~)2- 4co~ co~-I. (3.2) The electrical signal of the transducer is proportional to the relative displacement of the transducer to the bar end
(t) = y (0 - x (t).

1/6~\3[

05) ~[.~.[ ~+L/2\

where Ci and Si are the integral cosine and integral sine functions. In the approximation L/2 ~ 6 a rough estimate of F is provided by the simpler expression F-~-

We have checked that the direct action of the rotor on the transducer mass is negligible in the actual experimental set up, thus by applying the Fourier transform to (3.1) and setting f = 0, we find

Y(0,)-x (co)= (~)

where

12R z L2
7c~

D(0,)

co~

(3.3)

(2.15)

Using (2.7), (2.8) and (2.15), we conclude that the forcing term in the equivalent motion equation of the detector (2.3) can be expressed as g(t)=go cos 2o5t-72Go L M rc2 65 cos 2o5t (2.16)

D (0,) = [(0,~ - 0, 2 + ~ 0,})(0,} _ 0, 2) _ ~ co# _ co2 co, o,b/1 Q, 0 d L

(~Ob 2 2 (Dr +*0,[Q (co,-co)+Q~(0,~2 - c o ~"

corco21

(3.4)

where M, as specified by (2.10), depends on the geometry of the rotor (see below). At distances 6 ~ L this formula does not hold and one has to use the full relation expressed in (2.14) in which it is only assumed that the transversal dimensions of the antenna are negligible.

The function D (co) can be rewritten in terms of the quantities co+, co_, Q +, Q_ (merit factors of the modes), that can be measured directly, as:
90,0,+1[ ~ .coco-

(3.5)

3 The antenna response to the monochromatic gravitational field

The gravitational wave antenna of the Rome group (Explorer), installed at CERN, is equipped with a resonant capacitive transducer. The resonant transducer can be modeled as an ascillator with angular frequency 0,, and reduced mass mr that is coupled to the bar which, in its first longitudinal mode, has angular frequency cob and reduced mass rob. The transducer and the bar make a system of two oscillators which are well coupled if 0,r ~-0,b. The equations of motion for the two oscillators are: 9 0,b x.+ ~ x +90 , b 2 x+g~tr(x--Y)+g0,}(x--Y)=fbrnb Y + ~ @ - 2 ) + 0 , } ( y - x ) = ft mt (3.1a) (3.1b)

which shows that in the neighbourhood of each mode the frequency response closely approximates that of a Lorentzian (second order) system. When the rotor turns with o5- 890, + or o5-~ 89 0,_, the corresponding sinusoidal excitation is expressed by (2.16) as an equivalent acceleration of the oscillator modeling the antenna normal mode OCb(t)/mb=g(t)). The steady state amplitude of the mechanical output signal, after a transient with a time constant z_+ = 2 (Q__/co-+), is therefore:
0,2

tl m_go (0,2 __ 0,2) E(0,2 __ 0,2)2 jr_ (0, 0, _+/Q +-)2"11/2

(3.6)

with co=2(5. The transducer capacitor, biased at constant charge, converts the mechanical signal q into an electrical one which is proportional to its static electrical field E. The voltage signal at the capacitor is C Vt(t) = YrE ~ t/(t) Ct~p (3.7)

where x is the displacement of the bar end, y the displacement of the transducer, fb a n d f indicate the forces which act respectively on the bar and on the transducer, # =mJmb, and Qr and Qb are the quality factors of the two oscillators. As it is well known, these two coupled oscillators have two vibration modes, whose angular fie-

where 7t is a geometrical factor, C is the active capacity of the transducer, Cp its parasitic capacity (see Table 1). This voltage is usually amplified by the DC SQUID,

25

Table 1. Characteristics of the experimental apparatus

Frequency of the bar (first longitudinal node): Mass of the bar: Frequency of the transducer (first flexural mode): Mass of the transducer (disk): Ratio of the two equivalent masses: Capacity of the transducer: Decoupling capacity: Parasitic capacitance of the transducer: Transducer geometrical factor: Quadrupole Moment Rotor Mass Antenna length Antenna-Rotor Distance Transducer gap vb= 915.765 Hz M b= 2270 kg vt =916.6 Hz Mt = 0.402 kg #=3.064x10 4 C=3891 pF Cd= 94 nF Cp= 500.0 pF 7t=0.865 M=5.55.10 3kgm 2 mR= 8.75 kg L = 2.97 m S= 3.52_+0.05 m do = 5.4 gm

4 The dynamic gravitational field generator

The generator, conceived and constructed at CERN, is a rotor of 8.75 kg mass made of the aluminium alloy E R G A L 65. It has a cylindrical shape with 20 cm diameter and 12 cm height; at 7.03 cm from its axis two vertical slices have been removed breaking the cylindrical symmetry. Along the axis an inner concentric hole is made in which an elastic rotation support is inserted. Using the geometrical data given above, we evaluate, on the basis of the definition (2.10), the quadrupole moment of the rotor M as 5.55.10 3 kg. m 2. The support has been constructed in such a way that its fundamental mechanical frequency is much lower than the stationary rotation frequency. As a consequence, the center of mass of the rotor is made to lie on the rotation axis. The precession motions, which might arise during the acceleration phase, are magnetically damped by means of four magnetic coils which are symmetrically set around a small iron rod installed in the rotor top center. The contact between the rotor and the elastic support is made with a ruby sphere of 5 mm diameter, this is continuously lubricated. A turbine is mounted on the rotation axis. It is driven by two high pressure oil jets, and it rotates dragging the rotor. The driving torque is rather small: in fact it takes about one hour for the rotor to reach a rotation frequency of about 450 Hz. The pressure oil variations are kept within 7% by means of a suitable hydraulic circuit. The motion of the rotor is monitored using a small magnetized needle (which is balanced on the symmetric side with an identical non - magnetized needle). The needles are located on the bottom of the rotor at 2 cm from the center. During the rotation the magnetized needle comes close to a fixed pick-up coil providing a voltage signal which is used for the rotor phase control. The pick-up signal is hardware processed and transformed in a logic signal which has a period equal to the rotation period of the rotor. An aluminium plate of 2 cm of thickness and 20 cm in diameter is rigidly connected to the rotor bulk. Under this plate there are 10 copper coils, uniformely distributed along the rotor base in a plane parallel to the plate. This system is used as an electromagnetic brake, which is driven by the control system described below in order to keep the angular frequency (5 of the rotor at the designed value as well as to control the phase of this motion. The rotor and its driving system are located inside a cylindrical vacuum chamber of 3 cm of thickness made of stainless steel. The chamber is rigidly connected to a large support, located near one end of the gravitational wave cryostat. F o r safety reasons, we have interposed a wooden wall of 10 cm of thickness between the rotor and the cryostat and assembled a concrete " blockhouse" all around the rotating source. The motion of the rotor can be approximately described by the following law

but the final run of measurements reported in Sect. 6 was done using a F E T amplifier. Introducing the electromechanical transfer function W between the relative displacement q and the voltage at the amplifier output, the r.m.s, value of the output voltage due to the monochromatic near field is
Vr.....

1/2

(3.8)

and using an F E T amplifier

W=7,E c+c~A

(3.9)

where A is the gain of the F E T amplifier. When using a S Q U I D amplifier a simplified expression of the transfer function is

1 C w(co) = Asquid IZ(co)l ~' E C + ~ v" r

(3.10)

the impedance of the R-L-C electrical circuit where flows the signal current generated by the transducer and Asquid is the transfer function between this current and the output voltage of the SQUID. The fact that Z (co) is now complex introduces the additional complication of having a third oscillator of electrical nature coupled to the two mechanical oscillators. The coupling, which depends on the intensity of the electrical field E, is weak and can be neglected in the first approximation; however changes in the operating point of the SQUID, which are accompanied by variations of its input inductance, can produce very small variations of the mode frequencies v+ and v_. Due to the high Q values of the modes (106 - 107), even frequency variations of 1 part in 107 strongly affect the response to a monochromatic excitation. This effect is, of course, absent when using the F E T amplifier because in this case we do not have a third oscillator.

Z(co) is

I d~=Ap-(B ~-p+H+Ki2)c5

de5

(4.1)

26 where p is the pressure of the oil, i is the current in the coils of the brake and I, A, B, H and K are constant quantities. The current i is the sum of a constant term io, that coarsely balances the effect of the driving pressure p (which is kept approximately constant during the operation of the system), and of a variable term 6 i, whose value is computed by the control system. When the rotor angular frequency (5 is close to the desired value (5o (one half of the antenna mode that we want to excite), it can be expressed as (5=(5o+6(5. In this condition, the above equation (4.1) can be linearized: neglecting the product 6 056 i, we obtain: ~ t 6(5= where 6(5 To 2ioK(5o 6i I (4.2) 448 Hz, that is close to one half of the resonance frequencies of the modes. The phase OR of the rotor, measured with reference to the master synthesizer (as we did in all the phase measurements reported in what follows), after an initial transient settled to a value of 0 +_ 10~ The square wave obtained from the phase pickup was sent to a lock-in amplifier driven by the master synthesizer. The lock-in output, sampled at 1 s for 10 hours, was recorded on magnetic tape and analysed off-line. In Fig. 4, we show the fast fourier transform (FFT) of this signal obtained with all the data of this run. As shown in the figure, the peak lies in just one bin of the discrete spectrum, defining an upper limit of 30 gHz for the spectral width of the rotor signal.

5 The detector of the dynamic gravitational field I TO=B ]/~+ H + Ki2

(4.3) A detailed description of the gravitational wave antenna of the Rome group (Explorer) installed at the CERN laboratory in Geneva has been given in previous publications [13, 15]. We recall here the basic scheme of the experimental apparatus. The antenna is a cylindrical bar (60 cm in diameter and 297 cm in length) of A1 5056 with mass M b-- 2270 kg and reduced mass mb= Mb/2 for the fundamental longitudinal vibration (Vb=915.765 Hz at liquid helium temperature). The antenna is located into the vacuum container of a cryostat which has a liquid helium capacity of 3300 liters and it is suspended with a titanium alloy cable wrapped around the circle containing the center of gravity. The titanium cable constitutes the first of a multiple stage system filtering external mechanical disturbances with a total attenuation of 200 dB. The resonant capacitive transducer is located at one end of the bar. Its moving part has the shape of a mushroom and a mass of 0.4 kg. It vibrates in its first fundamental flexural mode with reduced mass mt= 0.34 kg and frequency vto=916.6 Hz (with zero bias electrical field). The gap between the plates of the transducer capacitor is d = 51 gm; its bias electrical field can be adjusted up to 8 x 106 V/m by means of an external battery. The transducer frequency v~ depends on the electrical field in the gap, thereby affecting the frequencies of the modes: for example a variation of 2.104 V/m on the bias electrical field gives a frequency change A v=2 mHz at E = 8.105 V/m, A v = 30 mHz at E = 8- 106 V/m. The electrical signal of the transducer is applied to the amplifier through a decoupling capacitor Ca. When using the SQUID, the transducer is disconnected from the external battery and a superconducting transformer provides the required impedance matching to the low input impedance of the amplifier. When using the FET amplifier [16], we maintained the transducer connected to the battery through a 10 Gf~ resistor. As shown in Fig. 2, the signal output V(t) from the amplifier is sent to lock-in amplifiers which demodulate it at the mode frequencies (v+, v_). The outputs of the various lock-in amplifiers (V-, V are sampled at time intervals of A t = 1 s, processed by a 12 bit analog to digi-

is the natural time constant of the rotating system. Its value ranges between half an hour and one hour, according to the values of p and io. The pick-up signal provides the information required for controlling the motion of the rotor. It is compared with the square wave generated by the master synthesizer of the system (driven by a very stable Rb oscillator) whose frequency is set at one half of one of the resonance frequencies of the detector. The phase difference ~PR between the pick-up signal and the synthesizer output, that is the error signal, is processed by the control system which is implemented in software with a Mac II personal computer and in hardware with VME electronic modules. More specifically, the error signal is obtained by sending the pick-up signal and the reference square wave to a VME time to digital converter (TDC). Its digital output, representing the phase q0R, is processed by the computer which controls, through a 16 bit VME digital to analog converter, the current of the brake of the rotor. The control law follows the proportional integral derivative model (PID):

6 i=acpR + b S ~~ dt +c ddq~ t

(4.4)

where the coefficients have been chosen on the basis of the above equation (4.2) to provide the required accuracy as well the necessary stability in closed loop conditions. The derivative term is necessary to increase the damping of the closed loop system, because of the high value of the intrinsic time constant of the rotor. The integral term is necessary to obtain a good accuracy in spite of any small variation of parameters such as the oil pressure or the friction. When the control program is running, after the initial transient, the rotor turns at the same frequency of the master synthesizer, with the same phase within _+10~ The performance of the rotor system, in particular its phase and frequency stability were finally checked experimentally. We drove the rotor at the frequency of

27

b~vb
....... r q [ : --~--:~ ]1
I

time RUBIDIUM CLOCK ~ - ~

ELECTRONIC CLOCK [ IMAGNETIC TAPE

F~
I '

LOCK-IN , ~ I
~v+ ' I

~ frequency
MASTER SYNTHESIZER at v / 2

UN1TI

',

pS.N .ESIZER

COMPUTER DATA ACQUISITION AND STORAGE Ix(t) lY(t)

Z-._---_--A SUPERFLUID

--

--HEL~MT~:M._.P~._a2T.._UR._E. ]

insulator I I

1-I

FREQUENCY I

II II

I PHASE PICKUP I

J LOK.I i
AMPLIFIER DRIVEN at v AMPLIFIER I ( SQUID or FET ) ]

[ osc, R\To.

..T9

C
TRANSD ER

Fig. 2. Scheme of the gravitational wave detector when the FET amplifier is used. The part contained in the dashed section is cooled at low temperature

Fig. 3. Schematic representation of the experimental setup

tal converter and sent to a Nord 100 computer for real time analysis and recording on magnetic tape. On the same tape we also record the Universal Time, as provided by an electronic clock controlled by a Rb frequency standard, which provides also a stable reference to the synthesizers that drive the lock-in amplifiers.

[ V/(Hz)'/2]
2800 2400 2000
I I I I [ I I I I

6 Experimental

results

1600 1200 8O0 400 0 4.8

i i i i i i E i i

We report here on the experimental results, obtained during various measurement runs done with the goal of performing an absolute calibration of the gravitational wave antenna. We used the experimental setup schematically shown in Fig. 3: the first runs with the DC SQUID, the last with the F E T amplifier. F r o m the two output x(t) and y(t) of the lock-in it was possible to obtain the amplitude V = ~ 2 and the phase ~0s= tg-1 (.V/X) of the oscillations of the antenna relative to the master synthesizer. The phase of the oscillations of the antenna relative to the exciting rotor are then be obtained as q) = qoR + (Ps. This phase information contains an arbitrary additive constant due to the delays introduced by the electronic circuits which have not been measured. As regards the antenna response to the sinusoidal excitation produced by the rotor, we remark its strong dependence on both the merit factor of the excited mode and the value of the driving frequency. In particular the high Q values of the antenna imply that the bandwidth of the resonances (160 gHz for a Q -~6 x 10 6) is very small and therefore the resonance frequencies should be known within a few tens of gHz. Furthermore, the time constants of the system are of the order of half an hour and both the excitation frequency and the resonance frequency must be stable well within the bandwidth for times of the order of a few time constants. The excitation frequency can be maintained stable well within these limits as has been shown in Sect. 4. The mode frequencies, however, are not stable with time, since they depend on various quantities: besides a very small temperature effect, they usually exhibit a small drift

4.88

4.96

5.04

5.12

5.2

[ mHz ]
Fig. 4. Fast fourier transform of the rotor pick-up signal9The resolution of the spectrum is 30 gHz

(of the order of i mHz/day), which we attribute to slow variations of the bias voltage of the transducer, as well as to minute sudden variations. The latter effect is mainly contributed by changes of operating point of the SQUID, which are accompanied by variations of the input inductance of this device. We made some preliminary runs using the SQUID. The effect of the rotor on the antenna was clearly visible and the excitation of the antenna raising with about the appropriate time constant up to amplitudes of twentythirty times the Brownian motion has been observed. However it was not possible to attain steady conditions and the observed amplitude was always below the expected amplitude at the resonance peak. We attribute this results to the above mentioned instabilities of the mode frequencies. On the basis of the above results we decided to perform the final run at v + using the F E T amplifier, whose operation is more stable than that of the SQUID. We

28 biased the transducer with a low electrical field, E =4.51 x 105 V/m, in order to minimize the frequency drift effects mentioned above and to limit the back-action effect of the noise amplifier on the Brownian noise level of the detector [17]. At the beginning of the run w e carefully measured the resonance frequency of the v+ mode (v'+ =924.516358 Hz with an error of ~ 8 gHz) by exciting the mode and observing the free decay of the antenna signal through the lock-in driven at v +, measuring also Q'+ = 6.2 x 106. Afterwards, when the rotor was running at v'+/2 with a relatively stable phase (~0R - - 10+ - 2 0 ~ the antenna signal increased exponentially. After a time of the order of 3z+, it reached an amplitude of 18.7 mV with a phase of ~0s~-- 100 ~ corresponding to a phase (p -~ - 115 ~ with respect to the exciting rotor. However, after 3 z +, the output signal was not longer stable and we observed a decrease in the oscillation amplitude for non obvious reason. In order to reproduce the above experimental condition, we systematically changed the excitation frequency and after two hours we found a new maximum of the oscillation amplitude at a frequency of the master synthesizer changed by 4 0 g H z , that is exciting the antenna at v'; =924.516438 Hz. During the plateau the values of ~o s ~_ - 120~ and OR - 0 + 10~ corresponding to q) ~ - t 15 ~ were measured. The antenna response as a function of the time is shown in Fig. 5. A fit of the data shows again the appropriate exponential behaviour with z = 2 0 0 6 s and a final value of 19.1 mV. Having observed the antenna oscillations for a time of 4-c+, we again measured the resonant frequency of the v+ mode. Having stopped the exciting rotor, the mode was mechanically excited and within 15 minutes we started to record the free decay which was followed for 1 hour. We found an exponential decay with Q'~_ =6.0-+0.2-106 and a resonant frequency coincident within 6 gHz with the last value v~_ of the excitation frequency. The experimental results obtained at v'+ and v'; can therefore be explained as due to a variation of the resoR

nance frequency of the modes during the 8 hours of the run. By means of a germanium thermometer located on the antenna, we checked that the frequency change due to the temperature variations of the bar cannot justify this drift. The most probable explanation is related to small changes of the polarization state of the dielectric of the capacitor Ca, that just before the beginning of the run was lowered from 200 V to 23 V. On the basis of these results we conclude that a) the oscillation amplitudes measured at v'+ and v'; were both at an excitation frequency equal to or very near to the resonant frequency of the mode at the time of the measurements (in the two cases the phase had the same value), b) a variation of 80 gHz of the frequency of the mode occurred during the measurements. As mentioned before, in the second measurement the time interval between the end of the external excitation and the beginning of the frequency measurement was only 15 min. We consider this measurement as our best estimate of the antenna response to a periodic gravitational excitation. F r o m the fit shown in Fig. 5 we obtain
Vm . . . . . . d =

19.1 --+0.7 mV.

The stochastic fluctuations of the data are mainly due to two different processes: the wide band electronic noise and the narrow band equivalent thermal noise of the detector. The wide band electronic noise is basically uncorrelated and it gives a neglegible contribution to the uncertainty on the final value Vme,surea evaluated using all the data in the fit, while the correlation time of the equivalent thermal fluctuations is the time constant z+ of the mechanical system [17]. As a consequence, the error on the measured value is deduced from the variance of the thermal fluctuations multiplied by the ratio between the thermal correlation time (z +) and the observation time (4z+). This result is to be compared with the value calculated from (3.6), (3.8) and (3.9):
Vexpeeled =

21.1 4- 0.6 mV

[mY]
25 20 15 10

where the uncertainty is mainly due to the error in the estimation of the Q factor. We think that the agreement is rather satisfactory, taking into account the mentioned instabilities. We now plan to set up an improved experimental apparatus with a more powerful rotor in order to repeat the measurements. But we deemed it appropriate to present the results we have obtained since this is the first time that an absolute calibration of a cryogenic gravitational wave antenna has been performed with a dynamic gravitational near field in 1 kHz range.

0
0 20 40 60 80 [min] 100 120 140

7 Conclusion
We have detected the effect of the gravitational field generated by a small rotating gravitational quadrupole on the 2270 kg cryogenic gravitational wave antenna of the

time

Fig. 5. The detector output R versus the time during the excitation with the rotor turning at v+/2. The continous line is the best fit of the data with an exponential law

29 R o m e group. The induced vibrations have an amplitude a b o u t 20 times the amplitude of the B r o w n i a n m o t i o n and could easily be detected w i t h o u t having to integrate the signal in time. W h e n the a n t e n n a was equipped with a F E T amplifier the amplitude of the detected signal was in agreement with the expectations and therefore represents an absolute calibration of the antenna. W e have encountered some difficulties in controlling the detector frequency to the accuracy required by its high Q value. The developement of an a p p r o p r i a t e " c o l d d a m p i n g " technique [12] should help to circumvent this problem. A r o t o r with a q u a d r u p o l e m o m e n t 100 times bigger is under construction to continue these measurements with a higher sensitivity which should allow a study of the r dependence of the generated field. 8. C. Jekeli, D.H. Eckhardt, A.J. Romaides: Phys. Rev. Lett. 64 (1990) 1204; M.E. Ander, M.A. Zumberge, T. Lautzenhiser, R.L. Parker, C.UV. Aiken, M.R. Gorman, M.M. Nieto, A.P.R. Cooper, J.F. Ferguson, E. Fisher, G.A. McMechan, G. Sasagawa, J.M. Stevenson, G. Backus, A.D. Chave, J. Greer, P. Hammer, B. Lyle Hansen, J.A. Hildebrand, J.R. Kelty, C. Sidles, J. Wirtz: Phys. Rev. Lett. 62 (1989) 985; J. Thomas, P. Kasameyer, O. Fackler, D. Felske, R. Harris, J. Kammeraad, M. Millett, M. Mugge: Phys. Rev. Lett. 63 (1989) 1902 9. P. Thieberger: Phys. Rev. Lett. 58 (1987) 1066; P.E. Boyton, D. Crosby, P. Ekstrom, A. Szumilo: Phys. Rev. Lett. 59 (1987) 1385 10. C.W. Stubbs, E.G. Adelbcrg, F.J. Raab, J.H. Gundlach, B.R. Heckel, K.D. McMurry, H.E. Swanson, R. Watanabe: Phys. Rev. Lett. 58 (1987) t070; V.L. Fitch, M.V. Isaila, M.A. Palmer: Phys. Rev. Len. 60 (1988) 1801; R.N. Coswich, N. Krishnan, S.N. Tandon, C.S. Unnikrishnan: Phys. Rev. Len. 61 (1988) 2179; P.G. Bizzeni, A.N. Bizzetti-Sona, T. Fazzini, A. Pergeo, N. Taccetti: Phys. Rev. Lett. 62 (1989) 1989; C.C. Speake, T.J. Quinn: Phys. Rev. Lett. 61, (1988) 1340; T.N. Niebauer, M.P. McHugh, J.E. Faller: Phys. Rev. Lett. 59 (1987) 609; K. Kuroda, N. Mio: Phys. Rev. Lett 62 (1989) 1941 11. J. Sinsky: Phys. Rev. 167 (1968) 1145 12. H. Hirakawa, K. Tsubono, K. Oide: Nature (London) bf 283 (1980) 184; Y. Ogawa, K. Tsubono, H. Hirakawa: Phys. Rev. 26 (1982) 729; K. Kuroda, H. Hirakawa: Phys. Rev. D32 (1985) 342; N. Mio, K. Tsubono, H. Hirakawa: Phys. Rev. D36 (1987) 2321 13. E. Amaldi, C. Cosmelli, G.V. Patlottino, G. Pizzella, P. Rapagnani, F. Ricci, P. Bonifazi, M.G. Castellano, P. Carelli, V. Foglietti, G. Cavallari, E. Coccia, I. Modena, R. Habel: Nuovo Cimento 9C (1986) 829 14. E. Amaldi R. Bizzarri, A. Degasperis, G. Muratori, G.V. Pallottino, G. Pizzella, M. Price, F. Ricci, C. Rubbia: Nota Interna del Dipartimento di Fisica, Universit/t di Roma ~ Sapienza", 866, 1986 15. F. Bronzini, E. Coccia, I. Modena, P. Rapagnani, F. Ricci: Cryogenics 25 (1985) 234 16. G.V. Pallottino, G. Vannaroni: IEEE Trans Instrum. Meas. 34 (1985) 676 17. G.V. Pallottino, G. Pizzella: Nuovo Cimento 4C (1981) 273

Acknowledgements. We express our thank to M. Michel Delattre for his valuable contribution to the construction of the rotor and for the technical support given during the runs of measurement.

References
1. A.H. Cook: In "300 Years of Gravitation", S.W. Hawking, W. Israel (ed) pag. 51, Cambridge: Cambridge University Press 1987 2. Y. Fujii: Nature (London) 260 (1976) 417 3. E.R. Cohen, B.N. Taylor: CODATA 63 (1986) 12 4. D.R. Long: Nature (London) 260 (1976) 417; I.L. Spero, J.K. Hoskins, R. Newman, J. Pellam, J. Schultz: Phys. Rev. Lett. 44 (1980) 1645 5. F.D. Stacey, G.J. Tuck: Nature (London) 292 (1981) 230 6. E. Fischbach, D. Sudarsky, A. Szafer, C. Talmadge, S.H. Aronson: Phys. Rev. Lett. 56 (1986) 3 7. D.F. Bartlett, W.L. Tew: Phys. Rev. Lett. 63 (1989) 1531