NIKHEF PROJECT PLAN: BENCHES Vibration isolation for external benches and suspension of internal benches for Advanced Virgo

Transcription

1 1 NIKHEF PROJECT PLAN: BENCHES Vibration isolation for external benches and suspension of internal benches for Advanced Virgo Responsible: J.F.J. van den Brand 1, 1 Nikhef, National Institute for Subatomic Physics, P.O. Box 41882, Amsterdam, the Netherlands June 21, address: jo@nikhef.nl Procedure start date 01/07/2010 Procedure end date 01/10/2013 Document version v01r01 Version date 20/06/2010

2 2 Abstract Several benches are used in Virgo to support optical systems. These benches need to provide sucient vibration isolation to obtain the sensitivity required for Advanced Virgo. Nikhef will implement vibration isolation for 6 benches that are outside the vacuum system, the so-called external benches. In addition, Nikhef is responsible for the suspension of the injection and detection benches, and the end-mirror payload for the input-mode cleaner. These systems are located inside the ultra-high vacuum system. Project collaborators Name Institution Thomas Bauer Nikhef thomas@nikhef.nl Mark Beker (PhD student) Nikhef mbeker@nikhef.nl Mathieu Blom (PhD student) Nikhef mblom@nikhef.nl Jo van den Brand (project leader) Nikhef jo@nikhef.nl David Rabeling (postdoc) Nikhef davidr@nikhef.nl Martin Doets Nikhef martind@nikhef.nl Eric Hennes (techn. coord. multi-pay) Nikhef e.hennes@science.uva.nl Richard Rosing Nikhef rrosing@science.uva.nl Willem Kuilman Nikhef y21@nikhef.nl Frans Mul (techn. coord. ext. benches) Nikhef/VU fmul@nikhef.nl Joost Rosier VU j.rosier@fbw.vu.nl Bert Clairbois VU h_e_clairbois@fbw.vu.nl Eric Genin (AdV INJ project leader) EGO eric.genin@ego-gw.it Piero Rapagnani (detector coordinator) Roma1 Piero.Rapagnani@roma1.infn.it Enrico Calloni (commissioning coord.) Napoli calloni@na.infn.it Franco Frasconi Pisa franco.frasconi@pisa.infn.it Riccardo De Salvo Caltech desalvo@ligo.caltech.it Alessandro Bertolini Albert Einstein Institute Alessandro.Bertolini@aei.mpg.de

3 3 Contents 1 Introduction 5 2 Technical description: External benches Overview of vibration isolation systems TMC - Stacis Damped springs Pneumatic Isolators The Minus-K system The EIB Seismic Attenuation System (HAM-SAS) Introduction Mechanical design for EIB vibration attenuation Sensors and actuators Control system Expected performance Installation procedure Action items Technical desription: Internal benches Multi-payload suspension Payloads for long-ndrc Action items Logistics Deliverables Involved Virgo subsystems Involved Nikhef and EGO infrastructure Planning Design nalization, call for tender EIB-SAS Production Integration EIB-SAS Implementation plan and maintenance Budget Manpower Nikhef manpower EGO manpower

4 4 4.7 Responsibilities Summary 39

5 5 1 Introduction Several benches are used in Virgo to support optical systems. The so-called external benches are located outside the vacuum system, while the internal benches are located inside the ultra-high vacuum system. External benches are used to support optical tables near the end mirrors: westend bench (WEB) and north-end bench (NEB), and for the injection (External Injection Bench, EIB) and detection (EDB) benches. These benches are in air, and connected to the ground via a supporting structure (legs). In addition, there are suspended injection (SIB) and detection (SDB) benches, and the end-mirror payload for the input-mode cleaner (IMC). These benches are placed in vacuum to provide an acoustic lter and are suspended via a multi-stage pendulum to provide ltering against seismic noise. Fig. 1 shows a layout of the various benches. Figure 1: Schematic overview showing the position and geometry of the various external and internal benches. The external benches carry telescopes, optical lters, and (quadrant) photo-diodes for beam monitoring and control. These benches need to provide sucient vibration isolation to obtain the sensitivity required for Advanced Virgo. Nikhef will implement vibration isolation for the (external) benches that are outside the vacuum system. In addition, Nikhef is responsible for the suspended injection bench, the suspended detection bench, and the suspension of the end mirror of the input-mode cleaner. Fig. 2 shows the laser and injection system layout for Virgo. The beam originates from the laser bench and is conditioned in the external injection bench. The beam enters the Virgo vacuum

6 6 system and falls on the suspended injection bench. In order to lter motions from the external benches, the beam resonates in the input-mode cleaner. Subsequently, the beam is injected into the Virgo interferometer (ITF). Figure 2: Schematic overview of the laser and injection system layout for Virgo. Components on top and below the suspended injection bench are shown. In order to demonstrate the diculties with the support of the external benches, we start the discussion with the external end benches. Fig. 3 shows pictures of the external bench at the West end-mirror (WEB). Seismometers are placed on the bench top and on the oor in order to determine the transfer function of noice from oor to optical table [?]. Fig. 5 shows that the spectra of both seismometers agree at low frequencies. The microseismic peak from sea waves (so-called microseismism) is clearly visible around 300 mhz. However, around 10 to 20 Hz the noise on the table is signicantly higher than on the oor. The mechanical resonances of the support table (most probably the legs) amplify the ground motion. Finite-element analysis [?] has been carried out in order to understand the motion of the support structure. The rst mode is predicted at 25 Hz: the table `rocks' around the main beams which bend in the middle. This is in agreement with the measurements. These main H-shaped beams are the `soft' elements in the current structure (see Fig. 3: we are concerned with the horizontal H-shaped support beams shown in the right picture). It is unavoidable that some light is diused by optical components placed on the bench. When the transmittance of the end-mirror is denoted T mirror, a fraction T mirror of the power circulating in the cavity leaves the end-mirror. This light can back-scatter into the ITF, adding noise to the

7 7 Figure 3: Photographs of the external bench located near the West end-mirror. A seismometer is visible on the left picture at the left-bottom. main beam. The probability for back scattering of the light intensity is denoted ɛ. Then again a fraction T mirror of this power will enter the cavity, where it is boosted with the nesse F. Thus, the amplitude of this back-scattered beam can be written as A = T mirror ɛ 2F π A 0 (1) where 2F/π is the bounce number of light in the cavity. The situation is sketched in Fig. 4. Figure 4: Model for the interaction of diused light from the end benches with the stored beam. The phase of this light is modulated by the displacement of the scattering surface (i.e. the movement of the bench), x(t) as follows The phase noise added to the main beam amounts to δφ = A sin φ A 0 ( ) 2π φ(t) = 2 x(t). (2) λ = T mirror 2F π ɛ sin φ. (3) Phase noise is related to length noise via δφ = 2π L/λ. The denition of strain h 2 L/L and in a Fabry-Perot cavity the sensitivity is enhanced by the light bounce number 2F/π. In addition, there is a factor 2 gain in sensitivity due to the two interferometer arms. The noise seen at the ITF output is then h noise = λ 4F L δφ = λ ( T mirror 4π ɛ sin φ = G sin ). 2L 2πF λ x(t) (4)

8 8 The coupling factor is determined by the transmittance of the mirror T mirror (about 40 ppm), the fraction ɛ of back scattered intensity (about 10 8 ), and the nesse F of the cavity (888 for Advanced Virgo [?]). The coupling factors have been determined [?] and are given in Table 1.1. Table 1.1. Coupling factors for the various external benches. NEB WEB EDB EIB G [ ] The model has been tested [?] by articially exciting the resonances with a shaker put on top of the bench. The displacement of the bench x(t) is measured with a seismometer and the ITF strain data can be tted to provide a measurement of G. This value can then be used to predict the noise Virgo will experience in quiet conditions. Figure 5: The movements of the external bench near the West end-mirror has been measured [?] with seismometers. The External Injection Bench (EIB) is a source of considerable noise between 16 and 100 Hz in the sensitivity curve of Virgo (see Virgo notes [1, 2, 3]). Fig. 7 shows the projected coherences for 2 frequency ranges between 7 and 25 Hz (Fig. 7a) and between 32 and 67 Hz (Fig. 7b), together with the design sensitivities for Virgo and Virgo+MS including the Monolithic Suspension. From this gure it becomes obvious that the seismic noise of the EIB exceeds the sensitivity of Virgo in the frequency regions around 18 Hz, and between 40 and 55 Hz. New mirrors using the newly developed `monolithic suspension' have been installed and should be commissioned by summer The monolithic suspension will result in a further improvement of the sensitivity, especially below 100 Hz. As a general rule, no individual noise source should contribute more than 10% to the total noise level. This means, that the seismic noise of the EIB must be improved by roughly a factor of 100, before the full potential of V+MS in the low frequency domain can be exploited. This means that the seismic insulation of the EIB must be improved as soon as possible.

9 9 Figure 6: Left: a photograph of the suspended detection bench; right: external detection bench. Figure 7: Projected coherences for 2 frequency ranges: a) around 18 Hz (Fig. 1a, left) and between 40 and 60 Hz (Fig. 1b, right), together with the design sensitivities for Virgo and Virgo+MS including the Monolithic Suspension.

10 10 Figure 8: The Transfer functions between ground and the top of the EIB, horizontal (left, Fig. 2a) and vertical (right, Fig. 2b). Fig. 8 shows the transfer functions for the horizontal and vertical directions of the EIB. Strong resonances are visible in the frequency benches mentioned before. A new support of the EIB must eliminate these excitations, and must provide isolation characteristics of about a factor of 100 around 18 Hz compared to the present values. This implies that an isolation against seismic noise of better than about 0.5 must be reached at 18 Hz. The elimination of seismic noise of the EIB forms the most stringent requirement for this project: the noise level has to be decreased by about two orders of magnitude in the horizontal direction at 18 Hz, and in the vertical direction at about 45 Hz. No signicant new noise may be introduced at higher frequencies. Excitation of natural modes at low frequencies must be small such that the beam handling system remains capable of compensating such movements. The requirements for the EIB are more stringent for Virgo+MS, than for Advanced Virgo. This implies that the solution for the EIB has to be installed as soon as the new mirrors and the monolithic suspension are installed and the combined LIGO-Virgo run is nished, which will be the case in October The requirements for the other external benches are less stringent than for the EIB.

11 11 2 Technical description: External benches Dierent options are available. One nds on the market passively and actively damped systems diering widely in prices and in performance. In section 2.1, an overview is given of the available options that are under consideration. Various vibration isolation systems (based on Stacis and damped springs) have been constructed and tested at Nikhef. In the following, we also give a brief discussion of our experience. 2.1 Overview of vibration isolation systems TMC - Stacis TMC oers the system `Stacis' [?] which combines passive damping at frequencies above roughly 20 Hz, with active damping based on sensors and piezo activators at frequencies below this value. Fig 9 shows the damping provided by this system (specications provided by the manufacturer), which is close to the required factor of 100 at frequencies above approximately 10 Hz. Among the open questions is the region of low frequencies (below 1 Hz) and at frequencies beyond about 200 Hz, possible drift of the system (is it stable to 1 micrometer over a period of days/months?), and the possibility for an external error signal into the controlling electronics. Figure 9: Transmissibility for the system STACIS, vertical (Fig. 3a) and horizontal (Fig. 3b). We have installed a test set-up of the Stacis system at Nikhef. Stacis is not inexpensive - an entire set-up is evaluated at around 60 keuro - and TMC has agreed that the test set-up is on loan, and will only be purchased if and when we nd that it will attenuate the vertical and horizontal oscillations of the benches without introducing new problems, e.g. in the form of drift. We measured the transfer functions in the dierent dimensions. As TMC does not provide any data for frequencies higher than 250 Hz, we performed measurements over a wide frequency range. The Stacis system performs best when the payload is in the upper range of the allowed weight. For this reason we added an additional layer of granite between the Stacis elements and the tabletop. The Stacis system is foreseen to support the tabletop at 4 `canonical' points, and has been installed accordingly to the specications provided by TMC. A rst attempt to measure the transfer functions between the supporting oor of the Nikhef building and the table top has shown a few interesting details (see Fig. 9): The system attenuates indeed strongly (xxx

12 12 db) at low frequencies (xxx Hz). However, at higher frequencies, f > 300 Hz, considerable excitation of the table top is found. More importantly, it seems that even the ground is shaken quite considerably, which will possibly cause problems for the neighbouring benches with ground contact, such as the laser bench. In addition, not only the controller produces a humming sound, but each of the activators of the Stacis system, which are mounted underneath the table top, are producing considerable acoustic noise. Given the experience with acoustic noise and its impact on table tops and optical elements, this is highly undesireable, and we therefore recommend not to use this system. We intend to complete our study, including a measurement of the stability of the system over long times (drift). The experience will be reported in a Virgo note Damped springs Dierent suppliers (e.g. Rosta [?]) oer damped springs of varying stiness (see p. 26, `Vibrachoc'). They can be chosen such that the intrinsic resonance lies in the range of a few Hertz, and they have the advantage of being purely passive systems. On the other hand, the damping material must not introduce high frequency noise, and the stability at frequencies below 1 Hz (and drift) must be acceptable. Figure 10: Attenuation as function of frequency with the sping system. Note: Preliminary data! Note: these data have been taken with the high-stiness springs, which leads to eigenfrequencies which are too high. This will be corrected in subsequent measurements. The data points below 60 Hz are believed to pose upper limits, as no subtraction of acoustic noise eects have been done. The data at frequencies above 60 Hz are believed to be dominated by acoustic noise. First measurements at Nikhef revealed a high sensitivity to acoustic noise. We have shown that we can excite the tabletop at a discrete frequency equally well by a oor-shaker (driven with an harmonic signal) and with a loudspeaker; in the latter case, the sensor on the ground does not show a signal at the excitation frequency. Preliminary results for the vertical transfer function are shown in Fig. 10. In these measurements the springs with the high spring constant have been used, but only with the table top (and without with the granite plate) which leads to too high an intrinsic frequency of 8 Hz. Nevertheless, the attenuation reaches values of almost 10 and

13 at frequencies of 18 and 45 Hz, respectively. Note that these are absolute values, which are almost a factor of 1000 below the curves shown in Fig. xxx, and thus well below the requirement indicated above. Figure 11: Left: photo of two damped springs; right: detailed view. We have purchased two sets of damped springs, which have been assembled in two sets of supports (Fig. 11). One set of springs is dimensioned such that they can accommodate the additional weight needed for the Stacis system (see preceding paragraph), the other is optimized for the weight of the EIB alone. The spring assembly provides high stability in the longitudinal direction, and - at least as long as it is not connected to any other structure - little resistance in the transverse direction. This changes as soon as the supports are connected to the table top. FEM analysis yields intrinsic resonances between 2.5 and 5 Hz. We will continue our study of damped springs to establish their use as vibration isolation for external benches. Special attention will be payed to creep of the material, long-term drift, and temperature eects. Long term drift measurements will be done with the RASNIK 1 system (developed at Nikhef). In essence, RASNIK consists of a coded illuminated mask, a lens and a CCD (Fig. 12). Any change in (relative) alignment between the mask, the lens and the CCD can Figure 12: The principal layout of the RASNIK system. be traced with high accuracy (< 0.4 µm) in the transverse directions (Ref. [?]). The conguration will be that two elements of the RASNIK system, e.g. mask and lens, are positioned solidly on 1 RASNIK has been developed more than 20 years ago at Nikhef and is continuously being rened. It has widespread use in high-energy experiments; ATLAS alone uses more than 8000 channels to keep the alignment of dierent detector parts under control.

14 14 the table top, whereas the third element (the CCD) is xed to the wall of the building. Two such systems are sucient to determine shifts in all three dimensions. These measurements can be done in real time, or can be logged via a PC, and can cover long periods (e.g. several days). The Rasnik system can also be used to verify that during an intervention the bench position has not changed Pneumatic Isolators Many suppliers have pneumatic systems in their product lines; see e.g. TMC or PSE. LIGO has tested the pneumatic isolators S-2000 provided by Newport. The transmissibility is shown in Fig. 13. The specications of the tested system seem to be slightly less favourable than those of Stacis, but do not show the `jumpy' behaviour, which is characteristic for an active system. On the other hand, it appears in general that isolation in the horizontal direction is less ecient than in the vertical one. In addition, accordingly to tests done at EGO, drifts are probably too important to full the requirements of the benches long and short term stability. Figure 13: Transmissibility for the system S-2000 (Newport), left: vertical, right: horizontal The Minus-K system Minus K applies a `negative stiness mechanism' in order to provide a very low net stiness. The transmissibility is shown in Fig. 14. Isolators typically use three isolators stacked in series: a tilt-motion isolator on top of a horizontal-motion isolator on top of a vertical-motion isolator. Figure 14: Typical transmissibility for a system (BM-6) of Minus-K, left: vertical and right: horizontal.

15 The EIB Seismic Attenuation System (HAM-SAS) Introduction For isolation of the EIB we propose a passive solution called `HAM-SAS: Horizontal Access Module Seismic Attenuation System' [4]. It has been developed originally at Caltech for LIGO. It is based essentially on specially formed cantilevers which are put into such a position that the pressure they excert on each other results in an `anti-spring eect' with a very low stiness. This results in natural frequencies of less than 0.5 Hz and in transmissibilities of -60 db and better. The optics suspensions have their resonances in the 1-10 Hz band and therefore it is important to strongly reduce the amplitude of the motion of the optical tables at those frequencies to avoid overloading the optics control systems causing noise injection and limiting the sensitivity achievable in the experiments. In the SAS the seismic attenuation is obtained passively using the properties of the mechanical oscillators which attenuate, as second-order low pass lters, above their natural frequency. The tables operate with low natural frequencies, 0.05 Hz in horizontal and 0.1 Hz in vertical, allowing large vibration suppression ratios to be achieved. The aimed isolation of 60 db at 3 Hz in both directions. An overview of the AEI system is given in Fig. 15. The SAS is a completely Figure 15: A 3-D model of one of the isolation tables for the 10 m prototype interferometer at AEI. The top panel houses a mm optical table, located in the vacuum. custom device capable to provide in a single stage vibration isolation factors not achievable, to our knowledge, by any existing commercial or laboratory developed system. Its design is derived from a prototype device built by Caltech in 2006 in the framework of the R&D for the gravitational wave detector Advanced LIGO. The system has recently been adapted at the Albert Einstein Institute in Hannover, in order to be incorporated in their 10 m prototype interferometer. Their optical table, consisting of a stainless steel honeycomb welded structure mm, sits on three vertical vibration isolators (named lters) providing mechanical compliance along the vertical, pitch and roll degrees of freedom. Fig. 16 shows a picture of a HAM-SAS system.

16 16 Figure 16: HAM-SAS systems have been realized for LIGO as suspension systems. Recently, such systems have been improved and are under construction for the 10 m prototype interferometer at AEI Hannover Mechanical design for EIB vibration attenuation The performance of HAM-SAS systems is wellknown. Therefore, such systems can be employed without too much testing, while minimizing risks. Due to its impressive perfomance, we are currently developing a HAM-SAS set up. The design has been made in collaboration with scientists from AEI and Caltech. The system has been stripped down in order to reduce costs and delivery time. For example, no UHV compatibility is needed. Also, the available EIB laser bench will be mounted. Various views are shown in Fig. 17. Figure 17: Front, side, top and perspective view of Nikhef's HAM-SAS solution for Virgo's EIB.

17 17 The AEI design will be adapted for the EIB support. We intend to employ three standard GAS systems (which can support a 300 kg each). The estimated weight of the EIB is about 700 kg. Since the system needs tuning to the 1 kg level, a separate table, hosting dummy weights, will be used. Also the bottom plate that houses the inverted pendula will be simplied. Views with and without the optical table of the EIB are shown in Fig. 18. Figure 18: View of Nikhef's HAM-SAS solution without (left) and with the optical table of Virgo's EIB. A bottom view with the optical table is shown in Fig. 19. Figure 19: Bottom view of Nikhef's HAM-SAS solution with the optical table of Virgo's EIB. Each lter is a tunable spring made by a crown of curved cantilever blades compressed each against the other: the constrained radial stress creates an anti-spring eect (geometric anti-

18 Figure 20: Composition drawing of the Seismic Attenuation System to host Virgo's EIB. The base plate (gray) holds three inverted pendula (blue), that are connected to an intermediate plate (brown). The intermediate structure (green and brown) supports three vertical GAS systems. The top frame (grey) rests on the GAS springs. The EIB optical table will be mounted to this top frame. 18

19 19 spring) [5] that allows a low eective stiness to be achieved in the vertical direction at the nominal load. A photo of a geometric anti-spring developed at Caltech is shown in Fig. 21. Figure 21: Photograph of a geometric anti-spring vertical isolator developed for LIGO at Caltech. The diameter of the Nikhef springs will be xxx mm and the blades will be cut from xxx mm thick maraging steel. Natural frequencies down to 0.2 Hz can be obtained by mechanical adjustment of the compression rate; longer natural periods can be achieved by applying positive feedback (so-called electronic anti-spring). The inertia of the blades limits the minimum vibration transmissibility to -60 db, which can be improved up to -80 db by using the built-in compensators (so-called magic wands). Figure 22: Horizontal platform holding the inverted pendula. The three lters are mounted on a rigid plate supported by three inverted pendulum (IP) legs: low natural frequencies (around 0.05 Hz) are achieved for the horizontal (X,Y) translational modes and for the yaw mode, by suitably sizing the IP bottom exures to compensate for the load gravitational torque which leads the system to the instability [6]. The IP top short exure hinges allow the lter plate to wobble in the horizontal plane preventing its pitch/roll movement. The inertia of the legs limits the IP minimum vibration transmissibility to -70 db; up to -90 db

20 20 can be achieved by tuning the built-in counterweights. A sketch of the horizontal isolation stage is shown in Fig. 22. The amplication of the seismic noise at the SAS resonances would make the motion of the tables uncomfortably large at low frequencies and therefore the attenuators are instrumented with a collection of sensors (linear variable dierential transformers and horizontal accelerometers) and voice coil actuators to perform electronic modal damping. The SAS control is made using elementary feedback lters and is limited to a few Hz bandwidth Sensors and actuators The GAS springs allow the optical bench to move in three degrees of freedom (z, pitch and roll) with respect to the spring box. The inverted pendula allow the latter to move in the remaining degrees of freedom (x, y and yaw) with respect to the oor. Sensors and actuators are employed to provide information on the 6 6 positioning and control matrix. Due to the action of the GAS springs and the pendula, this matrix naturally splits into two independent three degree-of-freedom matrics. Figure 23: Photograph of a horizontal voice coil actuator and LVDT on the LIGO system. A total of 6 LVDTs are employed whose measured voltage should permit a 1% linearity over a region of a few cm range for the low gain setting. At high gain the range is signicantly reduced, but the sensitivity is improved to the nm level (exceeding the stability of the oor). Three LVDTs are axially located in the GAS lters and measure the vertical position of three points of the optical table with respect to the spring box. Three other LVDTs are located at the inverted pendula legs and measure the horizontal displacement of the spring box. It remains to be determined whether additional LVDTs should be mounted on the top plate that supports the optical table in order to measure horizontal and vertical displacements with respect to the base. Each of the LVDTs is co-located with co-axial voice coil actuators that allow control of the system as well as damping of modes. A photo of a horizontal coil actuator is shown in Fig.

21 The geometry of the race-track coil and magnetic yoke are designed to deliver constant force (within 1%) over a 10 mm movement range in the horizontal plane. The actuators are capable of positioning the optical table within the resolution of the LVDTs. Three geophones will be placed on the optical table (type xxx) in a pin-wheel conguration. Geophones are relative velocity sensors and will be calibrated to provide displacement information. Note that the horizontal geophone response also depends on the tilt motion of the optical table (important at low frequencies). Remote controlled stepping motors, that interface with the structure using micropositioning springs, are used to null the static current of the dynamics actuators and allows to maintain alignment to within a few microns even in the event of power loss. For monitoring purposes we plan on installing various temperature sensors. In the testing of the sensor and actuator system we will use seismometers (two Trillium 240) and quadrant photo diodes with transimpedance ampliers (which we developed for Virgo's linear alignment system) Control system The LVDTs are operated with sinusoidal signals with frequencies around 20 khz. These signals are generated with an LVDT driver board that allows individually set tunable coil excitation levels. The position measurement is obtained by sensing with a lock-in amplier the amplitude and sign of the voltage generated in the receiving coils. The gains for the sensor coils can also be tuned individually. Schematics for these boards are available, while AEI oers an unpopulated PCB that can be used. We will investigate to employ Virgo LVDTs Driver PCBs for the stepping motors must be developed. It must be seen whether schematics from AEI or Virgo are available. HAM-SAS control and data systems for LIGO have been developed in EPICS. The code has been written in C in a Matlab Simulink interface. Again it must be transferred to Virgo standards. This involves developing the controls based on Virgo style ADC, DAC PCBs interfaced with their DSP cards. The control system provides positioning of the optical table near the optimal working point of the lters. This involves in addition to the static DC position control, a velocity control in order to apply viscous damping of the system. The so-called EMAS (electro-magnetic anti-spring) strategy [?] has been applied in the LIGO system through a digital feedback control system. This allows stiness control and resulted in lowering the vertical resonant frequencies below 100 mhz Expected performance The performance of HAS-SAS has been determined in the context of LIGO R&D and also at AEI for the 10 m prototype. The vertical attenuation is due to the GAS springs and can be improved by using so-called magical wands [?] and the performance is shown in Fig. 24. Most probably magic wand technology will not be needed to achieve the seismic attenuation performance needed for the EIB. The horizontal attenuation is due to the inverted pendula and the performance as measured on the LIGO prototype [?] is shown in Fig. 25. At high frequency the noise is aected by acoustics

22 22 Figure 24: Vertical transmissibility of the GAS lter with overcompensating wands. The gure shows the comparison between the model and the measurement performed using a GAS lter with three blades and three overcompensating wands. We can clearly see there is quantitative agreement below the frequencies of the structure's internal resonances. Figure 25: Horizontal transmissibility along x (left panel) and y (right panel). The coherence is often low but -70 db is reached at 4 Hz. Along the y direction, the coherence is higher than for x and the measurement is less noisy. The geophone noise oor is reached at about 20 Hz.

23 23 and perhaps angular ground tilt. Reliable tilt meters would be an asset for this purpose. We will tune the system at Nikhef and completion of the tests transport it to EGO and initially install it at the 1500 m laboratory. There the system will be integrated with the Virgo controls and subjected to a 2-3 week test program. After succesfull completion of these tests, we intend to set-up the system at the EIB Installation procedure A system such as EIB-SAS requires a solid support at the correct height. The actual height of the elements is less than the distance between the oor surface and the lower level of the EIB. The best would be to heighten the concrete oor to the level such that the table top of the EIB, sitting on the SAS system, would be at the present height. Since introducing several tons of concrete is not at all practical - the dust and dirt in the clean room would become a major problem - we propose to install support blocks of granite cut to the required dimensions. The concrete oor on which the granite blocks have to be positioned has to be equalized carefully. For this, we will use Sikadur 35LV. In order to prevent that the blocks are literally glued to the concrete oor we propose to use a thin layer of Sikadur 35LV within two thin sheets of plastic or aluminium to avoid that it does not binds neither to the granite nor to the concrete oor; tests of this technique will be performed at Nikhef. Evidently, the granite blocks will only be ordered once their precise height is known, depending on the system chosen. It has been veried that the oor at the location of the EIB is strong enough for the additional weight.

24 Action items The following actions should be carried out for Nikhef's evalution of TMC-Stacis and damped springs systems. 1. Finalize the evaluation of Stacis. Produce a Virgo note (2 weeks). 2. Investigate damping for the springs (2 weeks). The seismic attenuation of a spring set up depends not only on the natural frequency f 0, but also on its Q-value. Best attenuation for frequencies f f 0 is achieved for high Q- values since the transfer function depends on (f/f 0 ) 2. However, - and evidently - a high Q value has dramatic eects for f f 0, so that a compromise has to be found. A simple damping method is based on Eddy currents. 3. Set-up a Rasnik system for monitoring creep, drift and thermal eects (2 weeks). 4. Finalize the study of damped strings. Produce a Virgo note (2 weeks). The following actions should be taken for the development of the EIB-SAS system. 1. Produce 3D design for EIB SAS (2 weeks). 2. Produce 2D production drawings for EIB SAS (2 weeks). 3. Write production specications for EIB SAS (1 week). 4. Evaluate modes of inverted pendula with FEA. Design a scheme to push rst (banana) mode above 1 khz (2 weeks). 5. Produce LVDTs and voice coil actuators for EIB SAS. Obtain hardware specications from Virgo (4 weeks). 6. Populate driver LVDT PCB board (obtain from AEI Hannover and from Virgo) (4 weeks). 7. Populate driver accelerometer PCB (obtain from AEI Hannover and from Virgo) (4 weeks). 8. Populate driver stepping motors PCB (obtain from AEI Hannover and from Virgo) (4 weeks). 9. Set-up control and data acquisition for LVDTs and actuator system with CPU, DSPs, ADCs, DACs and BIO. Model hardware after Virgo new DSP control system. Hardware for analog systems should be Virgo compatible (4 weeks). 10. Develop models for the EIB-SAS based on a Lagrangian approach, Simulink and Matlab/Mathematica. Tests agains measurements (8 weeks). 11. Set-up control and data acquisition system for accelerometers (4 weeks). 12. Design and construct a set-up for tuning GAS lters (4 weeks). 13. Tune GAS lters (4 weeks). 14. Tune tilt correcting springs (1 week). 15. Provide load equalization of inverted pendula legs (2 weeks).

25 Measure the inverted pendula load curve (1 week). 17. Install counter weights for inverted pendula (1 week). 18. Leveling procedure for the optical table (1 week). 19. Characterize performance of LVDT driver board (1 week). 20. Calibrate the LVDTs (1 week). 21. Calibrate the transfer function of the geophones (1 week). 22. Develop and install the optical lever based on laser and quadrant photo diode (4 weeks). 23. Determine transfer functions with Trillium T240 seismometers (1 week). 24. Develop optical table control software. Establish velocity control for viscous damping. Realize stiness control by using the EMAS electromagnetic anti-spring strategy (4 weeks). 25. Measure the sensing matrix (1 week). 26. Measure the driving matrix (1 week). 27. Experimentally diagonalize the sensors and actuators (2 weeks). 28. Calibrate the voice coil actuators (1 week). 29. Calibrate stepping motor positioning system (1 week). 30. Determine passive seismic attenuation (2 weeks). 31. Determine active seismic attenuation (2 weeks). 32. After completing Nikhef tests, the EIB-SAS should be moved to the 1500 m laboratory at Virgo (2 weeks). 33. Test EIB-SAS at 1500 m laboratory. Measure seismic attenuation (various transfer functions) by using local controls only (2-3 weeks). 34. Install temporary reference frame for the YAG beam near EIB (1 week). 35. Put EIB-SAS in place and put EIB back in the same position (1 week). 36. Commissioning of EIB-SAS by using passive attenuation only. Execute calibration program for LVDTs and actuators, and for stepping motors for EIB positioning (3 weeks). 37. Commissioning of EIB-SAS by using active feedback from the interferometer (4 weeks).

26 26 3 Technical desription: Internal benches 3.1 Multi-payload suspension According to the optical reference solution two mirror suspensions will be hosted onto each of the internal benches (INJ, DET). The main issue in this case is the room on the bench and this development is highly interlaced with the development of those benches. Fig. 26 shows the present outline of the suspended injection bench, Figure 26: Left: optical layout of the suspended injection bench; right: a photograph of the SIB (rotated 180 with respect to the left diagram). Figure 27: Possible layout of the SIB when accommodating two power-recycling mirrors (PRM1 and PRM3). while Fig. 26 shows possible new outlines of the suspended injection bench. These systems will certainly require a full chain of local controls dedicated to control and recover their position

27 27 with respect to the sensors placed outside the vacuum chambers. ISC will dene the reference solution the strategy of NDRC longitudinal control. From the perspective of PAY the main issue is related to evade the recoil due to the action on NDRC mirror sitting on benches and the related breadboards. Once the mechanical scheme will be even preliminarily designed a standard optical lever system will be easily deduced. Preliminary investigations into the dynamics of multi-payload suspensions have been carried out [?]. The models shown in Fig. 28 has been studied. Figure 28: Left: a generic multi-mirror payload to be mounted on the Beam Splitter is analyzed for conceptual studies. Non-trivial issues concern for instance also the Reference Plate torque, applied through a mechanical arm of m. Right: the design of the bench multi-payload is constrained by serious problem of available space (not solved). In any case, the solution represents a strong violation of the symmetric design of the payload and new categories of normal modes, not controllable within the AdV frame, appear. The preliminary conclusion of these studies [?] is that the investigated possibilities are not worth of a deeper study. The complexity of the multi-payload systems, with many couplings aecting the dynamics of the suspended bodies in many DOFs, makes their control very challenging, with the use of a huge number of sensors and actuators and the implementation of Multi-Input Multi-Output control strategies. The only possibility to have a nite probability to reach a full simultaneous control of all the payload items, is to control actively the upper plate from which the bodies are suspended, a solution that is not implementable in Virgo without increase the payload RMS swing by a huge factor (one hundred). Other ideas, alternatives to the present Superattenuator, to create parallel chain of lters (each suspending a single payload) hanged from a rigid platform have been discussed, and are not worth of a dedicated R&D. We have initiated a study of a potential solution for the multi-payload suspension along the following philosophy: each object (mirror, bench) should have its own reaction mass; object and reaction mass should have the same center of mass;

28 28 Figure 29: Multi-payload suspension model for the suspended input bench. The model shows the input bench with its reaction mass and one of the power-recycling mirrors with its reaction mass. The mirror and its reaction mass have a dedicated marionette. Both bench and mirror payloads are suspended from a common platform that connects to the superattenuator. objects to be manipulated should be as much as possible on the central axis in order to minimize the eects on the superattenuator; the mirror and its reaction mass are suspended from a dedicated marionette; the marionette and the bench (with its reaction mass) are suspended from a platform; this platform is connected to the superattenuator (lter 7). The proposed solution is shown in Fig. 29 and has been studied using MSC Marc. At present we consider only the suspension of one of the power-recycling mirrors together with the injection bench. The second power-recycling mirror either will have a dedicated suspension based on a HAM SAS in combination with a triple suspension system, or it will be suspended from the platform with a dedicated marionette and recoil mass. This will be studied later. For the system under consideration, the important couplings (x θ y and y θ x ), that introduced o-diagonal elements in the previously studied solution [?], are absent in rst order due to the on-axis suspension approach. Since the modeling of multi-payload suspension concerns large objects such as mirrors and benches in combination with small geometries such as suspension wires, it is paramount to crosscheck the results. We have checked the MSC Marc approach for simple pendula by explicitely writing down the Lagrangian of the system and solving the system for modes. The resonant frequencies found by MSC Marc in some cases perfectly agree with the Lagrangian approach, but in some cases deviations are found. It is important to track down such deviations. This is underway by using FEA with Abaqus. Also ANSYS and Comsol analysis in combination with Simulink, Mathematica and Matlab based simulations will be carried out.

29 29 Figure 30: Multi-payload suspension model for the input optics. The model shows various modes of the system that demonstrate the high degree of decoupling of the mirror payload from the injection bench payload. Left: marionette x rotation, middle: marionette y rotation, right: mirror z translation. Fig. 30 shows various modes of the system. It is seen that the mirror can be manipulated by using its marionette and recoil mass in order to obtain positioning in the z direction as well as angular control for rotations along the x and y axes. During these displacements it can been that reaction forces and moments on the injection bench payload and superattenuator are small. Figure 31: Multi-payload suspension model for the input optics. Left: bench x translation, right: complex internal mode. A mode that shows the couplings in case of an x translation of the suspended bench is shown in Fig. 31. Again it is seen that the bench and its reaction mass act in such a manner as to minimize eects on the movements of the mirror payload and superattenuator. Again a high degree of decoupling is present. Presently, Virgo contains mirror payloads (a marionette with mirror and reaction mass) and injection and detection payloads (again marionette and bench) that are suspended separately from superattenuators. The solution studied by us can be seen to be as much as possible a superposition of these proven solutions. However, in the superposition

30 30 new modes arise that involve coupling between these systems. An example is shown in Fig. 31 (left panel) where mirror and bench payloads are in a mutual oscillation. Since we are concerned with high quality oscillators it is important that such complex modes are identied and that the control system is developed such that it can identify such modes and moreover can damp them. In the case at hand in Fig. 31 (left panel) such a mode cannot be handled by the mirror and its recoil mass actuations and also not by the bench and its recoil mass attenuators (since it is a common mode for these systems). However, it can be damped using the mirror marionette (e.g using the controls between the mirror marionette and its recoil mass). Figure 32: Transfer function for the multi-payload suspension model for the θ y rotation due to a moment M y on the platform. The FEA models allow the determination of transfer functions for various forces and moments from actuators. This is shown in Fig. 32. Such information is crucial for a proper design of the sensing, actuation and control system. Note that it is important that the reaction forces between mirror and bench payload, and on the superattenuator are small, since it must be proven that the system can achieve the AdV requirements, in terms of control capability, lock-acquisition, robustness and noise. While the present system using separate suspension for mirror payloads and benches has been already tested on a real interferometer, and the results are already compliant with AdV, a multi-payload suspension clearly represents a risk. The studies of the multi-payload solution for bench and a single mirror will be continued. Also in the case of payloads for the long NDRC such solutions may play a role. At present, studies are ongoing to investigate the constraints from Virgo's optical design in order to determine the exact sizes and position of mirros and to establish to specications. For example, it is expected that the power-recycling mirror must be movable along the z axis over ±10 mm in order to allow tuning of the PR cavity length. Also issues as astigmatism must be addressed. So far we did not study

31 31 Figure 33: Triple suspension for the second power recycling mirror. The triple suspension system is placed on top of a HAM-SAS table placed in the injection tower. the incorporation of the second power-recycling mirror. It may be suspended from the platform, or from the bench or other PR mirror. However, in that case it is clear that additional couplings and complex modes will appear. A more favorable solution maybe to completely decouple the suspension of this second mirror from the other mirror and bench. It will be studied whether this can be accomplished by placing a HAM-SAS system (see Fig. 16) inside the vacuum system of the injection tower and then to suspend the second PR mirror from a triple suspension. Such triple suspension have been developed for GEO600 and maybe suitable for the suspension of the relatively small (about 150 mm diameter) second PR mirror. The situation is outlined in Fig. 33. We will study the impact of such a solution on seismic attenuation, space, and tower access in the short term. 3.2 Payloads for long-ndrc Any long-term R&D program on multi-payload suspensions (necessary in any case), even if will be concluded with success, is not sucient to exclude that thin spurious mechanisms appearing on the real interferometer (so small to not be detectable by lab ground-based sensors, aected by seismic noise) could represent a showstopper for the project. The unique viable solution, free from a dedicated long R&D, is to design an optical conguration allowing the implementation of separate payloads, with a single superattenuator for each of them. In case this solution will be selected, additional funding (about 6-7 MEuro) will be required for Advanced Virgo. Nikhef intends to participate in this option in case it will materialize. First design discussions will be carried out in collaboration with the Roma La Sapienza group in the second half (H2) of The emphasis of this activity will be on modeling of suspension systems (nite-element analysis, Simulink, etc.)

32 Action items The following actions should be carried out for Nikhef's evalution of payloads for the suspended injection and detection benches. The studies for Advanced Virgo are still in the design phase. At present it is unclear which solution will be chosen by the collaboration. Consequently, we restrict the planning to Modeling of suspended multi-payload systems (5 months). 2. Study of control issues for suspended payloads (5 months).

33 33 4 Logistics 4.1 Deliverables We intend to complete our studies of TMC-Stacis and damped springs. In 2010 Nikhef should deliver a seismic attenuation system (SAS) for the External Input Bench (EIB). Other systems will follow. In addition, Nikhef will continue its study of multi-payload in vacuum suspension for injection and detection benches. Deliverable 1 Deliverable 2 Deliverable 3 Deliverable 4 Deliverable 5 Deliverable 6 EIB-SAS system EIB-SAS integration and commissioning SAS for other external benches Design study multi-payload suspension for SIB and SDB. Production SIB and SDB Integration and commissioning SIB and SDB The baseline solution for the vibration isolation of the EIB is the SAS system. Isolation for the other external benches is signicantly less critical and solutions based on damped springs can be considered. At present the timeline for installation of the vibration isolation for the EIB is set at October Obviously, this represents an agressive timeline. This target is set in order to benit as much as possible from the expected improvements during science mode with the monolitic suspensions. The EIB-SAS deliverable includes the following: 1. mechanics consisting of inverted pendula, GAS systems, spring box, base and top tables; 2. motion system for positioning in 6 degrees of freedom; based on springs, stepping motors, LVDTs, voice coil actuators; 3. control system with integration in Virgo's slow control. The design of the suspended injection and detection benches is underway. The present design study is concerned with multi-payload suspensions and involves detailed modeling based on Lagrangian methods, nite-element techniques, and other modeling packages (e.g. Comsol, matlab and simulink).

34 Involved Virgo subsystems In the following table we describe the subsystems that are involved in the realization of vibration isolation systems for Advanced Virgo. The subsystems are listed in order of decreasing involvement (# 1 is the subsystem that we intend to modify) describing the typeofconsequence on each subsystem. Project table 3.1. The subsystems that are involved in the realization of vibration isolation for Advanced Virgo. # Subsystem name Description of the involvement 1 External OSD optical constraints for geometry (diused light issue, etc.) benches DAQ electronics and software compliance IME infrastructure (cable trays, hydraulic pipes, water, etc.) 2 Internal OSD optical constraints for geometry (diused light issue, etc.) benches TCS accommodate thermal compensation systems PAY project is part of PAY working group SAT inuence on superattenuator DAQ electronics and software compliance IME infrastructure (cable trays, hydraulic pipes, water, etc.) 4.3 Involved Nikhef and EGO infrastructure In the following table we describe the type of infrastructure needed at EGO. Project table 3.2. The infrastructure at Nikhef needed for the realization of vibration isolation for Advanced Virgo. # Infrastructure Description of the involvement Clean room External benches prototypes Internal benches prototypes Workshop Prototype tests Electronic department Prototype control systems Project table 3.3. The infrastructure at EGO needed for the realization of vibration isolation for Advanced Virgo. # Infrastructure Description of the involvement Clean room HAM-SAS installation preparations Multi-payload installation preparations Workshop Installation tooling Vacuum department Vacuum equipment Control Slow control integration

35 Planning The planning in 2010 is dominated by the EIB-SAS project. The following considerations have to be taken into account: 1. The EIB-SAS must be installed as soon as possible after the end of VSR3 in order to take full advantage of the improved sensitivity of Virgo below 50 Hz. 2. The performance of EIB-SAS must be compliant with the requirements of Virgo+MS, as Advanced Virgo maybe less demanding on the characteristics of the EIB. 3. The system must be integrated into the Virgo control system. The steps that must be taken in the realization of EIB-SAS are summarized in section 2.3. The dierent steps are described below, grouped in three parts: `nalization', `production' and `integration' Design nalization, call for tender EIB-SAS A Virgo change request must be prepared in the last week of June The request will be reviewed by a dedicated committee on June 30, 2010 in Cascina. The meeting starts at 3:30 pm and it is possible to participate via EVO. The committee is chaired by Franco Frasconi (Pisa) and other members are Enrico Calloni (Napoli, commissioning coordinator), Eric Genin (EGO, coordinator INJ group), Irene Fiori (Pisa), Piero Rapagnani (Roma1, detector coordinator). preparation of `Virgo change request document' for EIB: Q preparation of `technical specication docs' for EIB: Q call for tender and contract assigned for EIB: Q production EIB: Q Production Production of GAS and inverted pendula involves the use of maraging steel. This material is of strategic importance (e.g. used in blades of ultra-centrifuges, missile heads) and procurement can take a long time. For this reason we intend to have components of SAS produced at Galli & Morelli in Lucca, Italy where this material is in stock. We are in contact with the directorate of this company. The estimated production time for the mechanical part for the EIB-SAS is 2 months. production must start in July In the same period the various electronic and control components will be obtained. We expect that EIB-SAS can be realized in October The Integration EIB-SAS The installation time is estimated at 20 days and is currently scheduled for the second half of October At present we aim at transporting EIB-SAS to Cascina in the rst week of October.

36 36 The system will be rst commissioned in the 1500 m laboratory. After passing acceptance tests, it will be integrated into Virgo. Nikhef will be involved in commissioning the EIB-SAS in Virgo (presently a Nikhef PhD student, Mathieu Blom, is member of the Virgo commissioning team and stationed in Cascina). The installation of the isolation of the other external benches which are less critical than the EIB will be detailed in a new release of this document. The same holds for the multi-payload suspensions for SIB and SDB Implementation plan and maintenance The planning is detailed for the EIB-SAS. Decisions on the SAS for other external benches still have to be taken. Also multi-payload solutions for SIB and SDB are under study and important design decisions have to be taken. Figure 34: Implementation plan. In Fig. 34 we show the implementation plan. In addition, the timetable for the deliverables is given. An agressive timeline is evident.