Sardinia Radio Telescope: General Description, Technical Commissioning and First Light

In the period 2012 June - 2013 October, the Sardinia Radio Telescope (SRT) went through the technical commissioning phase. The characterization involved three first-light receivers, ranging in frequency between 300MHz and 26GHz, connected to a Total Power back-end. It also tested and employed the telescope active surface installed in the main reflector of the antenna. The instrument status and performance proved to be in good agreement with the expectations in terms of surface panels alignment (at present 300 um rms to be improved with microwave holography), gain (~0.6 K/Jy in the given frequency range), pointing accuracy (5 arcsec at 22 GHz) and overall single-dish operational capabilities. Unresolved issues include the commissioning of the receiver centered at 350 MHz, which was compromised by several radio frequency interferences, and a lower-than-expected aperture efficiency for the 22-GHz receiver when pointing at low elevations. Nevertheless, the SRT, at present completing its Astronomical Validation phase, is positively approaching its opening to the scientific community.


Introduction
The Sardinia Radio Telescope (SRT) (Lat. 39 29 0 34 00 N -Long. 9 14 0 42 00 E; 600 m above the sea level) is a new Italian facility for radio astronomy whose commissioning was completed at the end of 2013. Its formal inauguration took place on 2013 September 30. The antenna is a fully steerable, wheel-and-track dish, 64 m in diameter, located 35 km north of Cagliari, on the island of Sardinia. It completes a set of three antennas devoted to radio astronomical science in Italy (Fig. 1), all managed by the Italian Institute for Astrophysics (INAF).
The SRT is a general-purpose radio telescope aimed at operating with high aperture e±ciency. Once all the planned devices are installed, it will observe in the frequency range from 300 MHz to 100 GHz and beyond (from 1 m to 3 mm in wavelength). The antenna gain is expected to vary from 0.50 to 0.70 K/Jy for the frequency range 0.3-50 GHz and to be around 0.34 K/Jy in the 3 mm band .
A key feature of the SRT is its active surface, in total composed by 1116 electromechanical actuators, able to correct the deformations induced by gravity on the primary surface (or \mirror"). E®ort is underway to employ this facility to also correct for non-systematic errors, such as temperature/ wind-related e®ects.
The SRT is capable of hosting many microwave receivers, located in four di®erent antenna focal positionsprimary, secondary and two beamwaveguide fociable to cover almost continually its frequency range. The SRT will operate in singledish (continuum, full Stokes and spectroscopy), Very Long Baseline Interferometry (VLBI) and Space Science (Ambrosini, 2011a) modes.
Thanks to its large aperture and versatility (multi-frequency agility and wide frequency coverage), the SRT is expected to have a major impact in a wide range of scienti¯c areas for many years to come. A full description of the potential SRT science applications is beyond the scopes of the present paper, as it is provided in a separate paper dedicated to the Astronomical Validation activities (Prandoni et al., in preparation). Here we illustrate some of the main areas where we think the SRT can play a major role in the next future.
Operations in the framework of international VLBI and Pulsar Timing networks are of top priority for the SRT. SRT is going to be one of the most sensitive European VLBI Network (EVN) stations, together with E®elsberg and Jodrell Bank. Its large aperture is also of extreme importance for Space VLBI observations with RadioAstron. Thanks to its active surface, the SRT will also represent a sensitive element of the mm-VLBI network operating at 7and 3-mm bands, where a substantial improvement in collecting area and in the coverage of the sky Fourier transform plane is of vital importance for increasing the number of targets accessible to the array and the quality of the images. Once the¯ber optic connection to the site is completed, the SRT will also participate in real-time VLBI observations (e-VLBI). The availability of three antennas ( Fig. 1) will moreover allow the constitution of a small Italian VLBI network, exploiting a software correlator (DiFX, already operating). The SRT will be also included in the geodetic VLBI network.
The SRT is one of the¯ve telescopes of the European Pulsar Timing Array (EPTA), which, together with the North American Nanohertz Observatory for Gravitational Waves (NANOGrav), and the Parkes Pulsar Timing Array (PPTA) share the goal to detect gravitational waves. Being the southernmost telescope of the EPTA collaboration, the SRT will allow a better coverage of pulsars with declinations below À20 , hence a better overlap with the PPTA. Thanks to its dual-band L/P receiver, SRT will be of great importance in measuring accurate dispersion measure variations, crucial to obtain ultra-precision pulsar timing, and search for signatures of space-time perturbations in the pulsar timing residuals. The SRT is also part of the Large European Array for Pulsars (LEAP), a project which consists in using the EPTA telescopes in tied-array mode, obtaining the equivalent of a fully steerable 200-m dish.
The SRT is expected to have a major impact also for single-dish observations. In particular, we aim at exploiting its capability to operate with high e±ciency at high radio frequency. Equipped with multi-feed receivers the SRT can play a major role in conducting wide-area surveys of the sky in a frequency range , which is poorly explored, yet very interesting.
For instance the¯rst-light K-band 7-beams receiver will be exploited to obtain extensive mapping of the NH3 in Galactic star-forming regions, in close synergy with existing IR/sub-mm continuum surveys of the Galactic Plane. The ammonia molecule, through its (1,1) and (2,2) transitions, is a good tracer of dense cores of molecular gas and is considered an excellent thermometer. Another interesting application of the K-band multi-feed receiver is a H2O line survey of nearby galaxies. This will increase the number of detected water masers and derive distances, dynamical models and total masses of (luminous þ dark) matter of the galaxies in the Local Group.
Wideband multi-feed receivers operating at higher frequency (40-90 GHz) -currently under developmentwill allow us to get access to unique molecular line transitions in our own Galaxy, like for example those associated with deuterated molecules (e.g. DCO þ ð1À0Þ, N2D þ ð1À0Þ), crucial to constrain the kinematic and chemical properties of pre-stellar cores, as well as to uncover the cool molecular content of the Universe in a crucial cosmic interval (redshift $0.3-2), through the mapping of redshifted CO low-J transitions. Wideband radio-continuum (and spectro-polarimetry) mapping of extended (low-surface brightness) Galactic and extra-galactic sources (e.g. SNRs, radio galaxies, nearby spirals), on the other hand, will permit resolved studies aimed at a better understanding of the physics of accretion and star formation processes. In addition high-frequency receivers will allow us to conduct (high spatial resolution) follow-up observations and monitoring experiments of AGNs, GRBs and other transient events in connection with high-energy experiments (FERMI, MAGIC, CTA), a research¯eld where the Italian astronomical community is very active.
Due to an agreement between INAF and ASI on the use of the instrument for space applications, SRT will also be involved in planetary radar astronomy and space missions (Tofani et al., 2008;Grue® et al., 2004). This paper presents an overview of the SRT system in the light of the results obtained during the commissioning phase (Ambrosini et al., 2013). Section 2 describes the overall SRT system, including technical speci¯cations of the main antenna modules. Section 3 shows the results of the tests performed in the commissioning phase and the timeline for the main milestones. In Sec. 4 some conclusions are drawn on the overall results of the commissioning.
For the convenience of readers, Table 1 provides a glossary of many acronyms used throughout the paper.

Antenna technical characteristics
The antenna design, whose schematic view is shown in Fig. 2, is based on a wheel-and-track con¯guration. The main re°ector (M1) consists of a backstructure that supportsthrough actuatorsthe mirror surface, itself composed of rings of re°ecting panels. A quadrupod, connected to the back-structure, supports the sub-re°ector (M2) and the primary focus positioner and instrumentation. The main and secondary mirrors and the quadrupod lie on the alidade, which is a welded steel structure standing on a large concrete tower that forms the bulk of the antenna foundation. Three large rooms were built behind the primary mirror to contain the secondary focus receivers, the beam-waveguide mirrors and several electronic instrumentation and cable distributions, respectively. A fourth room is located in the lower part of the alidade, where the power drivers for the motors, the antenna control unit and the cryogenic compressors are installed. The alidade also supports the elevation wheel, which is a conical truss anchored to the re°ector back-structure through a massive pyramidal structure.
The radio telescope is supported by a reinforced-concrete foundation excavated into rock; an outer ring beam sustains the azimuth track, while a central building accommodates the azimuth pintle bearing support, the azimuth cable wrap and the encoder system. The wheel-and-track structure consists of 16 wheels lying on the rail, which is a continuous welded ring 40 m in diameter. This is connected to the foundation by 260 pairs of anchor bolts and by a¯ber-reinforced grout. The overall structure of the antenna weighs approximately 3300 tons.
The antenna is steerable around the azimuth and elevation axes. The rotation around the two axes is managed by a servo control system, consisting in 29-bit absolute encoders (peak-to-peak position error 0.8 arcsec), 12 brushless asynchronous motors (eight in azimuth and four in elevation) and an ACU available on a BECKHOFF hardware platform employing an IRIG-B generator. A proper torque bias is applied to the motors to overcome the gearbox backlash and to improve the antenna pointing accuracy.
The primary re°ector surface consists of 1008 individual aluminum panels divided into 14 rows of identical panel types. Each panel has an area ranging from 2.4 to 5.3 m 2 . It is built using aluminum sheets glued, by means of a layer of epoxy resin, to both longitudinal and transversal Z-shaped aluminum sti®eners. The basic back-structure is composed of 96 radial trusses and 14 circumferential trussed hoops supported by a large center hub ring. The sub-re°ector surface consists of 49 individual aluminum panels with an average area of about 1 m 2 , whereas its back-structure is formed by 12 radial trusses and three circumferential trussed hoops supported by a center hub ring. Three of these trusses are directly connected to a triangular steel frame, which has the function of a transitional structure to the six sub-re°ector actuators. These actuators de¯ne the sub-re°ector position, and provide for sub-re°ector motion with¯ve degrees of freedom.
Further mirrors beneath the Gregorian focus, arranged in a Beam WaveGuide con¯guration, allow the addition of four more focal points with magni¯ed and de-magni¯ed F =D ratio in the intermediate frequency bands (Fig. 3).
Two of the four designed BWG layouts have already been constructed. They use three mirrors: M3 as a shared mirror, M4 for the¯rst layout and M5 for the second one. The re-imaging optics for BWG layout I were designed for maximum focal ratio reduction (right side of Fig. 3); whereas BWG layout II was designed so that the output focal point F4 lies beneath the elevation axis of the antenna (left side of Fig. 3). By an opportune rotation of M3, which takes 1.5 min, the desired BWG layout is selected. All the mirrors are portions of ellipsoids and present quite large apertures, as shown in Table 2. The two missing BWG layouts will be added in a later stage to the present con¯guration using two more mirrors and an appropriate rotation of M3; they will be dedicated to Space Science applications.
A rotating turret (Gregorian positioner assembly) is mounted eccentrically on the focal plane of the antenna dish, and can house eight separate cryogenic receiving systems and the associated feed horns for operating up to 115 GHz. A drive system can rotate the turret (within 2 min for a complete rotation of 340 ) so that any of the feed horns can be positioned on the focal plane. The servo control system consists of two brushless servo motors with drivers and a position-control computer.
A mechanism selecting among di®erent receivers exists also in the primary focus. Several receiver assemblies, the exact number depending on their dimensions, can be allocated at the secondary mirror back-structure side. An arm controlled by two servo motors can, in less than 4 min, place the required receiver box in the primary focus position (Fig. 4), which can accommodate low-frequency receivers whose dimensions are not mechanically compatible with the Gregorian positioner assembly.
The remote and automatic control of all these movements makes the antenna available in frequency agility, switching among all the observing bands in a fast and unmanned way.
Suitable atmospheric conditions (see Table 2 for the environmental speci¯cations) together with the active surface will extend the use of the instrument up to 100 GHz. This capability also requires an advanced metrology system for accurate antenna pointing (see Sec. 2.3). The availability of a Water Vapor Radiometer (WVR) and a Weather Station, already in operation, will also allow the dynamic scheduling of the observations, further improving the telescope productivity.
A complex helium 5.5 (99.9995% pure gas) plant consisting of seven pairs of supply-and-return lines assures that the microwave receivers, distributed in the various positions, can operate at cryogenic temperatures. The system is dimensioned so that each line can serve up to three microwave receivers. The total length of the lines reaches 1.5 km, and they are composed of both rigid and°e xible tubes. For the radio frequency links (Fig. 5), more than 3 km of coaxial cables connect the focal positions to a central point located in the elevation Equipment Room. The highest frequency useable with the coaxial cables is 4 GHz, with a maximum attenuation of 15 dB (at 2 GHz) for the longest path (100 m) and with a matching coe±cient better than À20 dB over the whole frequency band. An optical¯ber cabling, based on both single-mode and multi-mode¯bers, was deployed within the radio telescope for digital and analog data transmission. The optical¯bers are responsible for the connection of the radio telescope to the external infrastructures. Some customized optical links allow the transport of the astronomical signal, in the IF frequency band (0.1-2.1 GHz), from the EER to the remote control and data processing room. This solution, thanks to a link gain of 0 dB associated with a dynamic range better than the dynamic range of the receiver itself, ensures that the overall speci¯cations of any receiver are maintained. In order to connect the radio telescope to the Internet, mainly to transfer data during real-time VLBI observations, SRT will make use of a 10-Gbps link infrastructure, currently under construction.

Microwave receivers and back-ends
Notwithstanding the telescope's enormous collecting area and its advanced systems, able to ensure high e±ciency at all the operating frequencies and over the whole elevation range, state-of-the-art microwave receivers and digital and analog back-ends are necessary in order to perform breakthrough astronomical discoveries. The microwave receiver guides, ampli¯es,¯lters and down-converts the incoming astronomical signal re°ected by the mirrors. It consists of a cascade of di®erent microwave passive and active components, some of them cooled to cryogenic temperatures. The receiver mainly de¯nes the frequency bandwidth, the antenna illumination, the receiver noise temperature and the cross-polar purity. The radio receiver group of INAF is responsible for the design of the whole receiver chain, both from an electronic and mechanical point of view, for the assembly and testing and¯nally for the integration in the radio telescope.
The¯rst-light of SRT was obtained using the following three receivers: the L-and P-band dual-frequency coaxial receiver for primary-focus operations (Valente et al., 2010), the C-band monofeed receiver designed for the BWG layout I (Nesti et al., 2010;Orfei et al., 2011;Poloni, 2010;Peverini et al., 2011) and the K-band multi-feed receiver (seven corrugated feeds) for the secondary focus (Orfei et al., 2010).
For the very¯rst astronomical tests, the C-band receiver was installed in the Gregorian focus with a cone section added in front of the feed-horn to match the di®erent F/D numbers. This choice was made to avoid possible BWG mirror misalignments in the¯rst pointing tests of the antenna. Table 3 describes the main technical characteristics of the receivers currently installed, together with those foreseen to be completed in the next three years. For each receiver the table lists the observable sky frequency bandwidth, the focal position, the number of simultaneous beams in the sky (specifying the number of polarizations), an estimation of the expected antenna gain and of the system noise temperature when the antenna points to the zenith. Finally, the last column indicates the status of the receiver.
All the multi-feed receivers (S-, K-and Q-band) are equipped with a built-in mechanical rotator, whose aim is to prevent sky¯eld rotation while the telescope is tracking. The W-band receiver, which was decommissioned by the Plateau de Bure radio telescopes operated by IRAM, is mainly aimed at testing the SRT active surface, and at acquiring know-how in the 3-mm band. However, even though it is a singlepixel, one-linear-polarization and narrow-instantaneous-bandwidth (600 MHz) receiver, its installation on the SRT could play a relevant role in the mm-VLBI network, signi¯cantly extending its baselines.
The front-end outputs are connected to the back-ends either by coaxial cables (in case they are placed within the radio-telescope, in particular the Total Power and XARCOS back-ends) or via opti-cal¯ber in order to reach the distant control room.
Back-ends represent key elements to the success of the observations. They can be either generalpurpose or tailored to speci¯c scienti¯c activities. All the back-ends listed in Table 4 were installed and are now either commissioned or under test at the telescope (Melis et al., 2014).

Optical system and active surface
The SRT optical system is based on a quasi-Gregorian pro¯le with shaping applied to both the primary and the secondary surfaces. The present geometry results from a trade-o® between two goals: minimizing the overall system noise temperature,  mainly due to spill over, and reducing the standing wave pattern (due to internal re°ections between the primary and the secondary mirrors and thus detrimental to wideband spectroscopic observations) without excessively sacri¯cing the Field of View available from the Gregorian focus (Cortès-Medellin, 2002). The shaped parabola-ellipse pair, in fact, provides a wider focal plane than fully shaped re°ectors; the FOV's radius is 20 at 10% of aperture e±ciency loss. The design placed more emphasis on reducing the standing wave than on maximizing the aperture e±ciency.
One of the most innovative features of the SRT is the active surface that consists of 1116 electromechanical actuators mounted in the backup structure, beneath the primary mirror panel corners, and distributed along radial lines. Its¯rst aim is to reshape the mirror to compensate for the repeatable deformations due to gravity. Exploiting advanced real-time measurements, it will also correct for wind and thermal e®ects (Orfei et al., 2004). Each actuator moves either upward or downward at the corners of four adjacent panels, in the direction normal to the local surface, with a maximum stroke of AE15 mm. Such a large stroke was required in order to achieve the second aim, i.e. to modify the shaped pro¯le back to a surface that is parabolic enough to increase the maximum operating frequency observable from the primary focus (Bolli et al., 2014). A further application of the active surface is also to recover the manufacturing deformations of the secondary mirror (Bolli et al., 2003).
In order to perform high frequency observations the surface accuracy must be around 150 m. The contractual requirement for the alignment of the primary re°ector was 500 m with a \goal" of 300 m. Successive accurate photogrammetry measurement campaigns have shown a lower limit in the total RMS of the primary mirror equal to 290 m at 45 of elevation (Süss et al., 2012). These measurements, carried out in six di®erent elevation positions, allowed the production of a look-up table, used to correct the gravity deformation with the actuators in an \open loop" con¯guration. In order to further improve the measurements of the primary mirror pro¯le, microwave holography measurements will be implemented by using the hardware setup successfully tested at the 32-m Medicina radio telescope (Serra et al., 2012).
Moreover, several solutions are being studied and implemented so as to perform real-time measurements of the re°ector deformations and the misalignments errors in the subre°ector position, with the¯nal aim of correcting them in a \closed-loop" strategy (Pisanu et al., 2012;Prestage et al., 2004): (i) improved FEM analysisto understand the e®ects due to gravity, temperature and wind on the behavior of the mechanical structure of the antenna, and to guide the optimized design of the sensor systems; (ii) inclinometersto monitor the status of the track, the inclination of the azimuth axis and the variations of the alidade deformation as a function of the temperature; (iii) position-sensing devicesto optically monitor the lateral shift of the secondary mirror; (iv) optoelectronic linear sensorsto measure the deformations of the primary mirror; (v) optical¯ber range¯nders -to measure the deformations of the legs of the quadrupod in order to control the sub-re°ector position; (vi) thermal sensorsto map the distribution of temperature in the most sensitive areas of the radio telescope and predict the deformations induced by thermal gradients.

Antenna control software
The SRT consists of a large number of modules and devices to be managed and controlled in a timely and precise fashion. The many observing modes available must be commanded and harmonized with the data acquisition and the recording of timestamps and housekeeping data. The overall control software performing all of these actions is called Nuraghe, a software package exceeding half a million lines of code. It runs in the framework of the ALMA Common Software (ACS, a distributedobject framework in turn based on CORBA), which provides a general and common interface for highlevel software, hardware and other software parts above the operating system. For single-dish operations, various modes are supported such as raster scan, On-The-Fly (OTF) cross-scan and mapping, sky dipping, position switching and focusing. The system is also able to operate all the three Italian antennas under a common interface (user interface and scheduling system), as well as to accommodate external applications such as VLBI or pulsar observations. All data coming from the integrated back-ends are stored in a FITS¯le. The current release of Nuraghe (Orlati et al., 2012), which is written in Cþþ and Python, enables the operator to control most of the subsystemse.g. the antenna mount, the receiver chains, the back-ends, the active surface and the minor servo (Buttu et al., 2012).

Auxiliary instrumentation
We now discuss several associated ancillary activities such as radio spectrum monitoring, weather parameters measurement and time-frequency standard distribution, as they are fundamental aspects for the optimal operation of the SRT.
A mobile laboratory is available at the site to carry out dedicated measuring campaigns to monitor and identify Radio Frequency Interference (RFI) in the frequency range between 300 MHz and 40 GHz . Special attention is paid to the frequency bands observed by the¯rst-light receivers and to those allocated by the Italian Frequency Allocation Plan to the Radio Astronomical Service (see Sec. 3.3).
A rather sophisticated ground-based set of weather measuring instruments are deployed at the SRT site. Besides conventional sensorsambient thermometer, hygrometer, wind speed/orientation measurer and barometera multi-frequency radiometer is employed. It generates real-time estimates of di®erent microwave brightness temperatures, opacity, zenith water vapor and liquid content. The measurements conducted so far show a 40-45% probability of¯nding an integrated water vapor column density of less than 10 mm in winter time. The absence of cloud cover can be found 50% of the time in winter and 80% during summer; typical liquid water values range between 0.2 and 0.7 mm.
In winter, the opacity at 22 GHz is lower than 0.15 Np for 90% of the time (40% during summer). At higher frequencies, in the 3-mm band, the winter opacity is lower than 0.15 Np for 35% of the time.
The Time and Frequency laboratory consists of an active hydrogen maser producing the reference signals for the receiver local oscillators, the 1 PPS (Pulse per second) timing for data acquisition and all the related equipments. A dedicated GPS receiver derives the local clock o®set (as a di®erence from UTC GPS) for VLBI and provides the time to the antenna for its pointing via an IRIG-B generator (Ambrosini et al., 2011b).

The¯rst radio source
SRT detected its¯rst radio source in Summer 2012. The observation was performed with the C-band receiver, installed in the Gregorian focus at the time. The main uncertainties at that time were the encoder alignments of the two main movement axes. Using the Moon, a wide and bright target, as a¯rst reference radio source and moving the antenna in steps of a half beam-size in a cross-scan, we measured the following encoder o®sets: þ1.3 in azimuth and À0.5 in elevation. The successive OTF scans were performed across 3C218, chosen because, being not too far from the Moon on that day, it allowed us to use the just-measured \local" o®sets, in the absence of a pointing model. The antenna response is illustrated in Fig. 6; the noticeable pattern asymmetry is due to the employment of¯xed optics, observing far from the elevation of their mechanical alignment (45 ).

Pointing accuracy
The SRT software was designed to manage a complete and independent model for each receiver and each focus. Three models were thus processed, initially considering 11.3 0 , 2.7 0 and 0.8 0 as reference Half-Power Beam-Widths (HPBWs) for the L-band, C-band and K-band receivers, respectively.
To achieve a preliminary pointing model, we applied the encoder o®sets obtained during the¯rst radio source observation and produced many raw maps over bright and compact sources in order to measure the displacement of the optical axis with respect to the ideal one. This procedure permitted the creation of a basic set of parameters, which was used as a basis for the pointing models for all the three foci.
As a next step to improve the pointing model accuracy, we performed several series of spot observations (OTF cross-scans along the Horizontal frame) of selected pointing calibrators, each time measuring the pointing o®sets. In order to minimize thermal e®ects on the structure we usually performed the observation during the night, in any case avoiding the sunrise. Once the whole azimuth/elevation plane in the sky was satisfactorily sampled, the pointing model polynomial was¯t to the dataset (Maneri & Gawronski, 2002;Guiar et al., 1986). This process required several sessions to converge to an acceptable solution, for which the model residuals are required to be lower than one tenth of the beam size.
The RMS residuals after the¯t are presented in Table 5 and show considerably less than one tenth of the beam size up to 23 GHz. The pointing performance at higher frequencies (the beam dimension will be around 12 arcsec at 3 mm) will take advantage of metrological systems (as discussed in Sec. 2.3).

RFI
Dedicated campaigns were carried out, both with the mobile laboratory and through the telescope, in order to detect and identify the interfering signals. Generally speaking, the RFI environment was found to be fairly quiet in the C-band (especially above 6400 MHz) and the K-band (almost everywhere in the band), while the P-and L-bands appeared more polluted. Figure 7 shows the spectral acquisitions performed with the various SRT receivers. These scans were carried out spanning the whole azimuth range, repeating such 360 circles for di®erent elevations (every 3 above the horizon). These charts were obtained in \max hold" mode: all the instantaneous acquisitions are overplot and, for each frequency bin, the maximum recorded value is displayed, thus the diagrams represent a sort of \worst case scenario" of the signals detected in the above speci¯ed az-el ranges.
The L-and P-band environment is the most \polluted" at the site. Under these conditions, in order to¯nd a suitable frequency con¯guration for the Total Power back-end, whose narrowest built-in band is 300 MHz, we installed external tuneable (5% of the central frequency) band-pass¯lters. For L-band acquisitions, an e®ective compromise was found and we successfully observed between 1696.5 and 1715.0 MHz. No solution, allowing reliable and repeatable measurements in the P-band, was found. Almost all the interfering signals were identi¯ed to be self-generated by the apparatus installed in the telescope. Among these devices we can mention the VOIP phones in Alidade and elevation Equipment Rooms, the XARCOS back-end, the encoders of the PFP, and some devices of the control electronics of the K-band receiver.
As concerns the C-band receiver, the general panorama was quite satisfactory; we discovered the presence of only a few interfering signals, mainly concentrated in the lower part of the receiver bandwidth. The signals at 5900 and 6016 MHz have been identi¯ed to be self-generated. The former is produced by one of the local oscillators of the multifeed system, the latter by the device that guarantees Internet connection to the station through a satellite link. The RFI signals coming from external sources were all due to digital links, and they appeared to be much attenuated when observing southward.
The K-band turned out to be almost free from RFI. All the identi¯ed polluting signals came from external sources were related to¯xed links for mobile operators. These signals can be practically neglected by avoiding known azimuth and elevation positions. The bands allocated to radio astronomical service (22-22.5 and 23.6-24.0 GHz) were con¯rmed to be fully usable. We emphasize that several interfering signals turned out to actually be \self-RFI", i.e. they were produced by devices installed at the SRT site. Since most of those devices are going to be moved into a properly shielded room in a very near future, we expect that the RFI environment will greatly improve and we thus do not consider the acquisitions performed during the commissioning to be representative of the¯nal site conditions. It must also be stressed, for the sake of the future users of the SRT, that spectral back-endsand speci¯c mitigation techniquesare being developed in order to cope with RFI and are expected to be operative within 2016 (see Table 4).

System temperature
The C-band receiver measured system temperature versus the elevation position is shown for both polarizations (LCP and RCP) in Fig. 8. Values were obtained for every 10 of elevation in the 6.7-7.7 GHz sub-band. The atmosphere opacity was estimated by performing a skydip scan ( 0 ¼ 0:014).
The single expected contributions to the system noise temperature are listed in Table 6. The sum of these elements (23.7 K, 22.0 K) is comparable to the obtained measures, 27 and 25 K, respectively for LCP and RCP. The 2-K di®erence between the experimental curves is consistent with the di®erent receiver temperatures (T rx ), whereas the slight differences between theoretical and experimental values can be explained with a higher-than-predicted Fig. 7. Spectral plots illustrating the RFI a®ecting the various receiver bands. These \max hold" plots show all the signals received while observing in the whole azimuth range, for several elevations above the horizon. spill-over temperature (mainly coming from the BWG mirrors). Table 6 includes an extra-noise term caused by the cover of the Gregorian room, a 1-mm Te°on¯lm 1.5 m in diameter, which protects the apparatus from the external environment. The¯lm trade name is Virgin PTFE, and the refraction index (not provided as a function of frequency) declared by the manufacturer is 1.35, yielding a re°ection amounting to 4%. We measured the Tsys increase by temporarily removing the cover; results are summarized in Table 7, where the average of the recorded increments is shown for di®erent frequency bands. As shown in Table 7, such a contribution appears in the C-band measurements, but it is de¯nitively more signi¯cant in K-band.

Gain curve
The e±ciency measurement and the gain curves were produced exploiting the continuum back-end, performing cross-scans in the Horizontal frame on bright, point-like calibrators (Table 8). Atmospheric opacity was extrapolated from skydip scans.
This campaign was repeated in two phases, re°ecting two di®erent telescope con¯gurations. During the¯rst phase, the main mirrors were¯xed to the reference mechanical alignment. The second session was carried out when the active surface and the sub-re°ector were aligned according to photogrammetry, in order to compensate for the gravitational deformations that vary with elevation.
The antenna gain in the 7.0-7.7 GHz band is shown in Fig. 9 for the¯xed optics. The plot compares the theoretical values (triangles) with the experimental measurements for the LCP channel (circles). The expected curve was obtained taking into account both the receiver and the antenna efciency parameters, including the surface e±ciency computed from the RMS of the mirrors, in the hypothesis that the optics were aligned at all the elevations (not achievable in real measurements, as the subre°ector was not tracking). The trend of the measured curve re°ects the expected one, however a gain o®set (about 10%-15%) is evident at all elevationsincluding 45 , where the antenna optics were supposed to be e®ectively aligned.
The plots of Fig. 10 show the BWG receiver gain curves for both the circular polarizations, obtained with active surface and subre°ector tracking. The°atness of the curves along all the elevation range demonstrates a great improvement due to the compensation for structural deformations; on the other hand, the peak is slightly below expectation (0.61 instead of 0.66 K/Jy). We might Table 6. Contributions to the Tsys in the 6.7-7.7 GHz band.

Contribution Notes
T rx LCP 8.5 K RCP 6.8 K

Lab measurements
Gregorian cover 1.4 K Observations Atmosphere þ CMB 4.1 K ¼ 0:014 at zenith Ground spill-over 2.7 K Simulation BWG spill-over 7.0 K Simulation then conclude that the optical alignment of the main and secondary mirrors is quite satisfactory for this wavelength, and that the employment of the active surface is able to deliver an almost constant gain. Figure 11 shows the beam deformation (") with respect to the estimated HPBW (beam size, e ), for both telescope con¯gurations (¯xed and active surface). The beam size is a reliable indication of how the optics of the telescope re°ect the theoretical design; accordingly, the deformation of the beam along the elevation span could be a symptom of a°a wed telescope structure. The beam deformation is computed through Eq. (1), where az and el are the beam sizes measured along the azimuth and elevation axis, respectively. The comparison of the two curves led us to the same conclusion inferred through the gain curves: the e±ciency reduction observed with the¯xed alignment is almost completely recovered using the active surface and the tracking subre°ector.

Gregorian focus commissioning
The characterization of the K-band receiver, located in the Gregorian focus, was obviously complicated by the impact of weather conditions on radiation at these wavelengths. It was thus decided, so as to achieve very accurate and e®ective measurements, to limit the commissioning activities to the time intervals showing low and stable opacity. During these clear-sky periods the opacity at the site turned out to be particularly favorable.

System temperature
The K-band receiver Tsys measurements were performed following the same procedures employed for the BWG receiver, for both the polarizations, repeating the procedure for each 2-GHz sub-band. Figure 12 shows the LCP values for the central feed.  The atmospheric contribution was evaluated by means of skydip scans, whose results were compared to the radiometer measurements. Table 9 lists the main Tsys contributions.
Even taking into account the extra system temperature contribution due to the cover¯lm, wē nd that the measured values do not match expectations. The sum of these contributions reaches 49 K, quite distant from the observed value of 73 K if we refer to the 18-20 GHz band. The cause of such a di®erence between measured and theoretical was not completely understood and therefore is still under investigation.

Gain curve
The list of calibrators observed during the e±ciency measurements is given in Table 10. Each calibrator was sampled after performing preliminary crossscans to focus the telescope and achieve an optimized pointing. The atmospheric opacity was again estimated by means of skydip acquisitions.
A preliminary gain curve, achieved in the 22-24 GHz band, was acquired with the¯xed optics con¯guration. The experimental and the theoretical results, in Fig. 13, are comparable. Measurements show lower values with respect to simulations, as the latter were computed taking into account constantly alignedyet non-activeoptics, while measurements were carried out keeping the mirrors in their reference positions, i.e. the positions obtained with their mechanical alignment at 45 of elevation. In this reference position, where the optics alignment should be ideal, a slight loss in the peak gain is still noticeable (0.52 versus 0.55 K/Jy). This can be explained as a residual inaccuracy in the mechanical alignment.
Similarly, the gain curves for both LCP and RCP were measured enabling the active surface and the subre°ector tracking. We expected the curve to be°a t at 0.66 K/Jy, taking into account the RMS of the surfaces of mirrors, the alignment of M1 -estimated with photogrammetryand other involved parameters such as: accuracy of the alignment of the subre°ector panels, errors in the measurements on panels, gravitational, thermal and wind e®ects on panels, positioning accuracy of the actuators. The plots in Fig. 14 clearly show that, in terms of peak gain, the data match the predictions. On the other hand, even if the overall telescope e±ciency bene¯ts from the active surface and the tracking of the subre°ector, there is still a substantial decrease in the e±ciency below 45 of elevation. This excess of gain loss at lower elevations can easily be translated in a surface accuracy of about 500 micron (El ¼ 20 ). Our investigations suggest that gravity deformation measurements were not su±ciently accurate at low elevations. In view of the installation of higher frequency receivers, this problem shall be solved; however, this will be done in the re¯nement phase by using the microwave holography technique.

Beam size and bidimensional pattern
Measurements of the beam size and pattern were performed alternatively enabling and disabling the active surface, in order to verify its e®ectiveness and to roughly evaluate the quality of the optical alignment. The outcome of these experiments was also meant to be exploited to draw further conclusions about the observed gain curves, and to prove whether there was a relation between potential alignment inaccuracies and the gain drop observed at low elevations. Figure 15 compares the beam deformation " (Sec. 3.4.3, Eq. (1)), measuring the deviation from the nominal HPBW, observed in the two telescope congurations (¯xed and active surface) by means of cross-scans on non-resolved sources. Figure 16, instead, shows two raw maps in arbitrary counts, acquired on one of these sources. They illustrate the antenna beam patterns, respectively obtained with   the¯xed and active mirrors, in the same elevation range.
The telescope in¯xed optics con¯guration showed a beam size rapidly deforming as the pointed elevation deviated from the mechanical alignment elevation (45 ). Astigmatism and coma lobes, revealed in Fig. 17 (top), explain such a distortion, at least for the upper range of elevation (El > 60 ), that is compatible with the gain drop shown in Fig. 13. The pattern in Fig. 17 (top) shows also a distortion tilting at about 45 with respect to the az-el axes. This deformation is not aligned with the gravity vector, so it must be a contribution of several di®erent ones. The compensation of gravity deformations obtained employing the active surface of the primary mirror, on the other hand, allowed us to obtain a particularly e®ective alignment for elevations above 45 . Figure 17 (bottom) demonstrates that most of the distortions of the antenna beam pattern had disappeared in this range. An asymmetry in the sidelobes is still visible, likely due to some residual squint.
Finally, the map in Fig. 18, taken at the average elevation of 22 , highlights a signi¯cant asymmetry and deformation, visible both in the main beam and in the¯rst sidelobe. This deformation can be considered in agreement with the characteristics of the gain curve presented in Fig. 14, as it can be explained by the current status of the optics.
3.6. Primary focus commissioning 3.6.1. System temperature After the preliminary tests, the primary focus receiver turned out to have an issue in the microwave chain responsible for the signal phase shifting producing circular polarizations from the native linear ones. The problem a®ected the RCP only: the noise calibration signal was not detectable for this channel. The cause, although immediately identi¯ed, could not be¯xed before the completion of the commissioning, thus no further measurements could be performed. For this reason, this paper discusses only the results for the LCP. Table 11 presents the theoretical contributors leading to an estimated temperature of 27.0 K when the telescope is parked at zenith and up to 44.0 K when the elevation angle is 5 . Even considering all the caveats, e.g. the intrinsic inaccuracy in the prediction of the noise calibration signal levelthe system temperature is higher than expected possibly due to an underestimation of the simulated spillover and/or other noise sources. Figure 18 for example provides hints about the in°uence of RFI on the measured Tsys values: it shows clustered measurements, in particular in the elevation range from 15 to 50 , creating \jumps" in the Tsys trend. This likely derives from the presence of variable RFI a®ecting the signal level.

Gain curve
The L-band gain curve is given in Fig. 19. It shows slightly scattered measurements but, in this case, the gain value equal to 0.52 K/Jy is entirely in agreement with the expectations (0.50-0.55 K/Jy for the band under test). These measurements were performed using the active surface, yet only to re-shape the primary mirror in order to obtain a parabolic pro¯le. Elevation-dependent corrections were not applied, as it is not necessary to compensate for gravity deformations at these low frequencies. The parabolic pro¯le allowed us to increase the antenna gain with respect to the original quasi-Gregorian shaping. Using the latter, the antenna gain would rapidly decrease for frequencies above 1.3 GHz, reaching 0.4 K/Jy at 1.7 GHz. The re-shaping of the surface might allow us to pro¯tably employ receivers up to a frequency of 22 GHz in the primary focus.

Summary of the commissioning measurements and milestones
The entire process of the telescope commissioning took almost 18 months, from mid 2012 to the end of 2013. Table 12 shows the schematic timeline of the major steps carried out, including examples of the technical advancement that has been taking place in parallel to the AV activities. Table 13¯nally summarizes the main telescope measured parameters.

Conclusions
During the technical commissioning phase, all the devices foreseen for the telescope early activities were successfully tested. The SRT overall performance was proven to be close enough to expectations. Three receivers (L-, C-and K-band) and a Total Power back-end were fully characterized. The new antenna control system was continuously improved and updated during the activities, allowing for the execution of test observations in the most common single-dish modes. The capabilities of the primary re°ector active surface and of the tracking subre°ector were widely demonstrated, as the measurements performed enabling these devices turned out to be almost elevation-independent. Re¯nements regarding the optics are still needed only for K-band observations; further improvements, also in anticipation of the installation of even higher frequency receivers (up to 100 GHz), will be achieved utilizing microwave holography and metrology techniques, at present under investigation and testing. As concerns the environmental conditions of the telescope site, our experiments con¯rmed that the location meets the requirements for high-frequency observations. The occurrence of RFIaggravated by the local presence of temporarily unshielded apparatuseswas found to be an impediment only for P-band observations; the C band was mostly employable and the K band was con¯rmed to be particularly unpollutedespecially above 20 GHz.
The telescope was hence made available to the Astronomical Validation team, in charge of assessing its scienti¯c potentials, while proceeding with the installation and testing of additional devicessuch as digital spectrometersin view of the shortcoming opening of the SRT to the worldwide community.