HISTORY OF CHANGES. SHiP TP 2015 A RELEASED. Page 2 of 43 REV. NO. DATE PAGES DESCRIPTIONS OF THE CHANGES 0.1

Transcription

1

2 Page 2 of 43 HISTORY OF CHANGES REV. NO. DATE PAGES DESCRIPTIONS OF THE CHANGES Version 0.1 completed, distributed for checking Version 0.2 complete included all comments Updated erroneous proton loss data from Update of figures. PDF conversion problem.

3 TABLE OF CONTENTS Page 3 of 43 Extraction and beam transfer for the SHiP facility Introduction to beam transfer for SHiP Electrostatic extraction septa ZS performance limits Slow extraction ZS electrostatic septa Potential ZS performance limitations at SHiP intensities WIRE HEATING SEPTUM BEAM CONDITIONING HIGH VOLTAGE FEEDTHROUGHS AVERAGE BEAM POWER AND ANODE HEATING HIGH SPARK RATES AND CATHODE DAMAGE Improvements and testing Conclusion on ZS performance limits Beam Losses During Extraction Beam Loss Mechanism During Extraction Comparison with SPS Operation During WANF HighLevel Dosimetry Activation Impact of Increased Activation Possible mitigation measures Conclusion on beam losses during extraction Beam Line Design Beam Line Design Conclusion on beamline design Beam instrumentation for SHiP extraction and beamline Existing instrumentation POSITION AND PROFILE MONITORING BEAM INTENSITY MONITORING BEAM LOSS MONITORING Changes needed to the existing instrumentation New instrumentation needed for the new section of SHiP beamline Conclusion on beam instrumentation Design and powering of new splitter magnets Design requirements Magnet coil design Magnet yoke design Splitter powering Schedule and lead times Conclusion on new splitter magnet and powering Beam Dilution Sweep profile over target Dilution magnets... 30

4 Page 4 of Conclusion on dilution sweep Interlocking Hardware interlocking design requirements Hardware interlocking concept Magnet current surveillance Fast Magnet Current Change Monitor Interlocks Hardware interlock user inputs Software interlocks Conclusion on interlocking Spill structure and control Slow extraction servo spill EFFECTIVE SPILL DURATION LOW FREQUENCY HARMONIC CONTENT OF SPILL FROM MAINS RIPPLE MEDIUM FREQUENCY VARIATIONS RESIDUAL 200 MHZ BEAM STRUCTURE Conclusion on spill structure Machine Development Studies Foreseen program of studies Conclusion on machine development studies Conclusions and R&D activities References... 41

5 1. Introduction to beam transfer for SHiP REFERENCE EDMS NO. REV. VALIDITY Page 5 of 43 The SHiP experiment requests an intense beam of protons with a momentum of 400 GeV/c delivered to its target station foreseen to be located in the North Area at the SPS. After consideration of different extraction techniques, a slowextracted spill over 1 second was chosen as the baseline scenario considering the maximum acceptable instantaneous particle flux at the experiment and other constraints given by the maximum beam power accepted by the design of the SHiP target. A nominal intensity of p/cycle will be extracted as part of a 7.2 s SPS slowextraction cycle with a 1.2 s flattop, transferred via TT20 and a new beam line, and diluted onto the SHiP target with an rms beam size of at least 6 mm in both horizontal and vertical projections. For machine protection arguments, the proton beam momentum will vary from 400 GeV/c by at least 510 GeV/c, with 390 GeV/c a likely choice. In this baseline scenario, the percycle intensity would offer a total of protons on target per year, while providing to the other North Area targets, thus meeting the required number of protons on target for SHiP of in 5 years. The main challenges in meeting the integrated intensity requirements of SHiP lie in the beam extraction system of the SPS. The accelerated intensity of p/cycle is well within the reach of the SPS, which has previously accelerated up to p/cycle; however the slowextraction of p/cycle is equal to the historical maximum extracted intensity per spill. When combined with a quicker spill and a factor two increase in the total number of protons extracted per year the challenges start to become evident. Clearly, with the use of conventional slowextraction small losses are unavoidable on the wires of the electrostatic septum (ZS) that must necessarily intercept the beam. Increased activation of the extraction region will lead to challenges for the reliability of the system, with its maintenance, intervention and dose management becoming more important. The behaviour of the ZS under the pressures of an increased spill rate and total extracted beam intensity are still to be checked and understood in future machine development studies. The issues faced by the extraction system are outlined at length in this annex along with the mitigation techniques needed to realise the SHiP experiment at the SPS. In subsequent sections, the design of the new beam line is outlined, including the new splitter/switch magnet that is required to operate both the existing North Area facility and SHiP in parallel, the dilution system required to dilute the beam power density impacting the SHiP target, the interlocking system, the beam instrumentation and the expected time structure of the spill. 2. Electrostatic extraction septa ZS performance limits The beam intensity needed for the SHiP cycles are p+ extracted at 400 GeV/c with a 1 second long flattop, every 7.2 seconds, through the existing extraction channel in LSS2, to TT20 and the new section of SHiP beamline.

6 Page 6 of 43 The high extracted intensity per spill and the short spill length are both concerns for the performance of the electrostatic ZS septa, which are very special technology highvoltage systems used to split the beam with as few losses as possible. In the following sections the ZS septa design and operating conditions are described, together with the performance concerns. Data from comparable operational periods are presented to estimate the realistic operational performance limits, together with recommendations for improvements in instrumentation, controls and interlocking to minimise the potential impact on the ZS and the overall machine availability. 2.1 Slow extraction Slow extraction from the SPS in LSS2 is routinely operational and used for supplying beam to the North Area. The slow extraction is accomplished with a set of suitably located extraction sextupoles used to create a stable area in horizontal phase space. This initial phase space area is larger than the area occupied by the beam. A dedicated servoquadrupole consisting of 4 short QMS quadrupoles moves the tune towards QH = shrinking the stable phase space area. The beam is debunched and the chromaticity set to a large negative value. For a given momentum and thus tune, protons with coordinates outside the stable area move away from the beam core along the outward going separatrices, and eventually cross the wires of the ZS septum, into its high field region. s the particles into the magnetic elements of the extraction channel consisting of thin MST and thick MSE septum magnets, which move the beam into TT20 proper. A servo quadrupole is used in combination with a beam current transformer in TT20 to modulate the rate at which the beam is extracted. The high chromaticity means that the extraction is effectively made in a combination of momentum and betatron space, with highest momentum offset particles coming into resonance and being extracted first. There is therefore a momentum change through the spill, which via the dispersion in the transfer line will couple into position changes in time in the transverse plane. 2.2 ZS electrostatic septa The ZS septa were designed and built in the 1970s [1]. Each unit consists of two parallel electrodes of approximately 3 m in length. The circulating beam passes through the fieldfree region within the hollow anode, which is bounded toward the cathode by an array of this W / Re alloy wires, Figure 1, and when particles cross the wires they experience the high electric field between anode and cathode that deflects them across the thin magnetic septum downstream. Five such ZS units are needed to extract the beam at 400 GeV/c. The main parameters are given in Table 1. The design has a high complexity with the array of 2080 anode wires aligned precisely by the anode support, to within 20 m straightness. The use of Invar for the anode support of the first 3 ZS anodes in the extraction channel minimises mechanical deformation with beam heating. Clearing electrodes with a few kv potential are needed inside the anode to sweep any ions produced from residual gas, which can cause sparking if allowed to drift into the high field gap.

7 Page 7 of 43 Figure 1. ZS electrostatic septum used for slow extraction from SPS Clearly the electrostatic septum must be as thin as possible to minimise beam losses, and must contain as little material per unit length as possible, for the same reason. In addition it must be compatible with UHV and high voltage and must also withstand the intense heating by the beam. This determines the use of 60 m diameter W76 / Re24 alloy wire for the septum anode plane for the first, most upstream, 2 ZS anodes. Table 1. Parameters of electrostatic septa ZS Parameter Value Electrode length [mm] 2997 Cathode material Anodised Al Applied voltage [kv] 220 Interelectrode gap [mm] 20 Operating field [MV/m] 11 Sparks per year [per unit] <10,000 Decoupling resistance [M] 400 Anode support material Anode wire material Invar W / Re Anode wire diameter [m] 60 Number of anode wires per unit 2080 Anode wire spacing [mm] Potential ZS performance limitations at SHiP intensities The performance of the ZS with these very high extracted beam intensities in a short spill is likely to be a key factor in the overall SHiP performance. For very high extracted

8 Page 8 of 43 beam intensities, the ZS septa are liable to experience increased sparking, vacuum pressure rise, and eventually damage of the wires through beam heating. There are also secondary effects like high voltage feedthrough damage from beam losses Wire heating Tests and operational experience in the past have shown that the ZS anode wires can withstand slow extraction spills up to and above the SHiP value of p+. The half integer extraction used to the West Area Neutrino Facility extracted a much higher instantaneous flux of protons, with p+ extracted in about 5 ms [2], which is around a factor 75 higher in p+/s. Similarly, slow third integer extractions have been made for operation to the North Area with p+ extracted in 10 seconds. Studies for worstcase impact of a fastextracted LHC beam were also made in the past [4], which indicated that temperatures of around 2300 C would be reached by the direct impact of an LHC type beam, with zero divergence and protons impacting the wire. For SHiP, about protons are expected to impact the wire per spill, which even without any heat loss mechanism (dominated by radiation), the wires would reach C, which is below the operationally assumed limit of 1000 C and well below the 3100 C melting point of the material. Tests made in the SPS with ZS wires have confirmed these simulations [5], with a measurement of protons at 450 GeV/c impacting a 60 m wire before it broke through melting. This is almost an order of magnitude above the number of protons expected to impact the ZS wires per extraction during SHiP operation. In the past, damage to the ZS wires has occurred when operational errors have led to the circulating beam core impacting the wires [9], and this must be prevented by adequate interlocking both with fastresponse BLMs but also with settings and function tracking and locking capability Septum beam conditioning Operation with high extracted intensity will require a period of conditioning of the septa, with progressively increasing beam intensity. This is for vacuum and high voltage reasons, as the hot wires outgas significantly high vacuum pressure can degrade the high voltage performance significantly, for instance through the production of ions which penetrate the anode wires and are accelerated onto the cathode, causing a spark. It is expected that the conditioning period with beam could take of the order of a week, before the full operational intensity is reached. Beam conditioning will need to be repeated after each venting of the ZS system High voltage feedthroughs The recurring damage to the main ZS HV feedthroughs encountered during WANF operation was due to ionisation and bubble formation in the dielectric liquid of the ZS feedthrough [6], shown together with the plug and cable in Figure 2, which resulted in discharges in the feedthrough and eventual punchthrough of the structural and insulating alumina. Mitigation measures against this weakness have already been deployed on the operational ZS, notably with significant improvements in the hydraulic system used to circulate and regenerate the dielectric liquid. Since these measures were

9 Page 9 of 43 taken, the number of broken feedthroughs has dropped enormously, from a total of ~23 in the years 1983 and 1984 [7], to 0 since Average beam power and anode heating The SHiP cycle of p+ extracted in 7.2 s represents a doubling of the extracted beam power compared to past operation at p+ extracted in 14.4 s. This will double the power deposited in the anodes of the septa, the first three of which are made of Invar with a low coefficient of thermal expansion. The effect on the alignment of the septa needs to be verified with beam, with the risk that extra heating can result in worse septum alignment and an increase in specific beam loss per proton. The higher power may also result in a higher dynamic vacuum level which could result in more sparks, as a direct result of the higher pressure or as a result of more positive ions escaping into the high field gap. Again, studies with beam will be needed to determine whether these effects occur and whether they can be mitigated e.g. by modification of the ion trap voltage or with the extra pumping capacity for the ZS planned as part of the LIU project. Figure 2. Alumina ZS HV feedthrough (top) and associated plug High spark rates and cathode damage The final concern is about excessive sparking of the high voltage with very high intensities, which results in a collapse of the field seen by the beam and eventual beam

10 Page 10 of 43 loss on the downstream ZS or MST. Very high spark rates can also irredeemably damage the cathode or even lead to damage of the feedthrough, and as such are protected against by detection systems and interlocking which cuts the high voltage and can as well dump the beam [8]. 2.4 Improvements and testing Some sparking of the ZS during slow extraction is an inevitable feature of proton operation, however, and improvements to automatic surveillance, trending and the on line analysis of operational extraction parameters like specific beam loss per proton and spill quality are all measures which will both minimise beam losses and activation, and reduce the spark rates and risk of equipment damage. A full analysis of requirements in terms of ZS protection interlocks and surveillance should be made at an early stage, which will allow time for development and deployment of such tools. The instrumentation used for setting up of the slow extraction needs a serious review and upgrade, to allow more precise control of the process, faster setup and an easier optimisation. These developments will need to be tested with beam, together with the SHiP cycle and the 1 s extraction spill. Machine Development tests with increasing intensity extracted over 1 s to the North Area on the SHiP cycle are planned at the end of the 2015 proton run to probe experimentally the limits, at least in terms of scaling of the behaviour, measurements of stability and of spill quality. 2.5 Conclusion on ZS performance limits The proposed intensities for SHiP appear to be within the acceptable operational envelope for the ZS septa, based on considerable previous operational experience and the total and peak proton intensities extracted. Nonetheless, given the high activation expected for the ZS and the serious consequences of faults in the event of damage, a serious effort needs to be made to improve interlocking, diagnostics, surveillance, trending and longterm control of the extraction quality. Such developments will also directly benefit the Fixed Target program through increased availability and reduced beam losses in the extraction region. It must be borne in mind, in case of a ZS failure, and taking into account the ALARA procedures and restrictions imposed by RP, that the cool down time (without beam towards North Area and SHiP) may be significantly longer than the repair itself. 3. Beam Losses During Extraction The annual extracted beam intensity of p.o.t. proposed for SHiP is approximately a factor two higher than the previous record achieved at SPS during the 5 year operational period of the West Area Neutrino Facility (WANF, ) through LSS6. Operation of the existing system at higher intensities will inevitably lead to higher activation of the extraction region and its components, and will provide challenges for its operation and maintenance.

11 Page 11 of 43 In this section historical measurements of activation and dose are examined as a function of the number of extracted protons and the results are put into the context of SHiP by scaling to provide preliminary loss estimates. The activation and dose estimates provide the first data for assessing the impact of SHiP operation, which will be important for understanding system reliability and intervention planning. Finally possible mitigation measures to reduce losses and activation are discussed and the necessary actions outlined. 3.1 Beam Loss Mechanism During Extraction During slow extraction protons with coordinates outside the stable area move away from the beam core along the outward going separatrices and eventually cross the wires of the ZS septum into its high field region. Unavoidable beam losses induced by protons intercepting the septum wires represent less than 1% of the extracted beam intensity for a wellestablished cycle. 3.2 Comparison with SPS Operation During WANF A total of protons were extracted at 446 to 450 GeV from the SPS through LSS6 during the 5year operation of WANF between 1994 and 1998 [10]. Concurrently, a total of protons were also slow extracted to the North Area via LSS2. During this period the intensity of the SPS proton beam increased from 3.5 to protons per cycle shared with slow and fastslow extraction to LSS2 and LSS6. The extraction mechanism for the neutrino program was a halfinteger fastslow extraction with a spill length of 6 ms and an intensity per spill of up to protons. The slow extracted spill length was 2.3 s shared between LSS2 and LSS6, with an intensity of up to protons per spill. The proton yields on the fixed targets (T1, T2, T4, T6 and T9) during WANF are collected in Table 2. It is expected that loss estimates based on fastslow extraction will be conservative in comparison to the slow extraction technique foreseen for SHiP. This is because the fastslow extraction technique is intrinsically more lossy owing to the higher proton density at the septum wire. It is expected that the losses per proton for SHiP will be lower than those for WANF the exact factor needs to be verified with beam tracking simulations. The extraction system that was in place in LSS6 during WANF is essentially the same as is presently installed in LSS2 to be used for extraction to SHiP in the North Area.

12 Page 12 of 43 Table 2: Extracted proton intensities [ p.o.t.] during the operation of WANF. Year T2 T4 T6 North Area T1 T9 West Area (slow) (slow) (slow) Total (slow) (fastslow) Total (LSS2) (LSS6) Total HighLevel Dosimetry The dose levels measured during SPS operation are documented annually [11] with readings taken from dosimeters placed at regular intervals around the SPS and at key positions in the LSS regions. Further details on the exact position and procedure for the dose measurements can be found documented at length in the aforementioned reference. The data from a dosimeter placed on a cable tray approximately 1 m from the middle ZS tank was used to assess the linearity of the dose as a function of the number of extracted protons. The results are presented in Figure 3. The time over which the measurements were integrated varies from annual and biennial measurements, with one dose reported for the entire WANF period. Caution must be taken when using the absolute dose measurements because during operation prompt losses can be highly directional and vary depending on the exact measurement position in the tunnel. Nevertheless, the results show a reasonable linear correlation as a function of extracted proton intensity. On comparison of the two datasets in LSS2 and LSS6 there is some evidence that slow extraction is indeed a cleaner extraction mechanism per proton than fastslow extraction. Scaling the dose out to a total of p.o.t. an estimate of the total dose to cables close to the ZS is found at the level of ~0.4 MGy over 5 years of nominal SHiP operation. In the event of a nonoptimized extraction one could expect larger dose rates. An attempt was made to compare the highlevel dosimetry data with more recent data but due to a recabling programme in the SPS since WANF and the movement of dosimeters it is difficult to make such a comparison. There are signs that the specific dose measured by a similar dosimeter close to the ZS is a factor of 4 larger in recent years. The reasons for this are not clear, and remain to be investigated.

13

14

15 3.5 Impact of Increased Activation REFERENCE EDMS NO. REV. VALIDITY Page 15 of 43 The higher extracted beam intensity will increase the activation of the extraction region with the likely consequence of increased radiation damage of sensitive equipment and cables combined with increased cooling times needed to make interventions [13]. To help characterise the activation of the extraction region an empirical model developed during the WANF period [1415] was revived to aid the understanding of the activation process. At this time in the mid90s dedicated radiation monitoring equipment was installed to monitor the activation of the region after beam stoppages. The data presented below consists of a sum of counts on all radiation detectors placed in LSS6, totalling about 10 detectors. The model fits a sum of exponential functions with three fitting parameters, and starting each day with an amplitude proportional to the number of protons extracted that day subscripted as day j: The model ( and ) fits very well data taken at least 5 hours after beam stoppage, which avoids the fast lived activation products that are complicated to model. The potential of the model to predict activation is shown in Figure 6 by the quality of the fit to 3 years of activity measurements made during WANF. The impact on access for intervention becomes immediately clear when scaling to SHiP intensities, as is done for the 1995 WANF run shown in Figure 7, after normalisation using the activation survey measurements. The model predicts that 10 msv/hr would be measured at the ZS during the activation survey after 30 hours, which can be considered as rather conservative based on fastslow extraction losses. For longterm high intensity operation the buildup of longerlived isotopes must also be included in the model. This work could be extended using the PMI detectors installed in the SPS to monitor the cool down and aid dose planning for interventions. 3.6 Possible mitigation measures The most obvious mitigation is a reduction of the extracted beam intensity. A factor of two would give about the same activation as WANF operation in This would clearly double the number of days of operation needed for the facility to reach p.o.t. Reduction of the loss per proton extracted is a more attractive option. From the activation and dosimetry data, there is a rather large dispersion in the beam loss per proton, which presumably comes from a combination of different setting up of the extraction, stability of the extraction and machine, and the beam quality. It is clearly highly important to understand the reasons for the variations and to formulate methodologies to keep the operation at the lower end of the range. To this end, MD studies and simulation efforts will be required, together with the development of more sophisticated surveillance and interlocking software and upgraded instrumentation,

16

17 Page 17 of 43 Figure 7 Scaling of the induced radiation measured in LSS6 during the 1995 physics run to protons to compare the cool down with equivalent SHiP operation. 3.7 Conclusion on beam losses during extraction Dose and activation estimates for the nominal operation of SHiP at SPS were made based on historical data of past SPS performance. The upper limit for the dose at approximately 1 m from the ZS is estimated in the region of 0.4 MGy for the extraction of protons. The activation survey data predicts the peak activation measured during a survey made 30 hours after SPS operation is halted at 12 msv/hr, at approximately 1 m from the ZS. Increased activation leads to longer cool down times that can be modelled empirically and used to plan interventions. Possible mitigation measures are discussed, and a serious effort will be needed to investigate these and deploy the most effective for SHiP operation. It should however be noted that most of the failure scenarios will not affect the LHC beams significantly, as the LSS2 extraction channel is not used for LHC beam. 4. Beam Line Design The SHiP target location in the North Area allows the reuse of about 600 m of the present TT20 transfer line, which has sufficient aperture for the slowextracted beam at 400 GeV/c. The installation of new laminated and bipolar splitters [19], replacing the existing MSSB2117 splitter magnets, will permit switching the beam into a new line symmetric to the T6 beam line but on the left of the T2 instead on the right, see Figure 8. The new splitter will switch the beam to the left on the SHiP cycle but maintain the existing functionality and split beams to the rest of the North Area for Fixed Target cycles. The new beam line is approximately 360 m in length and will transport and dilute

18

19 Page 19 of 43 Figure 9 Optics from SPS extraction, TT20 and the new beam line to the SHiP target. The MSSBS splitter magnets are shown in white. The dilution magnets will be placed immediately downstream of the last dipole magnet but are not shown. x,y = 3000 m but can be varied in both planes to match the SPS emittance to the target, if required. A beam size at the target of x,y = 6 mm has been assumed. The cycletocycle variation of the TT20 quadrupole gradients x,y < 150 m through the splitter is shown in Figure 9. The gradients required in TT20 for this modification are smaller than presently applied, and all extra downstream quadrupoles are well within the maximum quadrupole strength of 80.8 T for QNL type magnets. The cycletocycle powering of TT20 will be tested in a dedicated MD and the transmission through the splitter aperture verified. Without suppression of the dispersion function it would grow as large as 40 m at the target, which has the potential to cause the beam to move during the spill as the tune in the SPS is varied and the momentum of the extracted beam changes. Although the power converters in the transfer lines of the North Area are scaled linearly to suppress this effect in normal operation, large dispersion at the target should be avoided to reduce beam position sensitivity on the extracted beam momentum. It is possible to suppress dispersion at the target with the 5 new quadrupoles, as shown in Figure 11. The numbers and types of magnets needed for the beam line are given in Table 3, along with the number of converters and the maximum required current. The most critical aperture in the line is the vertical gap in the last MBB magnets. The ±312 with a conservative normalised rms emittance value of 8 mm mrad. The aperture restriction will need careful attention, with error studies and measurements of the extracted vertical beam emittance for the SHiP cycle to be done. Indeed, if the power converters in the North Area beam lines can be reliably scaled as will be tested in future MDs, then the constraint on having zero dispersion at the target can be relaxed and the quadrupoles used to control further the beam size. The dilution kickers were added after this vertical aperture restriction to ensure the acceptance is maximised. Table 3 List of new magnets required for the beam line to SHiP.

20

21 Page 21 of 43 Figure 12 Vertical beam size and aperture from the SPS extraction point to the SHiP target assuming a conservative rms emittance of 8 mm mrad. 4.2 Conclusion on beamline design The position of the SHiP facility and the geometry of the new beam line was fixed by modifying the first splitter magnet in TDC2 and adding 17 MBBtype dipole magnets to deflect away from the TT20 beam line. The TT20 line is reused with modifications made to the powering of the 9 quadrupoles directly upstream of the splitter for the SHiP cycle. There is ample space for the installation of dilution kickers after the last active beam line element before the target. The vertical aperture in the MBBs is critical and will need to be carefully followed up. 5. Beam instrumentation for SHiP extraction and beamline With the high intensities needed for SHiP excellent setting up and control of the extraction and beam transfer will be essential to minimise activation and also risk of damage to equipment. The quality of the setting up and of the subsequent monitoring and control will depend to a large extent on the beam instrumentation, which will also be used in some cases for interlocking and active surveillance of the extraction quality. 5.1 Existing instrumentation The existing instrumentation in the LSS2 extraction channel and TT20 [21,22] was developed in the 1970s to fulfil the difficult requirements of instrumenting the beam losses, position and profile in the highly activated and high prompt dose region of the slow extraction septa. The basic monitor technology has proven robust but there are several improvements which could help with reducing the setup time and also which could help to better extraction and spill quality, and to control beamloss Position and profile monitoring The slow extracted beam is debunched and has no RF structure, which makes position measurements with standard SPS BPM electronics impossible. The slow extracted beam position is therefore measured with Secondary Emission Monitors (SEMs) that are made

22

23 Page 23 of 43 smooth and steady extraction rate. The data is digitized at 10 khz, allowing observation and correction of harmonics up to about 5 khz. has been used in the past for diagnostic purposes [22], and is presently being upgraded with new technology. It will allow the user to monitor the intensity evolution within an SPS turn, i.e. 23 s, and turnbyturn during a full proton or ion extraction. In the past this instrument received signals from a PhotoMultiplier observing a Quartz screen and sampled them every 100 ns. It was hence possible to observe the intensity evolution of the extracted beam with a resolution of 230 acquisitions per turn. The recorded turns can be consecutive to check the turn byturn stability or separate to check the evolution over the whole extraction. As for the Servo Spill SEM instrumentation, a Fast Fourier Transform facility is available to analyse the frequency structure of the spill up to 5 Mhz. The new system based on diamond BLMs should provide similar functionality Beam loss monitoring Beam losses are monitored with standard SPS type ionisation chambers. These give a loss reading at a 20 ms time sampling, and are usually displayed with the loss per beam cycle, readout at the end of the extraction. Figure 14 shows beam losses measured with ionization chambers in the LSS2 extraction channel for slowextraction of p+. The frontend electronics of the computers controlling the monitors generate beam interlocks, with a single threshold per monitor for total losses through the cycle. To monitor the losses in the TT20 beamline there are a total of 9 standard BLMs on selected elements, for surveillance and interlocking. Figure 14. Beam losses measured with ionization chambers in the LSS2 extraction channel for slowextraction of p+. The red line represents in the interlock level, above which the beam is dumped.

24 Page 24 of 43 the TPST, the MST, the MSE, the enlarged quadrupoles and the TCE shielding block. The first 16 monitors in the extraction channel are connected to the fast loss beam dump (BLD) system electronics, where the interlock is generated directly in hardware. This produces an interlock with a s response time if the thresholds are exceeded, to protect against the failures which can happen on the timescales of ms or faster [24]. 5.2 Changes needed to the existing instrumentation Although the basic functionality of the existing instrumentation is adequate the systems are ageing technology, designed to norms that are now superseded in terms of handling and quick connection and disconnection in a radioactive environment, and sometimes with limited spares. Targeted improvements would address some of these issues, increasing the reliability and reducing the risk of failure. Operational efficiency and beamloss management reasons would also favour extending the functionality somewhat. The following upgrades are suggested for TT20 instrumentation: Installation of some detectors which can measure the position of the beam without 200MHz structure at key locations in the line, for initial steering during setting up, and stability and extraction trajectory monitoring; Replacement of some of the existing split foils with SEM grids, again at least in a subset of key locations, depending on expected radiation dose and cabling; Improvement of the BLM coverage, with the addition of monitors at more regular intervals along the line; Upgrade of the BLM electronics to LHC standard, with integration and readout times of the order of 20 s. Here a clear synergy exists with the Consolidation project, where this work is currently requested for LS3; Replacement of SEM devices in extraction septa with modernised devices. Here a clear synergy exists with LIU project where an upgrade of the ZS pumping modules housing the SEM devices is planned in LS2; It is also suggested to replace as many BSP monitors as possible with BSGs to make steering of TT20 more deterministic, as the BSPs require a guess of the emittance/beam profile which is not very reliable. Cabling needs to be investigated. 5.3 New instrumentation needed for the new section of SHiP beamline The new section of SHiP beamline will need to be equipped with instrumentation for setting up and surveillance of the beam transfer. Position, loss and current monitoring are all required, and a method of characterising the swept beam on the target and also analysing the sweep after each shot is proposed. The main items and approximate numbers of individual devices are given in Table 4. Again, as for TT20 the positioning measurement needs to work with unbunched beam. The detailed specifications for each item remain to be developed however, the functionality needs to be at least as good

25 Page 25 of 43 (in terms of precision, dynamic range, radiation resistance, lifetime) as the equivalent existing instruments. Table 4. Number and types of instruments for new section of SHiP beamline. Function # items Beam position monitors 7 BTV screen 1 Sweep monitoring screen 1 Beam loss monitors 12 Beam current measurement 1 Fast beam current measurement (ns resolution) 1 The fast beam current measurement will be used to make an online characterisation of the spill, which will be essential for the target interlocking and may also be needed for the experimental veto or offline reconstruction. Detailed requirements must be developed for this functionality. A large dimension (approximately 40 cm width/height active region) screen will be needed to characterise during setting up, and possibly continuously monitor, the beam sweep upstream of the target. 5.4 Conclusion on beam instrumentation The beam intensities and spill structure required for SHiP can be instrumented by the existing monitoring systems in the LSS2 extraction channel and TT20 beamline. However, a set of moderate improvements are proposed to improve the operability of the beamline, and to also allow effective interlocking of the extracted beam. For the new section of beamline, a set of new instruments will be required, most of which replicate the functionality of the existing devices. 6. Design and powering of new splitter magnets One of the machinerelated challenges for the proposed location of the SHiP experiment is the 400 GeV/c switch out of TT20 to the new beamline, due to the high beam rigidity, absence of available drift space and necessity to retain compatibility with the present type of North Area Fixed Target operation. The proposed solution is to replace the three existing MSSB2117 splitter magnets with newly built MSSBS splitters, which allow negative polarity powering on a cycletocycle basis. This would provide the new switch functionality and retain the existing splitter mode for the North Area. In the following sections the main requirements and proposed technical solutions are described.

26 Page 26 of Design requirements The three existing MSSB magnets , and need to be replaced by similar MSSBS magnets which allow enough aperture for the beam deflected in the opposite direction. The main requirements for the new magnets are: Replicate existing splitter functionality for present NA beams; Polarity reversal possible within about 2 seconds; Adequate goodfield region around both sides of fieldfree septum hole; physical length as present MSSB. The present MSSB design [1] is an invacuum Lambertson septum, with a vacuum separation to keep the coils and water connections in air, built with radiation robust materials and lowmaintenance assembly. The magnets operate with 0.8 T in the gap, and have a limit Bmax of 1.6 T in the steel at the point of the septum element. For the North Area splitter design, the wedge angle of the septum is 36, and since Bmax.sin Bgap, it would therefore be possible by running at higher current to increase the gap field to about 0.95 T without a major effect on the field quality the alternative of using a higher saturation steel like FeCo would gain something, but would be much more expensive, mechanically tricky and lead to activation issues. The gap is 75 mm high, which requires about 48 ka.turns to reach 0.8 T. The maximum offset for the switched beam at the exit of the 3 rd MSSB is around 90 mm. Allowing another 40 mm for the beam size and orbit, alignment tolerances, the good field region (and pole width) needs to be extended by 130 mm only, although an extra 150 mm would make the septum hole symmetric to the pole. The requirements are summarised in Table 5. Table 5. Summary of functional requirements for upgraded TT20 splitter magnets. Parameter Unit Value Comment Number of magnets 3 Plus spare(s) Physical yoke length mm 5.2 Integrated field T.m 3.76 Gap field T 0.8 At 400 GeV Vertical aperture mm 75 For deflected beam Horizontal aperture mm +150 / 130 Presently +150 mm Switching time s 2.0 From +ve to ve polarity, at full field Flattop length s 20 Need to be able to run in DC mode 6.2 Magnet coil design The present coil scheme of 48 turns and 1 ka can be retained for the MSSBS. The coil technology is special, using compacted MgO powder around a central copper current carrying watercooled tube, mechanically supported by an external grounded copper sheath. The MgO is evacuated to avoid moisture degradation the maximum voltage to earth is 1 kv.

27

28 Page 28 of 43 Table 6: Parameters of existing MSSB and new MSSBS magnet. Parameter MSSB New MSSBS Magnetic length [m] Gap field [T] Stacking factor [%] Coil turns Current [A] Vertical gap [mm] Pole width [mm] Magnet inductance [H] Coil resistance [m] Number of magnets in series 3 3 Minimum risetime [s] 10 (?) 2 Maximum voltage to ground (3 magnets in series) [V] ~ Splitter powering For the powering of the new MSSBS magnets, a variant of the Linac4 transfer line converters (APOLO family) can be used. These can be assembled in 4 modules to deliver the requested current/voltage, in order to allow all three magnets to be powered in series. If needed, the maximum voltage of around 400 V could be reduced by a factor of two with balancing the voltage in +/, however this does not seem necessary to foresee at this stage as 400 V is well within the voltage limit of the MgO coil conductor. 6.5 Schedule and lead times The main technical issue is the design and construction of the new splitter/switch magnet, which will require a longer leadtim An initial R&D phase will be needed to verify that the design can achieve the very tight mechanical tolerances needed in the septum region, critical for beam losses. This R&D should start in mid2015 to be ready for the LS2 installation date, as the overall schedule for R&D, design, production and testing of these magnets is estimated at 48 months. It is essential to install the new splitters in LS2, as they are needed for normal North Area operation, and can only be installed after a long cooldown period to allow removal of the existing highly activated devices. 6.6 Conclusion on new splitter magnet and powering The construction of a new magnet MSSBS in place of the existing splitters will allow the switching of the full 400 GeV/c beam to the SHiP beamline and target. For this mode of operation there should be no significant beam losses at the new magnet; however, for the present mode of operation where beam is split to the T6 target the losses and activation will continue to be high. The technical design of the magnet appears feasible,

29 Page 29 of 43 based on an extension of the present technology with punched laminations. The powering of the magnet can be accomplished with a variant of standard CERN power converters. The R&D for the magnet needs to start in mid2015 to be ready for the planned installation in LS2. 7. Beam Dilution With the average beam power deposited on the target over the spill close to 2.5 MW the maximum beam energy density is crucial for the target design. To relax the demands on the target, a pair of orthogonal conventional magnets with a fast Lissajous powering function is foreseen to dilute the beam energy density during the spill by maximising the path length of the sweep on the target block. 7.1 Sweep profile over target A full optimisation of the sweep is required taking into account the possible magnet and powering characteristics, as well as the limitations of the target in terms of protons per mm 2. Some idealised sweeps have been proposed for evaluation. An idealised Archimedean spiral [29] has been investigated and used as baseline to develop target R&D [30], as shown in Figure 18: a constant transverse separation s, starting at a radius of 5 mm and finishing at 35 mm. The frequency of the oscillation is varied to maintain a constant sweep speed on the target, implying that the waveform powering the orthogonal dilution kickers must increase in frequency as the beam spirals in, or vice versa. In this case the sweep speed on the target is approximately 0.65 m/s. Figure 18 Reference beam trajectory on SHiP target. The variation in frequency and amplitude as a function of time during the spill is visualised in Figure 19, where the product of frequency and kick amplitude remains constant at a value of 0.7 Hz mrad.

30 Page 30 of 43 Figure 19 Horizontal and vertical beam positions on the target as a function of time, proportional to the current powering the dilution magnets. 7.2 Dilution magnets With a drift distance of ~120 m available to the target and a sweep radius tentatively fixed at 30 mm, a maximum deflection of ~0.25 mrad per plane is needed. The parameters demanded for the Archimedean sweep presented in Figure 18 are collected in Table 8. Table 8 Dilution system specification Parameter Value Number of magnets 2 (H and V) Dilution magnet to target distance [m] 120 Integrated field strength per magnet [T m] 0.33 Maximum kick amplitude [mrad] 0.25 Frequency x kick [Hz mrad] (constant) 0.7 Maximum frequency (min. amplitude) [Hz] 18.0 Minimum frequency (max. amplitude) [Hz] 3.0 Length of sweep on target [m] 0.67 Possible magnet types could be similar to the MPLH (SPS extraction bumpers) that can ramp with a di/dt of around 1300 A/s. With a current of ~80 A corresponding to 0.25 mrad, a maximum sweep frequency of approximately 3 Hz would be possible using the present type of magnet and power converter. Starting with larger amplitude and spiraling inwards would probably be easier for the dilution magnet powering. The monitoring and interlocking of the sweep currents will be very important for protecting the target. 7.3 Conclusion on dilution sweep The time scale of the sweep during the spill is relatively slow and the available drift space downstream of the final active element in the beam line before the target is

31 Page 31 of 43 sufficient such that conventional magnet technology can be used, which is already existing and in operation at SPS. 8. Interlocking The beam intensity needed for the SHiP cycles are p+ extracted at 400 GeV/c every 7.2 seconds, through LSS2, TT20 and the new section of SHiP beamline. In addition to measures needed to protect against beam loss in LSS2 and the TT20 beamline with this high beam power, there is a new section of beamline to interlock, new switchsplitter elements between the TT20 and SHiP beamline, and a dilution sweep system that must operate correctly to prevent damage to the target. The SHiP beamline will therefore require a new beaminterlocking system, in addition to the normal equipment interlocks for the new magnets and power converters. In this section the requirements, concept and proposed inputs for this beam interlocking are described. 8.1 Hardware interlocking design requirements The hardware beam interlocking for SHiP must fulfil the following requirements: Accept logical inputs (user inputs) from different equipment systems; The system should be extendable in case new additional inputs arise; High level of reliability, with redundant user inputs; Dump SPS beam in case of fault condition, within a reaction time which should be less than a millisecond; A subset of user inputs to be maskable, for settingup with low intensity beam; Masks must be automatically removed when intensity is above the setup limit; An interlock condition for the SHiP beamline should prevent injection into the SPS for the SHiP cycle, but not for other cycles; 8.2 Hardware interlocking concept The interlocking can be based on the standard Beam Interlock Controllers (BICs) used elsewhere in the SPS [26]. Each controller accepts 14 redundant inputs, of which 7 are mple AND of all unmasked inputs. The BICs are connected to the SPS Beam Interlock System (BIS) by means of optical fibres, and also to the Safe Machine Parameters system that provides several flags such as the setup beam flag. The inputs surveilling the extraction, lines, target etc. will require dedicated BICs with the outputs connected as inputs to the master BIC. The master BIC will be connected to the ring BA2 BIC. Three separate BICs are needed: one annel and TT20 beamline, one SHiP SHiP section of the beamline, the splitter (with SHiP elements used for the North Area beamline, targets and splitter settings for the NA targets.

32 Page 32 of 43 Two new Safe Machine Parameter (SMP) Flags will be needed to differentiate between NA fixed target beam and beam to SHiP: stable 390 GeV and stable 400 GeV (ST_390 and ST_400, respectively). These flags should be generated if the energy corresponds to the given value for more than several 10s of ms. The exact value needs to be defined, after analysis factors such as the ramp rates and power converter settling times. The four logic equations for the master BIC are then: NO_SLOW_EXTRACTION SHiP NA (1) NO_SLOW_EXTRACTION =! ST_390 &&! ST_400 (2) SHiP = ST_390 && TT21 && toship BIC (3) NA = ST_400 && TT21 && tona BIC (4) Figure 16. Schematic of new interlocking for SHiP and NA. Four additional BICs are needed, of which one is a Master BIC connected to the existing SPS ring BIC in BA2. The SPS SMP system supplies four flags to the extraction BICs: 2 x E_LHC (450GeV) = go to LSS4 and LSS6 extraction (for LHC) 1 x E_CNGS ( GeV) = goes to LSS4 extraction (for AWAKE) 1 x E_HIRADMAT (440 GeV) = goes to LSS6 extraction (for HiRadMat) Presently it is not possible to make more than four flags from the SPS SMP controller. In addition, an extension to the SMP concept is needed, in that the ST_390/400 flags need to be generated ONLY when the machine energy is stable at the specified energy, NOT when the magnets are ramping through the energy. This is needed to avoid generating a spurious beam dump e.g. when ramping the SPS to 450 GeV for LHC beam,