Presentation on theme: "Introductory comments on: “ How will the low level rf systems required for bunch compression, cavity tuning, machine protection, etc. be designed so as."— Presentation transcript:
Introductory comments on: “ How will the low level rf systems required for bunch compression, cavity tuning, machine protection, etc. be designed so as to perform reliably enough not to compromise machine operation?” Giorgio Bellettini ITRP Meeting May 26 th, 2004
Cold machine, 1 They address first the stability of acceleration phase in order to ensure stable bunch timing and energy, and no luminosity loss: “Each klystron in the main linac has a LLRF system that controls the phase and amplitude of the vector-sum of 32 cavities....typical phase stability of 0.3 deg. has been demonstrated at TTF I during beam operation. ….. For bunch compression …the specification…. may be as tight as ~ 0.1 deg.” “Potential for further improvement in phase stability exists with use of additional amplitude detectors, better isolation between channels, and temperature stabilization of the detectors.” COMMENT: plausible, but additions cost money.
Cold machine, 2 They address next the Mean Time Between Failures of RF controls: “..based on JLAB experience where 338 LLRF systems have been operated for more than 10 years…the MTBF of the digital feed-back control for amplitude and phase is expected to be of the order of 72 months…. A redundant simple feed-forward system …will control the rf at slightly reduced performance… the probability of both failing in the same month is 1/72*1/72 = 1/5000…and the probability of one failure in a linac (in a month) is 1/5000*280 (klystrons) ~ 5%. The probability of two such rf failures is ~ 2.5e-3. …. ◦. Since one rf station is permitted to fail, the availability of the LLRF is > 99.75 % (based on failure of 2 stations)… Mechanical frequency tuners have a lifetime of >20 years ….even so, as there are a large number of these mechanical tuners, it is imperative that their MTBF be well understood and sufficient…” COMMENT: plausible. Good the reference to experience gained at JLAB. Less strong the argument on the probability of double failure.
Cold machine, 3 They address very briefly the machine protection system: “For machine protection, exception detection (quench etc.) and handling will be implemented. A fast rf gate (redundant) is provided as input for interlocks from other subsystems.” COMMENT: information is not detailed. It looks to me that they did not design the machine protection system in detail yet. It sounds as if they consider this – which is a problem of major importance - to be of limited complexity.
Warm machine, RF phase control, 1 They describe first their control system of the acceleration phase: “The low-level RF system will use a common master source of timing and phase information for all regions and subsystems, and distribute this information via fiberoptic links to each sector of the collider. Each fiber optic-link is point-to-point, which allows the phase length of each link to be individually stabilized. … All key system design are based on the performance of prototypes and on other R & D results.” They quote the level of phase stability needed by the various RF systems and describe how to control and stabilize the phase to the required levels: “…. small changes in the low-level RF parameters… will cause luminosity reduction… the key issue is the system’s stability and whether it can reliably detect changes of the parameters…”
Warm machine, RF phase control, 2 How the controls are distributed locally: “Main distribution to Remote Receivers: …a prototype system using a 15 kilometer fiber was constructed and tested… demonstrated < ±1 picosecond variation over 1 month with a 10 degree C temperature variation of the transmitting fiber. This meets the requirements for systems other than the crab cavities and the bunch compressors… Distribution from Sector Receivers to Local Users: The trigger system is very similar to the existing SLAC trigger system……Two 11.424 GHz signals with equal amplitude… form the drive signal for the klystrons. By varying the relative amounts of the two signals the phase of the klystron with respect to the master source can be adjusted…”
Warm machine, RF phase control, 3 “Local Adjustment of Klystron Phases via Beam-Based Signals: …Over longer periods the klystron’s phase is adjusted by directly detecting the phase offset between the beam and the klystron’s RF power, and adjusting it to the desired value….this measurement relies on the fact that the NLC bunch train generates a very large RF power signal in each structure, and the phase of this signal is always purely decelerating with respect to the beam (ie, its phase is 180º)…. …At low current there is not enough signal from the beam to perform this operation….the relative phase between klystron and beam can be determined by measuring the phase of a structure’s output power under nominal conditions….and then measuring the output power with the klystron disabled for one pulse…this system was tested at S-band in the SLC where it demonstrated approximately 1.5 degree accuracy.”
Warm machine, RF phase control, 4 “Bunch Compressors and Crab Cavities: The bunch compressor relative phase tolerance is tighter than the main linac …(0.1 versus 0.25 psec)…but it applies only for a few seconds…the phase stability of the fiberoptic distribution prototype over such short times was not measured but is expected to be adequate…. The crab cavity tolerance is even tighter (0.025psec). Since the two crab cavities are quite close to one another, a single klystron provides the main power for both cavities, with a small “trim” klystron providing pulse-to-pulse corrections to one of the two cavities…” COMMENT: the phase control system is well understood and described, clearly based on great experience at SLAC. However, the required stability seems difficult to reach in some areas (note that crab cavities at IR are not needed for head-on collisions).
Warm machine, machine protection They outline their machine protection system: “…if a large number of stations fail simultaneously or simultaneously change their phases and/or amplitudes by a large amount a beam loss event may occur… drive the beam into the vacuum chamber or the irises ….Such events will be prevented by….redundancy of the master source, main distribution, and local distribution systems… …At the RF station the two reference signals are compared …. if the two master source phases are wildly different from one another… the phase of the Phase Comparison Unit will be used for RF genneration on the present linac cycle and a…inhibit signal will switch off the beam prior to the next linac cycle…” COMMENT: machine protection seems accurately conceived but difficult to realize.
spares Response: I. Introduction Uncontrolled variation in the amplitude and phase of linear collider RF systems can compromise machine operation in the following ways: 1)Small or uncorrelated changes in main linac RF parameters can change the energy and/or energy spread at the IP, leading to luminosity reduction 2) Small changes in bunch compressor RF parameters can change the arrival times of the two beams at the IP, leading to luminosity reduction
3) Small changes in the phase difference between the two crab cavities can lead to horizontal deflection of one beam with respect to the other, which will cause luminosity reduction 4) Large changes in bunch compressor and/or main linac RF parameters will change the beam energy by an amount sufficient to drive the beam into the vacuum chamber or the irises of accelerator structures, either of which would be damaged by the encounter. Issues (1) through (3) relate to small changes in the low- level RF parameters; here the key issue is the system’s stability and whether it can reliably detect changes of the parameters at the specified levels. Issue (4) is qualitatively different in that it pertains to preventing large changes from occurring in a relatively short time. The majority of this answer will address the stability and sensitivity questions; Section IV will discuss machine protection issues.
The low-level RF system for the X-band linear collider includes the following components: 1) A master source near the IP which generates a sinusoidal reference signal for the entire accelerator complex. The phase and frequency of the signal are controlled with high precision, since all RF systems site-wide will ultimately derive their own phase and frequency information from this waveform. The master signal also includes a “fiducial” which indicates the start of each linac cycle. Every device in the linac which needs to perform an operation at a fixed moment in each linac cycle will determine that “the moment has arrived” by detecting the fiducial (indicating start of a new linac cycle) and counting the number of oscillations of the waveform which follow the fiducial, until a fixed number have elapsed. Thus, the signal from the master source generates the baseline timing used by all timed operations in the site, as well as generating all RF signals used throughout the site.
2) A main distribution system which transports the master signal from its source to remote points throughout the complex. Receivers for the master signal will be positioned at 250 to 500 meter intervals throughout the complex. 3) A second distribution system which transmits the master signal from the receivers described in (2), above, to every device which requires RF or timing system information (klystrons, modulators, beam position monitor processing units, pulsed magnet controllers, etc.). 4) An assortment of local control equipment: programmable delays, phase shifters, frequency multipliers, etc. The most difficult specifications of the low-level RF system are all related to phase stability. Specifically:
1)The relative stability of the two crab cavities must be at the level of 0.025º of Sband (0.025 picoseconds) over a period of a few seconds. Over longer periods, a beam- based feedback can correct the relevant error by detecting a horizontal beam beam offset and correcting same. 2) The relative stability of the two bunch compressors must be at the level of 0.4º of X-band (0.097 picoseconds) over a period of a few seconds. Over longer periods, a beam- based feedback can correct the relevant error by detecting a change in the relative arrival times of the two beams at the IP and correcting same. 3) The stability of each klystron with respect to the master source signal must be at the level of 1º of X-band (0.25 picoseconds) over a period of one minute. Over longer periods, a beam-based feedback can correct the relevant error, as described in Section II.D, below.
4) The stability of the phase reference signal at each main receiver with respect to the master source signal must be at the level of 5 picoseconds over periods of one month. II. Subsystems and Performance The major components of the timing and phase distribution system from the master source to the individual klystron LLRF is described in this section. Figure 2 provides an overall schematic which summarizes the entire system. A) Master Source The master source is an oscillator which generates a sinusoidal waveform at 714 MHz. Once per 120 Hz linac cycle the phase of the oscillation is shifted; this phase shift is the fiducial, which is used by the local systems to indicate the beginning of the next linac cycle. The specifications and requirements on the source are not particularly challenging, and shall not be further discussed here.
B) Main distribution to Remote Receivers The phase reference system uses a point to point stabilized fiber link from the interaction region to each sector to distribute the 714 MHz reference and fiducial generated by the master source. The long fiber links use a diode laser modulated at the phase distribution frequency. Part of the light is reflected from the far end of the fiber back to the transmitting station. The phase of the RF on the reflected light is stabilized relative to the forward phase by using temperature to control the phase length of a fiber in series with the transmission fiber.A prototype system using a 15 kilometer fiber was constructed and tested. Phase measurements (with an independent read back) demonstrated < 0.2 picoseconds RMS noise for 1 minute (in a 10kHz measurement bandwidth), and < ±1 picosecond variation over 1 month with a 10 degree C temperature variation of the transmitting fiber. This meets the requirements for systems other than the crab cavities and the bunch compressors. Stabilization of the these systems is described in Section III.
C) Distribution from Sector Receivers to Local Users (“Fanout”) Within a sector, the reference signal is transmitted on coaxial cables using an RF interferometer system similar to that for the 3 km long SLAC main drive line. The maximum expected length of a local coaxial connection is approximately 500 meters, far shorter than the successful SLAC main drive line system. 1) Trigger Generation The system triggers are derived from the RF distribution system. Timing devices detect the fiducial described in II.A, above, and count cycles of the 714 MHz RF until the correct number have passed, at which time the trigger is generated. The generation and detection of a monocycle, low bandwidth fiducial has been tested. The trigger system is very similar to the existing SLAC trigger system with the exception that the reference signal is 714 MHz rather than 476 MHz, and it uses a lower bandwidth fiducial.
2) Klystron RF Generation and Phasing At each RF station, the 714 MHz reference is multiplied to produce the 11.424 GHz main RF signal. Two 11.424 GHz signals with equal amplitude are generated at each RF unit; one of the two signals is in phase with the incoming 714 MHz signal, while the other is 90º out of phase. The two signals are recombined to form the drive signal forthe klystrons. By varying the relative amounts of the two signals the phase of the klystron with respect to the master source can be adjusted. The existing “8-pack” LLRF uses this system, but with modular, rather than surface mount RF components.
D) Local Adjustment of Klystron Phases via Beam-Based Signals Over short periods (up to 1 minute) the klystron phase stability depends upon the phase stability of the incoming signal from the master source, and this stability is assured by techniques described in (B) and (C) above. Over longer periods the klystron’s phase is adjusted by directly detecting the phase offset between the beam and the klystron’s RF power, and adjusting it to the desired value. This measurement relies on the fact that the NLC bunch train generates a very large RF power signal in each structure, and the phase of this signal is always purely decelerating with respect to the beam (ie, its phase is 180º). The RF from the klystron and the beam mix in the structure, resulting in a change in phase of the structure’s stored energy; this phase change can be detected by measuring the phase of the power at the output coupler. By routinely performing this measurement, slow drifts of the klystron phase with respect to nominal can be detected, and a correction applied by changing the mixture of the two signals which determine the klystron phase (see (C) above). The klystron phase feedback described above relies on the potent RF signal generated by the full bunch train. At low current, for example single-bunch operation, there is not enough signal from the beam to perform this operation. In such a mode of operation, the relative phase between klystron and beam can be determined by measuring the phase of a structure’s output power under nominal conditions (thus measuring the klystron phase), and then measuring the output power with the klystron disabled for one pulse (thus measuring the beam phase).
This is the technique that will be used for initial phasing of the linac, and also for routine phasing during low-current operation. This second system was tested at S-band in the SLC where it demonstrated approximately 1.5 degree accuracy. III. Bunch Compressors and Crab Cavities The bunch compressor relative phase tolerance is tighter than the main linac phase tolerance (0.1 versus 0.25 psec), but the latter tolerance must be held for approximately one minute while the former only applies for a few seconds. The phase stability of thefiberoptic distribution prototype over such short times was not measured but is expected to be adequate.The crab cavity tolerance is even tighter than the bunch compressor tolerance (0.025psec). Since the two crab cavities are quite close to one another, a single klystron provides the main power for both cavities, with a small “trim” klystron providing pulse-to- pulse corrections to one of the two cavities. The design of the system is described in detail in .
IV. Machine Protection The failure of a single RF station will not change the beam energy sufficiently to cause a beam loss event, but if a large number of stations fail simultaneously or simultaneously change their phases and/or amplitudes by a large amount a beam loss event may occur. Such events will be prevented as follows: 1) Redundancy of the master source, main distribution, and local distribution systems. In the event of complete loss of a master source, a fiber, a main receiver, or a coaxial cable from a main receiver to an RF unit, all RF units will still be able to operate. 2) At the RF station the two reference signals are compared to each other, and to the RF phase from the previous linac cycle. This is accomplished by Phase Comparison Units, which are high-Q phase-locked loops which do not respond to rapid changes in phase. If the two master source phases are wildly different from one another, or if they are both wildly different from the phase expected by the Phase Comparison Unit, then the phase of the Phase Comparison Unit will be used for RF generation on the present linac cycle and a machine- protection inhibit signal generated which will switch off the beam prior to the next linac cycle. The Phase Comparison Units must maintain X-band phase to a few degrees over 8 milliseconds – well within the performance of standard commercial oscillators.
3) A request for a large change in phase or for a large change in the complement of active klystrons will generate a machine protection inhibit signal. This protects the machine against the possibility of inadvertent switching of off vast numbers of RF stations between linac cycles. 4) Variations in beam energy due to slow changes in RF system performance will be trapped by the machine protection orbit monitoring described in the answer to question 10. 5) To eliminate single points of failure in the compressor high power rf system, short sections of accelerator structures are individually powered with separate modulators, klystrons and LLRF. This ensures that the change from the loss of a single section is too small to produce a beam which could damage downstream systems.