Measurement of PEP-II Luminous-Region Parameters

1. L(z) data Boost Method Data BPM Measurements with Coupling Phase Advance Data Measurement of PEP-II Luminous-Region Parameters B. Viaud, Université de Montréal, Québec, Canada, H3C 3J7 C. O’Grady, J. Thompson, M. Weaver, SLAC, CA 94025, USA W. Kozanecki, CEA/Saclay, F91191 Gif-sur-Yvette, France The 3-D luminosity distribution at the IP of the SLAC B-Factory is monitored using e+e-→ e+e-,μ+μ-events reconstructed online in theBaBar detector. The transverse centroid and spatial orientation of the luminosity ellipsoid reliably monitor IP orbit drifts. The longitudinal centroid, sensitive to small variations in the average relative RF phase of the beams, provides a detailed measurement of the phase transient along the bunch train. Variations in horizontal luminous size are detectable at the micron level. The longitudinal luminosity distribution is sensitive to the e+,- overlap bunch length and the vertical IP β-functions. In addition to continuous on-line monitoring of all the parameters above, we performed detailed studies of their variation along the bunch train to investigate a temporary luminosity degradation. We also compare βmeasurements, collected over a year of high-luminosity operation, with HER/LER lattice functions measured by resonant transverse excitation. Our also measurements provide direct evidence for bunch-length modulation. Luminous-Region Monitoring At a luminosity of 7.1033 cm-2s-1, ~1000 e+e-→ e+e-,μ+μ- events are reconstructed per minute. The 3-D centroid, orientation (tilt) and size of the luminous ellipsoid, determined from the spatial distribution of event vertices, are routinely provided to the accelerator, every 5 to 10 minutes. Also provided online are the horizontal angular divergence, as well as the e+e- crossing angle, calculated from the horizontal tilt of the luminous region combined with the average horizontal boost vector of μ+μ-pairs. Two examples are illustrated on Fig.1: the longitudinal centroid (Z, black points) and vertical tilt (dy/dz, black points) of the luminous region. The longitudinal centroid is compared to RF phase difference measurements (blue points), while the vertical tilt is compared to PEP-II BPM measurements (blue points). There is an arbitrary zero-offset in the vertical scales. Vertex Data BPM Data Vertex Data RF Phase Data Fig 1. β*y measurements Neglecting the x-y coupling and dispersive effects, the differential luminosity distribution L(z) is: where fc is the collision frequency, Zc is the longitudinal position of the collision point, N+,- are the e+,- bunch populations, σt +,- are the bunch lengths (with t=x,y). The dominantly Gaussian shape of L(z) is modified, through its y dependence, by an hourglass factor function of β*y+,- (β*x dependence is negligible). This can be expressed in terms of an effective β-function, provided the two optical waists occur at the same location zy+-=zyeff (eq.1). Σz, β*,eff, zc and zyeff are extracted from a fit of L(Z) to the Z-distribution of event vertices (Fig.2). The long term history of the measurement of β*,eff is shown in Fig.3 (left, black squares). Two other methods provide statistically-independent information about the vertical IP β-functions during high-current operation: the vertical luminous size σy (right) increases as a function of the distance to the waist, at a rate that depends on β*y+,- and ε*y+,- ; the longitudinal dependence of the vertical boost angular RMS spread y'B (left, green circles) also depends on β*y+, because y-y' correlation intensifies as one moves away from the waist. These β*y estimators are also compared with a common set of single-bunch phase-advance measurements (red open circles). The apparent inconsistency between the luminous-region methods and the phase-advance measurements can be largely resolved by taking x-y coupling into account (left, purple triangles): in all three cases, fitting luminosity and boost distributions simulated using BPM-based measurements and a fully-coupled formalism results in β*,eff values much closer to the luminous-region results. Eq 1. Curve: Fit Histogram: Data Fig 2. σy(z) Data Phase Advance Fig 3. Z-centroid and Bunch length measured along the bunch train The variation, along the bunch train, of Σz 2 and Zc is shown in Fig. 4. The time stamp of the Babar trigger relative to the accelerator turn signal, combined with the more precise drift chamber information, locates in the train, within 2 ns, the exact bucket where the event occurred . The individual minitrains are clearly visible. The Zc variation (bottom) along the full train reflects the differential RF-phase transient between the e+ and e- rings. The Babar measurements are compared with an analytical prediction of ΣzRF (computed from the individual HER/LER bunch-length predictions) and of the differential RF-phase transient. This prediction takes into account the actual bunch pattern, bunch currents and RF complement to compute, in each ring separately, the expected RF loading, phase transient, RF-voltage and bunch-length variation along the train. Because this model yields only the zero-current bunch lengths, it should be corrected for impedance-induced bunch lengthening, which is expected to remain approximately constant along the train. + Data Theory Fig 4. Variation of the IP beam parameters along the bunch train For several months in mid-2005, PEP-II observed a systematic variation of the luminosity along the bunch train. This is illustrated in Fig. 5a, where the rate of reconstructed e+e- / μ+μ- events is displayed as a function of where along the train the event occurred. The individual minitrains are clearly visible. The event rate, equivalent to the specific luminosity Lsp (uniform bunch currents), drops by ~15% from the head to the tail of each minitrain. This accounted for a ~10% deficit in average luminosity. Extensive studies were performed to understand these observations. In this particular dataset, a 1-2 mm Zc variation occurs within each minitrain. Geometrical considerations (ignoring beam-beam effects, if any) predict its impact on Lsp to be less than 2%. The most significant signal is a significant 2μm variation of the horizontal IP spot-size σxLalong the minitrain (Fig.5.c), which, depending on the IP spot size ratios, might explain most of the luminosity drop. Unfortunately, the statistical precision of the vertical spot-size (Fig.5d) is insufficient to reveal any equivalent systematic degradation, as might be induced for instance by electron-cloud effects. Studies of variations in Σz or β*,eff along the train provided no additional clue. Empirical tuning ultimately eliminated most of the σxL and Lsp variation along the train. Fig 5.