Barrier bucket gymnastics and transversely split proton beams: Performance at the CERN Proton and Super Proton Synchrotrons

During the 2018 proton run, a new radio-frequency beam manipulation has been studied and successfully implemented at the CERN Proton Synchrotron (PS) for the first time. This technique is used to deplete a well-defined fraction of a continuous longitudinal beam distribution by creating a so-called barrier bucket. We propose a new application of the barrier bucket gymnastics in the multiturn extraction scheme used at CERN. These two exotic techniques are combined into a highly sophisticated procedure that dramatically reduces the beam losses at PS extraction, thus paving the way to high-intensity proton beams for future fixed-target experiments at the CERN Super Proton Synchrotron (SPS). In this paper, the expected performance of the PS and SPS is analyzed in detail to define a road map for making this novel extraction scheme operational.

Figure 1

The rf voltage normalized by the peak amplitude (top) of the pulsed rf system generating a barrier bucket with the reflection region ${\varphi }_{\mathrm{r}}$ and drift space ${\varphi }_{\mathrm{d}}$. The bottom plot is the illustration of the corresponding trajectories in the longitudinal phase space, including the separatrix (dashed line).

Figure 2

Diagram of the quarter areas of the phase space before (purple) and after (yellow) a particle reflection off a moving barrier comparing fully filamented cases. The buckets are shifted and centered around the edge of the reflection region for before and after a reflection to allow an easy comparison of the area changes in the reflection region and the drift space. The two trajectories phase spaces are shifted such that the main parameters used in the calculation are visible.

Figure 3

Left: phase-space area ratio estimate for low relative barrier speeds. Note that at zero barrier speeds the growth per reflection is exactly zero. Right: limiting case of the relative barrier speed at half of the phase velocity of an outer particle in the bunch; on this scale, all curves converge to the singularity regardless of the ${\varphi }_{\mathrm{d}}/{\varphi }_{\mathrm{r}}$ bucket aspect ratios.

Figure 4

Evolution of the bunch profile during fast expansion and compression at a normalized speed of $\dot{{\varphi }_{\mathrm{b}}}/\dot{{\varphi }_{\mathrm{f}}}\approx 0.25$. The oscillations after 60 ms are a clear indication for a nonadiabatic manipulation. Simulated (left) and measured (right) profiles are shown with their corresponding bunch lengths.

Figure 5

Evolution of the bunch profile during slow expansion and compression at a normalized speed of $\dot{{\varphi }_{\mathrm{b}}}/\dot{{\varphi }_{\mathrm{f}}}\approx 0.01$. The smooth transitions indicate an adiabatic manipulation. Simulated (left) and measured (right) profiles are shown with their corresponding bunch lengths.

Figure 6

Beam reflecting off a potential barrier, shown over half a PS turn. The amplitude of the $h=16$ system was lowered and then subsequently turned off at 70 ms, and the amplitude of the barrier bucket rf system was increased to reach the maximum at the same time. The transverse beam splitting was disabled during these measurements. Simulated (left) and measured (right) profiles are shown.

Figure 7

Examples of the longitudinal beam structure measured in the transfer line without a barrier bucket (top), with an asynchronous barrier bucket (middle), and with a synchronous barrier bucket (bottom) compared to the rise of the PS extraction kickers. The 200 MHz modulation is clearly visible.

Figure 8

The summed profiles from the fast beam current transformer in the TT2 transfer line compared with the WCM trace before extraction.

Figure 9

The maximum of the cross-correlation sequence of the last acquisition before extraction from the PS and the summed five-turn profiles measured in the transfer line, standardized. Each thin, horizontal line represents a pair of observations. In this systematic analysis of the profiles similar to Fig. 8, when the main feature of the profiles is the gap itself, the similarity is larger. This means that the losses at extraction are a smaller fraction of the profiles compared to the beam displaced by the barrier during one turn. Therefore, the loss reduction when the barriers are synchronized with the kickers and the SPS injection is expected to be effective even at moderate gap size.

Figure 10

Comparison of the low-pass filtered, normalized longitudinal line-density profiles at extraction at the same time in the rebucketing process for different acquisitions. The effect of the beam intensity on the line density and especially on the shoulders at the barrier edges can be seen for the same 180 ns gap size setting.

Figure 11

Normalized line-density modulation along the bunch, showing a decreasing trend with increasing intensity. The colors refer to the different gap duration settings as indicated in the legend, which match the observed gap width well. Independent of the gap duration, the line-density modulation improves with increasing intensity. 63 and 76 acquisitions for 180 and 250 ns are shown, respectively. The least-squares linear fits are plotted to highlight the downward trend.

Figure 12

Evolution of beam losses at PS extraction, measured by means of beam loss monitors installed in SS15 and 16, together with the programmed width of the gap generated in the longitudinal beam distribution by the barrier bucket during the experimental tests. The sudden drop of the BLM15 signal and simultaneous saturation of the BLM16 signal correspond to the copper blade set to parking position.

Figure 13

PS relative extraction efficiency measured during the tests performed with the barrier bucket for several configurations. Datasets including more than ten samples are shown using violin plots.

Figure 14

Example of beam intensity in the SPS as a function of time together with the evolution of the beam momentum. The vertical lines indicate the different stages at which the relative transmission efficiencies are evaluated.

Figure 15

Relative losses for the first and second injections (upper left) and during the flat bottom of the cycle (upper right) as a function of the width of the longitudinal gap in the beam distribution generated in the PS. No trend is visible up to a gap size of 250 ns, while a slight increase is observed for larger gaps. The lower plots show relative losses for front porch (left) and transition (right) as a function of the width of the longitudinal gap. A clear dependence on the gap width is observed.