Feynman diagrams of diffractive-production mechanisms of $J/\psi$ mesons at the LHC, where the double gluon system being emitted from the beam proton constitutes the pomeron. (a) is the pure CEP process, (b) has additional gluon radiation, and (c) and (d) involve proton dissociation. (Taken from Ref. [5]).

Layout of the active areas of the HeRSCheL stations around the LHCb interaction point (IP8), where for illustration the HeRSCheL stations have been magnified by a factor of 20 with respect to the rest of the LHCb detector. $z$-axis not to scale.

Energy deposit in the scintillators as a function of the pseudorapidity of the parent particle that caused the shower. The grey areas indicate the nominal pseudorapidity coverage of LHCb.

Schematic design of the scintillator and light guide of a single quadrant of B0/B1-type (47 mm inner radius), accompanied by a photograph of a B2/F2-type quadrant with the PMT and `clipping cable' attached.

Visualisation of a HeRSCheL station in its nominal data-taking position and in its parking position, when the detectors are retracted and rotated.

Photographs of the backward HeRSCheL stations B0, B1, and B2, respectively.

Photographs of the forward HeRSCheL stations F1 and F2, respectively.

Block diagram of the readout, powering, and control of the FSC detectors. The diagram shows only one quadrant and one of the two front-end crates.

(a) shows the signal per particle as a function of PMT high voltage for one of the quadrants of station B0. The absolute scale is determined using measurements with cosmic muons, and the evolution with high voltage using a pulsed LED setup. The analogue signal of one quadrant of station B0, recorded during injection tests (`TED shots') in November 2014, is shown in (b).

Average signal in $pp$ collisions as a function of integrated luminosity in 2015, for the four quadrants of station B0. The high-voltage settings were kept the same throughout the entire 2015 data-taking period.

Subtraction of common noise. (a) illustrates the correlated noise for the two integrator channels reading out one detector quadrant on station B0 in 2015 end-of-fill data, with respect to the signal in the spare cable (B0 reference), not connected at the detector end. (b) compares the raw signal in end-of-fill data taken during 2015 (red/darker filled histograms) with that after common-noise subtraction (blue filled histogram) and the raw signal RMS with the new adapter board in 2016 (dashed histogram). Each station houses four PMTs, read out via the dual-channel VFE board, and there are therefore eight channels to consider for each station. The relatively large contribution of uncorrelated noise to the signal in the B2 station was the result of imperfect grounding; this was resolved at the end of 2015.

Correlation between sum of raw HeRSCheL ADC counts on (a) the B side and (b) the F side of the interaction point, with reconstructed tracks on the same side.

Activity registered, after calibration, in one integrator attached to one counter for each HeRSCheL detector station during beam-beam crossings in the solid histogram, showing only the range up to 100 ADC counts. The empty-detector signal recorded after a bunch train is represented by the dotted histogram.

Distribution of HeRSCheL metric for events immediately following a bunch train (black line), representative of the response of the detector to CEP signal. The efficiency for an upper limit on the metric for retention of these events is also shown (red line).

Selected continuum-dimuon sample. In (a) is shown the invariant mass distribution of dimuon candidates. The shaded regions of resonant production are excluded. In (b) is shown the full $p^2_{\rm T}$ distribution of the selected dimuon candidates. Since the signal is expected to be concentrated chiefly below 0.1 $ {\mathrm{ Ge V^2 /}c^2}$ , the candidates below this threshold are shown in the inset histogram. For both regions, an additional histogram is superimposed, corresponding to the distribution of candidates in a background sample.

In (a) the distribution of $\ln(\xi_{\rm HRC})$ is shown for CEP (solid line) and non-CEP (dashed line) continuum-dimuon candidates. In (b) an example fit to the continuum-dimuon $p^2_{\rm T}$ distribution is shown, with the CEP signal template and double-exponential background indicated. In (c) the CEP signal efficiency and, for this process, background rejection are shown as a function of the limit chosen on $\ln(\xi_{\rm HRC})$.

Effect on the $p^2_{\rm T}$ distribution of $ { J \mskip -3mu/\mskip -2mu\psi \mskip 2mu}$ CEP candidates of applying a limit of 4.9 on $\ln(\xi_{\rm HRC})$. This cut is expected to retain close to 100% of CEP signal which have no additional tracks due to pile-up or single diffractive events in the combined LHCb and HeRSCheL geometrical acceptance. This corresponds to about 84% of the events which would be accepted if only the LHCb geometrical acceptance was used. At this working point, the yield of non-CEP $ { J \mskip -3mu/\mskip -2mu\psi \mskip 2mu}$ candidates falls to nearly a third of the original value.

HeRSCheL in the LHCb software trigger. Distributions of the sum of ADC counts recorded by the four counters in each station. The portion of events rejected by the limit in each station is indicated by the filled region.

Animated gif made out of all figures.