The results of the NanoDST timing analysis of the BSC triggers showed that they performed exceptionally well over a large range of instantaneous luminosities and filling schemes. The minimum bias threshold 1 trigger, perhaps the most important signal sent from the BSC to the Global Trigger, fired in-time with>90%
of the Level-1 accepted events until the bunch spacing was reduced to 5BX (125ns) when the rate of albedo particles incident on the detector caused a degradation in the timing accuracy of the trigger. The minimum bias threshold 2 was much more stable over the increased luminosity range but possibly at the expense of triggering efficiency. The efficiencies of the triggers are studied in section 4.3 using the CMSSW framework.
The BSC OR trigger, which fires when one of more single channels detects a particle, was set up such that the output pulse to the GT was 2 bunch-crossings (50ns) long. This was to improve the coincidence efficiency with the BPTX AND.
Further studies of this highly sensitive trigger must be done with the BPTX in order to get any quantitative information about its efficiency.
The High Multiplicity trigger was close to 100% accurate in its timing of triggering L1A events, creating very few (<1%) late triggers and has therefore provided an excellent tool for selecting high multiplicity events as well as providing a way of confirming the timing of the BSC triggers with the correct bunch crossing time, provided from the BPTX.
The beam halo triggers aimed to trigger on beam background and ideally, should show a very low contribution within the L1A dataset but is dependent on the amount of halo in the LHC and the timing is dependent on the nature and origin of the beam halo. As can be seen in table 4.1, the halo triggers fire 30 - 60% in the center of the readout window and 20 - 50% in the following two bunch crossings.
The beam halo triggers were not enabled in the L1 accept path and appear here only if they fired within an event which was triggered by some other means.
4.3 BSC Minimum Bias Trigger Turn On Curves
The BSC trigger data is recorded with all other technical and Extra-Algo triggers into the CMS data stream via the GT. Each recorded event contains a vast amount
Triggering Performance During 2010 LHC Running 71 of information from all CMS sub-detectors. Exploiting this fact, the following plots show how the BSC minimum bias triggers performed during early-mid 2010 run-ning. Figure 4.11 compares the BSC minimum bias threshold 1 and threshold 2 triggers and the equivalent HF minimum-bias trigger, the HF_Coincidence trigger which requires the detection of at least one particle over 4 GeV in a single tower in both calorimeters simultaneously. The trigger efficiencies were calculated as a function of the number of tracks in the inner tracker with events selected by the zero-bias trigger. The number of pixel tracks scales linearly with lumi-nosity and therefore, a low number of tracks can be considered as a typical low luminosity event, whereas many pixel tracks can be equated to a typical event seen at higher luminosity. pp runs from May - July 2010 were processed during which time the initial luminosity rose from 3.11×1028cm−2s−1 to 9.6×1029cm−2s−1 (L= 31.1 mb−1s−1 - 960 mb−1s−1) and a√
senergy range from 900 GeV to 7 TeV.
For an event to be deemed minimum bias, a BPTX coincidence (AND) trigger was required as well as≥ 1 track in the inner tracker. For each event passing this cut, the values of the BSC minimum bias triggers and the HF coincidence trigger were checked to see if they fired on the event. The efficiency was calculated by the ratio;
= NTriggers NCut
(4.1) where NCut is defined as events in which the BPTX_plus_AND_minus trigger fired and with ≥1 track seen in the tracker. NTriggers is defined as those events which satisfy the cut and also were fired by the BSC Minimum Bias trigger or HF Min-imum Bias trigger. Beam background is excluded by the requirement that the BPTX XOR trigger, which signals the arrival on non-colliding beam, is false.
Although the geometrical coverage of the BSC tiles is smaller than that of the HF, the BSC minimum bias triggers out performed the HF coincidence trigger for low multiplicity events, which are common at low enegies (900 GeV) and luminosities (<1030cm−2s−1). The 4 GeV lower energy threshold and broad energy resolution at low energies restricted the HF Calorimeter’s ability to act as an efficient trigger during the start-up phase of the LHC, which is where the BSC system excelled.
Number of Pixel Tracks
Figure 4.11: Calculations of the detection efficiency of the BSC minimum bias triggers (threshold 1 & threshold 2) for increasing number of tracks recorded by the CMS tracker provides insight into the turn on performance of these triggers
relative to the HF Minimum Bias trigger. Data from runs 136035 - 141881.
4.4 BSC trigger status report for 2011
The BSC Minimum Bias and OR triggers performed exceptionally well over a wide range of luminosities. However, the limiting factor of the BSC triggering performance in 2010 was the bunch spacing and channel multiplicities (the num-ber of particles interacting with each of the scintillator tiles). As the time between bunches approached 5 bunch crossings (125ns), the effectiveness of the BSC min-imum bias and OR triggers declined. The underlying causes for this were the
Triggering Performance During 2010 LHC Running 73 presence of albedo and the occurrence of pulse pile-up in the individual channels where each signal is unable to return to the baseline before the following bunch arrives. This caused a drop in the baseline below the -30mV discriminator thresh-old and some triggers failed until the baseline was restored. This problem was able to be compensated for by reducing the high voltages to the PMTs, restoring the signal amplitudes and reducing the time required for the signals to return to their normal zero volt baseline as shown in the photographs of figure 4.12. How-ever, this reduction in PMT voltage also decreases the efficiency of the individual channels and therefore, the efficiency of the trigger. At the beginning of 2011, it was decided to inform the CMS collaboration the BSC triggers were no longer operating at the required efficiency during nominal running and the system was configured to operate primarily as a beam monitoring detector.
Figure 4.12: Effects seen in the BSC due to pulse pile-up during April 2011.
(Left) The bunch spacing is 3BX (75ns). The analog signals were unable to return to zero volts before the next pulse arrived, resulting in a drop in the baseline below the −30mV discriminator threshold. (Right) Adjustment of the high voltages by ∼ 20 - 40V temporarily solved the problem at the cost of single channel efficiency. The adjustments needed to be repeated as the LHC
luminosity and bunch occupancy increased.
As the LHC filling schemes developed to bring the experiments up to full design luminosities, the BSC was no longer be able to cope either as a trigger, nor as a meaningful monitoring device. It’s large channel size lead to a large channel occupancy and large quantity of energy deposited in each tile per collision. In turn, this resulted in large amplitude signals which were unable to recover in less than the 25 ns or even 50 ns between bunches. The simplicity of the design also meant that it was impossible to filter out background noise due to ‘albedo’ and HF activation. An upgrade of the BSC must be capable of filtering out such random signals if it is to provide accurate triggers. The design of an upgrade depends very
much on the requirements of CMS. Intended as a monitoring detector capable of providing commissioning triggers, the BSC became a primary triggering detector.
Many of the triggering needs are now supplied by other CMS subdetectors such as the HCal and Muon systems. Minimum bias triggering is no longer required at a time when the LHC delivers high luminosities resulting in dozens of events with each colliding bunch. Beam timing measurements are currently done by the BPTX whilst the monitoring of beam losses is carried out by the BCM1L, BCM1F and BCM2. The only tasks not well covered by any BRM subdetector are the monitoring beam backgrounds during collisions and online measurements of luminosity in CMS. It is feasible that the BSC upgrade could fulfil one or both of these tasks.