Browsing by Author "Naaranoja, Tiina"

The Large Hadron Collider (LHC) at CERN is currently being started up after a long shutdown. Another similar maintenance and upgrade period is due to take place in a few years. The luminosity and maximum beam energy will be increased after the shutdowns. Many upgrade projects stem from the increased demands from the changed environment and the opportunity of installation work during the shutdowns. The CMS GEM collaboration proposes to upgrade the muon system in CMS experiment by adding Gaseous Electron Multiplier (GEM) chambers. The new GEM-detectors need new Front-End electronics. There are two parallel development branches for mixed-signal ASICs; one comes with analog signal processing (VFAT3-chip) and another with analog and digital signal processing (GdSP-chip). This Thesis covers the development of the digital signal processing for the GdSP-chip. The design is described on algorithm level and with block diagrams. The signal originating in the triple GEM-detector sets special challenges on the signal processing. The time constant in the analog shaper is programmable due to irregularities in the GEM-signal. This in turn poses challenges for the digital signal processing. The pulse peaking time and signal bandwidth depend on the choice made for the time constant. The basic signal processing techniques and needs are common for many detectors. Most of the digital signal processing has shared requirements with an existing, well-tested Front-End chip. Time pick-off and trigger production was not included in these shared tasks. Several time pick-off methods were considered and compared with simulations. The simulations were performed first using Simulink running on Matlab and then on Cadence tools using Verilog hardware description language. Time resolution is an important attribute determined jointly by the detector and the signal processing. It is related to the probability to associate the measured pulse with the correct event. The effect of the different time pick-off methods on time resolution was compared with simulations. Only the most promising designs were developed further. Constant Fraction Discriminator and Pulse Recognition, the two most promising algorithms, were compared against analog Constant Fraction Discriminator and Time over Threshold time pick-off methods. The time resolutions obtained with noiseless signal were found to be comparable. At least in gas detector applications digital signal processing should not be ruled out of fear for deteriorated time resolution. The proposed digital signal processing chain for GdSP includes Baseline Correction, Digital Shaper, Integrator, Zero Suppression and Bunch Crossing Identification. The Baseline Correction includes options for using fixed baseline removal and moving average filter. In addition it contains a small memory, which can be used as test signal input or as look-up-table et cetera. Pole-zero cancellation is proposed for digital shaping. The integrator filters high frequency noise. The Constant Fraction Discriminator was found optimal for Bunch Crossing Identification.