Development of the digital read-out system for the CERN Alice pixel detector

In order to gain new experimental insight at the TeV energy scale, CERN (Geneva) will build the Large Hadron Collider (LHC), a new collider machine operating at a maximum center-of-mass energy of 14 TeV (in the p+/p+ interactions). The accelerator can operate in a heavy ion collision mode achieving a center-of-mass energy of ~5.5 TeV. The experimental environment at LHC is characterized by a high crossing rate of the particle bunches (one every 25 ns for p+/p+) and high levels of radiation. Therefore stringent requirements are imposed on the performance of detectors at LHC. Such a particle physics environment calls for dedicated hardware/software solutions with specific constraints, such as radiation tolerance, limited amount of material and limited power dissipation. One of the particle physics experiments carried out in LHC is ALICE (A Large Ion Collider Experiment). The ALICE detector will face a very high density of tracks of particles (a multiplicity of 8000 charged particles per unit of rapidity, that implies a maximum density of ~90 tracks/cm2) and it comprises an enormous number of electronic channels (~ 2x107). Most of these channels (~15x106) come from the two layers of pixel detector, that produce a data rate of 75 Gbyte/s. The pixel detector system performs the bi-dimensional high-speed detection of the position of the tracks of ionizing particles, with a spatial resolution of ~ 12 mm. It comprises the silicon detector cells, the mixed front-end electronics (Pixel chip) and a digital module (Pilot system) located in the detector front-end, that accomplishes high level functions, typical of an external Data Acquisition stage. My main responsibilities were to contribute to the definition of the interfaces between the sub-systems and the design of the Pilot system. For such a complicated project like the ALICE detector, the definition of the design specifications and of the interfaces between the sub-systems is an important part of the study, in order to guarantee the feasibility of the project. Hence a closed collaboration between designers is often required, including the involvement of a designer in some design issues of the neighboring systems. This is why the pixel chip is extensively presented in chapter 2 (besides several contributions in the measurements discussed in the same chapter). The design of the Pilot system (described in chapter 3) challenges the processing of the huge amount of pixel data (in the experiment there will be 15 million pixels, managed by no more than 240 Pilot modules); moreover, the system plays the role of the master in the pixel fast-control protocol. The key idea to handle the pixel data is based on the low probability of a pixel cell being hit by a particle, due to the high granularity of the detector. A hit is represented by a logic value one in the pixel matrix received by the Pilot system: an on-line zero-suppression operation allows the full address encoding of every hit. This guarantees the required data compression rate, keeping a simple and reliable hardware implementation of the algorithm. After the encoding, an output stage running a CMI-encoded serial stream on a 40 meters copper cable transmits the data to the following stage (the router, that will be located outside the detector) at a bit rate of 155 MHz, thus minimizing the number of output links. The proposed architecture allows to reduce the clock frequency in the rest of the system. This avoids the risks and side-effects of the high-frequencies in such a harsh environment. This goal is reached also using state-of-the-art technologies, like the recent LVDS standard, for low-voltage differential binary transmission. As mentioned above, the interfaces between the detector sub-systems were an important part of the investigations, that are still in progress. As a consequence the design specifications of the single sub-systems are subject to several modifications in order to test and optimize the detector protocols. That is why, during the present stage of the preparation, a flexible implementation of the Pilot logic is required. Hence the system is prototyped on a board, based on programmable logic devices. In spite of the technology used, a 310 MHz clock has been successfully routed inside a programmable logic device. A dedicated set-up (based on LabView) for the testing of the board has been built, including two new custom boards (a receiver card for the test of the 155 MHz serial link, and a second one to interface the Pilot board with a system for logic testing). This set-up (described in chapter 4) allowed to check the correct behavior of the logic, in agreement with the Verilog simulations carried out during the design of the system. Once the specifications will be fixed, the final version of the Pilot system for the Alice experiment (supposed to begin in 2005) is foreseen to be on a single chip (ASIC). For the design migration to the ASIC technology, the use of automatic translation tools is under investigation. In addition, the board implementation already satisfies the requirements of other experiments (so far, the NA6i experiment at CERN-SPS).