Pasi Karppine - ProtoRhino
Kimmo Mustonen - ProtoRhino
Antti Meriläinen - ProtoRhino
Timo Hildén - Helsinki Institute of Physics
Erik Brücken - Helsinki Institute of Physics
Rauno Lauhakangas - Helsinki Institute of Physics
Matti Kalliokoski - Helsinki Institute of Physics
Eija Tuominen - Helsinki Institute of Physics
The Helsinki Institute of Physics (HIP) is physics research institute responsible for a Finnish research collaboration in the European Organization for Nuclear Research (CERN). HIP has an impressive record developing high-energy particle detectors—both semiconductor and Townsend discharge—for demanding international collaborators. During the Large Hadron Collider (LHC) project, HIP played a major role constructing the Compact Muon Solenoid (CMS) tracked detector hardware and software.
Developing highly sensitive detectors and faster readouts is not just important for big physics research. Their use also extends to applied problems in cancer treatments and international control of radioactive materials. Many potential applications require simultaneously sensing the position, time, and energy of a particle traversing the detector. At CERN, the LHC development also faced this challenge.
Similar systems are built for different purposes, only on a much smaller scale. One example is the Gas Electron Multiplier (GEM) that HIP developed. The GEM is an array-type detector composed of tens, hundreds, or even thousands of individual multiplier cells. In calibration as well as in development, we must sample each of the cells at a high rate to satisfy the requirements to detect the particles in space and time (spatiotemporal), which leads to extremely high data flows. Processing the data flow is a major engineering challenge, with up to ~60 Gb/s gathered from a single GEM array detector alone. The only technologically feasible solution is to analyze the data on the fly.
ProtoRhino Ltd. is one of the few companies in Finland that can develop systems based on NI FPGAs, which are the most promising technology for the implementation. We have an impressive track record not just in FPGA-reliant projects, but also in delivering tailored solutions for high-data-rate acquisition with real-time analysis capacity in broader scale. Our development team could thus support and contribute in low-level customization and system integration to streamline the project flow and improve the outcome.
As an NI Alliance Partner, ProtoRhino is a reliable subcontractor, so HIP chose our group to develop the required technological means.
GEM Detector Principles
The position-sensitive GEM detector is based on the same operating principle as a Geiger counter—the Townsend avalanche. The GEM detector is an array of small Geiger tubes (with a more complex structure), with embedded readout electronics instead of a circuit creating the rasping Geiger sound.
The following physical process generates the detector signal:
- A high-energy particle encounters the detector and is absorbed by the gas medium, exciting electrons from the atoms.
- An external electric field separates and accelerates these charged particles, ions, and electrons, exciting further atoms as they go (hence the name Townsend avalanche). The extent of the avalanche, and thus the total current on the collector electrodes, is proportional to the energy of the absorbed particle.
- The readout electronics amplify and record the current pulse, which lasts about 100 ns, as depicted (Figure 1 left).
Due to the short pulse duration, we have historically used an analogous amplification stage to elongate the pulse in time domain ([Field]Figure 2 right). While this procedure sets lower requirements for the readout speed, it also limits the particle incidence rate (the previous pulse must die out before the next incidence) and hinders the accuracy of the energy determination due to ill-defined amplification.
In the past, HIP relied on custom readouts developed at CERN. Such systems have an advantage of being suited for a given task, but are not supported beyond the lifetime of the academic project they were developed for. This has led to relatively short life cycles, rendering the amount of time and money invested unbearably high.
Before consulting ProtoRhino, HIP surveyed other technologies available on the market. HIP identified the FPGA products of Caen and Aptek as potential technologies. However, these FPGAs lacked the flexibility that is the very foundation of FlexRIO. This drawback would severely limit their use in a project that pushed the limits of the existing technology.
Data Readout With FlexRIO
HIP deemed NI products flexible with the longevity provided by a major vendor. In particular, NI recognized FlexRIO as an appealing platform that could match the varying requirements of all particle detector architectures and allow continuous upgrades in channel counts and acquisition rates up to Gigabit samples per second.
The ProtoRhino development team embedded the amplification stage into the GEM array and hardwired the amplified signals to the FlexRIO FPGA front ends (Figure 3). This solution met the low-loss linear amplification requirement on hundreds of parallel channels at up to a 50 million samples per second per channel acquisition rate. Furthermore, due to the hardware-level flexibility and the NI simulation tool kits, we could develop and test each part of the system separately, saving time and money during the final implementation.
Benefits of Using NI Products
We implemented the GEM readout front end with a pair of NI 5752 32-channel digitizer adapter modules, which provided an excellent channel-price ratio for the high data rate of 50 MS/s. We connected the 32-channel digitizer adapters to a pair of PXIe-7965R FlexRIO modules for the on-the-fly data analysis.
We used LabVIEW and the LabVIEW FPGA Module for programming and used the NI FlexRIO Instrument Development Library to accelerate code development. The available high-level toolkits and flexible GUI made the FPGA programming faster. This also supported the implementation of complex FPGA functionalities that would otherwise have been extremely difficult and time consuming.
In addition, the NI sales personnel actively participated in the process by helping choose the best suited hardware and arranged opportunities for hardware proof-of-concept demonstrations before launching the actual development project. This ensured a smooth and low-cost transition from pilot to the development phase.
Considering the complexity of the implementation, the final system has a relatively small size so we can use it in mobile operations such as controlling the flow of radioactive materials through international borders or in-depth mapping of the radiation environment in various operational conditions.
Users can access the raw signals and extract case-specific figures from the data because of the flexibility of NI products. FlexRIO is intrinsically scalable, so we can add further data channels to the hardware backbone. However, the main advantage of using the NI products was the beneficial cost per channel ratio, which scales down with increasing channel count.
The success of the project relied entirely on finding the right hardware provider and a subcontractor. In this respect, choosing NI hardware was an excellent choice. Choosing NI satisfied hardware-level requirements and the project benefitted from access to the NI Alliance Partner network database when seeking the right partner.
Tel: +358 50 435 794