Data Acquisition and Analysis

Instrument Control:

We use Python, C++ and LabView code for instrument control in our laboratories. This allows us to automate tasks ranging from environment monitoring to high quality data acquisition and on and offline processing.

Examples of our working this area include the design and implementation from scratch of a visual capture system for the ATLAS Tracker Upgrade with 1μm resolution for reception tests of silicon microstrip detectors for production. This solution is essential for reception tests of the UK share of microstrip sensors for this project, and our researchers have designed this visual capture system from scratch to project driven specifications. Another example is the development of a suite of LabView tools to drive Keithley Source Measure Units and multi-channel National Instrument Compact RIO data acquisition equipment for testing sensors in our laboratories.

Databases:

We have expertise in website and data base design and have developed bespoke SQL-based data base for internal websites for archiving device characteristics and documentation. This allows us to monitor the details of every step of our organic semiconductor detector development programme. From the raw material source, through to end results, internal notes and published papers. This capability ensures that valuable data and reports remain available for members the project at all times.

Data Processing:

Testing sensors in the laboratory quickly generates vast quantities of raw data that needs to be processed in a versatile way. We have expertise in handing data volumes from Mb to Pb in the Particle Physics Research Centre. No data set is too large for us to consider processing. Examples of our capability in this area include:

• An offline data analysis toolkit to emulate the functionality of a multi-channel analyser (MCA). This allows us to explore advanced signal processing techniques and to apply software filtering (for example to remove mains frequency noise) and evaluate emerging technology in detail that would otherwise be lost with a hardware solution. The input to these tools is a time series of data from some conventional acquisition system (for example the output of an oscilloscope). These tools are designed to be run in parallel to allow for scalability from single to multi-channel system tests. Our MCA tools allow for peak height and rate analysis of individual as well as sets of files, are automated to provide batch reports on sets of data acquisition runs and have simulation elements incorporated to facilitate understanding the performance of emerging technology while working with low TRL solutions.
• Our group also has extensive experience in the design of fast triggering systems for particle physics experiments.

Analysis Methods:

While data is the most asset for any organisation, what you do with that data really matters. Building on our fundamental science programme we have capability for sophisticated analysis including fitting (using chisquare and maximum likelihood approaches) right through to machine learning. We have decades of experience in using modern machine learning algorithms including support vector machines, decision trees, and artificial neural networks. In addition to the more traditional approaches, we also use modern deep learning approaches. Our work in this area is incorporated into our teaching activities and we can provide bespoke training courses on request.