A user configurable data acquisition and signal processing system for high-rate, high channel count applications

Salim, Arwa and Crockett, Louise Helen and McLean, John and Milne, Peter (2012) A user configurable data acquisition and signal processing system for high-rate, high channel count applications. Fusion Engineering and Design. ISSN 0920-3796 (https://doi.org/10.1016/j.fusengdes.2012.03.047)

[thumbnail of A user configurable data acquisition and signal processing system for high-rate, high channel count applications]
Preview
PDF. Filename: ASalim_FUSION_2012_in_press.pdf
Accepted Author Manuscript
License: Unspecified

Download (787kB)| Preview

Abstract

Real-time signal processing in plasma fusion experiments is required for control and for data reduction as plasma pulse times grow longer. The development time and cost for these high-rate, multichannel signal processing systems can be significant. This paper proposes a new digital signal processing (DSP) platform for the data acquisition system that will allow users to easily customize real-time signal processing systems to meet their individual requirements. The D-TACQ reconfigurable user in-line DSP (DRUID) system carries out the signal processing tasks in hardware co-processors (CPs) implemented in an FPGA, with an embedded microprocessor (μP) for control. In the fully developed platform, users will be able to choose co-processors from a library and configure programmable parameters through the μP to meet their requirements. The DRUID system is implemented on a Spartan 6 FPGA, on the new rear transition module (RTM-T), a field upgrade to existing D-TACQ digitizers. As proof of concept, a multiply-accumulate (MAC) co-processor has been developed, which can be configured as a digital chopper-integrator for long pulse magnetic fusion devices. The DRUID platform allows users to set options for the integrator, such as the number of masking samples. Results from the digital integrator are presented for a data acquisition system with 96 channels simultaneously acquiring data at 500 kSamples/s per channel.