Digitizer can capture extremely long and weak signals
Spectrum Instrumentation’s new package uses SCAPP (Spectrum’s CUDA Access for Parallel Processing) together with the company’s latest digitizers to harness the parallel structure of a CUDA graphics card for data processing. The data is sent directly to a GPU using Remote Direct Memory Access (RDMA) transmission, where fast time and frequency domain signal averaging is possible and limitations associated with other approaches are avoided. The new averaging package is interesting for all users who work with weak signals. These include applications such as mass spectrometry, LIDAR, radio astronomy, automation, radar, biomedicine, sonar and many more. The package is available now.
Until now, users who wanted to perform signal averaging had three basic options: First, they could buy a digital oscilloscope that included signal averaging. Another possibility was the purchase of a digitizer card, where the digitized data is sent to a PC and the host processor performs the calculations. As a third option, a special digitizer could be purchased, which takes over the signal averaging itself with FPGA technology (Field Programmable Gate Array). However, all these three techniques have severe limitations when averaging for long signals has to be performed.
Digital oscilloscopes normally use 8-bit analog-to-digital converters (ADCs), which results in an unfavorable signal-to-noise ratio (SNR). In addition, averaging capability is typically limited by small processors and limited onboard memory. Fast digitizers can generally offer much higher A/D resolutions and more memory. However, the acquisition speed of the digitizer card is usually limited by the PC’s data transfer rate, while the host processor also has many other tasks to perform in addition to averaging. Digitizers that use FPGA technology, on the other hand, are usually expensive and offer limited recording lengths.
In contrast, Spectrum Instrumentation’s new approach can perform averaging even on extremely long signals while offering users unmatched flexibility. The package works with the fast PCIe M4i series digitizer cards and the mid-range M2p series digitizer cards. The M4i series offers digitizers that can sample signals up to 5 GS/s at 8 bit resolution, 500 MS/s at 14 bit resolution or 250 MS/s at 16 bit resolution. The M2p cards offer sampling rates from 20 MS/s to 125 MS/s, all with 16 bit resolution and up to 8 channels per card. This allows users to select a level of performance that best suits their specific signal acquisition needs. Since data is transferred directly to the GPU card using RDMA transmission without host processor intervention, averaging can be performed for signals of almost any length.
The averaging offer is part of the SCAPP driver package and includes the extension for RDMA transmission to enable direct data transfer from the digitizer to the GPU. The package also includes a number of examples of interaction with the digitizer and examples of CUDA parallel processing with basic averaging functions. The examples include block averaging along with noise reduction methods and continuous averaging of signals in the frequency domain. The use of the provided, tested and optimized examples leads to immediate results. The entire software is based on C / C ++ and can be easily extended with normal programming knowledge. Own specific algorithms for averaging can be easily integrated. The SCAPP package enables RDMA transfer directly to the GPU for PCs with LINUX operating systems or transfer via the CPU for PCs with Windows-based operating systems.
More information: https://spectrum-instrumentation.com/