7 SETI Program at the Medicina INAF Radioastronomy …
59
• configuration of the station facilities in use;
• selection of the output data formats;
• monitoring and diagnostics of the system status;
• offline visualization of data output files.
Both the processing and the control application were developed to be easily adapted
to other or future hardware components as well to be expanded with new functionality
for other researches or applications.
7.3 The New Spectrometer Line
In the starting phase of the new spectrometer line, the NewSpec hardware was not
available. Even if the system was out of order as a spectrometer, it was working as a
high-end PC, and its computing power was used in other researches with very tight
deadlines.
So it was decided to start building a new system with available spare or used
components. This allowed the development of the pilot projects to carry out the first
software tests to choose the components to use for the final system.
7.3.1 First Pilot Project
The first pilot project was carried out in 2015 using old hardware already present in
the laboratory (see Fig. 7.6). The supporting system was a used PC equipped with an
Intel i5 650 dual-core CPU, 4 GB RAM, and an Nvidia Quadro 5000 GPU [28]. The
data acquisition was performed by a Picoscope 3204 digital oscilloscope connected
by a High-Speed USB interface working up to 50 MS/s at a 12-bit resolution [34].
Tests carried out on the Picoscope showed an excellent RFI rejection, but its
internal memory buffer of 256 kS limited the spectrometer performances at a
maximum of 128 k channels (that we rounded up to 125 k). The memory buffer
dimension allowed only a minimum frequency resolution of 200 Hz at the maximum
signal bandwidth of 25 MHz with a spectra rate of 24 spectra per second. Moreover,
this instrument cannot manage a continuous data flux because of the delay between
successive acquisition introduced by its software drivers. Finally, further tests showed
that the CPU efficiently processed a so reduced input data rate without delaying the
other software operations. So we removed the GPU and converted the software to
use the FFTW library returning to the NewSpec project system architecture (see
Fig. 7.3).
This spectrometer was initially used as a test instrument dedicated both to software
development, and Nvidia and AMD GPU tests. At the moment, it always connected to
a cylindric parabolic antenna of the N/S arm of the Northern Cross Radio Telescope,
and it is mainly used as a test system for software development and debug.
59
• configuration of the station facilities in use;
• selection of the output data formats;
• monitoring and diagnostics of the system status;
• offline visualization of data output files.
Both the processing and the control application were developed to be easily adapted
to other or future hardware components as well to be expanded with new functionality
for other researches or applications.
7.3 The New Spectrometer Line
In the starting phase of the new spectrometer line, the NewSpec hardware was not
available. Even if the system was out of order as a spectrometer, it was working as a
high-end PC, and its computing power was used in other researches with very tight
deadlines.
So it was decided to start building a new system with available spare or used
components. This allowed the development of the pilot projects to carry out the first
software tests to choose the components to use for the final system.
7.3.1 First Pilot Project
The first pilot project was carried out in 2015 using old hardware already present in
the laboratory (see Fig. 7.6). The supporting system was a used PC equipped with an
Intel i5 650 dual-core CPU, 4 GB RAM, and an Nvidia Quadro 5000 GPU [28]. The
data acquisition was performed by a Picoscope 3204 digital oscilloscope connected
by a High-Speed USB interface working up to 50 MS/s at a 12-bit resolution [34].
Tests carried out on the Picoscope showed an excellent RFI rejection, but its
internal memory buffer of 256 kS limited the spectrometer performances at a
maximum of 128 k channels (that we rounded up to 125 k). The memory buffer
dimension allowed only a minimum frequency resolution of 200 Hz at the maximum
signal bandwidth of 25 MHz with a spectra rate of 24 spectra per second. Moreover,
this instrument cannot manage a continuous data flux because of the delay between
successive acquisition introduced by its software drivers. Finally, further tests showed
that the CPU efficiently processed a so reduced input data rate without delaying the
other software operations. So we removed the GPU and converted the software to
use the FFTW library returning to the NewSpec project system architecture (see
Fig. 7.3).
This spectrometer was initially used as a test instrument dedicated both to software
development, and Nvidia and AMD GPU tests. At the moment, it always connected to
a cylindric parabolic antenna of the N/S arm of the Northern Cross Radio Telescope,
and it is mainly used as a test system for software development and debug.