Data acquisition and display for the PVLAS experiment

Author(s):
E. Milotti - UNIVERSITÀ DEGLI STUDI DI TREITES I.N.F.N.

Industry:
University/Education

Products:
Data Acquisition, LabVIEW, Dynamic Signal Acquisition

The Challenge:
Developing a data acquisition system that replaces a traditional lock-in amplifier – filter – signal analyzer chain and that permits further data processing to eliminate spurious artifacts due to uneven sampling in the experiment. Simple and effective visualization tools must also be provided to check data integrity.

The Solution:
Using LabVIEW both to acquire and visualize data online and to display and check them offline.

The PVLAS ellipsometer: a test of Quantum Electrodynamics and a search for dark matter.
The PVLAS experiment has been built to perform a very accurate check of Quantum Electrodynamics, and is currently also used for a dark matter search (axion-like particles). A powerful superconducting dipole magnet polarizes the vacuum inside a Fabry-Perot cavity, and interactions in the cavity are detected as subtle changes in the polarization state of a laser beam that traverses the cavity (for a more complete description of the experiment, see E. Zavattini et al. (PVLAS Collaboration): “Experimental Observation of Optical Rotation Generated in Vacuum by a Magnetic Field” Phys. Rev. Lett. 96 (2006) 110406, and references therein; see also the website of the experiment http://www.ts.infn.it/experiments/pvlas/). The physical signal is picked up by a low-noise photodiode, after an upstream modulation has shifted it to higher frequency. Part of the polarization modulation comes from a birefringence modulator driven by a precise function generator (HP3324A) at 506 Hz or 1733 Hz, while another contribution comes from the mechanical rotation of the superconducting magnet, at about 0.3 Hz, which modulates the actual physical signal. The mechanical modulation is quite critical, because it is affected both by the slight periodic unevenness associated to the sampling of the electrooptical trigger that detects the magnet rotation and by the intrinsic drifts of the electromechanical driving system. For this reason we need high-speed data logging (8200 Hz) of the photodiode signal, which is processed offline to remove the artifacts from uneven sampling, and simple tools to visualize the signal from the photodiode and from additional control sensors (e.g., Hall probes that monitor the fringe fields).

The data acquisition chain
The traditional data acquisition chain in an experiment such as this includes a preamplifier, a lock-in amplifier (with the clock provided by the function generator), a low-pass filter and low-speed data logging. And indeed we do have such a chain in the experiment, where the sampling rate of about 10 Hz is easily handled by a NI PCI-6032E board. However this setup is affected by the random drift of the mechanical modulation, and moreover the original diode signal is lost, and as a result it is impossibile to diagnose data acquisition problems in the offline analysis.
For these reasons we have set up another data acquisition chain which uses a NI PCI-4472 board. This board ideally fits our needs, because it has a very high dynamic range and introduces only a negligible quantization noise, the channels are simultaneously sampled and thus there are no cross-talk concerns, and finally the sigma-delta ADC’s provide alias-free sampling and therefore we can do without additional low-pass antialiasing filters.
The samples are written on disk using a simple LabVIEW data-logging VI, and later on they are analyzed offline by a program written in C, which fully replaces the lock-in demodulation and compensates for uneven sampling (for further details see E. Milotti: “Sine-fit procedure for unevenly sampled, multiply clocked signals”, J. Comp. Phys. 202 (2005) 134)

The suite of visualization tools
We have developed several LabVIEW programs to display the signals and the results of the data analysis, and the most important ones are called with the self-explanatory names DisplayFile and ReadResults. Figure 1 shows the main window of DisplayFile: since we must condense many views in a user-friendly form, we have used several nested tab structures, that allow easy and fast switching between different displays, and provide a very intuitive user interface. The program acts both as a sort of multiple-screen offline oscilloscope, where the user can slow down the data rate to improve visibility, and as a spectrum analyzer that carries out a rough – but very useful – frequency-domain analysis (i.e., without correction for uneven sampling). Since a very high frequency resolution is needed (of the order of 1 mHz over the whole range from 0 Hz to 4100 Hz), the records used for the FFT calculation are very long: they can be as long as 221 samples = 2097152 samples, and this is handled seamlessly by LabView. The actual data acquisition runs can be considerably longer and in this case we use vector averaging to reduce the noise background in the FFT spectra. All the graphs – both in the time domain and in the frequency domain –are enabled to use the highly useful LabVIEW selection tools that allow zooming, and in this way may highlight problems and glitches in the data acquisition process.
DisplayFile also shows some additional useful plots and we plan to slowly improve the program so that it may possibly replace the analysis step implemented in C.
The other tool, ReadResults, is used to display the results of the analysis step and to condense in a few numbers the output of the data taking. The display includes high-resolution FFT spectra which also show the results of the more sophisticated analysis step. This program also uses nested tabs and zooming tools, and figures 2 and 3 show a couple of graphs displayed by the program.

Conclusions
We have developed a data acquisition system that uses National Instruments hardware and software. The close integration of hardware and software and the excellent programming environment have greatly reduced development time and have led to a very flexible set of data analysis tools which are still being improved and adapted to the changing experimental needs.

Browse All Case Studies »