Search

The small signal-to-noise ratio (SNR) of conventional laser induced fluorescence (LIF) measurements using a continuous wave laser, either diode or dye, is typically overcome by amplitude modulating the laser at a specific frequency and then using lock-in amplification to extract the signal from measurement noise. Here, we present LIF measurements of the neutral helium velocity distribution function in an rf plasma using frequency modulated (FM) laser injection. A pulse train of 100% amplitude modulation is generated synthetically with a random sequence of pulse lengths. The FM signal then drives an acoustic optic modulator placed in the path of the injection beam in an LIF measurement. The signal from a fast photomultiplier tube is digitized and cross-correlated with the known modulation signal. The resultant FM-based LIF signal outperforms a conventional lock-in-based LIF measurement on the same plasma in terms of SNR and precision. [ABSTRACT FROM AUTHOR]

Laser induced fluorescence is used to measure argon ion heating during magnetic reconnection in the PHase Space MApping experiment (PHASMA). Sufficient signal-to-noise ratio (SNR) of the processed signal with pulsed laser injection is a delicate balance between saturation of the absorption line and injecting enough laser power to overcome the spontaneous emission of the plasma at the fluorescence wavelength. Averaging over many laser pulses and integrating over the fluorescence lifetime improves the SNR of the processed signal (processed SNR) when the SNR of the laser pulse time series is small (pulse SNR), but for laser powers small enough to avoid saturation, averaging over hundreds of pulses is needed to obtain an appreciable processed SNR over the entire Doppler-broadened absorption line. Here, we describe a matched filter processing method that significantly improves the SNR of the final measurement with fewer shots averaged. Investigation of simulated measurements validated by experimental results suggests that the matched filter method provides up to a 20% improvement in the processed SNR, resulting in less uncertainty in distribution function fits. [ABSTRACT FROM AUTHOR]