Since laser intensity noise often sets the performance limit of an experiment, it is important to characterize it to get good results. When making this kind of measurement, a low-noise, high-sensitivity photoreceiver is invaluable. The New Focus line of photoreceivers offers the ideal solution for this application. Their combination of shot-noise limited performance, high conversion gain, and wide bandwidth enables noise measurements to be made simply and easily.
The ﬁgure shows a typical setup for measuring laser intensity noise. The laser output is attenuated, so as to not saturate the photodetector, and focused onto the photoreceiver. It is important to be sure that the photodetector is not overﬁlled. This can lead to unwanted noise coupling as mechanical vibrations cause the photodetector to see varying amounts of laser light. These acoustic frequencies will give false peaks in the amplitude-noise spectrum.
The photodetector output is then sent to a spectrum analyzer. It is quite common for the spectrum analyzer to have an equivalent input noise that is greater than the shot noise of the photoreceiver. In this case, adding a low-noise preampliﬁer is helpful. Test that the spectrum analyzer is not limiting your measurement by turning off the photodetectors power. The noise level should drop by at least 10 dB. To test that the measured noise is laser noise (and not photodetector noise) block the light from the photodetector. The noise level should again drop by at least 10 dB.
Measuring intensity noise is easily performed with a spectrum analyzer. The best way to display laser intensity noise is as a linear spectral density. In this measurement, the amount of noise produced in a given measurement bandwidth is measured and normalized to the square root of that bandwidth. Some spectrum analyzers automatically convert to spectral density units (V/). Unfortunately many high-frequency models do not. If the spectrum analyzer does not directly measure a spectral density, you will have to divide the spectrum by the square root of the noise-equivalent bandwidth of the measurement. The noise-equivalent bandwidth as a function of the resolution bandwidth is often speciﬁed in the spectrum-analyzers users manual.
The measured noise spectral density, SV(), can now be converted to either power ﬂuctuations (SI() in W/), or the commonly seen relative intensity noise (RIN() in 1/). Power ﬂuctuations are related to the measured spectral density of voltage noise measured with the spectrum analyzer by
where G is the conversion gain of the photoreceiver at the wavelength of the laser (in V/W) and AV is the preampliﬁer gain (in V/V). To achieve the greatest accuracy, these parameters should be independently measured.
RIN is related to the spectral density of voltage-noise by
where VDC is the DC voltage at the spectrum analyzer input. While RIN is a useful measure of laser technical noise, it is a deceptive measure of quantum-limited noise (such as shot noise). The reason for this is that laser-power ﬂuctuations, like relaxation oscillations in solid-state lasers, that are inherent in the lasers output will be accurately measured after optical attenuation. In contrast, the shot-noise-limited RIN (equal to where q is the charge on the electron, and IDC is the DC photocurrent) is dependent on the (arbitrary) photocurrent at which the measurement was made. Therefore, stating that an optical signal is shot-noise limited at a particular frequency is meaningless unless the detection current is speciﬁed.