Noise Specifications | previous | next | feedback |
The strength of noise e.g. of the output of a laser often needs to be quantified – particularly in the context of applications like frequency metrology, sensitive spectroscopic measurements, or optical fiber communications, where the performance of devices or systems is limited by noise.
Examples for Noise Specifications of Lasers and Optical Amplifiers
Intensity noise is often measured by analyzing the laser output with a photodiode and related electronic equipment such as an electronic spectrum analyzer. It can be specified with a power spectral density of the relative intensity noise as a function of noise frequency. For some purposes, a root-mean-square (r.m.s.) value, essentially the square root of the integral of the power spectral density over some frequency range (e.g. 1 Hz to 1 MHz), is sufficient. However, r.m.s. values without specification of the measurement bandwidth are nonsensical. Even worse are noise specs like "±1%".
Optical phase noise can be quantified by the power spectral density (PSD) of the phase fluctuations. Alternatively, the PSD of the fluctuations of the instantaneous frequency can be specified. Such power spectral densities often diverge for zero frequency, so that integration down to zero frequency is not possible. For simple random-walk processes, the specification of a coherence time or coherence length or of a linewidth value can be appropriate. Note that linewidth values often depend on the measurement time.
Frequency noise is directly related to phase noise; it is the noise of the instantaneous frequency, the latter being related to the temporal derivative of the phase.
Timing jitter of a pulse train can be quantified as the power spectral density of the timing deviation (e.g. from some noiseless reference) or the timing phase. r.m.s. values are also common in cases with stationary noise, e.g. of actively mode-locked lasers, or for given frequency intervals.
In frequency metrology, the use of a van Allen deviation or van Allen variance as a function of time is common. These quantities can be calculated from a power spectral density, whereas the opposite calculation is ambiguous.
The noise figure of an electronic or optical amplifier quantifies the amplifier excess noise.
In frequency metrology, it is common to use normalized phase fluctuations x(t) = δφ(t) / (2π ν0), i.e., phase fluctuations divided by the mean angular frequency. The time derivative of the normalized phase fluctuations then delivers the normalized frequency fluctuations y(t), i.e., the fluctuations of the instantaneous frequency relative to the mean frequency. For a comparison of the phase noise of sources with different mean frequencies, it is appropriate to compare the power spectral densities Sx(f) of normalized phase fluctuations or Sy(f) of normalized frequency fluctuations, rather than of the phase or frequency excursions themselves, because these normalized fluctuations are what determines the precision and accuracy e.g. of a clock.
Common Problems
For various reasons, correct noise specifications are often not achieved:
- The mathematical description of noise is somewhat sophisticated, and often not properly treated in physics or engineering courses. As a result, nonsensical noise specifications are widespread in product data sheets and even in the scientific literature.
- In laser physics, there is a variety of different types of noise, which are conceptually and physically related in non-obvious ways (example: optical phase noise and timing phase noise of mode-locked lasers). Therefore, physical insight is as essential as the mathematical knowledge.
- Noise measurements are subject to many non-trivial technical issues. For example, detailed know-how on the inner workings of electronic spectrum analyzers is required for obtaining correct results from measurements with such devices. A black-box treatment (using the device readings without knowing how they are generated) easily leads to wrong results e.g. via inappropriate device settings or the failure to apply certain correction factors (e.g. for logarithmic averaging in the spectrum analyzer).
See also: laser noise, intensity noise, relative intensity noise, phase noise, frequency noise, power spectral density, linewidth, coherence, coherence time, narrow-linewidth lasers, Spotlight article 2006-10-09
Category: fluctuations and noise
