Experimental demonstration of radiation flux measurement accuracy surpassing the Nyquist limit

In this paper we present an extensive and detailed experimental assessment of the performance of homodyne detection vs direct detection for determining the flux of an incoherent light source. The intensity of a laser reference signal is measured, before and after contamination by photon bunching noise (transimpedance gain of 20k), simultaneously. Moreover, the measurement containing photon bunching noise was done using two schemes: direct and homodyne detection, also with no appreciable time delay. The sampling rate of all measurements is the same, and is always higher than the frequency limit of the bunching noise. By carefully comparing the resulting three time series, it is found, surprisingly, that the homodyne detection of the incoherent signal resembles more closely the variance and variability pattern of the original coherent signal, and the effect is most prominent at the highest sampling frequency. It therefore appears flux estimates made using the shot noise of the field were able to overcome the Nyquist theorem, viz. by combining the homodyne technique with sampling at a rate faster than the bunching noise fluctuations it is possible to surpass the sensitivity limit of the radiometer equation. This is, to the best of our knowledge, the first time that such a detailed analysis of shot noise fluctuations has been investigated experimentally. Although a full theoretical understanding is still lacking, the experiments appear to show an unambiguous effect that requires further investigation.