Figure 3.

Reduction of intensity noise. (A) Steady-state normalized power of the trap laser with (green) and without (red) stabilization. Data was taken after 1 h of equilibration time. (B) A step increase in laser power from 150 to 300 mW with and without feedback. Such steps led to a slow (>10 s), substantial (>3%) systematic error. Intensity stabilization effectively minimized this error. Traces offset for clarity. Inset is a zoom-in of the transient noise where the data was normalized. (C) Records of power versus time during rapid (5 s) changes in power with and without stabilization. Without stabilization, the final measured power was only 55% of the intended value (750 mW). Only after ∼10 min at the desired power did the unstabilized trace equilibrate to within 1% of the requested power (data not shown). (D) The short-term (5 s) fractional standard deviation (σ) as a function of each power as in panel C. The intensity-stabilized trace shows approximately three orders-of-magnitude improvement over the unstabilized trace and is <0.1% (black dashed line) over the full range. Traces in panels A–C are at 2 kHz and color in panels B–D is the same as in panel A.