Contents I. Introduction
Polarization controllers are the critical adaptive element in polarization mode dispersion (PMD) compensators for high bitrate systems and have numerous other applications including polarization multiplexed systems and coherent systems. In the latter case, an incoming random state of polarization (SOP) must be aligned with the polarization of a local oscillator. Fundamentally, the controller must achieve the desired polarization transformation, whether it is between a fixed SOP and an arbitrary SOP or from an arbitrary SOP to an arbitrary SOP. This transformation must be accomplished in a continuous, or endless, fashion so that a smooth change in the input SOP can be accommodated without causing abrupt changes in the output. In other words, the polarization transformation operation must be transparent to a data-loaded signal. Polarization controllers operating in this transparent manner are referred to as reset free. In addition to transparent operation, the physical control signals such as the voltage applied to the device to affect the polarization transformation must be bounded. If one of the control signals should reach a bound, any transformation requiring the control signal to exceed its bound could not be achieved. Thus, it is helpful to think of two levels of control: 1) the physical device control to produce some intermediate polarization transformation such as that afforded by a waveplate and 2) the overall polarization control using multiple intermediate transformations, typically requiring a cascade of waveplates. A continuous, reset-free polarization controller has previously been described using the electrooptic effect in lithium niobate [1], [2]. The electrooptic effect was used to vary both the TE–TM coupling and the relative TE–TM phase shift [3]. A control circuit with bounded drive signals and employing dithering and a gradient search algorithm was used to demonstrate polarization tracking. In this paper, we address the physical control aspects using a device implementation that requires only phase shifters, avoiding the need for active TE–TM couplers, and show how to achieve reset-free physical tuning of our device with respect to a continuously rotating quarter-wave plate (QWP) and half-wave plate (HWP) realization. We do not address the second aspect of reset-free overall polarization control involving multiple waveplates, since a more extensive treatment is needed to define reset free in this context in a mathematically rigorous manner and is the subject of a forthcoming paper. For this paper, we assume that the polarization control is performed using the well-known waveplate approach cited by Heismann. As Heismann described reset-free physical control for lithium niobate realizations, we show that reset-free physical control can be obtained in the important case of isotropic planar waveguide realizations.
