The Space Interferometry Mission's (SIM) shared-baseline astrometric interferometer System Test Bed 3 (STB3) has been constructed at JPL. STB3's objective is to use two of its interferometers (guides) for low frequency (0 to 1 Hz) fringe stabilization in the third one (science). This approach - being proposed for the first time in the context of space based observatories - is needed given the dim nature of science stars to be observed by SIM. Fringe stability is mostly affected by the low frequency attitude motion of the test bed's instrument table, with the inevitable exception of instrument vibration, thermal drift, and atmospheric fluctuations. Relative changes in table attitude cause optical path changes in the guide interferometers, which are tracked, linearly combined and fed forward to the science interferometer's active delay line to stabilize its optical path. This technique for tracking fringes in the science interferometer is possible because the position of the guide stars relative to the science star is well known. This open loop fringe tracking technique is dubbed Path-length Feed Forward, or PFF. In STB3, current fringe stability in the science interferometer using the PFF technique is at 50 to 60 nanometers RMS (from 0 to 500 Hz). Compare this to 15 to 20 nm RMS fringe stability in the guide interferometers, which operate in closed loop mode. Vibration, thermal drift and atmospherics in the science and guide interferometers are largely eliminated with the use of an internal metrology system. By design, mechanical vibrations are above the bandwidth of the interferometer system, and are passively rejected. Nevertheless, the internal metrology system can easily reject current low-level vibrations in STB3 down to the 6-nanometer RMS level.
Fringe tracking error in the science interferometer due to atmospherics is currently about 40 nanometers RMS at frequencies below 1.0 Hz. In SIM, the error in this low frequency band must be no more than 6 nm RMS. This error arises because the optical path stabilized by the internal metrology system is not equal to taht of the starlight, so not all atmospheric fluctuations in the starlight path can be stabilized. Therefore, there is a need to reduce the strength of atmospheric fluctuations or to filter them from the PFF command. In STB3 the strength of atmospheric fluctuations is already reduced with the use of optical path enclosures, which brought these fluctuations down from ~170nm RMS to their current levels of ~66nm RMS with a spread of 20nm. Simulations show that signal to noise ratios are generally not sufficient to filter atmospheric errors on-line.