E.G. Adelberger, J.H. Gundlach, B.R. Heckel, K. Mauldin, S.D. Penn and H.E. Swanson

We are currently building a new torsion balance designed to provide improved Equivalence Principle tests at length scales ranging from 1m to . The design addresses the major statistical and systematic errors that limited our previous work¹ at the level a 10 ^-12cm/s².

Statistical noise will be reduced by operating the torsion balance vacuum at pressures below 10^-7 Torr. We are fitting the apparatus with a corotating 30 l/s ion pump, for which a rotating high voltage feedthrough has been built and is currently being tested in the old Eöt-wash apparatus. Torsion noise originating in the torsion fiber itself and in its attachments has been identified as a dominant noise source. Since it is known that low temperatures improve material properties we plan to surround all susceptible parts with a LN₂-cooled jacket. (We expect that systematic uncertainties will limit the achievable precision so it is not necessary to operate at liquid helium temperatures.)

A high-performance turntable consisting of an air-bearing, with a direct-drive eddy-current motor in a feedback loop with a high-quality shaft encoder will provide smoother rotation than our old turntable. The new turntable was developed together with, and built by, the Professional Instruments Company.² It is nearing delivery and promises to perform to specifications. A special test torsion pendulum designed to be extremely insenstive to gravity gradients will be used to verify the smoothness of rotation. For this test the old Eöt-wash balance will be suspended from the new turntable.

The largest systematic correction in our previous rotating torsion balance experiments resulted from a tilt of the turntable (due to the varying tilt of the laboratory floor) that induced a proportional change in equilibrium angle of the pendulum. This will be eliminated in the new apparatus by suspending the pendulum and autocollimator angle readout from a gimbal located inside the vacuum can.

The effects of residual gravity gradients were a significant systematic error in our previous measurements. We have developed a new pendulum containing 8 test bodies. The test bodies are mounted radially in two planes onto a tubelike holder. With this design most of the low-order multipole moments, q_lm, vanish. The lowest-order gravitational torque arises from the q₄₄ moment of the pendulum (used to make a gravitational calibration); the lowest order effect at the signal frequency nominally occurs in l=7 multipole order. To reduce problems from non-repeatable test body placement, the test bodies are mounted with screws into conical receptors.

Gravitational gradients at the pendulum will be compensated with 1.5 tons of lead. These masses will be shaped to reduce the Q₂₁, Q₃₁ and Q₂₂ gradients separately. To measure and then minimize stray pendulum moments arising from mechanical imperfections, these compensators can be rotated by 180° to double the gradients instead of cancelling them. A stand for precisely positioning these massive compensators is completed.

We have constructed a hermetic thermal shield consisting of a double-walled cylinder with temperature-regulated water flowing between the layers and an insulating foam layer on the outside. A layer of µ-metal will be installed just inside the shields.

The electrical signals from the balance are led through a slip-ring assembly to the non-rotating side. This assembly is rotated by a feedback mechanism that follows the turntable and hence does not exert any torque on the main turntable.