Operation and Improvements of a Cesium Beam Standard Having 4-Meter Interaction Length

Some factors affecting a long cesium beam standard´s accuracy are discussed and the results of two year´s operation are given. Increasing the interaction length reduces the linewidth and thus the errors in measuring the line center, as long as the signal-to-noise ratio is sufficient. Errors due to nonlinearity of FM used in servo locking a crystal oscillator to the resonance are below 1×10^-11; other servo errors are smaller. The described variable C-field measuring technique and demagnetizable Armco iron shields reduce errors due to C-field uncertainty to 1.2×10^-12. Cavity phase shift effects are below 1.5×10^-11. Standard deviation of single measurements with 4.6 sec averaging time is 2.7×10^-11 under good conditions. Considering all effects, including those of microwave power and of cavity tuning, the stated figure of accuracy is ±3×10^-11. Two year´s VLF comparisons show agreement to about ±10^-10 with respect to USFS.