Logo of the Physikalisch-Technische Bundesanstalt
Panoramic view of the clock hall at PTB with the four caesium clocks CS1, CS2, CSF1 and CSF2.

Unit of Time

Working Group 4.41

The primary clocks CS1 and CS2 of PTB

The photograph shows PTB's atomic clock hall with the primary clocks CS1 and CS2. The clocks CS1 and CS2 were put into operation in 1969, 1985, and 1988, respectively. CS1 served as a model for later developments. Its design principle was new and differed in many details from that of the primary clocks of the national institutes of other countries and of commercial Cs clocks. At PTB, it was realised that an axially symmetric atomic beam and an axial C-field as well as the selection of slow atoms in a narrow velocity interval by means of magnetic lenses was particularly advantageous. Combined with other design details, these features allow the different systematic frequency-shifting effects to be determined with small uncertainty.

The estimate of the amount of the different uncertainty contributions was described in a number of publications (see literature).

To give an impression, the table below shows the frequency shifts determined for CS1 together with their uncertainties (State: August 2000).

Cause of frequency shift

Frequency shift



as relative quantities

in multiples of 10-15

Magnetic field along the atomic beam (quadratic Zeeman effect)





Electric field of the thermal background radiation in the clock (AC Stark effect)





Motion of the atoms in the atomic beam (relativistic time dilatation)





Transitions between different magnetic sub-states in the atomic beam due to static, inhomogeneous fields (Majorana transitions)







Transitions between different magnetic sub-states in the atomic beam due to incorrectly polarized microwave field







Population differences between the magnetic sub-states in the atomic beam





Electric losses and mechanical asymmetry of the microwave resonator





Frequency detuning of the microwave resonator



Wrong determination of the line centre of the atomic resonance due to electronic servo errors





The total uncertainty of CS1 is obtained as the root-mean-square of the numbers stated in the last column, assuming that the individual uncertainty contributions are uncorrelated. The mean duration of the second realized by CS1 should therefore deviate by not more than plus or minus 7.10-15 from the duration of the SI second with a probability of 67 %. Accordingly, CS1 should agree with an ideal clock to within ±0,22 µs in the course of one year. An uncertainty of 12.10-15 was determined for CS2, and an uncertainty of 14.10-15 for CS3

For many years, the PTB clocks were the world's most accurate clocks. Moreover, it was possible to operate them continuously over many years and to derive the time scale UTC(PTB) realized at PTB directly from the primary clocks. Between 1978 and 1995 when a basic renovation became inevitable, CS1 was operated practically without interruption. Between 1992 and 2010, the time scale of PTB was derived from CS2, since the one of the fountain clocks practically serves as the reference.