The CSAC design is unique in that the physics is vacuum-packaged to eliminate convection/conduction effects. This enables CSAC to resist harsh thermal environments.

Timing error due to those effects cannot be calculated by simply looking at a datasheet temperature specification (commonly referred to as temperature coefficient). Ramp rate, dwell time, airflow, number of cycles, and so on will all effect the outcome. The oscillator design and sample data must be considered to choose the best performer.

The CSAC design, shown in the following illustration, is unique in that the physics (laser, resonance cell, and photodetector) is vacuum-packaged to eliminate convection/conduction effects, with an overall thermal resistance of 7000 °C/W. Its tensioned, polyimide suspension further isolates the physics from outside vibration or thermal-induced mechanical stresses.

Thanks to the design of the CSAC, the temperature ramp-rate of its environment has little impact on its frequency response. The following two graphs show an example of an aggressive temperature profile. The oscillators are soaked at a hot temperature (50 °C) for over 1 hour before being rapidly cooled to –10 °C in about 12 minutes (5 °C/min). They are soaked at the cold temperature for 2 hours before they are rapidly returned to the hot temperature.

Both CSACs, as shown by the purple and black traces, varied <±3 x 10–10 Hz/Hz. For comparison, an OCXO with similar specified temperature coefficient was subjected to the same test. Its frequency response was 3x–4x worse. Note: The CSAC temperature coefficient specification is ±5 x 10–10 from –10 °C to 70 °C. OCXO is specified as ±4 x 10–10 from 0 °C to 70 °C. Measured factory temperature coefficient for CSAC_1 and CSAC_2 was 4.25 x 10–10 and 2.8 x 10–10, respectively.

The following table lists the corresponding time error. At ~1 μs of time error, the CSACs are clearly superior when compared to a commercial OCXO, which accumulated 6.5 μs. A CSAC’s thermal response approaches the performance of larger, more power-hungry Rubidium-based atomic oscillators.

Looking back at equation 1 (see part 2 of this series), we have demonstrated performance assuming zero initial phase and frequency offset. This time, however, the environmental effects are non-zero.

In the next post of this series, I’ll write about the “effects on timing error: initial frequency offset.”

To read the complete white paper, go here now.

Please connect with me on LinkedIn and also follow Microsemi on LinkedIn.

Tags: Chip Scale Atomic Clock, OCXO, Oscillators, Synchronization, Temperature Change, Timing