TL;DR
- ▸Atomic oscillators reference a hyperfine transition in the atom and are stable by physics rather than by manufacturing tolerance.
- ▸The result is several orders of magnitude better long-term stability than quartz, and dramatically tighter holdover during GNSS denial.
- ▸Telecom G.8275.1, defence-grade timing, regulated financial timestamping and metrology applications all routinely justify atomic oscillators. Most enterprise and broadcast deployments don't.
What atomic adds
Quartz crystal oscillators — even good OCXOs — drift at rates dominated by physical ageing of the quartz and by temperature sensitivity. The drift is bounded but real, and over multi-hour or multi-day holdover scenarios it accumulates faster than atomic alternatives. Atomic oscillators (caesium and rubidium) reference a hyperfine transition in the atom itself, which is stable by quantum physics rather than by manufacturing tolerance. The result is several orders of magnitude better long-term stability and dramatically tighter holdover during GNSS denial.
Rubidium is the more common atomic option in PTP grandmasters because it's cheaper, smaller, lower power and has a useful operational life of 8-12 years. Caesium primary frequency standards are more accurate and longer-lived but are large, expensive and overkill for almost any commercial deployment outside national metrology labs.
When it's worth it
When the deployment's worst-case holdover requirement exceeds what an OCXO or DOCXO can deliver. Telecom G.8275.1 grandmasters routinely use rubidium because the ITU-T accuracy classes effectively mandate atomic-grade frequency references. Defence-grade timing routinely uses rubidium because the threat model includes deliberate multi-hour GNSS denial. Regulated financial timestamping environments often use rubidium because a single multi-hour GNSS event during a trading day has direct regulatory and financial consequences.
Outside these specific scenarios, rubidium often becomes a vanity purchase. The cost premium is real (typically 5-10x an OCXO), the power and cooling costs are non-trivial, and the rubidium physics package itself has a finite life. If you don't actually need rubidium-grade holdover, you're paying for capability the deployment will never use. Specify against the documented holdover requirement, not against datasheet preference.
Frequently asked questions
What is the difference between rubidium and caesium oscillators?+
How much better is rubidium than OCXO for holdover?+
How long does a rubidium oscillator last?+
Related reading
Blog · Foundations
Quartz Crystal Oscillators: The Workhorses of Precision Timing
Quartz crystal oscillators — TCXO, OCXO, DOCXO — are the workhorse oscillators in almost every PTP grandmaster shipping today. How they work, where their limits are, and why they remain the right answer for most deployments.
Blog · Foundations
Atomic Clocks and GPS Timing
How atomic clocks define UTC, how GPS distributes that time globally, and what the relationship between the two means for precision timing infrastructure on the ground.