Running a holdover test

Why test?

The datasheet says "24-hour holdover OCXO" — but real performance depends on temperature, the age of the crystal and how long the unit had been within spec before the GNSS outage. Testing equals evidence.

Test procedure

1. Pre-condition: let the NTP server run on stable GNSS lock for at least 48 h, at the declared operating temperature
2. Baseline measurement: log offset against an external reference (a second GNSS server) for 1 hour
3. Disconnect: physically unplug the GNSS antenna
4. Continuous measurement: log offset every 5 minutes for 24-72 hours
5. Reconnect: reconnect the antenna, log recovery time
6. Analysis: plot offset over time, determine time-to-tolerance-limit

Tolerance limits per use case

• IT general: ±10 ms
• Broadcast: ±1 ms (sub-frame)
• Financial: ±100 μs (MiFID II)
• Power Profile: ±1 μs (IEC 61850)

In practice with Masterclock

• NTP100 (TCXO): holds ~1 day within ±1 ms — then drifts quickly
• NTP100-OSC (OCXO): holds 24-72 h within ±100 μs
• GMR5000/6000 + OCXO: 5-7 days within ±10 μs
• GMR6000 + Rubidium: months within ±1 μs

Reporting for audit

• CSV export of the offset log
• Time-vs-offset chart
• Statement: "Holdover test compliant with UTC ±X μs over Y hours, conform IEEE and RFC/IEC <spec>"

Delivery by Daylight

For compliance projects we deliver a holdover validation report as an optional deliverable. Discuss before ordering.

Contact: Daylight bv.