Time and frequency calibration
Time and frequency are the most accurately measured quantities, by many orders of magnitude. Calibration compares an oscillator or timer with a better reference, and the main limits are warm-up, gate time and cables.
Instruments and typical accuracy
| Instrument | Typical range | Typical accuracy | Where it is used |
|---|---|---|---|
| Quartz crystal oscillator (XO) | kHz to hundreds of MHz | ±10 to ±50 ppm | Clocks, microcontrollers, stopwatches |
| Temperature-compensated oscillator (TCXO) | 1 to 100 MHz | ±0.1 to ±2 ppm | Radios, instruments |
| Oven-controlled oscillator (OCXO) | 5 to 100 MHz | About 10⁻⁸ to 10⁻⁹ after warm-up and aging | Frequency references, test equipment |
| Rubidium standard | 10 MHz | About 10⁻¹⁰ | Lab reference, telecom |
| GPS-disciplined oscillator | 10 MHz, 1 PPS | About 10⁻¹¹ to 10⁻¹² long term | Calibration labs, networks |
| Cesium beam standard | 10 MHz | About 10⁻¹² to 10⁻¹³ | Primary reference, timing centres |
| Hydrogen maser | 10 MHz | Best short-term stability, about 10⁻¹⁵ | Timing laboratories |
| Frequency counter, stopwatch, timer | To GHz | Limited by timebase, resolution and trigger | Instruments in everyday use |
Manual stopwatch timing is limited by reaction time (0.1 to 0.2 s), not by the quartz clock. The accuracy figures are typical ranges to help you compare instruments. Always use the specification of the instrument in front of you.
What a frequency offset means in time
| Offset | Per day | Per year |
|---|---|---|
| 10 ppm | 0.864 s | 315 s |
| 1 ppm | 86.4 ms | 31.5 s |
| 0.1 ppm | 8.64 ms | 3.15 s |
| 10⁻⁸ (0.01 ppm) | 0.864 ms | 0.315 s |
| 10⁻¹⁰ | 8.64 µs | 3.15 ms |
| 10⁻¹² | 86.4 ns | 31.5 µs |
The time error accumulates if the offset stays the same. A clock 1 ppm fast gains 86.4 ms every day.
Reference standards and traceability
- The second. Defined by the cesium-133 hyperfine frequency, 9 192 631 770 Hz, realised by atomic clocks in national institutes.
- UTC. Coordinated Universal Time is distributed through GPS, network time and time signals, with traceability to the national laboratory.
- Working references. Rubidium, OCXO and GPS-disciplined oscillators are the usual lab references, checked against UTC.
- Quality system. ISO/IEC 17025, scope listing frequency and time interval with the uncertainty.
How it is calibrated
Typical procedure
- Warm up oscillators for the time in the manual (hours for OCXO, sometimes days after power loss).
- Connect the reference to the counter external reference input, with 50 Ω cables and terminations.
- Set the gate time for the resolution you need, for example 10 s for 10 MHz to reach about 10⁻⁸.
- Measure the device at its output, with the trigger level in the middle of the signal.
- Calculate the offset in Hz and ppm, and compare with the specification.
- For clocks and timers, compare a long interval against a reference, and use the longer interval for better resolution.
Where it goes wrong
- Reference not locked. A GPS receiver that lost its signal drifts silently.
- Too short a gate. The ±1 count error dominates and the result is noisy.
- Cables and terminations. Reflections cause trigger errors.
- Warm-up and aging. An oscillator that moved recently has not settled.
- Temperature. Even a good crystal moves with the room.
What goes in the uncertainty budget
- Reference uncertainty and its stability.
- Counter resolution (±1 count) and the trigger error.
- Gate time and averaging.
- Cables, delays and phase noise.
- Temperature, aging and the warm-up of the unit under test.
Combine the contributions as in the uncertainty budget calculator, then decide pass or fail with a guard band as explained on the calculators page.
Calculators
Everything runs in your browser. Nothing you type is sent anywhere.