Pendulum calibration laboratorytime and frequency traceabilityfrequency referenceFTR-210R

Time & Frequency Reference and Traceability for Calibration Labs

From a GNSS-disciplined rubidium reference and stand-alone high-stability sources to distributing 10 MHz to every bench: building a traceability chain you can explain.

In short

In a calibration lab, the key to time and frequency measurement is not instrument resolution but traceability: every frequency reading must state what it was compared against. The Pendulum FTR-210R disciplines a rubidium oscillator to GNSS for time accuracy better than 10 ns rms to UTC, with traceable calibration built in. Shielded rooms and basement labs with no GNSS reception are covered by the stand-alone 6688/6689 references with calibration uncertainty of 5×10⁻⁹ or 5×10⁻¹¹. The FDA-301 and DA-36 distribute the reference over fiber or coax to benches up to 2 km away, and the CNT-91/91R takes on the comparison measurements with 35 ps single-shot resolution. This article places each of the four in the traceability chain.

The questions a traceability chain has to answer

For every frequency reading on a calibration certificate, the assessor wants to know what it was compared against, where the uncertainty comes from, and when the last calibration was. Inside a lab the chain usually looks like this: a primary reference traceable to UTC, one or more backup references that hold when the primary fails, a distribution system that delivers 10 MHz to every bench without degrading it, and a counter for the comparison measurements.

Long-term accuracy and short-term stability actually pull in opposite directions. A GNSS receiver alone is accurate over the long term but noisy over the short term; a rubidium clock alone is stable over the short term but drifts slowly at its aging rate. The lab's reference architecture is an arrangement between the two.

Primary reference: the GNSS-disciplined FTR-210R

GNSS satellite in orbit — GNSS satellite timing

The Pendulum FTR-210R locks a rubidium oscillator to GNSS. While locked, GNSS provides long-term accuracy and the rubidium oscillator provides short-term stability and holdover once lock is lost. Time accuracy of the 1 pps output to UTC is better than 10 ns rms, typical holdover drift is about 1 µs per 24 hours, and the aging rate in manual holdover is below 5×10⁻¹¹ per month. A temporarily blocked antenna does not invalidate the measurements.

Standard outputs are 5 × 10 MHz, 1 × 5 MHz and 1 × 1 pps, expandable to 11; options add 2 × 100 MHz outputs and IRIG-B DCLS programmable pulses. Traceable calibration is built in, which is exactly what a calibration lab needs: the traceability documentation of the reference itself is on file. The item to confirm before installation is antenna placement, because reception conditions decide whether it reaches its rated figures.

When there is no sky to see: the 6688 and 6689 stand-alone references

Shielded rooms, basement labs, ships and vehicle racks receive no GNSS, so they need a stand-alone reference that depends on no external signal. The 6688 core is either an OCXO or a rubidium oscillator, with 5 or 10 outputs of 10 MHz plus one 5 MHz, so one unit drives up to eleven instruments in sync. The 6688 has a calibration uncertainty of 5×10⁻⁹ at +23±3 °C and a monthly aging rate of 3×10⁻⁹; the 6689 reaches a calibration uncertainty of 5×10⁻¹¹.

A stand-alone reference starts drifting at its aging rate the moment it leaves the calibration bench, so its accuracy is a function of time since last calibration. The central selection question is therefore how often you can afford to calibrate. The aging rate sets how quickly the unit leaves tolerance, and with it how dense a recalibration schedule the lab has to run.

Getting the reference to every bench: FDA-301 and DA-36

There is one reference source, but the instruments that need it are spread across rooms and sometimes buildings. Coax becomes unacceptable in attenuation and noise after a few hundred meters, and links between buildings introduce ground loops. The FDA-301 solves this stretch over fiber: a single reference is distributed to remote sites up to 2 km away, with up to 18 fiber outputs or 12 coax outputs, sine distribution from 100 kHz to 60 MHz, pulse distribution for 1 pps and IRIG-B DCLS, and dual master/slave inputs with automatic switchover.

Small point-to-point needs are served by the DA-36: one fiber and one coax input, one fiber and four BNC outputs, frequency range 200 kHz to 16 MHz. It is fully analog with no PLL, so it introduces none of a phase-locked loop's own phase noise or lock time. For phase-noise-sensitive comparison measurements, that trade-off matters.

Comparison measurement: CNT-91/91R

Calibrating an oscillator under test requires a counter to compare it against the reference. The CNT-91 has 35 ps single-shot time resolution, a measurement rate of 250,000 readings per second and internal memory for up to 3.5 million readings, so Allan deviation and time interval statistics can be completed with adequate sample counts. The standard measurement range is 0.001 Hz to 400 MHz, with an optional third channel to 3, 8, 15 or 20 GHz. The 91R has a built-in rubidium timebase, suited as a stand-alone reference for mobile calibration.

Evencat is the distributor for Pendulum in Taiwan and can explain the calibration certificate content and recalibration arrangements for each instrument. Lab expansions and short-term comparison projects can also be validated first through time and frequency standard rental.