In your FE-5650A rubidium frequency standard, being “locked” means the electronics have successfully tuned the oscillator so that its frequency is being controlled by the rubidium atoms, rather than merely running as an ordinary quartz oscillator.
A simplified picture is:
Power ON → warm up → find rubidium resonance → LOCK → accurate 10 MHz output
Inside the unit, roughly this happens:
- A quartz crystal oscillator generates the basic frequency. Quartz is stable, but it slowly drifts.
- The unit heats its rubidium vapour cell and lamp to their operating conditions.
- The electronics interrogate the rubidium atoms and look for their extremely precise atomic resonance.
- A feedback system—essentially a PLL/servo control loop—adjusts the quartz oscillator until it corresponds with that atomic resonance.
- Once the feedback loop has acquired and is maintaining that condition, the unit is locked.
So there is an important distinction:
10 MHz present ≠ necessarily locked
Your FE-5650A can produce a very nice-looking and apparently stable 10 MHz signal on your Rigol immediately after switch-on, while the rubidium physics package is still warming up. At that stage you’re principally seeing the quartz oscillator.
Once locked, you have:
Quartz oscillator → continuously corrected by rubidium → extremely stable 10 MHz reference
That’s also why your observation that the frequency looked stable on the oscilloscope immediately after power-up doesn’t prove atomic lock. An oscilloscope simply doesn’t have sufficient frequency resolution over a short measurement to reveal the tiny difference.
In your experiments, the particularly interesting moment has been when the 15 V power consumption falls substantially after warm-up and the LOCK/BITE voltage changes state. Those observations give us much more information about whether the rubidium system has actually acquired lock than simply seeing 10 MHz on the scope.
A useful analogy is a quartz wristwatch being continually checked against an extraordinarily accurate atomic clock. Before lock, the watch is keeping time by itself. After lock, someone is continuously comparing it with the atomic reference and making microscopic corrections.
And this connects directly with the PLL you were asking about earlier: the feedback loop is what keeps the oscillator tied—or “locked”—to the atomic resonance.
Following is reproduced from the mailing list of the Society of Amateur Radio Astronomers September 2026, available publicly on the internet:
In fact, the tracking loop in a Rb standard is not a PLL. A PLL locks the phase of an VCO/VCXO to an external signal.
But a Rb cell does not emit any signal, it just slightly attenuates light, when irradiated with the right frequency. So a PLL can not be used, because the Rb cell only gives amplitude information, no phase.
The loop in a Rb standard must track the slight dip in light intensity. Usually, the output of the 6.xxx GHz synthesizer is slightly frequency modulated, with a modulation frequency in the hundreds of Hz range. If the FM span includes the Rb frequency, there will be a dip in the amplitude at the light detector. When the FM span is centered on the Rb frequency, there will not be any FM modulation frequency fundamental in the output of the photodetector, just the second harmonic, because of the symmetry. If the FM is not centered on the Rb line, the fundamental will appear, its phase (0/180 deg) indicating whether the frequency is too low or too high.
Since the FM span is much smaller than the crystal oscillator thermal drift, a search circuit is usually included, which will “walk” the VCXO up and down, to find the Rb dip. This can be seen as the triangular waveform, which stops when the Rb cell warms up, and the loop can lock.
Manuals of some Rb standards can be found on the web, which describe this in more detail.
Marko Cebokli