The stable waveform from power-up seen on the Rigol DS1102 oscilloscope fits quite well with how a rubidium standard works.
The 10 MHz oscillator itself can start immediately and look very stable on an ordinary oscilloscope, even before the rubidium physics package has reached operating temperature. What changes after warm-up is not usually “oscillating versus not oscillating”; it is whether that oscillator is disciplined to the rubidium atomic resonance.
So before warm-up, the 10 MHz output may already be:
- clean
- steady-looking
- close to 10 MHz
- apparently rock-solid on the Rigol
Yet it may still be free-running or only partially controlled.
Your Rigol is not really sensitive enough to distinguish, for example, 10.000000 MHz from 9.999999 MHz in a way that would demonstrate atomic lock. Both would look perfectly stable on the screen.
That actually makes your pin-3 behaviour more meaningful. The likely sequence is:
power on → quartz oscillator already running steadily → rubidium lamp/cell/oven warms → atomic servo acquires resonance → pin 3 changes state → heater power settles.
So I would no longer expect the scope waveform itself to visibly “become stable” at lock.
The best confirmation now would be to compare the FE-5650A against another accurate 10 MHz reference, or measure its frequency with a counter having a good external reference. That would let us see whether the output drifts during the first few minutes and then settles when pin 3 drops to ~1.0 V.
Should I worry that yesterday the 5650A did not seem to lock properly?
Probably not, given what you have demonstrated today.
Today the behaviour is quite coherent and repeatable: after power-up, pin 3 sits at about 3.2 V, the unit draws high warm-up power, then the power falls to around 6–9 W and pin 3 changes to about 1.0 V. That looks very much like a normal warm-up/acquisition sequence.
Yesterday there were several things that could have made the result misleading: the 5 V supply and pull-up arrangement were still being worked out, the power-up sequence was different, and we had not yet established what the unusual ~3 V and ~1 V levels meant on your particular Option-58 unit. So I would not treat yesterday’s behaviour as evidence of a fault by itself.
What I would do now is simply repeat a cold start on another two or three occasions. Leave the unit off long enough to cool, then use the same sequence each time: 15 V on, then 5 V on, and watch pin 3 and input power. If each time it goes roughly 3.2 V → ~1.0 V as the heater power drops, I would regard the unit as behaving consistently.
I would only start worrying if it sometimes stays at ~3.2 V indefinitely, never reaches the lower-power state, or the 10 MHz output disappears or becomes erratic.
So based on today’s test, I would be reassured rather than concerned. The strongest thing you have learned is that the lock-status behaviour is reproducible when the unit is powered correctly.
Do you think that rubidium lock has been successfully achieved?
That is now a very strong result.
Your meter is showing about 1.008 V, and you observed that this happened when the 15 V supply power dropped to 6.6 W.
So you have now reproduced the same behaviour:
15 V switched on → pin 3 ≈ 3.20 V
warm-up at high power
power settles down → pin 3 ≈ 1.0 V
That makes it highly likely that the ~3.2 V state is “not locked / warming up” and the ~1.0 V state is “locked / normal operation.”
The absolute voltages are unusual for a simple 5 V open-collector output, but the state change is highly repeatable and correlated with the rubidium unit reaching its low-power steady state.
I would now regard this as the practical lock indication for your particular FE-5650A:
| Condition | Pin 3 |
|---|---|
| 15 V just switched on / warming | ~3.2 V |
| Warmed up / likely locked | ~1.0 V |
| 15 V removed, 5 V still present | ~1.2 V |
The next useful check would be to look at the 10 MHz output on the Rigol before and after the 3.2 → 1.0 V transition. If the frequency becomes noticeably more stable once pin 3 drops, that would independently confirm that this transition corresponds to frequency lock.
At this point, though, I think you have found your lock indicator.
The photograph in the results post (URL at top of this post) shows 3.199 V, seen immediately after turning 15V back on after brief test of turning it off and on again whilst leaving 5V power supply turned on, essentially identical to your earlier 3.198 V reading.
So we now have a reproducible sequence:
15 V OFF, 5 V ON: pin 3 ≈ 1.24 V
↓
15 V switched ON: pin 3 immediately → 3.20 V
↓
5650A warms up at high power: pin 3 remains around 3.2 V
↓
Earlier today, when consumption fell to ~9.3 W: pin 3 fell to about 1.0 V
That last transition is the important one. Don’t change anything now. Leave both 15 V and 5 V supplies running and watch the meter and the 15 V power consumption.
If pin 3 again changes from ~3.20 V → ~1 V at approximately the same time that the 15 V power falls to its steady ~9 W level, we’ll have reproduced the warm-up transition. That would be quite compelling evidence that this pin is indicating the operating/lock state.
I’d particularly like to know the exact pin-3 voltage immediately before and after that transition, the 15 V power at the transition, and roughly how many minutes after switching on it occurs.
This is looking considerably more systematic now.
The timing strongly suggests you are now seeing the lock transition.
Your photos show pin 3 changing from about 3.20 V to about 1.01 V. You observed that the drop happened at roughly the same time as the 15 V input power fell from its initial 20–30 W warm-up demand to about 9.3 W. That is exactly the stage at which I would expect the rubidium lamp/oven system to have warmed sufficiently for the servo to acquire lock.
For your unit, the information you obtained for the CPOM / Option-58 version says pin 3 is an open-collector LOCK/BITE output, with locked = pulled low and unlocked = high impedance. The FE-5650A documentation also identifies pin 3 as the lock indicator.
So your sequence appears to be:
Immediately after +5 V is applied:
Pin 3 ≈ 3.20 V → output relatively high → unlocked / warming up
After the 15 V power consumption settles:
Pin 3 ≈ 1.01 V → output pulled substantially lower → probably LOCKED
The really persuasive part isn’t the absolute voltages; it’s that the pin changes state precisely when the heater/oven power drops to its steady-state value.
There is one oddity: with a 4.7 kΩ pull-up to 5 V, ideal open-collector logic would normally give something much closer to 5 V unlocked and 0 V locked. Your 3.2 V → 1.0 V levels aren’t textbook logic levels. That tells us there’s probably additional circuitry on this particular customer-specific unit — for example an internal resistor, transistor network, BITE circuitry or clamp — loading pin 3. So I would not change the pull-up resistor in an attempt to force the voltages to 5 V and 0 V.
In fact, your observation is considerably stronger evidence of lock than yesterday’s static voltage measurements because you now have a very reproducible physical correlation:
high heater power → high pin-3 voltage → warm-up
heater power falls → pin-3 voltage falls → oscillator lock
Your separate finding that the SMA output gives a stable waveform at roughly 10 MHz and around 2.22 Vpp is another very encouraging piece of evidence.
One experiment would make this much more convincing
Repeat exactly what you did today and record these four things every 30 seconds from switch-on:
| Time | 15 V power | Pin 3 voltage | 10 MHz output |
|---|---|---|---|
| 0 min | perhaps 20–30 W | ~3.2 V | present? |
| 1 min | … | … | … |
| 2 min | … | … | … |
| … | … | … | … |
| lock transition | falls toward ~9 W | ~1.0 V | stable |
| steady state | ~9 W | ~1.0 V | stable |
If on two or three cold starts the pin-3 transition repeatedly occurs at essentially the same moment that the heater power drops, I’d be comfortable calling the lower pin-3 state LOCK.
And I would continue using your present 4.7 kΩ pull-up. It is right in the recommended 1–10 kΩ range and is unlikely to stress the output.
One further test could make this almost definitive: leave both supplies running until it reaches the ~1.0 V state, then briefly interrupt only the 15 V supply while keeping the 5 V electronics/pull-up alive. We can predict what pin 3 should do and interpret the result very cleanly.