The seller of the 5650A I purchased has come back to me with following information:

FE‑5650A (Rubidium Frequency Standard, DB‑9 Connector, CPOM / Opt‑58)

1. DB‑9 Pin Assignment (DB9 Male Housing)

表格

Note: Pins 8 and 9 are TTL‑level signals. Do not connect directly to RS232 serial ports. A TTL‑USB adapter board is required.

2. Power Supply Requirements: Both +15 V and +5 V must be supplied

‑ +15 V: Input range 15‑18 VDC; peak inrush current 2‑3 A at startup, steady‑state current approx. 0.5‑0.7 A
‑ +5 V: 5.0 V ±0.25 V, approx. 100‑200 mA. Must be externally supplied; the module cannot generate 5 V internally.
‑ The ground connections of the two power supplies shall be tied together.

3. LOCK Status Pin (Pin 3)

Output type: Open‑Collector. The internal circuit can only pull the pin low and cannot drive a high logic level.

Logic definition:
‑ Locked condition: Pin pulled low to GND
‑ Unlocked / Loss‑of‑lock condition: Pin high‑impedance (floating)

An external pull‑up resistor (typical value 1 k‑10 kΩ pulled up to +5 V) is mandatory for lock‑status detection. Without the pull‑up resistor, the pin floats and the MCU cannot read valid logic levels.
With pull‑up resistor fitted: Locked = Low level; Unlocked = High level.

What effect does the above information have on how I should operate the device?

Yes. This changes the diagnosis quite substantially, and it also explains several of the puzzling things you saw when testing your FE-5650A.

The crucial new information is that your particular CPOM / Option-58 unit requires BOTH 15 V and 5 V supplies. Previously you were powering only pin 1 with 15 V. So although the unit heated up and drew substantial current, it was not actually being powered according to its required configuration.

What I would do now

Wire the DB-9 as follows:

Most importantly, join the grounds together: pins 2 and 5 should be common at the power supply ground. Pin 7 is the signal ground and can share that common reference for your measurements.

So the basic power arrangement is:

             FE-5650A DB9

15–18 V +  ---------------- Pin 1
15 V GND  ----------------- Pin 2
                              |
                              +---- common ground
                              |
5 V GND   ------------------ Pin 5
+5.0 V    ------------------ Pin 4


Pin 3  ---- LOCK output
Pin 6  ---- leave open
Pin 7  ---- signal ground
Pin 8  ---- leave open
Pin 9  ---- leave open

Your 15 V supply should ideally be capable of at least 3 A, because the seller specifies a 2–3 A startup surge. The 5 V supply only needs a few hundred milliamps, although I’d use something rated around 0.5–1 A or greater.

This also explains your strange pin-3 measurements

Previously you measured pin 3 at about 0.063 V, and later around 2.74 V.

We were treating those voltages as potentially meaningful lock indications. With the information from the seller, we shouldn’t interpret either measurement that way.

Pin 3 is an open-collector output. It does not produce a proper high voltage itself. When the rubidium oscillator is unlocked, pin 3 simply becomes electrically open/floating. A digital multimeter can consequently show almost anything depending on leakage, internal circuitry and meter impedance.

You need to provide a pull-up.

I suggest starting with 4.7 kΩ:

              +5 V
                |
              4.7 kΩ
                |
                +---------- Pin 3 LOCK
                |
             voltmeter
                |
Pin 7 GND ------+

Then measure pin 3 relative to ground.

You should see approximately:

Unlocked: ~5 V
Locked: ~0 V

Do not connect pin 3 directly to +5 V. The 4.7 kΩ resistor is important because when the FE-5650A asserts LOCK it effectively switches pin 3 to ground. With a 4.7 kΩ pull-up, that’s only about 1.1 mA, which is very modest.

What I think happened during your previous tests

Your observations now make much more sense.

You supplied 15 V to pins 1/2, the unit initially drew about 0.68 A and subsequently reached around 1.5 A and became hot. That strongly suggests that at least the heater/rubidium physics package circuitry was operating.

But the missing 5 V supply may have prevented other electronics from operating correctly. That could explain why you never found the expected RF output while scanning from 1–20 MHz and why pin 3 didn’t behave as expected.

I therefore wouldn’t regard the previous SDRangel frequency scan as evidence that the FE-5650A is faulty.

Your next test

I’d now do this in a very controlled sequence:

1. With everything OFF, connect +15 V → pin 1 and its ground → pin 2.
2. Connect a regulated +5.0 V → pin 4 and its ground → pin 5.
3. Join the two supply grounds.
4. Put a 4.7 kΩ resistor between pin 4 (+5 V) and pin 3.
5. Leave pins 6, 8 and 9 completely disconnected.
6. Turn on both supplies.
7. Measure pin 3 relative to pin 7.
8. Watch the voltage for perhaps 10–15 minutes while the rubidium physics package warms up.
9. Once pin 3 changes from roughly 5 V to near 0 V, we have a proper indication that the oscillator has locked.
10. Only then go looking for the RF frequency-standard output on the SMA connector with your Rigol or SDR.

One particularly important point: don’t put 5 V onto pin 3 just to see what happens. Use the resistor.

This new information is actually encouraging. Given that your module was hot and drawing current but you had never supplied the now-confirmed mandatory 5 V rail, I would repeat the entire lock/output test before concluding anything is wrong with the 5650A.