---
HOST: What if I told you there's a device on this bench that keeps time accurate to better than one part in ten billion — and you can buy one used, on eBay, for about fifty bucks?
This is a rubidium frequency standard — sometimes called a rubidium clock, though technically it's an atomic frequency oscillator, not a clock that counts hours and minutes. These units were pulled from decommissioned telecom infrastructure and cell towers, where they served as the beating heart of network synchronization.
Today, we're going to tear one of these down, examine every subsystem in detail, understand the quantum physics that makes it tick, and then put it back together and test it — comparing it against a GPS-disciplined oscillator to see if this retired workhorse still has what it takes.
Let's get into it.
---
HOST: The unit we're working with today is a typical surplus rubidium module — this one happens to be a FE-5680A, one of the most common units you'll find in the secondhand market. Similar units were made by Symmetricom, Stanford Research, and Microsemi, but they all follow the same fundamental architecture. [HOST rotates the module in hand]
At roughly the size of a thick paperback book and weighing about a kilogram, these are remarkably self-contained. There's a DC power input — typically 15 to 24 volts — a 10 MHz or sometimes 5 MHz output, and a few control pins. Some units also expose a voltage-controlled frequency adjustment input and a lock indicator signal. [CLOSEUP: Rear connector pins and label with model number, serial, and voltage requirements]
The label tells us the model, serial number, and operating parameters. This particular unit wants 15 volts DC at about 1.2 amps during warm-up, dropping to around 0.6 amps once it's locked. That warm-up current is significant — it's mostly going to the physics package heater, which we'll explore in a moment.
Before we crack it open, let me give you a quick primer on how a rubidium standard actually works, because it informs everything we'll see inside.
---
HOST: Rubidium-87 has a ground-state hyperfine transition at approximately 6.834682610 GHz. This is the frequency associated with the energy difference between the F=1 and F=2 hyperfine sub-levels of the 5S₁/₂ ground state. It's an incredibly well-defined atomic property — the same everywhere in the universe — which is what makes it such a good frequency reference.
The trick is: how do you detect this transition and lock a crystal oscillator to it? The answer involves a technique called optical-microwave double resonance, and it relies on a beautifully clever interaction between light and atoms.
A rubidium lamp emits light at 780 nm — the D1 resonance line. This light passes through a rubidium vapor cell, and if the atoms in that cell are being driven at exactly 6.834682 GHz, something quantum-mechanical happens: the atoms become transparent to that 780 nm light. A photodetector on the far side sees a dip in light intensity — a sharp, narrow resonance peak. The electronics lock a microwave oscillator to the center of that dip. Simple in concept, extraordinary in execution.
Now let's see how the engineers packaged all of this.
---
HOST: Four screws hold the outer shield — this mu-metal casing provides magnetic shielding for the physics package inside, because the hyperfine transition frequency is sensitive to external magnetic fields. [CLOSEUP: Mu-metal shield being lifted away, revealing internal PCB and physics package]
And here we are. This is dense. Two major things dominate the interior: the physics package — that cylindrical assembly — and the surrounding electronics PCB. [CLOSEUP: The physics package — bulb, cavity, photodetector, heater wrap]
Let's focus on the physics package first, because this is where the quantum magic happens. [HOST points with tweezers to each component as described]
At the top, you have the rubidium lamp — a small glass bulb containing rubidium metal and a buffer gas. It's excited into plasma by an RF oscillator coil wrapped around it — you can see the coil windings here. This lamp glows with that characteristic rubidium emission at 780 nanometers. [CLOSEUP: RF excitation coil around the lamp housing]
The light passes through an interference filter to isolate the D1 line, then enters what's called the resonance cavity — a glass cell containing rubidium vapor at low pressure, surrounded by a microwave cavity structure, often a TE₀₁₁ mode cavity. This is the heart of the physics package. [CLOSEUP: The rubidium absorption cell inside the cavity, light path visible with slight vapor glow]
On the far side of the cavity sits a photodetector — a simple silicon photodiode — which monitors the transmitted light intensity. When the microwave frequency inside the cavity matches the 6.834 GHz hyperfine resonance, the rubidium vapor undergoes what's called coherent population trapping, and the transmitted light intensity shifts slightly. That tiny optical signal is what the entire system locks onto. [CLOSEUP: Photodetector diode and its leads going to the preamp section of the PCB]
Wrapped around the entire assembly you'll see heater elements and a thermistor — the physics package operates at a carefully controlled temperature, typically around 50 to 70 degrees Celsius, to maintain stable vapor pressure inside the cell. [CLOSEUP: Thermistor bead and heater traces bonded to the cavity housing]
It's an elegant piece of engineering. Lamp, cavity, detector, heater — all wrapped into one sealed, temperature-controlled package.
---
HOST: Now let's look at the electronics. The frequency synthesis chain is what takes the 10 MHz crystal output and generates the 6.834682 GHz needed to interrogate the rubidium atoms — and then uses the atomic resonance to discipline that same crystal. [CLOSEUP: The main oscillator crystal — a small HC-40 or similar package on the PCB]
It starts here — a 10 MHz quartz crystal oscillator, typically an OCXO — an oven-controlled crystal oscillator. This is the local frequency reference. [CLOSEUP: Frequency multiplier and mixer ICs on the board]
The 10 MHz signal is multiplied up in stages. A phase-locked loop synthesizer — often a custom ASIC or a chain of discrete multipliers and dividers — generates a signal at 6.834 GHz or at an intermediate frequency that gets mixed up to the final value. In some designs, a step-recovery diode is used as a frequency multiplier to reach microwave frequencies from a lower harmonic. [CLOSEUP: Step-recovery diode or microwave mixer component, identifiable by its package]
The synthesized microwave signal is coupled into the physics package cavity. The photodetector output — that tiny error signal — goes back into a servo loop. This is a classic feedback control system: the error signal tells the electronics whether the microwave frequency is too high or too low, and a correction voltage is applied to the crystal oscillator, nudging it back on center. [CLOSEUP: Op-amp and filter components in the servo loop section]
The loop bandwidth is deliberately narrow — typically less than one hertz — because the atomic resonance is extremely sharp (Q factors of 10⁷ or higher), and you want the crystal's short-term stability to dominate at fast time scales while the atom provides long-term accuracy. This is a key design philosophy: the crystal provides low noise, the atom provides low drift.
---
HOST: Remember I mentioned that the hyperfine transition frequency is sensitive to magnetic fields? This is both a problem and an opportunity. [CLOSEUP: The C-field coil — a solenoid or Helmholtz-style coil wrapped around or near the resonance cavity]
The problem is that stray fields shift the resonance. The mu-metal shield we removed earlier handles most of that. But the opportunity is this: by applying a precisely controlled, known magnetic field — called the C-field — you can select one specific Zeeman sub-component of the hyperfine transition and also fine-tune the effective resonance frequency slightly. [CLOSEUP: C-field current adjustment potentiometer on the PCB, if present]
In many rubidium standards, there's a trim adjustment — either a potentiometer or a DAC — that controls the C-field current. Adjusting this changes the magnetic bias field, which in turn shifts the lock point of the servo by a small amount. This is one of the methods used to calibrate the output frequency during manufacturing, and in some units, it remains user-accessible. [CLOSEUP: Zeeman sub-level diagram overlay showing how magnetic field splits the F=2 level into nine components, mF = -2 to +2]
The C-field also lifts the degeneracy of the mF sub-levels. Without it, the resonance is broad and multi-component. With a proper C-field — typically 30 to 70 microtesla — a single, well-defined Zeeman component is selected, giving a clean, narrow resonance for the servo to lock onto.
It's a subtle feature, but without the C-field, these devices simply would not achieve their specified performance.
---
HOST: Alright, let's button this back up. Four screws, reconnect the power and output cables, and now — the waiting game. [Time-lapse: unit warming up, multimeter showing current draw starting at ~1.2A and slowly declining over several minutes]
A rubidium standard typically needs 5 to 10 minutes for the lamp to ignite and the heater to bring the physics package to temperature, then another few minutes for the servo to acquire lock. During warm-up, current is high. You'll see it taper off as the heater reduces power. Some units have a lock indicator — a pin that goes high or low to tell you when the servo has settled. [CLOSEUP: Lock indicator on oscilloscope transitioning from low to high]
There — the lock line just went active. Our unit claims it's locked. But is it actually accurate? Let's test it.
---
HOST: For testing, I'm comparing our rubidium standard against a GPS-disciplined oscillator — a GPSDO — which derives its frequency from the GPS satellite constellation, traceable to UTC. The GPSDO is our reference — it's accurate to better than one part in 10¹². [CLOSEUP: Dual-trace oscilloscope showing two 10 MHz sine waves, nearly stationary relative to each other]
First, let's look at the two signals on the scope. Both are nominally 10 MHz. If there's a frequency offset, you'll see the phase drifting — one waveform slowly sliding past the other. Let me put this into X-Y mode or just watch the Lissajous pattern. [CLOSEUP: Lissajous figure on oscilloscope — slowly rotating, indicating a tiny frequency offset]
You can see a very slow rotation — this means there's a small frequency difference. Let's quantify it. [CLOSEUP: Frequency counter display showing the rubidium output measured against the GPSDO timebase, reading 10,000,000.00X Hz]
The counter reads 10,000,000.003 Hz — that's a fractional offset of about 3 × 10⁻¹⁰. That's 0.3 parts per billion. For a surplus unit of unknown provenance, that's honestly excellent. The specification for these units when new is typically ±5 × 10⁻¹¹, so we're within an order of magnitude of factory spec on a used, eBay-sourced module. [HOST overlays a frequency stability plot — Allan deviation vs. tau]
Let me also run an Allan deviation measurement over about an hour, comparing against the GPSDO. [CLOSEUP: Allan deviation plot on screen, showing characteristic curve — improving from ~10⁻¹¹ at 1 second to ~10⁻¹² at 1000 seconds]
At one second of averaging, we're seeing around 3 × 10⁻¹¹ — again, right in line with what you'd expect. At longer averaging times, the rubidium holds beautifully. That's the whole point: the crystal gives you short-term quietness, the atom gives you long-term truth.
---
HOST: And there you have it — a complete teardown and test of a rubidium frequency standard. Inside that unassuming metal box, you've got a rubidium lamp, a microwave cavity, a photodetector, a precision synthesizer chain, and a C-field coil — all working together in a servo loop that locks a humble quartz crystal to one of nature's most fundamental constants.
These surplus units are one of the best bargains in precision test equipment. For fifty or a hundred dollars, you get a frequency reference that would have been science fiction a few decades ago. Whether you're calibrating counters, building a home lab, or just satisfying your curiosity about atomic physics, a rubidium standard is an incredible tool to have on your bench.
If you enjoyed this deep dive, hit that subscribe button, and let me know in the comments what teardown you'd like to see next. Until then — stay precise. [END CARD with subscribe button, related videos]
--- [TOTAL WORD COUNT: ~1,530] [TOTAL CLOSEUP MARKERS: 22]
Need precision timing solutions? Get a quote from BRIDZA