Login
Resurrecting my FE-5680A Rubidium frequency reference
 |
| Fig 1: The Hammond 1590 aluminum case housing the FE-5860A rubidium osc- oscillator and other circuitry - the markings faded by time and heat. Click on the image for a larger version. |
The first of these - my Efratom LP-101 - fired up just fine, despite having seen several years of inactivity. After letting it warm up for a few hours I dialed it in against my HP Z3801 GPSDO and was able to get it to hold to better than 5E-11 without difficulty.
My other rubidium frequency reference - the FEI FE-5680A - was another matter: At first, it seemed to power up just fine: I was using my dual-trace oscilloscope, feeding the 'Z3801 into channel 1 and the '5680A into channel 2 and watching the waveforms "slide" past each other - and when they stop moving (or move very, very slow) then you know things are working properly: See Figure 2, below, for an example of this.
That did happen for the '5680A - but only for a moment: After a few 10s of seconds of the two waveforms being stationary with respect to each other, the waveform of the '5680A suddenly took off and the frequency started "searching" back and forth, reaching only as high as a few Hz below exactly 10 MHz and swinging well over 100 Hz below that.
My first thought was something along the lines of "Drat, the oven oscillator has drifted off frequency..."
 |
| Fig 2: Oscillogram showing the GPS reference (red) and the FE-5680A (yellow) 10 MHz signals atop each other. Timing how long it takes for the two waveforms "slide" past each other (e.g. drift one whole cycle) allows long-term frequency measurement and comparison. Click on the image for a larger version. |
As it turns out, that was exactly what had happened.
Note:
I've written a bit more about the aforementioned rubidium frequency references, and you can read about them in the links below:
Oscillator out of range
While it is the "physics package" (the tube with the rubidium magic inside) that determines the ultimate frequency (6834683612 Hz, to be precise) it is not the physics package that generates this frequency, but rather another oscillator (or oscillators) that produce energy at that 6.834682612 GHz frequency, inject it into the cavity with the rubidium lamp and detect a slight change in intensity when it crosses the atomic resonance.
In this unit, there is a crystal oscillator that does this, using digital voodoo to produce that magic 6.834682612 GHz signal to divine the hyperfine transition. This oscillator is "ovenized" - which is to say, the crystal and some of the critical components are under a piece of insulating foam, and attached to the crystal itself is a piece of ceramic semiconductor material - a PTC (positive temperature coefficient) thermistor - that acts as a heater: When power is applied, it produces heat - but when it gets to a certain temperature the resistance increases, reducing the current consumption and the thermal input and the temperature eventually stabilizes.
Because we have the rubidium cell itself to determine our "exact" frequency, this oven and crystal oscillator need only be "somewhat" stable intrinsically: It's enough simply to have it "not drift very much" with temperature as small amounts of frequency change can be compensated, so neither the crystal oven - or the crystal contained within - need to be "exact".
 |
| Fig 3: The FE-5680A itself, in the lid of the case of the 1590 box to provide heat- sinking. As you can see, I've had this unit open before! Click on the image for a larger version. |
The give-away was that as the unit warmed up, it did lock, but only briefly: After a brief moment, it suddenly unlocked as the crystal warmed up and drifted low in frequency, beyond the range of the electronic tuning.
Taking the unit apart I quickly spotted the crystal oscillator under the foam and powering it up again, I kept the foam in place and watched it lock - and then unlock again: Lifting the foam, I touched the hot crystal with my finger to draw heat away and the unit briefly re-locked. Monitoring with a test set, I adjusted the variable capacitor next to the crystal and quickly found the point of minimum capacitance (highest frequency) and after replacing the foam, the unit re-locked - and stayed in lock.
Bringing it up to frequency
This particular '5680A is probably about 25 years old - having been a pull from service (likely at a cell phone site) and eventually finding its way onto EvilBay as surplus electronics. Since I've owned it, it's also seen other service - having been used twice in in ground stations used for geostationary satellite service as a stable frequency reference, adding another 3-4 years to its "on" time.
As quartz crystals age, they inevitably change frequency: In general, they tend to drift upwards if they are overdriven and slowly shed material - but this practice is pretty rare these days, so they seem to tend to drift downwards in frequency with normal aging of the crystal and nano-scale changes in the lattice that continue after the quartz is grown and cut: Operating at elevated temperature - as in an oven - tends to accelerate this effect.
By adjusting the trimmer capacitor and noting the instantaneous frequency (e.g. adjusting it mechanically before the slower electronic tuning could take effect) I could see that I was right at the ragged edge of being able to net the crystal oscillator's tuning range with the variable capacitor at its extreme low end, so I needed to raise the natural frequency a bit more.
If you need to lower a crystal's frequency, you have several options:
- Place an inductor in series with the crystal. This will lower the crystal's in-circuit frequency of operation, but since doing so generally involves physically breaking an electrical connection to insert a component, this is can be rather awkward to do.
 |
| Fig 4: The tip of the screwdriver pointing at the added 2.2uH surface-mount inductor: It's the black-ish component at sort of a diagonal angle, wired across the two crystal leads. Click on the image for a larger version. |
- Place a capacitor across the crystal. Adding a few 10s of pF of extra capacitance can lower a crystal's frequency by several 10s or hundreds of ppm (parts-per million), depending on the nature of the crystal and the circuit.
Since the electrical "opposite" of a capacitor is an inductor, the above can be reversed if you need to raise the frequency of a crystal:
- Insert a capacitor in series with the crystal. This is a very common way to adjust a crystal's frequency - and it may be how this oscillator was constructed. As with the inductor, adding this component - where none existed - would involve breaking a connection to insert the device - not particularly convenient to do.
- Place an inductor across the crystal. Typically the inductance required to have an effect will have an impedance of hundreds of ohms at the operating frequency, but this - like the addition of a capacitor across a crystal to lower the frequency - is easier to do since we don't have to cut any circuit board traces.
 |
| Fig 5: The crystal is under the round disk (the PTC heater) near the top of the picture and the adjustment capacitor is to the right of the crystal. Click on the image for a larger version. |
