We think of crystals as the gold standard of frequency generation. However, if you want real precision, you need something either better than a crystal or something that will correct for tiny errors — often called disciplining the oscillator. [W3AXL] picked up a rubidium reference oscillator on eBay at a low cost, and he shows us how it works in the video you can see below. He started with a GPS-disciplined oscillator he had built earlier and planned to convert it to discipline from the rubidium clock.
The connector looks like a D-shell connector superficially, but it has a coax connector in addition to the usual pins. The device did work on initial powerup, and using a lissajous pattern to compare the existing oscillator with the new device worked well.
We think the output might have been a little cleaner with a 50-ohm terminator on the scope. The clock uses a rubidium vapor lamp and a glass cell containing rubidium. A voltage-controlled oscillator excites the cell with a 6.8 GHz signal. Rubidium atoms can exist in three states. Because of a filter in the system, the cell will eventually reach a steady state where all the atoms are locked in one state. But the voltage-controlled oscillator can change that if the frequency is exactly right.
When the cell is in the steady state, the light passes through it with very little loss. However, if some atoms are in a different state, the cell will block some light, and a photocell can detect this. By tuning the oscillator with feedback to reach a minimum output from the photocell, you get a very precise output.
By the end of the video, the GPS oscillator never totally locked with the rubidium standard, although they were close. However, there is no way to know which of them was further off the actual 10 MHz reference.
We are looking forward to part two, where he actually converts the GPS-discipline oscillator to use the new atomic clock. We’ve seen this done before, of course. Cesium is even more precise, but there are things still better than that.
I thought he was going to discipline the rubidium oscillator with the GPS signal? I have a couple of old cell site rubidium frequency references and that’s how they work. My understanding is that this will provide a frequency reference with a lower phase noise than a crystal, and would be more stable during a loss of satellite signal.
That’s part II which, presumably, isn’t out yet.
So, I haven’t worked with the Efratom unit here, but my SRS has a pretty detailed theory of operation in the manual, and they say it sweeps the oscillator across the dip, basically dithering it to get a more pronounced dip and then mathematically determining where the center is, rather than trying to hold the center with a continuous analog servo loop. (Which could possibly get stuck in a local minimum that might be an artifact of some other environmental factor or whatever.)
This actually produces a fairly noisy output, so the Rb package itself isn’t used directly, it’s used (with more software and math!) to discipline an internal OCXO, which actually provides the 10MHz output. Quartz has low phase noise as long as you don’t physically jostle it, and it gains the long-term stability of the rubidium.
Awhile back I looked at some rubidium oscillators on ebay and others sources; most of the ones I ran across were broken, out of cal, and/or over priced. The most interesting thing I’ve seen is the Leo Bodnar GPS disciplined programmable oscillator. It works and it’s affordable.
I have one of those Leo Bodnar GPS references and it’s great, I use it as my lab reference.
I also have a HP 5065A Rubidium Standard on the ‘to repair’ shelf that I’ll get to one day…
I have an almost finished design to convert it to 10MHz output.
Time is like pi.
How many digits are there?
How many parts can we split a hair into?
I just got visited by some aliens. Fom some far away galaxy.
They said “precision, schmicision – accuracy is where it’s at.”
By the way, their GPS had died and they were lost and stopped by for directions.
Heisenberg and Schrodinger are on a road trip, when a cop pulls them over.
The officer walks up and asks if they know how fast they’re going. Heisenberg replies that they do not, but know with high precision where they are. The cop thinks that’s weird, and begins to search the vehicle. He opens the trunk and asks, “did you know you’ve got a dead cat in the trunk?” Schrodinger says, “well, now we do.”
The cop goes on and says “You were going 95 MPH in a 45 MPH zone”. Heisenberg then shouts, “just friggin’ great! Now we’re lost!”.
There are two kinds of people in this world:
Those who like Heisenberg/Schrodinger jokes, and those that don’t understand them.
Back in the 80s I worked for Motorola in San Diego. I designed police and fire radio dispatch systems. The last project I worked on was a radio dispatch system for the Orange County Fire Department. The system was based on Mot’s 800 MHz trunking tech, in this case simulcast, with frequency reuse across multiple hill/mountain/tower tops. We used rubidium freq standards at each site and linked them with microwave hops. As I recall, it was a real b***ch to get it aligned, but it worked. I’m sure it has been replaced by now, so maybe some of those rubidium standards are popping up on ebay. Or maybe they did 20 years ago…
Nowadays, it’s a TRAK GPS disciplined reference for simulcast.
I remember as recently as 2000, for Quantar’s the TRAK’s were used.
As part of our field set up/optimization we used the FS-725 to slave
the 2670 or 3920 to measure frequency for troubleshooting.
Only thing that really needed tweaking were simulcast launch offsets
on the backhaul links (back in your day, probably the SpectraTac cards,
newer we’re using the ATAC3000 comparators, all in software).
GTR8000’s right off the line, are usually spot on freq wise.
80’s … the day of the 6809’s .. and the USCI cards, etc.
The old SMR days…. before the dark times, before P25, a more civilized age.
I wonder: are these rubidium standards used everywhere in the world in mobile phone towers? Because here in France, i asked about them twice to people installing 5G antennas, and both times i was looked at with big interrogating eyes and answered “what are you talking about?”.
If they are, they are not very precise. I used to see basestations up to 50ppb either way when I was in that end of the business.
The installers have no idea what’s in the hardware they’re bolting into place.
All the CDMA equipment I ever worked on, had ovenized quartz GPSDOs. There was a while that Nortel GPSTMs were super cheap in surplus. I’m out of the industry so I don’t know what they’re using in 5G.
I believe some of the older Lucent equipment used Rubidium sources, there’s a quartz version and a rubidium version of the RFTG module. But I believe those are even older and I’d have lamp-life concerns at this point.