FY 2019 budget requests shutdown of NIST time stations

If you run the 10 MHz throug a fast gate, then you get nice markers every 10 MHz.

Welcome to supply and demand. The cheap used FE-5680A boxes were mostly salvaged from telecom and cellular equipment. It was well known that they age, so these were also retired before they drifted out of spec. That's why there were so many of them on the market. The trend has been towards using GPSDO receivers for frequency sources, which don't age. The problem with these has been the need for an outside antenna, which was mostly eliminated by more HSGPS (high sensitivity GPS) for indoor use: Also, but making receivers that can utilize additional satellite constellations (GPS, WAAS, EGNOS, MSAS, Galileo, GLONASS, BeiDuo, QZSS, etc) as well as pseudolites when available. So, with the supply of surplus rubidium freq sources drying up, and the remaining units probably requiring replacement rubidium lamps, little wonder the prices have increased.

Sure. If you're a survivalist and own a secret underground laboratory with no view of the sky, an autonomous frequency standard might be a good thing. Otherwise, I prefer a GPSDO.

You're welcome, but which JL? John Larkin or me?

Jeff Liebermann wrote

Well I noticed GPS comes and goes, when I am testing something, then the last thing I want is some weard interrupt.

Now you got me there, never realized your initials are also JL. I meant John Larkin.

You may need a better antenna location. We usually see 5 to 7 satellites incoming.

I bought the Efratom mostly to see how it works. It's an old, all discrete parts design. It works from clever design and, apparently, some fortunate numerical coincidences.

Do you (or anyone) have any idea where the frequency standard for GPS lives? AIUI there are Rb and Cs clocks on each satellite, but do those also get 'conditioned' by some earth reference?

George H.

I suggest you get a better receiver, outdoor antenna, or both. If you are in Europe and near an airport, get something that does EGNOS: Or maybe just wait until 2020 when Galileo becomes fully functional.

My house is buried in the trees, on the side of a steep hill, with little view of the sky. From inside my house, I can see 5 of 9 visible GPS, and 4 of 7 Glonass birds. No Galileo because my receiver isn't setup to hear those, and no QZSS (Japan) or BeiDou (China) because I'm in the wrong part of the planet. The only time when I see no satellites is when it's raining or when I'm using an ancient GPS receiver.

However, you're worrying about something that is not a problem. There are numerous GPSDO designs, but most use a very slow phase or frequency lock loop to stabilize a TCXO (temp compensated xtal osc). Time constants of 100 to 1000 seconds are common. For frequency stability measurement, averaging over an entire day is common. Notice how the short term stability of a Trimble Thunderbolt GPSDO is awful, while the long term stability is excellent:

Here's a rather simple GPSDO using the 1pps output from the GPS to stabilize a 10MHz VCXO. (Errata and updates) The sample rate for the FLL (freq lock loop) is 16 seconds. Looking at the graph in Fig 7, it might take 16 hrs for the frequency to stabilize. You could loose signal for a few minutes and the loop would never notice.

With such a long time constant, a GPSDO can loose its input for many minutes because the TCXO will simply stay on frequency and not drift wildly. With a fast PLL or FLL, the frequency might be kicked around by noise generated by the phase/freq detector. If the input disappears, most phase/freq detectors are designed to hold their output to whatever it was before it lost the signal.

Thanks. I wasn't sure if I had performed such an un-natural act and recommended something other than my favorite, a GPSDO.

The GPS MCS (master control station) is at Schriever Air Force Base in Colorado Springs, Colorado, US. There are also 16 MS (monitoring stations) with atomic clocks that monitor the various satellites, and send any corrections to the MCS.

No. The satellite clocks do not get tweaked. I'm not sure but methinks this is because the relativistic effects would require almost continuous adjustment. Instead 16 ground stations compare the satellite transmissions with a local cesium frequency standard. The data gets sent to the MCS which some how processes it into satellite position corrections that are transmitted to the users GPS receiver. Exactly how the corrections work, I don't known.

Jeff Liebermann wrote

Yes, but I have the rubidium standard so I do not care.

My GPS receivers modules are normally indoors when I experiment, except for the ones in the drone and remote of course.

When GPS goes it is most likely local interference... Normally I have a very good indoors signal on GPS, and a good signal on GLONASS outside. I looked up and asked in the GPS group about the difference, and it seems the Russian sats have simply less transmit power.

Nice site. Looking at the block diagram of the system in that pdf the first thing that draws attention is that he divides the 1pps by 16! So now you have 1 pulse per 16 seconds, that does _not_ I repeat _not_ speed up locking now does it?

There must be a better way.

Nevertheless I have bookmarked it, downloaded the pdf, and will read it some more.

He uses a PIC 18F2220 micro, that is good ;-)

But the warm up time of my rubidium reference is at most a few minutes before the 'lock' LED comes on.

Yes, opinions differ, main point is you are happy with what you do / have.

But now when you are locked to the GPS (or other system), then you are basically locked to a rubidium reference up there I'd think, so it seems a bit complictiitatiated way to get a stable 10 MHz to me?

;-)

Anyway, what's the fun in having a scope with a 30-ps rise time if you never get to play with it?

Cheers

Phil Hobbs

You'd need some pretty long time constants otherwise!

Cheers

Phil Hobbs

There's nothing magic about codes for that job--you can get the same effect using a PLL with a nice strong analogue phase detector such as a CMOS mux. Some software is good for acquisition when the signal is that weak. Meself I'd probably mix down to 10 kHz or so and then softwarify that. Ten kilohertz gets you pretty well out of the drift and 1/f noise that can bite you when pulling signals out of the mud like that, and you can make arbitrarily narrow filters digitally.

Cheers

Phil Hobbs

And often dramatically quieter, iirc. Some years ago I looked at GPSDO phase noise specs, and they were the pits compared to a surplus rubidium.

Cheers

Phil Hobbs

I wonder why they used such a low dither frequency (127 Hz).

They'd have to use a pretty slow PLL in the synthesizer to avoid putting that modulation on the output.

Cheers

Phil Hobbs

The good GPS sources discipline a rubidium. It takes a long loop time constant to filter the jitter out of the 1 pps ticks.

One hazard of the old Tek 11800-series is that some (all?) sampling heads have terrible delay tempcos. Modern digital scopes are much better.

In a modern scope, when you trigger from one vertical channel and look at another, the exact trigger point is software interpolated to a fraction of a sample period. That not only gives very low jitter, but the symmetry of the input channel circuitry (as compared to the external trigger input) greatly reduces drifts.

One of our cheapish Rigol scopes has a channel-to-channel jitter measurement floor around 30 ps RMS, a small fraction of the sample period. Dr Shannon was a smart guy.