A few days ago, I wrote about using the best possible NTP settings for millisecond precision or better. Those settings sometimes often give my SDR rigs a time precision of tens of microseconds. That is great for running amateur radio weak signal modes like WSPR, FT8, etc., but it may not be sufficient for machines with nanosecond timekeeping requirements.
NTP is also not the answer if ultra precise frequency control or clock ticks are needed. For example, broadcasters and mobile phone stations typically use rubidium or cesium based atomic frequency standards in-house. Those things will accept time transfer from GNSS satellites, and then run based on their own, locally generated signals. With such precise timing, it is logical and normal to set up in-house stratum 1 NTP servers too, which can serve time to local machines which is good to a few tens of nanoseconds.
Note: Once you have a local stratum 1 time source for your computers, consider upgrading your timekeeping softare on those machines from NTP to PTP, which stands for "Precision Time Protocol." It makes better time transfer computations, accounting more accurately for netwwork latency.
The next step up in precision from getting the best possible time hacks from NTP is to drive your computer clocks and radio frequency references from steady crystal oscillators disciplined by GNSS timing signals. The bare crystal oscillators on your computer motherboard are not stable enough for this. They run fast or slow, drift with temperature, and start drifting away after each and every time hack.
You can set up an external clock, with a GNSS antenna, which can either reside in its own cabinet or within your computer on a PCIE card. These add-ons will use oven compensated crystal oscillators (OCXOs), housed in an isolated environment, which can be trimmed to the proper frequency and drift very little due to eternal influences. With a time hack from GNSS, these can keep your system tracking time within a few nanoseconds, and with low jitter. That matters a lot or processing wideband radio spectrum, high definition audio and video, and when precise timestamps are needed. This sort of lashup has a weakness, which may matter to some: loss of precision whenever GNSS is unavailable or degraded. OCXOs are quite a lot better than basic crystal oscillators, but they do drift away from proper time without active discipline based on GNSS.
When you have a need for 100% availability for clocks which are dead steady and precisely on frequency, for any of a thousand reasons, you will indeed need to set up your own rubidium or cesium time and frequency standard and lock it to timing distributed from the GNSS constellations. With such precise time and frequency references available locally, your systems can continue for days, if not weeks, if GNSS signals are intermittently lost. How long you can actually go without a time hack depends on your usage case.
With an in-house atomic reference, you can have signals as stable and steady in frequency and phase as you will ever need for amateur radio, high definition video, or high fidelity audio. Cesium references are the most accurate, but will cost thousands of dollars on the used and surplus markets. Rubidium standards are far better priced, with good surplus units available for one or two hundred bucks.
The basis for using atomic time and frequency standards is that elements all absorb and radiate energy in a unique pattern of frequencies when they are energized in the ground state and then give up that energy. That means they burn in certain colors and they likewise give and take at certain microwave frequencies. We believe this is true for all of the elements, and that spectra of any element is the same for all atoms of that element.
Rubidium-87 atoms, all of them, have what is called a "hyperfine transition frequency" of 6.834682610 GHz. The hyperfine transition frequency of cesium-133 is higher, at 9.192631770 GHz. If a device can generate a microwave signal at the hyperfine transition frequency and has a means to detect that the atoms are absorbing energy, there will be no doubt about that frequency or any other frequencies derived from it.
In fact, there are ways to know when the atoms are being energized at the right frequency. Rubidium standards use a laser, which shines through the gaseous Rubidium and onto an optical detector. When the microwaves are at the hyperfine transition frequency, the Rubidium becomes more opaque and the laser light intensity dips at the detector. Cesium beam standards have a different kind of physics package. They pass hot gaseous cesium through a vaccum chamber, in the presence of magnetic fields. Atoms in the desired energized state are deflected toward a detector, which provides the necessary feedback for tuning the microwave frequency.
A system which finds and stays on the frequency associated with the light dip is called a "frequency locked loop." We can derive very useful radio signals of various frequencies by using frequency division and phase locking methods. The time and frequency standards you bring into your radio shack or homelab will most likely have outputs at 10 MHz, maybe 5 MHz, and one pulse per second. For precise time, the only extra thing to add is a reliable source to acquire the real date and time of day. Your equipment will have its own precise measure of time intervals; it simply needs a reference for marking where it is in terms of wall clock time (or UTC).
Give consideration to how your computers will receive timing data from your sources. It may seem that USB is convenient, but inaccuracies may be introduced due to buffering through the data interface being used. It will be standard for quality sources to deliver the single pulse per second ticks by toggling a voltage somewhere on the serial bus, GPIO pins, or PCIE bus. Remember, it is the tick marking the start of seconds which is critical, not the date or which second of the day is elapsing.
For a quick plug and play solution using new equipment, consider something like the Leo Bodnar time and frequency standards. They are excellent performers, reasonably priced, and have excellent documentation.
There are also a bunch of atomic time and frequency standards which you may find in listings on Ebay or other platforms carrying broadcast, telecom, or military surplus electronics. My suggestion is to watch the listings, stalk newer and better conditioned gear, and be wary of junked gear priced super low or being sold for parts. Here are some makes and models of notable equipment:
Atomic time and frequency standards are becoming more and more miniaturized, due to demand or lighter, more portable, and less power hungry devices for mobile communications. Expect to see more of these listed for sale, and hopefully at better prices, as time goes on.
© 2015 - 2026 Skywavelinux.com, All Rights Reserved. Written and curated by Webmaster - Developer Philip Collier Contact, Privacy Policy and Disclosures, XML Sitemap.