Sub miniature atomic clocks enable greater precision
4 mins read
Every electronic product or system needs a clock to keep it working at the level of performance which its designers intended. And there's a number of approaches available, ranging from humble quartz crystals operating at 32.768kHz, to very high frequency devices at the other end of the scale.
But there are some applications where frequency stability is just as important as the frequency itself. In those applications, some designers are turning to atomic clocks.
Many will be familiar with atomic clocks as an essential part of a national standards laboratory. In the UK, the atomic clock system at the National Physical Laboratory (NPL) in Teddington is the guardian of time. Not only was the first atomic clock built at the NPL, but the site continues to develop even more accurate systems.
Atomic clocks are used in a wide range of applications – everything from communications to satellite navigation. By accessing an accurate source of time and applying it globally, these systems can work together. In telecommunications, it means data arrives when it should. In navigation systems, such as GPS, it provides the accuracy needed.
At this level, the high accuracy of a national standard plays a central role. But not everyone will need that accuracy; neither do they have the space for such devices. A number of companies have developed atomic clocks which bring better accuracy and which occupy little more space than a top quality oscillator. UK based Quartzlock, a specialist in the design and manufacture of high accuracy timing and frequency solutions, has developed atomic clocks based on the transition of rubidium atoms.
Clive Green, Quartzlock's managing director, said the devices can be used where an improvement is needed over the performance of a regular quartz oscillator. "Normally, a quartz oscillator is stable to something like 1x10-7 per year. Sometimes, however, it can be less accurate; more like 1x 10-6. The best quartz oscillator in the world will cost you £5000 and you still won't get an accuracy of 1 x 10-8. Those devices aren't normal; they are large boxes and most distributors of quartz oscillators will not have heard of them."
So why should designers worry about such accuracy? "An instrument which features a normal quartz oscillator drifting at this level will need to be calibrated regularly," Green said. "Whether it's a synthesiser, a signal generator, a digital signal oscilloscope or whatever, anything with a quartz oscillator will need to be sent away for calibration, unless your company is lucky enough to have its own calibration lab."
It is possible to get better accuracy through an approach called a gps disciplined oscillator. These devices combine a gps receiver with a high stability oscillator to provide an accurate reference for timing applications. But there are potential problems. "If you lose the antenna, you lose the reference," Green pointed out. "However, some systems have a hold over feature and if the antenna is lost, the frequency will be held for a period."
He also said timing issues are being encountered in the mobile phone network. "Basestations are getting old," he said. "Leads and connectors are giving up; it's sometimes like the cabling to the masthead of a yacht. The problems are all down to gps reception and the effect depends on how mission critical your application is."
He says it costs around £300 for what he terms a 'decent' quartz oscillator. "That can be a large part of a design's bill of materials," he continued, "and getting the price down means losing performance."
When performance is an important design parameter, Green points engineers towards rubidium oscillators. "Rubidium oscillators have 100 times better drift performance than quartz at 4 x 10-10 per year and the devices are available for only twice the price of an oven controlled crystal oscillator (ocxo)."
He believes these parts will be attractive in a range of applications, including telecoms networks and where synchronisation is needed. "And the price will come down in volume to the point where rubidium oscillators become competitive," he added.
So how does a rubidium oscillator work? Green said two rubidium gases are at the heart of the technology. "There's Rb85 in a lamp and Rb87 in the cell; all at precise pressures and quantities."
In order to generate the stable frequency, the gas in the lamp is excited with an rf field. Once excited, it emits radiation, which then passes through the Rb87 in the cell. Light passing through the cell is detected by a photodiode.
In fact, the rubidium lamp and cell are used to discipline a quartz oscillator, rather than generate the frequency directly.
What the photodiode is looking for is a slight dip in the light output from the cell, which indicates the Rb87 atoms are resonating at their natural frequency. When the output is approaching the natural resonant frequency, it's swept with an rf synthesiser. "When the change in light output is detected, the circuit locks that oscillator to whatever frequency it happens to be at. The output from the rubidium oscillator will be 10MHz, accurate and stable," Green continued.
Rubidium oscillators, when first developed, were supplied on 3U boards. According to Green, the devices became miniaturised when Rohde & Schwarz started looking at the technology, creating a device in a 2 x 3 x 4in package. Quartzlock's E10-MRX is now a similar size to conventional frequency components.
Green believes Quartzlock's long experience with rubidium oscillator technology has brought physical and electrical benefits to devices such as the E10-MRX, which he describes as a sub miniature atomic clock. "At the heart of the design is Columbian waveguide cavity," he said, "and this is accompanied by an ultra low noise synthesiser and d/a and a/d converters to reduce power consumption and mass significantly. It also brings higher resolution control of the C field, cavity and cell temperatures and lamp power."
He also says the device brings size and lock time benefits. "The E10 has been miniaturised to a size equivalent to an ocxo (50 x 50 x 20mm) and now has a lock time of a minute. The normal warm up time for an Rb oscillator is at least five minutes, which means valuable battery power is not wasted during warm up." The E10-MRX draws 6W from a 12V supply.
Green says the E10-MRX can be used as a standalone frequency source in UMTS and LTE applications and for extended holdover in CDMA, WiMAX and LTE basestations. It also brings stability to other communication and transmission applications.
Building on this, Quartzlock has developed the E10-LN. This module, which measures 91 x 55 x 30mm, takes the technology featured in the E10-MRX and adds the company's active noise filter technology. According to the company, this will enable new applications where size is restricted.
Green says applications such as defence, marine, remote tv repeaters, wired telecom infrastructure, portable wireless test instrumentation and bench/system synchronisation will all benefit from the E10-MRX.
For the future, Quartzlock has developed Coherent Population Trapping technology for rubidium oscillators. Green said this technology, aimed at 'next generation' applications, will halve power consumption to 3W.