Fullscreen HTML5Med Res (??MB)HTML4

Rubidium RUBIDIUM DISCHARGE LAMP IN TEMP-CONTROLLED OVEN FILTER CELL Rb-87 795nm, 780nm PHOTO DETECTOR OUTPUT RESONANCE CAVITY (6.834, 682,612,8GHz) Rb-87 780nm PHOTODETECTOR 6.834GHz RESONANCE CELL MICROWAVE FREQUENCY ANALOG TO DIGITAL CONVERTER Frequency Standards Rb-85 STEP RECOVERY DIODE WHICH MULTIPLIES VCO FREQUENCY BY 19 (19 x 359.72 = 6384.68) ‘PHYSICS PACKAGE’ LAMP EXCITING OSCILLATOR (~150MHz) SYNTHESISER WITH OUTPUT AT ~359.72MHz DIGITAL TO ANALOG CONVERTER How they’ve shrunk in both size & cost! ‘FINE TUNING’ PHASE ERROR CORRECTING MAGNETIC COIL OR ‘DISCIPLINING’ VOLTAGE MICRO CONTROLLER DIGITAL TO ANALOG CONVERTER LOW NOISE 10MHz CRYSTAL OSCILLATOR IN TEMPERATURE CONTROLLED OVEN 10MHz OUTPUT (±5 x 10 -11) Like their better known cesium-beam based cousins, rubidiumvapour frequency standards (AKA ‘atomic clocks’) have shrunk considerably in both size and cost since they were first developed in the early 1960s. In fact, rubidium standards have shrunk much further than the cesium type and are now down to the same size as an oven-controlled quartz oscillator or ‘OCXO’. Their cost has also dropped far below that of a cesium standard, too. By JIM ROWE A LTHOUGH THEY didn’t appear in practical form until a few years after the first cesium-beam frequency standards, rubidium-vapour standards proved to be much more suited for making smaller and more portable frequency references. This was partly due to the rapid developments in microwave technology that took place The smallest Rb-vapour standard to date: Quartzlock’s E10-MRX, shown here actual size. 36  Silicon Chip during and after World War 2, along with the dramatic developments in solid-state technology that happened at much the same time. There was also a chicken and egg effect. When both military and commercial communications began to move into the UHF and microwave spectrum, this generated a huge market for low-cost yet highly accurate frequency standards – and manufacturers of rubidium-vapour standards were able to take advantage of this demand. The growth in demand not only continued but almost became explosive as mobile phones first went ‘cellular’ and then morphed into CDMA and its related digital technologies. Of course, the development of the internet and digital data communications played a major role too, as the need for accurate frequency and timing references multiplied exponentially. The net result of these developments is that by the year 2000, there were many hundreds of thousands of compact, low cost Rb-vapour frequency standards in use all over the world. There were also many different firms manufacturing them, such as Quartz­ lock in the UK (www.quartzlock. com), Stanford Research Systems in California, USA (www.thinksrs.com), Symmetricom Inc, also in California (www.symmetricom.com) and FEI Communications Inc of Mitchell Field, New York (www.freqelec.com). Nowadays we’ve reached the stage where you can buy very small Rb-vapour frequency standards brand new for less than $2500. For example, the Quartzlock E10-MRX subminiature unit is currently available for $A2210 plus GST (or $2431) from Quartzlock’s Australian representatives Trio Test & Measurement (www.triotest.com. au), while the Stanford Research Systems PRS10 can be ordered online from their website for $US1645 plus $US224.20 for shipping and handling. But that ain’t all, folks. A quick scout around the web – and on eBay in particular – discovered many different ‘used’ Rb-vapour references available for less than $US250 and as little as $US99.00 (plus postage etc). Most of siliconchip.com.au How Rubidium Frequency Standards Have Shrunk In Size 100,000 Above: used FE-5680A Rb-vapour standards are available on eBay from China for less than $150 including postage. these were on offer from China and appear to be ‘reclaimed’ from surplus telecom gear – perhaps exported from the USA or Europe. There were many different FE5680A units from FEI Communications, plus a few FRK, FRS-A, FRS-C and LPRO-101 units from Symmetricom (the latest incarnation of Efratom, which became Ball and then Datum). Because they are ‘pre-loved’, these very low priced Rubidium frequency references are a bit of a gamble. That’s because the Rb-vapour discharge tubes used in these references have a relatively limited working lifetime – the latest generation tubes have a rated lifetime of 20 years, while the earlier generations were found to have somewhat less. So it’s quite possible that at least some of the el-cheapo units were junked because their RB-vapour tubes had reached the nominal end of their working life. But before you dismiss the idea of buying one of these ‘el cheapo’ frequency references, consider this: on the internet I also found some information suggesting that it’s possible to ‘rejuvenate’ tired old Rb-vapour discharge tubes – to bring them back to almost ‘as new’ condition. So buying one of them might not be so risky after all. Anyway, to cut this introductory preamble short, here’s an admission: I recently purchased one of the used FE-5680A units myself, with the idea of seeing how easy it is to get going. Hopefully, I may be able to tell you siliconchip.com.au Volume (cm3) 10,000 1,000 100 10 1 1960 1970 1980 1990 2000 2010 2020 Fig.1: Rb-vapour frequency standards have dramatically dropped in volume since their development in 1960. The first units were about 91,000cc, while the Quartzlock E10-MRX (facing page) is only 65cc – quite a drop. more about this later in the year. For the present though, let’s have a look at how rubidium-vapour frequency references or ‘atomic clocks’ actually work. How they work First of all, forget any thoughts about ‘atomic clocks’ (of either the Cs-beam or Rb-vapour type) having anything to do with nuclear power. They don’t – not at all. While they do involve atoms of cesium or rubidium gaining or losing energy, this is purely in terms of changes in the energy levels of electrons in the outermost levels of the atoms. There are no changes inside the nuclei of the atoms. Just electrons jumping from one energy level to another, as in normal electrical conduction. And to brush up on high school physics and chemistry, rubidium is a silvery-white metallic element in the alkali metal group. It has the atomic number 37 and an atomic mass of 85.4678. It’s also highly reactive, oxidising rapidly in air or water (like sodium, potassium and cesium). Rubidium also has a very low melting point; just 39.3°C (102.74°F). Naturally occurring rubidium is a mixture of two isotopes: Rb-85, which is very stable and Rb-87 which is slightly radioactive (with a half-life of 48.8 x 109 years). The isotopes are usually found mixed in the proportions 72.2% of Rb-85 to 27.8% of Rb-87. As the only difference between the two isotopes is the number of neutrons in their nuclei, this explains why the official atomic mass of natural Rb is given as 85.4678. Rubidium atoms in both of these natural isotopes have only a single electron in their outermost energy levels (ie, a single valence electron). But in the unexcited or ‘ground’ state of both isotopes, this valence electron can occupy one of two very slightly separated energy levels – depending on the electron’s spin. The operation of Rb-vapour frequency standards takes advantage of the fact that the two ‘hyperfine-split’ ground state energy levels of Rb-87 atoms differ by an amount (28μeV) corresponding exactly to the amount of energy carried by a microwave photon February 2014  37 RUBIDIUM DISCHARGE LAMP IN TEMP-CONTROLLED OVEN FILTER CELL Rb-87 795nm, 780nm Rb-87 780nm PHOTODETECTOR 6.834GHz RESONANCE CELL Rb-85 ANALOG TO DIGITAL CONVERTER SYNTHESISER WITH OUTPUT AT ~359.72MHz DIGITAL TO ANALOG CONVERTER ‘FINE TUNING’ PHASE ERROR CORRECTING MAGNETIC COIL OR ‘DISCIPLINING’ VOLTAGE MICRO CONTROLLER MICROWAVE FREQUENCY STEP RECOVERY DIODE WHICH MULTIPLIES VCO FREQUENCY BY 19 (19 x 359.72 = 6384.68) ‘PHYSICS PACKAGE’ LAMP EXCITING OSCILLATOR (~150MHz) PHOTO DETECTOR OUTPUT RESONANCE CAVITY (6.834, 682,612,8GHz) DIGITAL TO ANALOG CONVERTER LOW NOISE 10MHz CRYSTAL OSCILLATOR IN TEMPERATURE CONTROLLED OVEN 10MHz OUTPUT (±5 x 10 -11) Fig.2: this generic block diagram for a fairly recent Rubidium-vapour frequency reference shows how these units work. Earlier units, like the Efratom FRK and the original FE-5680A, had an analog frequency locking loop and synthesiser but worked in much the same way. at a frequency of 6.834,682,612,8GHz . So if a photon of this frequency meets an Rb-87 atom where the valence electron is in the lower ground state energy level, it can ‘bump’ the electron into the upper level. Conversely, if the Rb-87 atom has its valence electron in the upper ground state energy level and ‘relaxes’ (say as the result of an applied magnetic field), it emits a photon of this frequency. As it happens, atoms of the Rb-85 isotope also have two hyperfine-split ground state energy levels for the valence electron. However, quite fortui- tously, the lower of these two energy levels in the Rb-85 isotope is almost exactly the same as the upper ground state energy level of the Rb-87 isotope. Rb-vapour frequency standards also take advantage of this coincidence, as we will see shortly. Now take a look at the block diagram of Fig.2, which shows the typical configuration inside an Rb-vapour frequency standard. At its heart is the so-called ‘physics package’ at upper left, which essentially functions as a very high ‘Q’ filter, tuned to the Rb87 ‘hyperfine transition’ frequency Left: the PRS10 Rbvapour frequency standard from Stanford Research Systems. It measures just 50 x 75 x 102mm and can be purchased (new) for about $2200 including freight. 38  Silicon Chip of 6.83468GHz. By having this filter as part of a feedback loop based around the low-noise 10MHz voltage controlled crystal oscillator at lower right, the frequency of the oscillator is ‘disciplined’ to remain at exactly 10MHz ±5 parts in 10-11. As you can see, there are two main components inside the physics package. One is the rubidium discharge lamp at the left end, while the other is the resonance cell and microwave cavity at the right end. Although a third ‘filter cell’ is shown between the two in Fig.2, many of the newer Rb-vapour frequency standards have a simplified configuration where the filter cell is effectively combined with the resonance cell. The discharge lamp on the left is filled with a mixture of Rb-87 enriched vapour and a noble gas such as krypton. This gas mixture is excited by RF energy from an oscillator operating at about 150MHz, via both a pair of electrodes and a coil wound around the lamp. As a result of this excitation, a discharge plasma is established inside the lamp and it emits light in the ruby red part of the spectrum with two peak wavelengths at 795nm and 780nm. These correspond to the valence electrons in excited Rb-87 atoms dropping from an excited energy level to one or the other of the two hyperfine split ground state levels. The light from the discharge lamp siliconchip.com.au then passes through the filter cell, which contains Rb-85 vapour with its atoms mainly in one of their two hyperfine ground states. As a result of the coincidence of the lower hyperfine energy level for Rb-85 with the upper hyperfine energy level for Rb-87, the Rb-85 vapour effectively absorbs the light photons corresponding to the Rb-87 atoms dropping to their upper hyperfine energy level. So the light emerging from the filter cell mainly consists of photons corresponding to the Rb-87 atoms dropping from their excited states to their lower hyperfine energy level. In other words, the filter cell removes most of the 795nm light photons and leaves mainly those with a wavelength of 780nm to pass into the resonance cell. Now the resonance cell is filled with Rb-87 vapour, together with a ‘buffer’ gas. When the atoms of Rb-87 in the cell are in their ground state, there will be exactly the same number of valence electrons occupying each of the two hyperfine energy levels. However, when the 780nm light photons coming from the filter cell pass through the vapour, many of the photons interact with the Rb-87 atoms, which absorb their energy and shift their valence electrons up to an excited energy level. These excited atoms quickly relax again, dropping down to one of the ground state hyperfine levels. The nett result is that a ‘population imbalance’ is created between the two hyperfine energy levels: more electrons will be in the upper hyperfine level than in the lower level. This interaction of the 780nm photons with the electrons in the Rb-87 vapour atoms is known as ‘optical pumping’. Metal cavity As you can see, the resonance cell is housed inside a metal cavity and its dimensions are arranged to make it resonate electromagnetically at 6.834,682,612,8GHz. RF energy is fed into the cavity to excite it at this frequency, using the kind of system shown in Fig.2. A frequency synthesiser driven by the 10MHz crystal oscillator produces an RF output at close to 359.72MHz, and this RF is fed to a step recovery diode inside the resonance cell and cavity. The diode effectively multiplies the 359.72MHz signal by 19 times. So it provides enough energy at the nineteenth harmonic of the 359.72MHz siliconchip.com.au Two views of the Quartzlock E10-MRO Rb-vapour frequency standard. At top is the outside view showing the DB-9 connector (used for most external connections), plus the SMA connector for the 10MHz output at upper right. The lower view shows the inside, with the Rb ‘physics package’ at the rear. signal to excite the resonance cell at (19 x 359.72) = 6834.68MHz. The effect of this electromagnetic energy in the resonance cell is to cause many of the Rb-87 valence electrons to effectively transfer from the upper hyperfine energy level down to the lower level. This in turn makes it easier for the 780nm photons passing through the cell to re-excite the Rb-87 atoms once again, bumping their valence electrons up to an ‘excited’ level so that they can ‘relax’ again and fall back to one of the hyperfine levels. The result of this fairly complex interaction is that when the RF energy fed into the resonance cell is at the exact frequency which corresponds for transfers between the two hyperfine levels of Rb-87, there are fewer of the 780nm photons emerging from the rear of the resonance cell and reaching the photodetector to its right. At frequencies that are either higher or lower than this, more of the photons pass straight through to reach the photodetector. The overall effect of the rubidium physics package is therefore to provide a very accurate or high-Q filter, allowing the frequency of the 10MHz crystal oscillator to be ‘disciplined’ via a control voltage (applied to a varactor diode) to the exact frequency where the RF energy fed to the resonance cell results in maximum absorption of the 780nm photons. Since the 6.834GHz energy is derived from the 10MHz oscillator and hence locked to it, this means that the frequency of the oscillator is held very accurately to 10.000MHz (±5 parts in 10-11). So that’s the basic idea. However you might already be wondering how the frequency control feedback loop in Fig.2 can zero in to the correct frequency, if the RF energy fed into the resonance cell and cavity remains locked to a single frequency. How can it tell when everything is tuned for a dip or notch in the photodetector output? That’s done by introducing a small February 2014  39 Milestones in Atomic Clock Evolution 1944:  The concept of atomic clocks developed by Isidor Rabi of Columbia University (USA). Rabi wins a Nobel Prize. 1948: Harold Lyons and his associates at the US National Bureau of Standards (NBS) achieve the first atomic frequency standard, using the resonance of ammonia at 23.870GHz. 1951:  French physicist Alfred Kastler, working on a combination of optical resonance and magnetic resonance, develops the technique of ‘optical pumping’. This played a key role in the development of masers, lasers and Rubidium frequency standards. Kastler was awarded a Nobel Prize for this work in 1966. 1954:  J. R. Zacharias proposes the idea of an ‘atomic fountain’ clock, although this does not become feasible until 1989. 1955:  Louis Essen and Jack Parry, working at the National Physical Laboratory (NPL) in Teddington (UK), achieve the first working cesium-beam atomic frequency standard. 1956: The first commercial cesium-beam frequency standard, the Atomichron, was developed by Zacharias, Richard Daly and Joseph Holloway at the National Company of Malden, Massachusetts (USA), working together with MIT. Between 1956 and 1960, about 50 Atomichrons were delivered, mainly to US Government agencies. They were very large and bulky devices. 1958:  P. L. Bender, E. C. Beaty and A. R. Chi, working at the US Naval Research Laboratory, develop the concept of using optical detection of narrow Rb-87 hyperfine absorption lines, and also the idea of using Rb-85 vapour to filter out one of the hyperfine lines. Both concepts were the key to producing amount of FM (frequency modulation) into the output of the frequency synthesiser, so it swings cyclically above and below the correct frequency – usually at a low audio rate of about 70Hz. The amplifier following the photodetector is arranged to filter the detector’s output and deliver an output voltage that’s proportional to the second harmonic of the modulating frequency, at 140Hz. It turns out that this second harmonic signal peaks at the exact frequency corresponding to the notch in the photodetector’s DC output. As you can see, most modern Rb-vapour standards use a microcontroller to ensure that the 10MHz oscillator 40  Silicon Chip Rb-vapour frequency standards. 1960:  R. J. Carpenter and his colleagues, and also M. Arditi, developed prototype rubidium-vapour oscillators. 1960-65: Now at Varian Associates in Beverly, Massachusetts, Joseph Holloway worked with Richard Lacey and Norman Ramsey to develop a cesium-beam tube only 16” (406mm) long. This was offered as a component to firms considering the manufacture of compact commercial Csbeam frequency standards or ‘clocks’. Then in 1964, a 12” (305mm) long Cs-beam tube was developed for Hewlett-Packard, to use in their first Cs-beam frequency standard (the HP5060A). This became famous as the ‘flying clock’. 1964:  The first operational Rubidium frequency standard was developed by P. Davidovits and R. Novick. 1967: HP acquires the manufacturing rights for Varian’s cesium-beam tubes. Also in 1967, the 13th General Conference on Weights and Measures defined the second as “the duration of 9,192,631,770 periods of the radiation corresponding to the transition between the two hyperfine levels of the ground state of the Cesium 133 atom”. 1970: Matt Zepler, working for Plessey at Roke Manor in Hampshire (UK), develops an Rb-vapour oscillator that is small enough to fit as a component module in a 2U rack-mounting case. At about the same time, KVARZ, an institute in Gorky (Russia), developed a compact Rb-vapour frequency reference that was small enough to be fitted into mobile and airborne equipment. 1971:  Hugo Fruhof and his team at Efratom Electronik GmbH in Munich (Germany) remains locked. The microcontroller monitors the photodetector output via an ADC (analog-to-digital converter) and applies the phase error correction or ‘disciplining’ voltage to the oscillator’s varactor via a DAC (digital-toanalog converter). The micro is usually programmable in terms of the feedback gain and time constant, etc. In most cases, this can be done from a PC via an RS-232C serial cable. Referring back to Fig.2, you may be wondering about that second DAC controlled by the micro and the magnetic coil it drives – wound around the resonance cell inside the physics package. What is that for? develop the FRK – a very small Rb-vapour oscillator. Fruhof and Efratom then moved to the USA, where they began manufacturing a series of compact Rb-vapour frequency standards – evolving into the ‘FRS’ units which became an industry standard. The FRS units measured only 4” x 3” x 2” (102 x 76 x 51mm). Efratom was acquired by Ball, then Datum and then Symmetricom. 1989:  S. Chu, M. Kasevich and their colleagues at Stanford University in California demonstrate a Cs-fountain technique for measuring atomic resonance. Shortly afterwards, the first primary frequency standard based on this approach was developed by a group at the Laboratoire Primare du Temps et Frequences (LPTF) in France. 1992:  Leon Cutler and Robin Giffard of HP Labs develop the much-improved HP 5071A Cs-beam clock, smaller and lighter than its predecessors but still measuring 524 x 425 x 133mm and weighing 30kg. 2000:  Hewlett-Packard splits into two: HP (computers and printers) and Agilent Technologies (test and measuring instruments). The Precision Time and Frequency division becomes part of Agilent. 2005: Agilent sells its cesium frequency standard line to Symmetricom of San Jose, California. 2012:  Dr Thomas Cao, Clive Green and Dr Cosmo Little, working at Quartzlock (UK) Ltd in Devon, England, develop an ultra-miniature Rb-vapour clock measuring only 51 x 51 x 25mm and weighing less than 150 grams – the E10-MRX. 2013: Symmetricom of San Jose sold to Microsemi Inc., a semiconductor firm in Aliso Viejo (Orange County, Southern California). While the energy difference between the two hyperfine ground state levels of the Rb-87 isotope is very stable at the value corresponding to 6.834,682,612,8GHz, it can vary slightly in response to changes in the ambient magnetic field. As a result, the resonance cavity and the magnetic coil wound around it must be housed in a mu-metal shielding box to minimise external perturbations. The current through the magnetic coil is then adjusted by the micro to bring the apparent Rb-87 hyperfine transition frequency to the correct figure of 6.834GHz. The adjustment range is quite small – about ±2 x 10-9, which is why the function of the siliconchip.com.au magnetic coil is usually described as for ‘fine tuning’. But wait – there’s more! Now while the accuracy and stability of these rubidium-vapour frequency standards is much better than that of a temperature-controlled crystal osc­ illator or ‘TCXO’, they’re still not as good as a cesium-beam standard. That’s why Cs-beam standards are regarded as the primary references for time and frequency, with Rb-vapour standards relegated to secondary status. However, nowadays there is a fairly simple way to achieve greater accuracy from a Rb-vapour frequency standard. This is by using the 1pps pulses from a GPS receiver to ‘discipline’ the Rbvapour standard. This allows an Rbvapour frequency standard to achieve almost exactly the same accuracy and stability as a Cs-beam standard. As you may be aware, the 1pps pulses from a GPS receiver have excellent long-term accuracy and stability, because they are locked to Cs-beam standards in the GPS satellites. However they vary significantly in the short term, due to both propagation changes Two views inside the SRS PRS10 Rb-vapour reference. On the left, only the outer mu-metal shield has been removed, showing two of the four PCBs. On the right, the three upper PCBs have been removed, to show the Rb discharge lamp and microwave absorption cell assembly. and jitter in the GPS receiver circuitry. On the other hand, Rb-vapour standards have excellent short and medium-term stability, with a low ‘ageing’ drift rate (approximately 5 parts in 10-10 per year). So disciplining an Rb-vapour standard with the 1pps pulses from a GPS receiver (over a significant period of time) provides the short and medium-term stability of an Rb-vapour standard with the longterm accuracy of a Cs-beam standard. It also avoids needing to have the Rb-vapour standard recalibrated every couple of years, to compensate for its ageing drift rate. Many modern Rb- vapour frequency standards have the ability to lock to external GPS-derived 1pps pulses, while some have a GPS receiver built in. Either approach allows them to achieve this near Csbeam performance. For example, the Quartzlock E10GPS unit, after being disciplined to GPS 1pps pulses for 24 hours, is specified to have a frequency accuracy of less than one part in 10-12, with a short-term stability of less than three parts in 10-11 over a 1s period, less than one part in 10-11 over a 10s period, or less than three parts in 10-12 over a 100s period. SC That’s very impressive! Proposed Format for KitStop ¼ Page Ad Silicon Chip Magazine February 2014 And now for something completely different . . . Here’s something from the past that you will enjoy far into the future! Radio, TV & Hobbies April 1939-March 1965 Every article to enjoy once again on DVD-ROM This remarkable archival collection spans nearly three decades of Australia’s own Radio & Hobbies and Radio, TV & Hobbies magazines. Every article is scanned into PDF format ready to read and re-read at your leisure on your home computer (obviously, a computer with a DVD-ROM is required, along with Acrobat Reader 6 or later (Acrobat Reader is a free download from Adobe). For history buffs, it’s worth its weight in For anyone with even the vaguest Only available from gold. interest in Australia’s radio and television SILICON CHIP history (and much more) what could be Order now via siliconchip.com.au better? This is one DVD which you must or (02) 9939 3295 have in your collection! ONLY $ 00 62 plus P&P siliconchip.com.au Remote Control Made Easy The KSRC2 UHF set controls appliances, lighting, scoreboards & models over 40metres. Its two independent receiver relay outputs are rated to 500Watts Lots more Controllers on our website Special S.C Project Offer!!! $22.30 inc. GST Plus $7.50 P & P KSDVM-30 ULTRA-COMPACT 4.5V-30VDC Digital Panel Meter Here's a meter that is range-optimized for solar, automotive and trucking applications: Features:Bright 0.36” Red LED Digits, Value Snap-Fit Housing, 2 Wire Installation. inc. GST Plus $3.60 Pack & Post $6.70 Hours of Fun – Our 5 Kit Bonanza Kick 2014 off with this great set: A FK109 2 LED Flasher, FK908 Soil Moisture Indicator FK233 Emergency Vehicle Siren with speaker, FK602 2W Mono Audio Amplifier (Uses the FK233 Siren speaker) plus the FK401 LightActivated Switch. All for $25.50 inc. GST Plus $8.50 Pack & Post On Line at www.kitstop.com.au P.O. Box 5422 Clayton Vic.3168 Tel:0432 502 755 February 2014  41