Most Widely Used Type of Atomic Clock

Most Widely Used Type of Atomic Clock

• Most Widely Used Type of Atomic Clock

• Smallest, Lightest, Lowest Power, Least Complex, Least Expensive, Longest Life, Excellent Performance, Stability & Reliability

• Device of Choice When Better Stability Than a Crystal Oscillator is Needed

• Lower Aging, Lower Temperature Sensitivity, Faster Warm-up, Excellent Retrace, Lower Tip-Over Sensitivity, Lower Radiation Sensitivity

Mature

• Mature

• Basic Principles Little Changed Since Late 1950’s
• Optical Pumping, W Interrogation, Optical Detection, Synchronous Demodulation/Integration

• Highly Refined

• Stabilities Measured in pp1015
• Reduction of Light Shift, TC, Noise, Offsets, etc.

• Modern Techniques

• Digital Control, DDS, DSP, etc.

• Emerging Physics

• Laser Pumping, Cs Vs. Rb, CPT Interrogation

Vigorous Developmental Activity

• Vigorous Developmental Activity

• RFS Technology More Dynamic than for Other Atomic Clock Types
• New Products Driven by Rapid Growth in Highly- Competitive Commercial Market
• That Market is Large Enough to Support Continuing Technological Innovation

Commercial

• Commercial

• Small Size and Low Cost
• Moderate Performance

• Military/Aerospace

• Environmental Hardening
• Full Performance
• Trend Toward COTS and PEMs

• Space

• High Performance
• High Reliability

Small (X72 < 125 cc)

• Small (X72 < 125 cc)

• Inexpensive (< $1.5k)

• Widely Used in Telecom Applications

• Often Combined with a GPS Receiver to Form an Accurate, High Stability Time and Frequency Reference

Small and Environmentally Hardened

• Small and Environmentally Hardened

• Widely Used for Secure Communications

• Classic Designs Used MIL Parts that are Now Hard to Obtain

• Military Programs are Now Using Ruggedized Versions of COTS Units

Military Programs are Also Using Modern Militarized RFS Units Designed Specifically for Rugged Applications that Use Lot-Qualified PEMs ( -40C) and Have Up-to-Date Features Like Digital Control Interfaces

• Military Programs are Also Using Modern Militarized RFS Units Designed Specifically for Rugged Applications that Use Lot-Qualified PEMs ( -40C) and Have Up-to-Date Features Like Digital Control Interfaces

Usually Designed Specifically for Space

• Usually Designed Specifically for Space

• Vacuum Environment an Advantage

• Radiation Hardening

• Control/Monitor Interface Required

• Isolated DC Power Return Usually Needed

• S-Level Parts Usually Required

• Extensive Analysis and Design Review

• Extensive QTP/ATP Testing

Pre-GPS (1)

• Pre-GPS (1)

• NTS-1 (Final Timation S/V)

• GPS ( 200)

• Blocks I, II, IIA, IIR, IIF (Underway) & III (Future)

• Milstar ( 10)

• Original and AEHF (Underway)

• Science ( 5)

• Cassini Huygens Probe

• Other ( 0)

• ETS, Galileo (Underway)

Provide Highest Performance

• Provide Highest Performance

• RAFS is Best GPS Clock where 24-Hour Stability is Critical

• Overcome Crystal Oscillator Limitations

• RFS Has Much Lower Aging
• RFS Has Better Environmental Stability (Temperature, Radiation, etc.)

• Use Proven Technology

• Mature, Well-Understood, Reliable

Digital Synthesis

• Digital Synthesis

• DDS for Tuning and Modulation
• Better Performance & Flexibility

• Digital Control

• DSP Servo Loop
• Eliminate Analog Errors

• Laser Pumping

• Potential for Higher S/N and Stability
• Diode Laser Noise & Reliability are Issues
• No Laser-Pumped Products on Market Yet

Early Units (e.g. HP 5065) Emphasized Performance as Laboratory Standards

• Early Units (e.g. HP 5065) Emphasized Performance as Laboratory Standards

• Integrated Cell Approach (e.g. Efratom FRS) Began New Evolutionary Path Toward Smaller, Cheaper, Lower Performance Units for Widespread Use

• GPS Satellite Clock Requirements Led to Highest Performance Space Rb Clocks (e.g PerkinElmer RAFS)

Early RFS designs were complex instruments with large physics packages, complex electronics, front panel controls, ovenized crystal oscillators. They generally had performance better than today’s commercial units, were made by the 100’s, and cost about the same as a family car.

• Early RFS designs were complex instruments with large physics packages, complex electronics, front panel controls, ovenized crystal oscillators. They generally had performance better than today’s commercial units, were made by the 100’s, and cost about the same as a family car.

Today’s commercial RFS units are much simpler to build (thanks to more sophisticated designs and better electronic parts). They are plain boxes with enhanced functionality (e.g. GPS syntonization) and performance tailored to the application. They are made by the 10,000’s and cost only about 5% of a family car.

• Today’s commercial RFS units are much simpler to build (thanks to more sophisticated designs and better electronic parts). They are plain boxes with enhanced functionality (e.g. GPS syntonization) and performance tailored to the application. They are made by the 10,000’s and cost only about 5% of a family car.

The design of military/avionics RFS units has evolved less, mainly because the environmental conditions remain harsh. The trend is to leverage the economies of scale of commercial units (COTS, PEMs) to the greatest extent possible. Typical quantities are 100’s with prices about 25% of a family car.

• The design of military/avionics RFS units has evolved less, mainly because the environmental conditions remain harsh. The trend is to leverage the economies of scale of commercial units (COTS, PEMs) to the greatest extent possible. Typical quantities are 100’s with prices about 25% of a family car.

Rb space clocks have evolved least of all from the original designs, but many small refinements have resulted in remarkable performance (pp1014 medium-term stability and daily aging rates). Each unit is individually crafted and tested at a cost equivalent to 10 or more family cars.

• Rb space clocks have evolved least of all from the original designs, but many small refinements have resulted in remarkable performance (pp1014 medium-term stability and daily aging rates). Each unit is individually crafted and tested at a cost equivalent to 10 or more family cars.

Integrated Resonance Cell

• Integrated Resonance Cell

• Discrete -> Integrated Isotopic Filter

• Smaller Microwave Cavity

• TE011 -> TE111 -> Loaded, etc. -> Jin Resonator

• Smaller, Hotter Cells

• Inch -> cm -> mm Cell Dimensions
• +65 -> +80 -> +95 Oven Setpoints

• Optical Filter

• Laser Pumping (Future)

The Rb-85 isotopic filter can be a separate cell, or combined with the Rb-87 absorption cell (likely using natural Rb which contains both isotopes).

• The Rb-85 isotopic filter can be a separate cell, or combined with the Rb-87 absorption cell (likely using natural Rb which contains both isotopes).

• Most RFS designs use the integrated approach because it is smaller and cheaper, and works well. Light shift can nulled by adjusting the isotopic ratio in the lamp. Temperature coefficient can be nulled by adjusting the buffer gas mixture in the cell.

The main disadvantage of the integrated cell approach is high RF power shift caused by inhomogenity in the cell (the degree of isotopic filtering varies as the light passes through the cell, and the position of optimum signal varies with the RF power).

• The main disadvantage of the integrated cell approach is high RF power shift caused by inhomogenity in the cell (the degree of isotopic filtering varies as the light passes through the cell, and the position of optimum signal varies with the RF power).

• The separate filter cell allows light shift to be nulled by adjusting its length and/or temperature before the light enters the absorption cell, thus reducing the inhomogenity and RF power sensitivity.

The absorption cell TC can be nulled with the buffer gas mixture, but the filter cell TC cannot.

• The absorption cell TC can be nulled with the buffer gas mixture, but the filter cell TC cannot.

• But, if the separate filter and absorption cells are placed in the same oven, their net TC can be nulled, thereby allowing an overall optimization (zero light shift, zero TC, low RF power sensitivity).

Classic GR TE011 ( 80 cm3)

• Classic GR TE011 ( 80 cm3)

• Classic EG&G TE111 ( 17 cm3)

• FEI TE101 Rectangular, Dielectrically Loaded ( 5 cm3)

• Temex Magnetron (L-C)

Northrop-Grumman TE201 Rectangular, Loaded, Scaled to Rb ( 3 cm3)

• Northrop-Grumman TE201 Rectangular, Loaded, Scaled to Rb ( 3 cm3)

• Jin Resonator (1 cm3) See U.S. Patent No. 6,133,800 (2000)

A closely related cavity

• A closely related cavity

• design was that of the

• highly innovative Circa

• 1995 Northrop-Grumman

• miniature Cs resonator.

• It used a 1.2 cm3 highly

• dielectrically loaded

• TE201 rectangular 9.2 GHz

• cavity in a 1.6 cm3 overall

• resonator assembly with

• a 0.1 cm3 cell. This

• scales to a 3.0 cm3 cavity

• for Rb at 6.8 GHz.

The cells in the latest commercial RFS designs have (along with their cavities) gotten much smaller. While this comes at the expense of a broader line, lower Q and poorer short-term stability, good performance e.g. y()= 1x10-11 at 1 second can still be realized.

• The cells in the latest commercial RFS designs have (along with their cavities) gotten much smaller. While this comes at the expense of a broader line, lower Q and poorer short-term stability, good performance e.g. y()= 1x10-11 at 1 second can still be realized.

The general trend has been toward higher Rb cell operating temperatures.

• The general trend has been toward higher Rb cell operating temperatures.

• Higher cell operating temperature is associated mainly with the need to maintain the same optical absorption with a shorter cell length.

• The main advantage of the higher oven setpoint is the ability to operate at higher ambient/baseplate temperatures (e.g. +85C for the Symmetricom X72).

• Disadvantages include broader linewidth (lower Q and stability) higher power, longer warm-up time, and reduced reliability.

Operation to high RFS baseplate temperatures was a difficult aspect of earlier MIL units. Various means were used to depress the Rb vapor pressure (natural Rb, K mixtures) or to actively cool the physics package.

• Operation to high RFS baseplate temperatures was a difficult aspect of earlier MIL units. Various means were used to depress the Rb vapor pressure (natural Rb, K mixtures) or to actively cool the physics package.

• The highest-performance GPS RAFS units use large cells operating at relatively low (+65C) temperature to achieve the best stability. Fortunately this is consistent with the S/V thermal conditions.

Complex Synthesis (Low Buffer Gas Offset)

• Complex Synthesis (Low Buffer Gas Offset)

• Simpler Analog Synthesis (e.g. 5.3125 MHz)

• DDS Synthesis

• Flexible
• Relaxed Cell Tolerance
• High-Resolution Digital Tuning
• Improved Servo Modulation

Early RFS designs used complex synthesis schemes to achieve good microwave purity (low sideband pulling) and low buffer gas offset (low barometric sensitivity and comfortable fill tolerance). Essentially all RFS RF chains use a high-ratio step recovery diode (SRD) microwave multiplier because of its simplicity and the relatively low W power required. In many cases, the SRD multiplier is also used as a mixer.

• Early RFS designs used complex synthesis schemes to achieve good microwave purity (low sideband pulling) and low buffer gas offset (low barometric sensitivity and comfortable fill tolerance). Essentially all RFS RF chains use a high-ratio step recovery diode (SRD) microwave multiplier because of its simplicity and the relatively low W power required. In many cases, the SRD multiplier is also used as a mixer.

Experience showed that poorer purity and larger offset was acceptable, which allowed simpler synthesis schemes. A popular one only involved division by 16. The resulting 4.6 kHz offset was handled in some cases by measuring the cell frequency as it is filled.

• Experience showed that poorer purity and larger offset was acceptable, which allowed simpler synthesis schemes. A popular one only involved division by 16. The resulting 4.6 kHz offset was handled in some cases by measuring the cell frequency as it is filled.

• Many other analog synthesis schemes have been used, often involving PLLs. An important consideration is how the servo modulation is applied (RC phase modulator, into PLL phase detector, direct VCO FM, etc.). In most cases, the modulation must be kept off the output signal.

The advent of the 1-chip direct digital synthe-sizer (DDS) has greatly simplified RFS RF chains, making the cell offset a free variable, removing its tight fill tolerances, and offering an ideal way to apply servo modulation. As long as the DDS signal is inserted additively, 32 bit resolution is adequate for fine tuning. This has the additional advantage that the C-field can be held at a fixed minimum value for low magnetic sensitivity.

• The advent of the 1-chip direct digital synthe-sizer (DDS) has greatly simplified RFS RF chains, making the cell offset a free variable, removing its tight fill tolerances, and offering an ideal way to apply servo modulation. As long as the DDS signal is inserted additively, 32 bit resolution is adequate for fine tuning. This has the additional advantage that the C-field can be held at a fixed minimum value for low magnetic sensitivity.

The latest generation of commercial RFS RF chains use synthesis devices intended for the wireless industry. For example, IC PLL, quadrature modulator, and VCO chips can operate at 2.3 GHz (1/3 of the Rb frequency), thereby eliminating the high-ratio SRD multiplier (a simple diode tripler can be used instead).

• The latest generation of commercial RFS RF chains use synthesis devices intended for the wireless industry. For example, IC PLL, quadrature modulator, and VCO chips can operate at 2.3 GHz (1/3 of the Rb frequency), thereby eliminating the high-ratio SRD multiplier (a simple diode tripler can be used instead).

The need for frequency multiplication can be entirely eliminated by new high-speed digital dividers. This makes it possible to obtain a very pure W interrogation signal essentially free from pulling by sub-harmonics and mixing products. This technique is already being used in high-performance atomic clock designs.

• The need for frequency multiplication can be entirely eliminated by new high-speed digital dividers. This makes it possible to obtain a very pure W interrogation signal essentially free from pulling by sub-harmonics and mixing products. This technique is already being used in high-performance atomic clock designs.

• All RFS RF chains must carefully consider the effect of phase noise on the interrogation signal. Noise at twice the servo modulation frequency is particularly important.

The majority of earlier-generation RFS RF chains generated the Rb hyperfine frequency of  6834.682611 MHz by multiplying a 5 or 10 MHz crystal oscillator to 6840 MHz and subtracting an offset of  5.3 MHz in a mixer. Often the microwave portion used an SRD multiplier as a combined multiplier and mixer.

• The majority of earlier-generation RFS RF chains generated the Rb hyperfine frequency of  6834.682611 MHz by multiplying a 5 or 10 MHz crystal oscillator to 6840 MHz and subtracting an offset of  5.3 MHz in a mixer. Often the microwave portion used an SRD multiplier as a combined multiplier and mixer.

• Early RFS designs synthesized the 5.3 MHz offset to high precision to accommodate a low absorption cell buffer gas offset. This synthesis was either fixed (e.g. AFS-81) or variable (e.g. HP5065). The latter added considerable complexity, but provided more convenient tuning, reduced the required C-field range, and eased cell fill tolerances.

Many later RFS designs (e.g. Efratom FRK, M-100, etc.) used a simple fixed synthesizer wherein 5.3125 MHz was generated by dividing 5 MHz by 16 to 312.5 kHz and adding it to 5 MHz. This approach has a large buffer gas offset (4.9 kHz), and often had poor W spectral purity but, in practice, worked quite well.

• Many later RFS designs (e.g. Efratom FRK, M-100, etc.) used a simple fixed synthesizer wherein 5.3125 MHz was generated by dividing 5 MHz by 16 to 312.5 kHz and adding it to 5 MHz. This approach has a large buffer gas offset (4.9 kHz), and often had poor W spectral purity but, in practice, worked quite well.

Better W spectral purity can be obtained by straight multiplication of an exact submultiple of the Rb frequency, usually from a phase-locked crystal oscillator in the 100 MHz region (e.g. The EG&G RFS-10 used 76 x 89.93007 MHz obtained as 90 MHz – 10 MHz/143). It is hard to make the offset variable because its steps are multiplied, reducing the resolution.

• Better W spectral purity can be obtained by straight multiplication of an exact submultiple of the Rb frequency, usually from a phase-locked crystal oscillator in the 100 MHz region (e.g. The EG&G RFS-10 used 76 x 89.93007 MHz obtained as 90 MHz – 10 MHz/143). It is hard to make the offset variable because its steps are multiplied, reducing the resolution.

The cleanest approach is to use a “natural frequency” configuration without any synthesis. This is rarely done because most systems require a standard (e.g. 10 MHz) reference. But a noteworthy exception is the PerkinElmer GPS Block IIR RAFS, which uses 13.40134393 MHz with a straight x510 (3·2·85) multiplier chain.

• The cleanest approach is to use a “natural frequency” configuration without any synthesis. This is rarely done because most systems require a standard (e.g. 10 MHz) reference. But a noteworthy exception is the PerkinElmer GPS Block IIR RAFS, which uses 13.40134393 MHz with a straight x510 (3·2·85) multiplier chain.

The 1-chip DDS has made variable RFS RF chain synthesis practical. For example, the Symmetricom 8130A uses the classic  5.3125 MHz final mix, but generates it with a 32-bit DDS (which also applies squarewave FM for the servo). A tuning resolution of  3.4x10-13 is obtained by separately adjusting both FM frequencies.

• The 1-chip DDS has made variable RFS RF chain synthesis practical. For example, the Symmetricom 8130A uses the classic  5.3125 MHz final mix, but generates it with a 32-bit DDS (which also applies squarewave FM for the servo). A tuning resolution of  3.4x10-13 is obtained by separately adjusting both FM frequencies.

Other modern RFS RF chain configurations can eliminate the high-ratio SRD multiplier by using new RF ICs and high-speed digital devices. The Symmetricom X72 uses a  2 GHz PLL chip, IC W VCO, SSB mixer, DDS, and diode tripler to generate its Rb interrogation signal. The latest Symmetricom Cs synthesizers use  9 GHz GaAs dividers with a phase-locked DRO, SSB mixer, and DDS to generate a very pure and stable signal.

• Other modern RFS RF chain configurations can eliminate the high-ratio SRD multiplier by using new RF ICs and high-speed digital devices. The Symmetricom X72 uses a  2 GHz PLL chip, IC W VCO, SSB mixer, DDS, and diode tripler to generate its Rb interrogation signal. The latest Symmetricom Cs synthesizers use  9 GHz GaAs dividers with a phase-locked DRO, SSB mixer, and DDS to generate a very pure and stable signal.

Analog Synchronous Detection (Circa 1960’s)

• Analog Synchronous Detection (Circa 1960’s)

• Analog with Digital Control/Filtration

• Analog In/Out with Numeric Processing (e.g. Integrate & Dump, P, DAC, VCXO) Common Today

• Fully Digital (ADC, DSP, NCO) Possible Today

The function of the RFS servo is to process the discriminator signal from the physics package to produce a frequency control signal for the overall frequency lock loop. This has traditionally been done with an analog synchronous detector and integrator, and has been refined to a point of great perfection. It generally includes 2nd harmonic lock detector and acquisition sweep circuits.

• The function of the RFS servo is to process the discriminator signal from the physics package to produce a frequency control signal for the overall frequency lock loop. This has traditionally been done with an analog synchronous detector and integrator, and has been refined to a point of great perfection. It generally includes 2nd harmonic lock detector and acquisition sweep circuits.

The analog servo is gradually evolving toward a digital embodiment, which offers a simpler implementation and eliminates analog errors.

• The analog servo is gradually evolving toward a digital embodiment, which offers a simpler implementation and eliminates analog errors.

• A first step in that direction is to use an integrate and dump circuit at the front end, digitize the resulting discriminator information, process it numerically, and convert the result back to an analog VCXO control voltage.

An important consideration is the need for a high-resolution (e.g. 20 to 24 bit) DAC.

• An important consideration is the need for a high-resolution (e.g. 20 to 24 bit) DAC.

• It is best applied for relatively slow modulation rates where there is no large 2nd harmonic content and the transients can be suppressed by the reset interval.

• That approach is used successfully in the Symmetricom 4410 GPS Block IIF Digital CAFS.

A further step toward RFS servo digitalization is to directly sample the discriminator signal, process it completely numerically, and convert the result to an analog VCXO control voltage. This approach is successfully used in the Symmetricom X72.

• A further step toward RFS servo digitalization is to directly sample the discriminator signal, process it completely numerically, and convert the result to an analog VCXO control voltage. This approach is successfully used in the Symmetricom X72.

An all-digital servo would apply the digital frequency control information directly to a numerically-controlled oscillator (NCO) without D/A conversion. That approach is under consideration for the NRL Advanced Technology rubidium clock.

• An all-digital servo would apply the digital frequency control information directly to a numerically-controlled oscillator (NCO) without D/A conversion. That approach is under consideration for the NRL Advanced Technology rubidium clock.

Most classic RFS designs use DC analog oven temperature controllers comprised of thermistor bridges, DC amplifiers, and heaters. The heaters have evolved from bifilar windings to Kapton foil layer(s) and transistors. Some designs have used power switching circuits for high efficiency and wide dynamic range, but noise can be a problem. Transistor heaters are efficient but can have residual magnetic field problems. Thermal insulation has evolved away from foam to G-10 and small air gaps as ovens have gotten smaller.

• Most classic RFS designs use DC analog oven temperature controllers comprised of thermistor bridges, DC amplifiers, and heaters. The heaters have evolved from bifilar windings to Kapton foil layer(s) and transistors. Some designs have used power switching circuits for high efficiency and wide dynamic range, but noise can be a problem. Transistor heaters are efficient but can have residual magnetic field problems. Thermal insulation has evolved away from foam to G-10 and small air gaps as ovens have gotten smaller.

The best classic designs used fairly large, high thermal mass ovens with low loss thermal insulation and a tightly-coupled thermistor sensor. They supported high temperature stabilization factors with simple DC proportional controllers with the dissipative device external from the oven. The bifilar wound or double-layer meander foil heater had very low residual magnetic field, and was within the C-field coil and inner magnetic shielding. The PerkinElmer GPS RAFS is a highly-refined example of this approach.

• The best classic designs used fairly large, high thermal mass ovens with low loss thermal insulation and a tightly-coupled thermistor sensor. They supported high temperature stabilization factors with simple DC proportional controllers with the dissipative device external from the oven. The bifilar wound or double-layer meander foil heater had very low residual magnetic field, and was within the C-field coil and inner magnetic shielding. The PerkinElmer GPS RAFS is a highly-refined example of this approach.

Later RFS designs have smaller, lower thermal mass ovens, little thermal insulation, and generally use transistor heaters. The C-field coil and inner magnetic shield may be inside the oven to reduce the residual magnetic field of the transistor heater.

• Later RFS designs have smaller, lower thermal mass ovens, little thermal insulation, and generally use transistor heaters. The C-field coil and inner magnetic shield may be inside the oven to reduce the residual magnetic field of the transistor heater.

Some current RFS designs use digital oven controllers, converting a thermistor bridge output to digital form, implementing the control algorithm numerically, and reconverting to analog form for the oven heater. Advantages include the ability to adjust for different supply voltages, warm-up times and soak temperatures, ease in implementing PID control algorithms, and elimination of analog errors.

• Some current RFS designs use digital oven controllers, converting a thermistor bridge output to digital form, implementing the control algorithm numerically, and reconverting to analog form for the oven heater. Advantages include the ability to adjust for different supply voltages, warm-up times and soak temperatures, ease in implementing PID control algorithms, and elimination of analog errors.

Warm-up time is one of the parameters that distinguished military RFS units. It is not sufficient to simply raise the oven demand power. The heat must get into the cell, and that requires specific (and expensive) design details (thermal bonding, etc.). Extremely fast RFS lock-up time (< 30 sec) has been obtained, but not with the current COTS-based generation of units.

• Warm-up time is one of the parameters that distinguished military RFS units. It is not sufficient to simply raise the oven demand power. The heat must get into the cell, and that requires specific (and expensive) design details (thermal bonding, etc.). Extremely fast RFS lock-up time (< 30 sec) has been obtained, but not with the current COTS-based generation of units.

Magnetic shielding materials and geometries have changed little over time.

• Magnetic shielding materials and geometries have changed little over time.

• Newer units can operate at fixed, low C-field since their frequency adjustments are made by a DDS. This greatly reduces their magnetic sensitivity. The C-field can be boosted during lock acquisition to avoid Zeeman lockup.

• Periodic reversal of the C-field to eliminate 1st order magnetic sensitivity is hard to implement because of switching transients.

Zeeman state interrogation to measure and control the C-field could be applied to advanced RFS designs to improve their stability. This requires microprocessor control and an agile RF chain.

• Zeeman state interrogation to measure and control the C-field could be applied to advanced RFS designs to improve their stability. This requires microprocessor control and an agile RF chain.

There is a fairly wide range of RFS stability available, depending on the class of unit and its operating environment

• There is a fairly wide range of RFS stability available, depending on the class of unit and its operating environment

• The typical level of RFS performance is a short-term stability of 1x10-11-1/2 for 1  103 seconds, with a floor of 2x10-13 depending on the thermal and barometric environment, an aging of 1x10-11 per month, and an overall temperature instability of 2x10-10. These are essentially unchanged from previous product generations.

The smaller commercial units tend to have somewhat poorer specified performance, but this is largely due to wider manufacturing margins.

• The smaller commercial units tend to have somewhat poorer specified performance, but this is largely due to wider manufacturing margins.

• The actual performance of RFS units in a military environment is determined mainly by the thermal and vibrational operating conditions.

• The stability of the latest generation of GPS Rb clocks is more than an order of magnitude better than ordinary RFS units.

The performance of the early large laboratory RFS units was remarkable good, even by today’s standards.

• The performance of the early large laboratory RFS units was remarkable good, even by today’s standards.

• The performance of the miniature (e.g. Efratom FRK) units that followed was (and still is) lower, but sufficient for most applications, particularly considering their practical advantages of size and cost.

• The military RFS units of that era had comparable performance, but delivered it under harsh environmental conditions (including radiation).

The current generation of commercial RFS units continues this trend toward somewhat lower performance with much smaller size and lower cost – an attractive compromise for most applications.

• The current generation of commercial RFS units continues this trend toward somewhat lower performance with much smaller size and lower cost – an attractive compromise for most applications.

• The current generation of GPS Rb clocks is based a much-refined version of the original RFS instruments with very high performance, and a size and price to match.

Classic Units Had Analog C-Field Tuning, Analog Monitors, and Lock Status Indicator

• Classic Units Had Analog C-Field Tuning, Analog Monitors, and Lock Status Indicator

• Frequency Adjustments Were Usually Made With C-Field Potentiometer

• Some GPS Clocks Had Digital C-Field Tuning Interface

• Most New RFS Units Have Digital Control and Monitoring Interface (e.g. RS-232)

• Preferred Tuning Method is Now by High Resolution DDS via Digital Interface

The operating environment for commercial RFS units is generally benign. The main issue is the upper temperature limit, for which the latest designs (e.g. X72) are excellent.

• The operating environment for commercial RFS units is generally benign. The main issue is the upper temperature limit, for which the latest designs (e.g. X72) are excellent.

• The application of commercial units in military environments can pose problems. Classic MIL-Spec RFS units were designed and qualified for harsh conditions. COTS units may need ruggedization to be successfully applied there. The most common problems are vibration and humidity/moisture. PEMs are limited to  -40C.

Phase noise and spectral purity remains a problem under vibration. Crystal g-sensitivity is always an issue. RFS physics packages can be (and were) designed to be highly vibration-immune (e.g. rigid optical path), but COTS units are often not. Nevertheless, commercial RFS units can be ruggedized to remain locked under severe vibration.

• Phase noise and spectral purity remains a problem under vibration. Crystal g-sensitivity is always an issue. RFS physics packages can be (and were) designed to be highly vibration-immune (e.g. rigid optical path), but COTS units are often not. Nevertheless, commercial RFS units can be ruggedized to remain locked under severe vibration.

• Radiation remains an important environmental factor in space.

Temperature instability is often the most important error term for an RFS. A few degrees of temperature change can equal a month’s aging.

• Temperature instability is often the most important error term for an RFS. A few degrees of temperature change can equal a month’s aging.

• The TC of early large RFS units was excellent (e.g. < 3x10-11 from –55C to +65 C the AFS-81).

• It became larger for the next generation of miniature commercial and military unity (e.g. FRK and M-100), which had typical excursions of 3x10-10 over their full temperature range.

Even worse, RFS TC often wasn’t very smooth or consistent. Several physics package and electronic factors contributed to the TC with varying signs and magnitudes.

• Even worse, RFS TC often wasn’t very smooth or consistent. Several physics package and electronic factors contributed to the TC with varying signs and magnitudes.

• The early GPS Rb clocks had TCs that severely limited their performance, even under small orbital temperature excursions. This was greatly improved by the use of baseplate temperature controllers (BTC).

The latest generation of commercial and military RFS units have somewhat better and more consistent TC, largely because of electronic improvements. For example, servo modulation distortion is eliminated as a contributor by using a DDS instead of analog means to apply the FM.

• The latest generation of commercial and military RFS units have somewhat better and more consistent TC, largely because of electronic improvements. For example, servo modulation distortion is eliminated as a contributor by using a DDS instead of analog means to apply the FM.

• The latest generation of GPS Rb clocks has exceptionally low TC. Not only is it inherently low ( 2x10-13/C), but their integral baseplate temperature controller (BTC) literally drives the TC into the noise.

Some new commercial and military RFS designs (e.g. X72 and 8130A) have provisions for internal digital temperature compensation. This, by storing a table of frequency corrections at many temperatures, or by curve fitting, can provide an improvement factor of 3:1 to 5:1, depending on the dynamics.

• Some new commercial and military RFS designs (e.g. X72 and 8130A) have provisions for internal digital temperature compensation. This, by storing a table of frequency corrections at many temperatures, or by curve fitting, can provide an improvement factor of 3:1 to 5:1, depending on the dynamics.

Temperature Compensation Requires No Additional Hardware in Some Modern Designs.

• Temperature Compensation Requires No Additional Hardware in Some Modern Designs.

• It Does Require More Test Effort to Measure Uncompensated TC, Calculate & Load Compen-sation Data, and Confirm Compensated TC.

• Dynamics Limit Amount of Improvement – Temperature Sensor Location and Response Time Are Factors.

• Compensation May Degrade Short-Term Stability Depending on Tuning Resolution and Compensation Algorithm.

The aging of rubidium frequency standards has, in general, not changed dramatically over time.

• The aging of rubidium frequency standards has, in general, not changed dramatically over time.

• For most well-stabilized units, the typical aging is 1x10-11/month, 1x10-10/year, and 1x10-9 forever.

• The PerkinElmer GPS RAFS is an exception: all units have shown a smooth negative aging of a few pp1014/day.

• Those units have large cells made with extremely clean processing, highly refined electronics, and individual adjustments to optimize their operating conditions.

The aging of most RFS units is not as consistent in magnitude or direction because of multiple effects, including contributions by their electronics.

• The aging of most RFS units is not as consistent in magnitude or direction because of multiple effects, including contributions by their electronics.

• The underlying cause of aging in an RFS physics package, when light shift is adequately nulled, is probably physical/chemical loss of N2 buffer gas into the Rb film and/or glass envelope of the absorption/resonance cell. RAFS data shows a good fit to a diffusion model.

• This mechanism, if understood, could likely be reduced, but little research has been done.

The fundamental limits of Rb gas cell aging have not been investigated because (a) RFS aging is good enough for most applications (e.g. much lower than a crystal oscillator, OK for telecom holdover, and corrected for by the GPS system), (b) other significant aging factors are also usually involved, and (c) this research would require an extensive, long-term effort.

• The fundamental limits of Rb gas cell aging have not been investigated because (a) RFS aging is good enough for most applications (e.g. much lower than a crystal oscillator, OK for telecom holdover, and corrected for by the GPS system), (b) other significant aging factors are also usually involved, and (c) this research would require an extensive, long-term effort.

• This issue may become more important as cells become smaller and surface-to-volume ratios became less favorable.

Internal Provisions for Synchronization and Syntonization to External 1 PPS (GPS) Reference.

• Internal Provisions for Synchronization and Syntonization to External 1 PPS (GPS) Reference.

• Internal Phase-Locked Crystal Oscillator to Provide Low Phase Noise Output

The life of a modern RFS is essentially unlimited a practical sense. The MTBF is determined mainly by electronic failures. There is no life-limiting wear out mechanism. The maintained service life of an RFS is usually determined by the obsolescence of its application.

• The life of a modern RFS is essentially unlimited a practical sense. The MTBF is determined mainly by electronic failures. There is no life-limiting wear out mechanism. The maintained service life of an RFS is usually determined by the obsolescence of its application.

• The breakthrough for RFS life came in the early 1980’s by T. Lynch’s idea at EG&G to apply calorimetry to the Rb lamp fill process.

• Wider VCXO tuning range has made periodic crystal oscillator trimming unnecessary.

• There is quite an active surplus market for used RFS units. 30-year + lifetimes are not unusual.

Primarily Associated with Electronics

• Primarily Associated with Electronics

• Total Gamma Dose

• Analog (e.g. Offset) Errors

• Neutrons

• Linear IC, Transistor & Photodiode Gain Loss

• Transient Effects

• Phase Error (RF O/P Flywheeling Required)
• Photocurrent Burnout (Current Limiting Needed)

• Less Emphasis Recently

• Still Needed for Space

Commercial RFS units have gotten smaller and cheaper, with only modest reductions in performance.

• Commercial RFS units have gotten smaller and cheaper, with only modest reductions in performance.

• Military RFS requirements are largely being met by ruggedized versions of commercial designs.

• Space RFS units have achieved extraordinary levels of performance and reliability.

As Yogi Berra said, “it’s tough to make predictions, especially about the future”. Here goes anyway…

• As Yogi Berra said, “it’s tough to make predictions, especially about the future”. Here goes anyway…

• While one can expect to see gradual improvements in commercial rubidium frequency standards, the recent wave of reductions in size and cost will not continue as rapidly. They were driven by a huge expansion of telecom usage, and that is likely to slow. In addition, further size reductions will begin to seriously impact their traditional performance .

It is likely that military rubidium frequency standards will continue their present trend toward the use of COTS to the greatest extent possible. The days of the full-up MIL-spec (M-3000 class) RFS are largely over. Application-specific military RFS units will continue to be needed, but they will leverage ruggedized commercial technology.

• It is likely that military rubidium frequency standards will continue their present trend toward the use of COTS to the greatest extent possible. The days of the full-up MIL-spec (M-3000 class) RFS are largely over. Application-specific military RFS units will continue to be needed, but they will leverage ruggedized commercial technology.

Further improvements in the performance of high-end rubidium frequency standards (e.g. GPS Rb clocks) are possible, but it will require a heavy investment in time and money to implement them.

• Further improvements in the performance of high-end rubidium frequency standards (e.g. GPS Rb clocks) are possible, but it will require a heavy investment in time and money to implement them.

• Coherent Population Trapping (CPT) offers attractive new opportunities for next-generation gas cell clocks.

• It is possible that the next breakthrough in gas cell atomic frequency standards will come at the “chip scale”, opening up entirely new applications for precise timing, especially in portable equipment.

General

• General

• C. Audoin and B. Guinot, The Measurement of Time – Time, Frequency and the Atomic Clock, Cambridge University Press, Cambridge, UK, ISBN 0-521-00397-0, 2001.

Rubidium Frequency Standards

• Rubidium Frequency Standards

• J. Vanier and C. Audoin, The Quantum Physics of Atomic Frequency Standards, Adam Hilger, Bristol and Philadelphia, Volume 2, Chapter 7, ISBN 0-85274-433-1, 1989.
• W. J. Riley, “The Physics of the Environmental Sensitivity of Rubidium Gas Cell Atomic Frequency Standards”, IEEE Transactions on Ultrasonics, Ferroelectrics and Frequency Control, Vol. 39, No. 2, pp. 232-240, March 1992.

Time & Frequency Metrology

• Time & Frequency Metrology

• D.B Sullivan, D.W Allan, D.A. Howe and F.L.Walls (Editors), "Characterization of Clocks and Oscillators", NIST Technical Note 1337, U.S. Department of Commerce, National Institute of Standards and Technology, March 1990.