The Continued Pursuit of Precision in Oscillator Technology
By David Meaney, Engineering Sales Manager, Fox Electronics
Oscillators would only be worth the value of the quartz and conductor plates they contain if not for the piezoelectric effect, first discovered over 130 years ago by brothers Jacques and Pierre Curie. The two are credited with being the first to detect the generation of a charge on a crystal that was in direct proportion to an applied mechanical stress. And, soon after, they observed the inverse effect, which involved applying voltage to the crystal yielding a proportional geometrical strain.
But, we’ll need to fast-forward nearly two generations before we see the development of the first practical application of the piezoelectric effect. In 1917, using a piezoelectric quartz element sandwiched between two steel plates, French physicist Paul Langevin made a transducer that could emit high frequency waves and then detect the reflected echo.
The application to which this quartz element was applied — Sound Navigation and Ranging or SONAR — resulted in an improved method for submarine ultrasonic echo detection. A timely discovery considering the preponderance of enemy submarines that plagued the Allies throughout World War I.
By 1926, quartz crystals were being used to control the frequency of radio broadcasting stations. And, in 1928, the first quartz crystal clock was developed by Bell Laboratories that demonstrated accuracies up to 1 second in 30 years! So began a string of countless applications of quartz crystal oscillators that have made — and continue to make — significant contributions to mankind.
Rapid Adoption Ensued
A significant property of quartz crystal oscillators is that they exhibit very low phase noise, allowing them to produce a signal that is pure tone. This attribute makes them particularly useful in telecommunications — which require stable signals — and in scientific equipment that require very precise timing references.
Other positive attributes include their high Q and temperature-stable properties that have contributed to the integration of quartz crystals as a core component in nearly every electronic device manufactured today.
And the Challenges Begin
In the beginning, there were conventional oscillators, which deliver a frequency determined almost entirely by the properties of the crystal. On the plus side, these make for simple and very cost-effective oscillators with good noise characteristics. A big minus however, is the relatively long production and delivery times they require. It can take up to eight weeks to fabricate the unique crystal that delivers the desired resonant frequency.
As a result, conventional oscillators are used only for custom-frequency applications where minimizing noise is paramount, long-range planning cycles are the norm, and there is a need for high-volume. Unfortunately, most if not all, of these conditions are less frequently encountered in today’s rapidly progressing and highly demanding frequency control market.
So, in the early 1990s, designers met this challenge head-on with programmable oscillators.
These employ a programmable phase-locked loop (PLL), post divider, output MUX and output buffer that endow the crystal oscillator with the capability to generate any required frequency — within reasonable technological constraints.
No longer requiring a unique crystal with a unique resonant frequency, production lead times were reduced significantly. But, due to those aforementioned “technological constraints,” programmable oscillators experienced what oftentimes were unacceptable inherent noise flaws. They also required higher production costs – due to larger die sizes – and they ended up serving a smaller market than first projected, yielding higher unit costs than conventional oscillators. All deal breakers for many applications.
Another Day at the Drawing Board
To eliminate the noise flaws, get production costs under control and still maintain mandatory quick deliveries, design engineers developed another new oscillator format — the configurable oscillator.
By employing yet another new innovation — a building-block fabrication process — a product was developed that maintained the desirable short delivery times, while providing control over the noise issues prevalent with the programmable oscillators.
The configurable oscillator starts out as a conventional oscillator, same as the programmable.
But the similarities end there. Through the use of the building-block modular approach, the designer is afforded the desirous advantage of being able to choose which crystal blank, fractional N PLL, Delta Sigma Modulator and output buffer will collectively yield the noise requirements, frequency and output each specific application demands.
By merging the best high Q quartz reference source with very sophisticated ASICs in a single package, configurable oscillators are capable of delivering the ideal custom engineered frequency in a matter of days. This greater timing flexibility gives designers the safety net of still getting on-time delivery, even if they submit last minute changes in specifications or requirements.
The ultimate outcome is an oscillator format that operates at any desired frequency, exhibits noise characteristics on par with conventional oscillators, and will satisfy the need for quick delivery. And, even more, due to the small die sizes used in production combined with the use of common frequencies, configurable oscillators have price points that are even less than those of conventional oscillators.
While the configurable oscillator quickly became the most cost-effective solution for low volume applications — and the ideal choice — when the need for noise mitigation conflicts with quick turnaround, the innovations didn’t end there. Due to the efficiencies and economies of scale resulting from the use of modular components fabricated using a building block process, the configurable oscillator is the preferred solution — even over conventional oscillators – for high-volume, low cost applications that cannot tolerate high noise levels.
Ever Higher Performance Demands
Industry applications and requirements continue to evolve so rapidly that there is perpetual pressure on design engineers to develop frequency control components that exhibit ever-increasing accuracy, lower noise and more stable performance, all while keeping unit costs in line and delivery times short.
Meanwhile, many electronic designs are trending towards the need for more timing-critical, compact and energy-efficient performance from their oscillators. Pile on top of that even more sophisticated demands from critical applications such as SONET or Ethernet networking that range from precise, high frequencies to low-power solutions, which are of particular importance in mobile wireless devices. All have created a huge gap between the need and what is available from normally inventoried components.
A Present Day Snapshot
Standard High-speed Complementary Metal-Oxide Semiconductor (HCMOS) crystal oscillators have well served the commodity frequency component market for decades, primarily due to no phase noise specifications. Also, integrated jitter tolerances are measured period-half period, cycle-to-cycle. Typical commodity products that integrate HCMOS oscillators include computers and peripherals, handheld and gaming devices, consumer electronics, and products used in security, multi-media and audio applications.
Moving up to the next level of the oscillator technology pyramid, we enter the arena of complex commodity products. Integrated jitter requirements are measured in picoseconds (10-12) with a range of 1 ps to approximately 2.5 ps. Technologies include SONET (OC-12), Fibre Channel (2G to 8G), LAN/WAN (10/100, Gig-E), and Storage (1.5G, 3G) SAS/SATA. Many products geared for industrial, military and medical applications fall into the complex commodity category.
Frequency components that serve this level of technology requirements include conventional oscillators, micro-electromechanical systems (MEMS) and configurable oscillators.
One level higher are precision technologies that measure integrated jitter ranges in femtoseconds (10-15), specifically 400 fs to approximately 350 fs. These precision technologies include SONET (OC-48), Fibre Channel (16G), LAN/WAN (10Gig-E), and Storage (6G) SAS/SATA. Surface acoustic wave (SAW) oscillators and low jitter configurable oscillators deliver on these demanding requirements.
Near the tip of the pyramid, we find technologies aptly named ultra-precision, since they require maximum integrated jitter of approximately 350 fs. Extremely demanding technologies and their specifications include SONET (OC-192), Fibre Channel (32G), WAN (40 to 100Gig-E), and Storage (12G) SAS/SATA.
Continuing the trend of successfully responding to the industry’s ravenous craving for ever greater speed, volume and the bandwidth to support it, design engineers have developed a new breed of ultra low jitter configurable XOs and VCXOs that offer lower phase jitter and a broader frequency range than currently available oscillators.
Performance of these new XOs — projected to operate with the same electrical performance and low levels of jitter found in fixed frequency oscillators — is intended to satisfy the ever-increasing demands of SONET and Ethernet networking disciplines, telecom, data storage, medical monitoring and measurement, military communications, and signal processing as well as test and measurement. (Figure 1)
At the very tip of the pyramid — for the moment at least — are applications, such as metropolitan area networks and long haul transmissions, which require integrated jitter measurements under 200 fs. TCXO, VCXO and oven-controlled crystal oscillators (OCXOs) are the frequency control components of choice for these exceedingly demanding applications.
A Vision of What’s Ahead
Because of the never-ending demand for greater improvements in frequency control components with respect to speed, volume, bandwidth and noise reduction, the ultimate challenge for design engineers will be to develop technologies that balance these with cost and speed of delivery.
After all, even the most optimally performing component is useless if it can’t be delivered on time and on budget.
But, if past performance is any indicator of future performance, far more than cost saving is on the side of the design engineers who have proven time and again that they can deliver no matter what performance criteria the consumers of frequency control components will hurl their way.
No doubt, these engineers will rely extensively on their ability to build even higher performing configurable oscillators using the highly modular design concept.
The balance of physical attributes and performance characteristics demonstrated by the configurable oscillators makes them far and away the most cost-effective and desirable option for very low volume applications. And, the configurable design is also the preferred technology when the need to lessen the impact of noise conditions conflicts with the need for quick turnaround and delivery.
Consider what happens when you factor in the efficiencies and economies of scale achieved when integrating common modular components into the build process for configurable oscillators. They not only become a quicker and more cost-effective alternative to conventional oscillators for high-volume, cost-sensitive applications, they also do it without compromising on low noise performance.
Although the original building block design approach for building configurable oscillators was first implemented over a decade ago, newer variations have extensively broadened its applicability for building the high-performing frequency control components required by today’s advanced technologies.
These more recent variations of building block concepts include quartz and silicon reference sources supplemented by proprietary electronic circuitry to better adapt them for required performance characteristics.
Now for a dose of reality. Just as Moore’s Law will someday hit a wall — after all, there are only a finite number of transistors that can ultimately fit on a circuit board — the frequency control component industry too must obey the laws of physics. Limitations will always be looming around the next corner whenever we attempt to achieve the next balancing act between greater performance and versatility than the previous application, versus cost and delivery.
But when you tally up all the technological challenges met or surpassed by developers of frequency control components since the piezoelectric effect was first observed, that corner could still be very far away.
this article to a friend!