Electronic Design

Make Reliable Interconnections In Rugged, High-Speed Applications

Optimizing Performance And Cost Requires Careful Consideration Of Usage, Materials, Data Rates, And Interconnection Methods.

It is possible to optimize miniature interconnect systems that are designed to operate at 1 GHz and above in mechanically demanding environments. However, doing so requires the consideration of many issues that wouldn't apply in slower, static applications. Control of issues such as signal integrity and EMI must be established up front in the design cycle, and thereafter maintained over the lifespan of the product.

Applications continue to challenge the designer in terms of size, weight, routing freedom, range of movement, and the number of signal lines. This makes it no easy task to maintain the integrity of these signal lines at the lowest applied cost.

Fortunately, technology improvements, combined with refined design practices, have helped. They've made sure that cabling solutions, optimized for price/performance, can be used well into the 2-GHz frequency range for the lifetimes of many systems. One important design practice involves the use of proper cabling and cable materials.

A cable-interconnection system that's longer than a tenth of the operational wavelength in analog systems is usually treated as a transmission line to ensure adequate signal integrity. For example, a 1-GHz analog signal with a velocity of propagation (VP) of 85% will fit this criteria when the interconnect-system length is 2.55 cm or longer. Digital interconnect systems should also be treated as transmission lines to help the square or trapezoidal data pulses maintain good shape. This typically requires that the tenth harmonic of the fundamental data rate be apparent.

A primary means to preserve signal integrity is through impedance matching of interconnect-system components. Because signal power is reflected by impedance discontinuities, all of the interconnection points in the transmission chain of source, cable, and load need appropriate degrees of matching. When matching of components isn't possible, physical features should be used to avoid abrupt transitions and minimize reflections.

As a rough guide, impedance-controlled cabling provides for data rates in excess of 1 GHz. Coaxial-cable impedances generally range from 50 to 100 Ω. At higher frequencies, coaxial cable approaches the performance of triaxial (double-isolated shielded) cable. That's because the signal tends to flow on the inside of the shield, and the noise on its exterior.

At high speeds, cable-interconnection points require controlled geometries and consistent adherence to processing standards and workmanship. Breakout geometries can significantly affect performance at higher speeds, since shields are typically removed to expose inner signal conductors for termination. This shielding gap creates a local inductive impedance discontinuity.

In addition, hand stripping and termination of ribbons may produce too much length variation, leading to excessive crosstalk and skew variation in timed signals. For this reason, the first-tier cabling houses use lasers and dedicated tooling to provide better length-tolerance control. Field rework and repair should generally not be attempted.

Using matched-impedance connectors and printed circuits is fundamental if they are not to degrade the cable's performance. When multiple signal paths converge on a common connector or printed circuit, sufficient grounding pins or traces must be allocated for signal paths that aren't otherwise separated by ground planes. Generally, increasing the ground-to-signal ratio of connector contacts or interdigitated grounds will support higher speeds by allowing closer impedance matching and decreasing capacitively coupled noise (crosstalk).

Separating terminations by using larger connectors or printed circuits also helps reduce crosstalk by the inverse of the separation squared. But when space is at a premium, this option proves unattractive. Maintaining low-resistance grounding paths is important in avoiding changes in reference voltages that can, like crosstalk, produce inaccurate data interpretation in digital systems.

Overall cabling-system rise time will be relatively unaffected if connectors and transitional pc boards are mismatched by a few ohms. Nor will the rise time be affected if they're very short in comparison to the cable length. The cable rise time will tend to dominate under these conditions.

When choosing the cable to be used, keep in mind that ribbon coax cables are especially suited to single-ended signals. This contrasts with shielded pairs, typically used for differential signals. By using the center conductors of two coaxes for signals and electrically connecting their shields, ribbon coax can be run differentially. But this doesn't take advantage of the coupling effect of a differential pair to provide better side-to-side skew (defined as a difference in delay time from one coax to another).

Ribbon coax can achieve lower skew than round coaxial cables because the length of each coax can be controlled more precisely. Round cables are made by applying a helical twist to the conductors as they're built up in bundles or layers. This increases skew due to the length variations inherent to the cabling process.

Realize that numerous variables affect skew. Since skew is a variation in propagation delay, it can be influenced by any variations in physical or electrical length, as well as changes in signal rise times. In effect, skew is a measure of the quality of all processes used in the assembly's fabrication. For this reason, more companies are contracting for complete assemblies from capable turnkey suppliers. Previously, they'd usually attempt to combine components and assembly labor from various sources into assemblies with low-skew requirements.

Controlling Skew
A major cause of skew is the variation in electrical length due to differences in the dielectric constant. For skew control, in-line measurement of time delay is important, but not always available from cable manufacturers. Figure 1 shows data from a process producing 38 AWG coax within an 8-ps/ft. tolerance.

Various system-level issues limit the practicality of higher speeds. These include the transmission distance, required signal-to-noise ratio, tolerance to signal degradation, system bandwidth, and packaging requirements. Within these constraints (and when carefully optimized), ribbon interconnect systems can support signal transmission rates of 2 GHz or higher (Fig. 2).

Fortunately, maintaining signal integrity and controlling EMI in coaxial systems tend to be complementary functions. Shielding with uniform coverage not only reduces susceptibility and radiation effects in signal lines, but provides more consistent impedance.

Short cable terminations maintain a complete shielded signal environment as long as possible.They also minimize a system's EMI windows. Full 360° shield terminations provide mechanical strength, as well as a more continuous and lower-inductance path to ground for shielding effectiveness. Conversely, inadequate shielding opens the signal path to incident EMI and can cause the interconnect system to radiate.

Other means can be used to help control EMI. Filters may be introduced at the signal receiver end to remove noise at frequencies appreciably higher or lower than that of the signals. Or, signal rise times can be slowed at the signal source to minimize the high-frequency components most prone to radiate.

Usually, cable shields should be grounded at both ends. By grounding the shield at one end only, the cable becomes susceptible to frequencies with wavelengths less than six times the cable length. In certain applications, single-ended grounding may be advantageous to eliminate ground loops--especially when EMI susceptibility is out of the signal band.

Maintaining signal integrity over time can be challenging, especially in applications requiring repeated cable flexing or exposure to harsh conditions. Cabling-system environments can be characterized as static, repetitive, or freely dynamic. Static conditions are typically found inside instrumentation to join circuit elements sharing a common structure.

Routing requirements for installation, however, can impart significant stresses to the cabling system. And even static conditions may include vibration, mechanical shock, and other environmental stresses during transportation, storage, and usage. Board-to-board jumpers would fit this description.

Repetitive conditions can be found in applications like those encountered in articulating mechanisms, hinged structures, or devices with highly predictable movements. For example, ribbon coax can be used in mobile electronics, where size and weight prove critical (Fig. 3)>.

Dynamic systems include probes in which an operator or machine can freely move one end of the cable to any desired location within its length limits. Test probes are examples (Fig. 4).

Vibration, repeated bending, and gross mechanical shocks can cause short- or open-circuit conditions. Conductor choices must be made carefully, since material degradation of the conductors themselves will eventually cause variable resistance, high impedances, or cracking at points of flexural stress. Choices of alloys typically require tradeoffs in terms of dc resistance, tensile strength, flex life, flexibility, and cost. It's also important to control and qualify the thin dielectrics and jackets incorporated into miniature coax so that shorting between conductors is prevented during the product's life.

To help reduce electrical and mechanical degradation, designers should use encapsulants to protect termination regions. Cable bend radii should be kept as large as practical (at least 10 times the coax diameter). And materials must be reviewed for compatibility with all anticipated usage conditions.

Interconnection points also must be protected against mechanical shock using suitable flex/strain relief mechanisms. Cable assemblies have to be rugged enough to withstand any mechanical stresses characteristic of the application, such as crush forces or abrasion. These stresses need to be anticipated to assure adequate performance over the life of the product.

Make sure the operational and storage requirements are thoroughly understood. These mainly include temperature, humidity, corrosive conditions, and exposure to processing or cleaning solutions. Choosing materials with the appropriate mechanical properties and compatibilities can help ensure that the cable interconnect system sustains its ability to perform over the product's intended life. Materials and assemblies may undergo relatively subtle changes during the course of repeated use. Because of this, designers can help avoid the risk of large numbers of downstream field failures through sufficient up-front qualification testing.

Smaller coaxial conductors can provide high-speed performance at spacings of 0.6 mm or smaller. A common design error is the failure to incorporate interconnect pad geometries to fit an optimized termination process, leading to unnecessary labor content and operator dependence. Usually, transitions to circuit geometries smaller than the coax can be most easily accomplished through printed circuits.

Conductor sizes may be mixed within the same ribbon in applications requiring varied power, signal, and shielding requirements. Ribbons can also be made to virtually any width to support the transmission of high-count data channels. These structures support demanding applications in which size, weight, flex life, the ability to route, and signal integrity are key considerations.

With the ability to apply mass-termination techniques, ribbon coaxial structures offer advantages in assembly. Ribbon structures support organized packaging geometries that are free of crossovers and make effective use of space.

TAGS: Components
Hide comments


  • Allowed HTML tags: <em> <strong> <blockquote> <br> <p>

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.