The rapid changes that have been occurring (and will continue to occur) in electronics systems are driving changes all the way throughout the supply chain for electronics components. This article will focus on the impact of increasing operational frequency (speed) on the materials used in the making of PCBs.
A short description of most PCB materials follows. These materials are composed of three components: woven glass fabric, plastic (resin), and copper. Multiple types of glass and resin are used in making PCB materials, and how they are mixed impacts the mechanical and electrical properties of the material. This results in some interesting effects on the materials-dielectric constant is not the same for all thicknesses of material, the dielectric constant and loss tangent may go opposite directions (e.g., thin core is low Dk and high loss, thick core is high Dk and low loss). A common measure to help quantify the changing electrical properties in PCB materials is the resin/glass ratio.
The key electrical properties of materials are dielectric constant or Dk (relative permittivity), and loss tangent (permissivity). Dielectric constant is related to how much charge two conductors can hold with a voltage applied between them. Low Dk separates less charge (capacitance), while a high Dk separates more charge. Dk also determines the velocity at which electricity moves down a conductor. (Note: Dk is measured as a relative value compared to a vacuum.)
Loss tangent-also known as dissipation factor Df-is a measure of how much electromagnetic energy is absorbed by the dielectric material. An easy way to think about this factor is microwave cooking. Things that heat up fast or cook fast in a microwave have a high loss tangent (water is a good example). Things that stay cool have a low loss (glass and ceramic are good examples). Loss is a function of frequency, where for the same loss tangent more energy is lost as the frequency rises.
Electrical Functional Needs
The rise in the frequency of signals being transmitted through a typical PCB is continuing to rise rapidly. In addition, the lower voltages being used (along with the short switching window) are making signal integrity issues increasingly important.
What are some of the applications that are leading the rise in frequencies used on boards?
The first category is optical networking. Optical networking is already well along in introducing 10 Gbit/sec per channel systems. These operate at 0.6 to 3 GHz on feeder channels, which are combined to achieve the final output rate, and the data is all being transmitted serially. This is immensely fast compared with parallel buses that run at 133-250 MHz. To operate at this frequency, the quality and the predictability of the signal have to be very good. To achieve their goals, the makers of these machines are utilizing enhanced electrical performance materials and system architectures designed to minimize the length and number of high-frequency signals on the PCB. Virtually all of the data being transmitted in these systems is done using differentially driven circuits.
Another area pushing the material performance is high-speed computing. Computing still uses more single-ended circuitry then telecom, making computer engineers more interested in the effects of synchronous switching and the overall greater sensitivity of single-ended circuits to crosstalk and noise. Frequencies running on boards are generally much lower on computers, so engineers are not as interested in the enhanced electrical performance materials. However, they are interested in switching noise, power plane impedance (ground bounce), and decoupling capacitance. This leads to an interest in methods that provide more effective current (i.e., power) to high-speed devices during switching. Embedded distributed capacitance has become an important material to this market.
Finally, high speed computing and its related components are looking at high-speed serial data transmission, very similar to the optical networking systems. This would replace or enhance the currently used wide busses, speeding communication on dedicated pathways as opposed to generic busses. The requirements for these types of circuits will be very similar to the optical networking needs.
Modifying the performance of the material to improve its functionality towards these attributes is the driver behind the trends in material performance.
In classifying materials by their end use, there are three categories of materials: FR-4, enhanced epoxy systems, and high performance materials (see Table 1).
FR-4 materials have the poorest electrical performance of all the commonly used material systems. This has begun to reduce the worldwide market share for FR-4 systems in the past few years. Most of the activity in this category has been on improving the processing cost of the material, focusing on reduced lamination cycle times and reduced hole drilling costs. The key activity for FR-4 to improve its impact on the electrical properties has been an increasing focus on thickness uniformity, particularly on thin cores. Current specification limits for thickness on thin core FR-4 take up most of the +/-10% impedance tolerance, forcing the PCB fabricator to closely monitor the material thickness shifts.
However, FR-4 remains the material of choice for most designs in North America, and is the benchmark material of the PCB industry. The dominant reason for this is the low cost and broad availability of the PCBs made from this material system. Several OEMs have internal development teams that are devoted to keeping their designs on FR-4 and avoiding specialized material systems. FR-4 will continue to be an important material for many years, and material suppliers will continue to upgrade their systems to make the material more competitive.
Enhanced epoxy systems have a slightly reduced Dk value, with a significant reduction in the loss tangent. Several systems in this category have been in use for years, including Getek, Megtron, 4000-13, and FR-408. This family of materials has been characterized by several factors: the material systems have been proprietary and don't directly match other materials in the category (the exception is Getek/Megtron which are a match for each other.); the materials are more difficult to process than FR-4 and require better processing controls; and finally, the materials have had good price/performance value. This family of materials has been growing rapidly and taking market share from the FR-4 family. Users of these materials have found the relatively small price increase from FR-4 was easily balanced by the improved design margin available. These materials became even more valuable as users crossed the 1 GHz barrier.
Enhanced epoxy systems will continue to gain market share, with much of the gain in the computer/server market as on-board speeds continue to rise. Backplanes in particular have used many of these materials. The top end of telecom applications are starting to move out of this materials category, but the mid-range will still see use. If you are operating between 1-5 GHz, and you are moving signals more then a few inches, enhanced epoxy materials could be valuable to you. I recommend qualifying at least two of the material systems for high-volume applications, as it will improve the supply line for PCBs.
High performance materials are seeing tremendous amounts of developmental activity. The growth in this category is just starting, but it appears to be strong. The low loss of these materials, along with being designed for high volume manufacturing, makes them the best choice for the transition to 5+ GHz serial data transmission. One of the key materials systems in this category is the A-PPE material. This material was developed by Asahi Glass of Japan and has been globally licensed to three large material suppliers: Isola, Nelco and Polyclad. This provides global multiple sources of very similar material. These materials are currently significantly more expensive then FR-4 materials and require special processing to manufacture them into PCBs. These materials provide a large performance improvement, but also will have a significant cost differential, making careful analysis of value important when making a material choice.
High performance materials improve nearly every category of the signal integrity attributes listed earlier. Lower Dk with improved control helps impedance control, x-talk, jitter (in certain conditions) and skew. Lower Df (loss) improves rise and fall time, and total attenuation. The only negative is that the low loss doesn't quickly dampen any noise that does occur.
There are several trends in high performance materials. Since the volume use of these materials is just starting, new materials are still being developed for this market. Expect to see new offerings in the next 12 to 18 months. Changes to the classical approach to PCB materials are happening: Rogers 4350 material has the same Dk and loss for all thicknesses, unlike standard materials. The 4350 has a thicker resin layer that can reduce micro Dk effects and helps with jitter. The Nelco 6000-21Si material uses a new type of glass fabric with lower Dk and loss (see Table 1 above). The Si glass better matches the resin Dk, making for better material Dk control and improved micro Dk variation. I expect to see a trend of developing materials that are easier to process, with improved electrical properties.
Other Material Trends
Embedded distributed capacitance (EDC) materials have long been dominated by ZBC, which is a 2-mil thick epoxy laminate. Newer materials designed for use as embedded distributed capacitance have been under development for several years. These include thinner versions of ZBC (1 mil) and ultrathin (.15-.4 inches thick) high dielectric materials (3M C-ply is an example). These materials range from 500 picofarads/square inch to >10 nanofarads/square inch.
Embedded distributed capacitance performs several functions:
EMI shielding - EDC provides both additional faraday shielding and helps quiet EMI generation. Testing has shown that the amount of capacitance has little impact on the effectiveness of the material for EMI shielding.
Ground plane impedance - The addition of EDC to PCBs is effective at smoothing the impedance (eliminating harmonics) over increasing frequency. This allows all the harmonics of the pulse to behave similarly. Testing has shown that higher capacitance and thinner materials perform better at smoothing out the impedance.
Power buss noise (ground bounce) - This function is the ability of power planes to provide high current during switching, and not exhibit voltage spikes. It does this through low inductance and high storage capacitance. Testing has shown that discrete decoupling capacitors are ineffective at frequencies above 500 MHz, while EDC was shown to be effective to at least 5 GHz (which was the limit of the testing). This function appears to be most related to thickness of the material.
Conductive anodic filament (CAF) growth-also known as dendritic growth-is a long term PCB failure mechanism where copper filaments grow between power and ground causing shorts and failures. CAF resistant materials are designed to greatly inhibit the formation of CAF. CAF formation is a function of time, voltage, and spacing between power and ground elements. Products susceptible to CAF growth may benefit from CAF resistant materials.
The rapid increase in frequency in electronics has driven changes throughout the supply chain. In only a few short years, digital electronics have moved through the frequency bands where design and materials have long been established, and are soon to move to a new frontier. This has forced changes in PCB materials to meet the needs of the final products. The rate of new material introduction is progressing at an unprecedented pace. Expect to learn about even more interesting materials in the future.
Doug Trobough is chief technology officer at Merix Corp.
Merix - www.merix.com
Nelco - www.parknelco.com
MEM - www.aromat.com
Rogers - www.rogers-corp.com
Isola - www.isolalaminatesystems.com
© 2001 CMP Media Inc.
6/1/01, Issue # 1806, page 17.
Reproduction and distribution of material appearing in PCD and on the www.pcdmag.com Web site is forbidden without written permission from the editor.
Contact information for Reprints.
NOTE: The articles presented here contain only the text originally published in Printed Circuit Design magazine. Any accompanying graphics and illustrations have not been recreated here. You may view the article in its entirety in each printed issue of PCD.