Printer Friendly

Optical waveguide: PCB interconnects: optical technology continues to displace copper for for high-speed data transmission, but wide spread adaptation is still five years away.

Increased computing power is the primary driver in a move towards the optical interconnect. (1) This drives miniaturization, high density circuits and high date rates resulting in a bottleneck at the electrical connections. Electrical (copper) interconnections, under these conditions, compromise performance and require increasing costs to shield and to condition signals to overcome electrical interconnection issues.

Optical technology generally continues to displace copper for data transmission over shorter distances. The higher the data rate required, the shorter the distance over which optical methods become superior. Telecommunications, from long haul down to the metro links, are a few kilometers in length and are dominated by single--mode optical technologies.

The Transition from Copper to an Optical Connection

FIGURE 1 is a graphic of the current generic data transmission environment with increasing distance horizontal and increasing bandwidth vertical. (2) It illustrates the general state of optical methods replacing copper with optics and beginning to challenge copper in backplane applications.

TABLE 1 illustrates the battlefront between optical and copper data transmission in more detail by including specific distance-bandwidth combinations and an indication with $$ signs of the cost to fill the application utilizing copper methods.

The general consensus is that a copper pair is a viable media for transmitting up to 18 Gb/s over a maximum distance of one meter using conditioning/detection circuits. Whether or not copper can be used at greater bandwidth/distance combinations, marked as "Frontier" in this table, is an open issue.

Converting to optical methods at any of the frontiers can be an economic or technical issue, or both. At the longer distances, the issue is usually economic with optical innovation continually overcoming the economic barrier, resulting in adoption.

At the shorter distances, technology is not nearly as developed, and the economic issues are not yet clear, implying that it will be longer until those frontiers are overcome.

PCB Applications

As stated earlier, it is believed that copper can be used to transmit data at rates up to 18 Gb/s for distances up to one meter if signal-conditioning circuitry is used. Beyond that, the losses from the skin effect and dielectric loss tangent become too great to compensate. Optical methods may be viable and, indeed, have been used in a few specialized applications over the years. Optical backplanes, however, have not yet entered the mainstream and are not expected to do so in the near future.



FIGURE 2 shows an AMP Tyco, LIGHTRAY OFX, a flexible solution for precision fiber management in high fiber count telecommunications network equipment applications.

Designed using the latest fiber optic technology, LIGHTRAY OFX consists of a number of optical fibers encapsulated in thin protective film substrates. As well as saving valuable space in equipment and equipment racks, LIGHTRAY OFX also provides optimum optical performance. Optical fiber "crossovers" allow the optical fibers to be routed in both regular and irregular designs. Fibers are routed on 250 [micro]m center lines.

The displacement of copper by optical technologies for data transmission continues. The growing demand for greater data transmission capacity, smaller physical size and lower power consumption generally favors optical methods and results in optical connections replacing copper connections as soon as the total cost of the optical solution is competitive. With the rising costs of energy and the power draw and data transmission rates increasing, optical technologies will continue to replace copper.

If we look at the use of optoelectronic technology across the spectrum, it might look something like:

* Massive use in long haul, metropolitan communications.

* Significant use in large LANs.

* Growing use within data centers for rack-to-rack data transfer.

* Copper OK to 20 Gbit up to one meter in backplanes.

* On circuit boards other than backplanes.

** Static; awaits need & demand.

** Much technology demonstrated but cost questionable.

* On-to and off-of chip.

** Exploring alternatives for lower power and more bandwidth.

* On-chip.

** Exploring alternatives for lower power and more bandwidth

** A modulated light source is needed.

Dr. Bruce Booth and I have discussed alternative waveguide manufacturing processes in detail in articles published in the March 2008 and April 2008 issues of PCD&F. (3) Ridge technology and Diffusion technology are the two main types of processes, TABLE 2 compares these technologies. In the table, technology maturity is assigned for each of several technologies to be considered and includes important attributes using a scale of 1 to 4; namely:

[1] literature, conceptual stage or early feasibility only.

[2] laboratory demonstration, preliminary proof of concept, evaluation and testing.

[3] prototypes constructed and delivered for evaluation, and/or demonstration or system design development, limited pilot production, initial application specific testing.

[4] commercial product deployment, extensive testing, manufacturability demonstrated.

TABLE 3 discusses typical process characteristics of the different types of waveguides. The same maturity rating used in the previous table applies.

Needs and Showstoppers

What are the gaps and showstoppers for optoelectronic PCBs in production? We need a major OEM to say what its next generation of product X is going to be. Before committing resources to a technology, OEMs want to see the complete optical interconnect system demonstrated.

Besides having a good waveguide technology, OEMs are looking to facilitate a complete optical interconnection system from off-board to the chip level, between chips on multiple boards and back out off board.

Capabilities for a complete optical interconnection system involve at least:

* Optical input and output connections at the edges of the motherboard or backplane.

* Right angle connectorization at the daughter board (DB) junction with the MB.

* Optical signal distribution on the DB or into Tx/Rx link packages.

These capabilities must be cost-competitive and low-risk versus electrical transmission. A low-cost, technically viable, modulated light source is also necessary.

There is another interesting difference between copper and optoelectronics having to do with standards. The PCB industry is structured on standards: design standards, manufacturing standards and testing standards. The optoelectronics industry is not this way; it is structured around Multi Source Agreements (MSAs). It's not clear that this is a showstopper or just a mild bump in the road, but it will be interesting to follow as we go forth.


Optoelectronic technology is currently a mixed bag. If we look at optoelectronics by segment, generally using distance as criteria, we see the following segmentation.

Long Haul Telecom. These are now moving to higher speed transceivers from 1.5 Gb/s to 10 Gb/s. Some heavily used links move to multiple bits per symbol, implying complex digital signal processing. Dense wavelength division multiplexing is growing. Once a fiber is "provisioned," meaning with multiplexers, demultiplexers and optical amplifiers, it is likely to be fully built out and utilized before another fiber is provisioned.

LANs. This segment is mostly data-com with people using Ethernet as a standard. Transceivers up to 10 Gbit are common. Some 4 Gbit transceivers are available but expensive and physically large.

From 1 Meter to 100 Meters. Plastic Optical fiber is growing in usage in this area, especially in automotive, and potentially aircraft applications. The attenuation of fluorinated POF is as low as 5 db/km and thus a candidate to replace glass in some applications. Multimode polymethymethacrylate POF offers low installation and assembly cost. Fiber can be cut with a razor blade like tool and simply inserted into connectors to form splices or interface to sources and detectors.

On-chip and Off-chip Up to 10 Meters. The frontier of optical methods replacing copper is on-to-chip and off-of-chip, with the leading edge being multicore microprocessors connected to memory with links of 1 meters to 10 meters.

1 Meter and Less. Backplane and circuit board applications still lack a good light source that works technically and is cost effective. Much effort is going into finding such a light source that, ideally, would be compatible with CMOS semiconductor manufacturing. Optoelectronic substrate connectivity is sufficiently advanced, practical and available today to be considered for some applications. The maturity of optical interconnectivity at the substrate level is sufficient to begin initial prototyping.

Practical multi-mode optical interconnection solutions using optical fiber and planar polymer waveguides are still being explored, developed and prototyped. A few years ago, the envisioned deployment was 2008/2009. Today, it appears that wide spread adaptation is still a few years off and that 2010 to 2015 might be closer to reality.


(1.) 2009 IPC International Technology Roadmap for Interconnections.

(2.) 2009 iNEMI Technology Roadmap.

(3.) Dr. Bruce L. Booth and Jack Fisher, "Optoelectronics Comes of Age": Printed Circuit Design & Fab No. 2008 February and March.


Dick Otte, President Promex, Inc. Santa Clara, CA. Dr. Bruce L. Booth, President, Founder and CTO. of Optical InterLinks Inc.

JACK FISHER is a consultant with InterconnectTechnology Analysis, Inc;
TABLE 1. Speed / Distance Status.

 5CM-1M 1-2M 2-4M

1 Gb/s Copper Copper $ Copper $$
3 Gb/s Copper $ Copper $$ Copper $$$
5 Gb/s Copper $$ Copper $$$ Copper $$$$
10 Gb/s Copper $$$ Copper $$$$ Frontier
20 Gb/s Copper $$$$ Frontier Frontier
40 Gb/s Frontier Frontier Optics

 4-10M 10-30M

1 Gb/s Copper $$$ Copper $$$$
3 Gb/s Copper $$$$
5 Gb/s Frontier Optics
10 Gb/s Frontier Optics
20 Gb/s Optics Optics
40 Gb/s Optics Optics

TABLE 2. Polymer waveguide technology process comparison.

 Technology based on: Technologies based on:
 Monomer Diffusion with Ridge Formation with Clad
 clad lamination Backfill

Technology-- Polymerization Image and develop
General induced monomer with aqueous ,
Attribute diffusion self etching (wet chemical
 development - 3 or RIE processing

Representative Optical CrossLinks, Inc. Rohm&Haas/Shipley,
practitioners IBM,

General Pre coated Siloxane based
process photosensitive acrylate negative acting system
description monomer &binder that processes like a
 polymer on temporary Photoresist.
 carrier 1 foot wide 100+ Compatible with CIVICS
 feet long. and PWB
 Contact photomask uv manufacturing and
 exposed waveguides. waste streams. Loss
 Monomer diffusion mass @ 850nm 0.025dB/cm.
 transfer creates higher Embedded waveguides
 density polymer guide. demonstrated with
 "Self-developed" guides minor increase in loss.
 are clad with similar Initial data shows good
 polymer, interdiffused, stability @85/85 for
 fully exposed, and bake over 2500 hrs [2.5]
 cured. [3]

 Technologies based on: Ridge Formation with Clad

Technology-- Screen print Embossed
General or mold

Representative Gemfire / OptoFoil-
practitioners Dow Corning Fraunhofer
 Institute, IZM

General Hot Embossing of
process commercial
description available polymer
 foil using special
 negative tool
 made by high
 quality lithography
 and galvanic
 Filling of grooves
 with polymer of
 higher index of
 refraction. UV
 curing. Cladding
 lamination with
 same foil as used
 for embossing
 Several drying &
 tempering steps.

TABLE 3. Analysis of Existing Optical Waveguide Technology.

 Technology based on:
 Monomer Diffusion with
 clad lamination

Technology-- Polymerization induced
 General monomer diffusion self
 Attribute development. (3]

Representative Optical CrossLinks, Inc.

Process, Characteristics, General Capability

General Pre coated photosensitive
process acrylate monomer &binder
description polymer on temporary carrier
 1 foot wide 100+ feet long.
 Contact photomask uv
 exposed waveguides.
 Monomer diffusion mass
 transfer creates higher
 density polymer guide.
 "Self-developed" guides are
 clad with similar polymer,
 interdiffused, fully exposed,
 and bake cured. [3]

Effective CTE & Robust outer layers w high
Tg -- impacts T Tg (>2000+) and low CTE
run out, solder <50ppm for T/mech prop. ;
reflow, and permits IR reflow 3000 30
bonded substrate sec. [3]
to flex link option

Field size / tool Largest photomask exposure
size to date - 18x12 inches.
Volume Typical 6 in. [3] Direct write
production guides demonstrated, large
 potential mask or mask on rotating
 cylinder options for very long

Polymer Optical Bulk material, SM, and MM
 material loss guides all same loss (no WG
 wall scatter) at 0.08 dB/cm at
 840nm; 0.26dBcm at 980nm
 System loss material /
 connectivity is key.
Guide size, 4 to 100+ um sq. or rect. with
index -- range 0.003 to 0.035 index to
 Typical MM WG's 35x40um
 wNA-.3 3
Density / pitch/ MM guides with 4 micron
stacking space or greater. 10:1
 hyperfine pitch Typical 50-
 200 um space. [3] Stacked
 for 2D arrays with 125 um
 vertical pitch in delivered
 Product. 3

 Technologies based on: Ridge Formation with
 Clad Backfill

Technology-- Image and develop with Screen print or
General aqueous , etching (wet mold
Attribute chemical or RIE)

Representative Rohm&Haas/Shipley, Gemfire / Dow
practitioners IBM, Corning

Process, Characteristics, General Capability

General Siloxane based
process negative acting system
description that processes like a
 Compatible with CMOS
 and PWB manufacturing
 and waste streams.
 Loss @ 850nm
 Embedded waveguides
 demonstrated with
 minor increase in loss.
 Initial data shows good
 stability @85/85 for over
 2500 hrs [2.5]

Effective CTE & Tg >2000, CTE ~
Tg -- impacts T 120ppm. Isothermal
run out, solder TGA shows no
reflow, and significant decomp up to
bonded substrate 50 hours. Initial tests
to flex link option pass solder reflow test

Field size / tool Meter long waveguide
size demonstrated. Typical
Volume substrate size
production processed 12x8.
 potential Material shows good
 litho on large scale
 PWB exposure tools.

Polymer Optical 0.025dB/cm @ 850nm
 material loss -0.07 dBb/cm @ 980nm
 Measurements made on
 meter long waveguides

Guide size, 5-100um with contrast
index -- range from 0.02-0.035.
 Typical wg 50x50 @-.25
 delta n [3]

Density / pitch/ Sum US through 50um
stacking Us demonstrated. 5;1
 aspect ratios
 demonstrated [3]

 Technologies based on: Ridge Formation with
 Clad Backfill

Technology-- Embossed

Representative OptoFoil-Fraunhofer
practitioners Institute, IZM

Process, Characteristics, General Capability

General Hot Embossing of
process commercial available
description polymer foil using special
 negative tool made by
 high quality lithography
 and galvanic
 metallization. Filling of
 grooves with polymer of
 higher index of refraction.
 UV curing. Cladding
 lamination with same foil
 as used for embossing
 (undercladding). Several
 drying & tempering steps.

Effective CTE & Tg 207[degrees]C
Tg -- impacts T CTE ca. 70 ppm
run out, solder
reflow, and
bonded substrate
to flex link option

Field size / tool 5" [3]
size Not for very large
Volume backplanes but for
production modules

Polymer Optical < 0,2 dB/cm @ 850 nm
 material loss [3],
 Lover loss with high
 quality embossing tools
 possible [2]

Guide size, MM typical: 50 [micro]m x 50
index -- range [micro]m
 NA: 0,1...0,3 tuneable [3]

Density / pitch/ MM typical 250 urn pitch
stacking [3], min. 50 [micro]m
COPYRIGHT 2009 UP Media Group, Inc.
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2009 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Fisher, Jack
Publication:Printed Circuit Design & Fab
Date:Mar 1, 2009
Previous Article:Dielectric performance in lead-free assembly: new test methods offer powerful tools to help understand circuit board failures and the role of...
Next Article:Designing with high frequency materials: understanding the primary and secondary performance requirements in a high frequency application can help...

Terms of use | Privacy policy | Copyright © 2021 Farlex, Inc. | Feedback | For webmasters |