SI
SI
discoversearch

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor.  We ask that you disable ad blocking while on Silicon Investor in the best interests of our community.  If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.
Technology Stocks : Son of SAN - Storage Networking Technologies -- Ignore unavailable to you. Want to Upgrade?


To: J Fieb who wrote (2990)3/21/2001 1:43:45 PM
From: J Fieb  Read Replies (1) | Respond to of 4808
 
From the archives....

Optical Device Division Key to Company Growth
SOMERSET, N.J.--(BUSINESS WIRE)--March 8, 2001--EMCORE Corporation (NASDAQ: EMKR - news), a leading provider of semiconductor technologies for global communications applications, reported that bookings for the first two months of the current quarter are running substantially ahead of the pace set at this time last year and total $41.1 million across all product lines.

As a result of the increased demand for new datacom and telecom products, nearly half of EMCORE's material products bookings are attributable to optical device products, where growth has increased more than 10-fold from the second fiscal quarter in 2000. The Company's Optical Device Division, located in Albuquerque, N.M., offers VCSEL (Vertical Cavity Surface Emitting Laser) products including bare die, arrays, packaged components, and optical subassemblies for integration into Gigabit Ethernet, Fibre Channel, Infiniband(SM), Wave Division Multiplexing (WDM), ATM systems, and high speed telecom applications that encompass very short reach OC-192 and high speed optical backplanes.

``We anticipate that the strong demand for EMCORE's optical device products will continue to grow as bandwidth requirements increase and next generation datacom and telecom networks emerge,'' said Reuben F. Richards, Jr., President and CEO. ``Our recent bookings demonstrate that EMCORE continues to have positive customer response to our entire datacom and telecom product portfolio. In addition to the products currently driving revenue growth, the Company plans several new product introductions in the next month, further expanding our presence in the accelerating fiber optic markets.''

According to ElectroniCast Corp., an independent market research firm located in San Mateo, CA, the VCSEL-based transceiver market will continue to rapidly expand, reaching $3.4 billion in 2004 and continuing on to $14.1 billion in 2009. The drivers for the dynamic growth of optical device products, as reported by ElectroniCast, are the rapid advancement of datacom standards, penetration of the telecom market, new opportunities within Infiniband switch fabrics, and the aggressive deployment of intra-system array-VCSEL interconnects.

EMCORE's 2.5 Gbps 850 nm oxide VCSEL offers the speed, performance, and reliability demanded of next generation data communications networks. This product technology forms the critical high-speed optical link for customers' next generation serial and parallel optical transceivers for OC-192 and other markets. These VCSELs provide low current thresholds and smoother slope efficiencies to meet performance requirements of existing and future Gigabit Ethernet standards.

The Company also provides 850 nm 1x4 and 1x12 Oxide VCSEL arrays and high-speed gallium arsenide (GaAs) photodetector arrays, capable of data rates up to 3.125 Gbps. These arrays are ideal for providing high-speed optical interconnects between switches routing Internet and electronic data by enabling faster data transmission without increasing the size of the switch.



To: J Fieb who wrote (2990)4/13/2001 1:30:28 PM
From: J Fieb  Read Replies (2) | Respond to of 4808
 
So it became known that Fidelity has over 10% of FNSR shares. FNSR has a new memo from Frank....

Frank's News « back to Frank's News

High Tech Woes and the Way Out
March 15, 2001
by Frank Levinson

High tech is in the midst of one of its periodic “slumps”. The NASDAQ is off from a high of nearly 5200 to now around 1900. A drop of more than 60%.

Some of the losses stem from the dot-com reality that companies do need customers, sales and profits in order to survive. Some of it stems from an over excited state of networking and storage companies that support the Internet revolution. Some of it is just the over-reaction that always occurs at time like this (and also occurred on the upside of this market as well a year ago!).

But the purpose of this column is not to provide insights into the stock market. People often ask me whether Finisar stock is going up or down. Now as Finisar is a public company and I am one of its officers, I do not have a public opinion. But more to the point … I have taken to making a note of what my response WOULD have been each time I have been asked and generally in the near term, less than 1-2 weeks, (isn’t everyone a day trader?) I have been WRONG. So, one interesting data point is that I do not know whether our stock is moving up or down, but perhaps there are some insights about our products and the general technological climate where we operate where there are insights …

So what technical non-developments are behind this collapse of share prices? It is surprising how little real technical improvement has happened over the last couple of years and tech sales are fueled by discernable differences in product performance. Let me list a few examples of imperceptible movement—
Most important is the failure of all recent PCs to be seen as different from PCs that are 2 years old. This note is being typed on a Dell Inspiron 7500 with a 500 MHz Pentium III processor, 128 MBs of memory, 20 GB hard drive, 56K modem and 10/100 Ethernet. If you compare that to even the top desktop system today with either AMD or Intel’s best processor (1.4 GHz, 256 MB RAM, 30 GB HDD, 56K modem, 10/100 Ethernet), you will find that both systems boot in about the same time, both launch applications in the same time, etc. Technology markets are propelled by the (usually) real perception that improved productivity comes from upgrading the hardware periodically. Compared to my previous machine, the Inspiron was much faster, more reliable.

More than 2 years ago, I met a new Intel hire who was working on the Merced marketing team. I asked her then why this new processor was so late. Well, as time has shown, late then was only beginning! Now called Itanium, the new processor family is still not widely released and the early reviews is that it is not a perceptible improvement. Supposedly the next generation version of it really is but it is also late.

Microsoft introduced Windows 2000 almost 2 years ago as well. I use it on the Inspiron mentioned above and it is truly a marked improvement from any previous version of any Windows™. It never crashes, it has improved device handling. But … it never crashes is pretty crummy praise for a product. It would be seriously funny the current Microsoft ads that proudly announce that the blue screen of death is now a thing of the past if it was not so pathetic.

And it is still slow. Microsoft must also take part of the credit for it taking 3 minutes to boot a machine. It rapidly fragments the disks and so you need to be a rather sophisticated user even to keep it running medium slow over time.

For the majority of applications most of us run MS Office 2000. This productivity suite works well but it seems to me that we must now question what features will come in the future to drive further upgrades. More of the same is not what is required. No news of new killer Aps in the S/W arena … and nothing that makes beautiful use of all of those additional MIPs from the new fast processors.

Networking improvement for the end user has not improved in the past 3 years. LAN connections are 10/100 Mb/s. Dial up is limited to 56K modem speeds. Worse yet, the world has reasonably fast LANs and the USA (and the world) have great transcontinental Gb/s links but we all generally rely on T1 or DSL tie these Gb/s structures together. We have a copper straw strangling the communications between our local and global communications networks! The result is fast local connections but hopelessly slow ties to the rest of the world.
My reaction to all of this is to throw open my bedroom window this evening, lean out and begin shouting to Silicon Valley where I live,

I am weary of all this and not going to take small meaningless upgrades anymore!

But this gesture stolen from the movie Network is not sufficient … so here is what Finisar is going to do to challenge this trend—
We are introducing 10 Gb/s modules to the market that are truly revolutionary. The modules have DFB lasers, on board clock recovery and serial to parallel interfaces (SERDES chips); they are small in size, stingy with power (~2.5 watts), simple to use and operate over singlemode fiber over distances from 1 meter to 10 kilometers according to the specification. (I will also let you in on a little secret … they work up to 20 km in every test we have made.)

The 10 Gb/s modules are based on manufacturable, scalable technology that will enable Finisar to help the Ethernet standard stay on the very successful track whereby a 10x increase in network speed costs only 3x more.

The modules will support the introduction of avalanche photodiodes (APDs) this year into the product thereby increasing the link distance up to 40 km. Additional WDM enhancements will come next year permitting this same module to be considered a foundational part of 100 Gb/s Ethernet! corporate-ir.net

Finisar is introducing very cost effective WDM technology through the GBIC and SFP (small form factor pluggable) product lines. Today we support 8 wavelengths on a CWDM grid from 1470 to 1610 nm with 20 nm spacing. These GBIC based parts are paired with Optical Add/Drop Multiplexer passive components that allow the very low cost deployment of metropolitan rings where every node on the ring can have its own Gb/s Ethernet drop. If the Finisar WDM technology is paired with off the shelf technology from Cisco, Extreme, Foundry and others then it is possible for the network to support 100 Mb/s connections in the metropolitan area markets for less than a total installed cost of $500 (not including monthly fiber rental charges which are shared by the whole network) quicken.excite.com

Finisar will make extensive use of avalanche photodiodes in upcoming products. APDs provide typically an additional 8 dB of link budget and they can do this for a relatively small cost impact. Today these diodes (which require rather high bias for operation) are considered exotic but they represent a terrific and simple way to enhance metropolitan area networking because the extra link budget can be used for WDM and switching.

Finisar will be supporting a number of MAN trials and even fiber to the home (FTTH) trials with the technology in 1, 2 and 3. This support will also come in the form of new passive optical devices and network models to support the wide scale deployment of the subsystem technology. What this means is that start up companies can be involved in this revolution where it was once thought to be the exclusive domain of companies whose capital was measured in the 10s of billions of dollars.
These are the things that I can specifically describe to you about today. But the level of innovation here at the moment is beyond what can be discussed or even described. Never in our 13 year history have we been more innovative or have pushed the boundaries of what is possible so rapidly. Never will we be more disruptive with the introduction of radical price/performance products!

Additional areas where innovations can be expected include—
Radical new optoelectronic packaging. What we generally have today traces its roots back for telecom devices to at least the early 1980s; for datacom the distance is certainly back to the mid-1980s. SMT, non-hermetic, integrated functions all are still in their infancy and inelegant. Packaging represents one of the most fundamental areas to look for radical cost savings in the future.

As optics begins to achieve the same level of total bandwidth use as modems did over twisted pairs, new line coding and signal processing will emerge. This is already happening but it has never achieved wide acceptance; in the next 2 years it will become commonplace. Finisar intends to be a leader here.

Fully integrated optoelectronic devices on InP or GaAs will be deployed. Arguably Finisar is already doing integration with the Sensors Unlimited array photodiode products. These devices have 256 or 512 photodiodes in a row and all must function within tight specifications. Unless the yield on an individual device is better than 99.7% then the array will not yield above 50%! So we already sell devices with up to 512 InP optoelectronic elements … and the InP camera chips which have 320 x 240 elements require integration of more than 72,000 InP opto parts on a single chip. In the future we expect the integration of lasers, amplifiers, photo detectors and amplifiers. sensorsdwdm.com
Care to join this revolution? Let’s get back to the days when new products in technology were “insanely great” and when first turned on by a new user, the “wow!” factor was palpable. That is what makes technology sell … killer apps, wow! and truly differentiable performance.

Write me back with any suggestions you have!



To: J Fieb who wrote (2990)4/30/2001 9:14:40 AM
From: J Fieb  Read Replies (2) | Respond to of 4808
 
The cc won't be for another week or so, but on the surface(a pun?),
19.6% sequential growth during these times, looks very, very good? If the visibility is good for next Q?

Emcore Corporation Announces Preliminary Results for Second Quarter
Revenues increase 100% to a record $47.9 million Reports operating profit before goodwill Schedules Conference Call for May, 8, 2001 at 9am EST
SOMERSET, N.J.--(BUSINESS WIRE)--April 30, 2001-- EMCORE Corporation (NASDAQ: EMKR - news), a leading provider of semiconductor technologies for global communications applications, today announced financial results for the second quarter of its fiscal year ending March 31, 2001.

Revenues for the quarter ended March 31, 2001 reached a fifth consecutive record of $47.9 million, up 100.2% compared to revenues of $23.9 million for the corresponding quarter of fiscal 2000 and up 19.6% sequentially compared to revenues of $40.1 million for the quarter ended December 31, 2000. The increase in revenues for the second quarter of 2001 as compared to the quarter ended December 31, 2000 was due primarily to increased shipments of fiber optic products, however, the Company experienced demand across all product lines. The Company's backlog also increased above the level of December's reported record $155 million.

Gross profit for the quarter reached $19.6 million, an increase of 97.1% compared to $9.9 million for the corresponding quarter of fiscal 2000, and an increase of 18.5% sequentially compared to gross profit of $16.5 million for the quarter ended December 31, 2000.

Operating profit for the quarter ended March 31, 2001 was $31,000, excluding goodwill amortization, compared to an operating loss of $3.6 million, excluding goodwill amortization, for the quarter ended December 31, 2000. Operating expenses declined as a percentage of sales due primarily to a decrease in research and development expenses to $12.0 million from $13.2 million in the preceding quarter. The completion and release of several new fiber optic products, early in the quarter including a 10 Gb/s serial device and 2.5 Gb/s LC/SC TOSAs (optical subassemblies), for various data communications network applications contributed to the decrease. EMCORE expects research and development expenditures to continue at levels similar to the second quarter's for the remainder of fiscal 2001 as the Company finalizes the development and commercialization of new fiber optic products, including long wavelength VCSELs (vertical cavity surface emitting lasers), optical subassemblies and modules.

Net income for the quarter including non-recurring items was $2.9 million or $0.08 per diluted share, compared to a net loss of $4.2 million, or $(0.14) per diluted share, for the corresponding quarter of fiscal 2000, and net loss of $7.0 million, or $(0.21) per diluted share for the quarter ended December 31, 2000. On a weighted average basis, diluted shares outstanding for the three months ended March 31, 2001 were 38.4 million compared to 29.8 million for the three months ended March 31, 2000.

During the quarter ended March 31, 2001, EMCORE also finalized a two-year, $20 million line of credit with First Union National Bank.

EMCORE will release complete results after the close of business on May 7, 2001 and will discuss the results further on a conference call to be held on May 8, 2001 at 9:00 a.m. EST. To participate in the conference call, local and international callers can dial 719-457-2657. A replay will be available beginning 12:00 p.m. EST on May 8, 2001 until 12:00 a.m. EST on May 12, 2001. The replay number is 888-203-1112, passcode: ``516714''.

The foreshadowing of long wavelength VCSELs will be another whole gear in the transmission for companies in this sector.
Bring on the 1310 VCSELs. It will spawn tweaked reactors that everyone will want to buy and those will crank out longer reach transceivers. Have read they can put something like 20,000 VCSEls on a 3 inch disc. The data centers will expand their diameters, whether FC or Iband.

Gonna keep some of these seats for a few years.

By 10G copper is all done even for cpu interconnects.
Optics doesn't care whether it is FC,10gE,10FC,iSCSI,Iband.



To: J Fieb who wrote (2990)7/4/2001 9:50:27 AM
From: J Fieb  Read Replies (1) | Respond to of 4808
 
VCELs are really agnostic when it comes to protocals. 10G FC and GE are serial. Short haul metro is parallel. A few companies are touting 1310 VCELs, that is a whole new market for these things...

From Compound Semiconductor Magazine Issue 6 No. 5 (July 2000)

VCSELs: Driving the Cost Out of High Speed
Fiber Optic Data Links

compoundsemiconductor.net

DAVE WELCH
W. L. Gore & Associates

Presented at the Key Conference on Compound Semiconductors
Key West, Florida
March 13-14, 2000

VCSELs offer both performance and cost advantages over traditional edge-emitting lasers. Following strong investment by the U.S. government, universities, and industry in the late eighties and early nineties, VCSELs are now being manufactured in high volume by a number of suppliers. The main application at present is gigabit-speed communications, while technology advances are opening up other markets at higher data rates and longer transmission distances.

The VCSEL Advantage

Essentially, VCSELs are very high performance lasers that can be produced much like low cost LEDs. VCSELs have a low divergence circular output beam and very stable performance characteristics over temperature. They can be modulated at high data rates, and can have threshold currents of less than a milliamp in some cases. In addition, VCSELs offer the manufacturing advantages of wafer-scale fabrication and test, and the practical ability to produce either small or large arrays.

Although VCSELs can be manufactured at relatively low cost, the real advantage comes in the ease of packaging these devices. For example, the low divergence circular beam enables improved coupling tolerances. Also, high-speed operation is achieved at low power, which minimizes problems with crosstalk and heat dissipation.

Another important aspect of VCSELs is that the stable performance over temperature allows the monitor photodiode feedback loop to be eliminated in many applications. This not only eliminates the monitor photodiode, it also simplifies the IC and package designs, with important cost implications. Furthermore, non-hermetic packages are now available and the ability to produce uniform, low cost arrays with good temperature stability has enabled the development of low cost, parallel optical modules (see Figure 1).

VCSELs were successful at entering the communications market because of the strong demand created by the need to transition from 100 Mb/s applications like Fast Ethernet to 1 Gb/s applications like Gigabit Ethernet and Fibre Channel. The performance of LED sources that had been used for 100 Mb/s applications was not able to match the speed requirements of the emerging standards, so a new source technology was required. One interesting point is that VCSELs went head to head with CD lasers in the early stages of the Fibre Channel and Gigabit Ethernet markets. At the time, VCSEL production volumes were very low, while the CD lasers were being produced in high quantities. Despite this, VCSEL manufacturers were able to enter the market and be cost competitive at an early stage. Customers were attracted by the performance and reliability advantages of VCSELs, as well as the promise of scalability to higher data rates. Today, the vast majority of high-speed, short wavelength transceivers use VCSELs, and production is now at the level of million of units per year.

Products and Markets

The first commercial VCSEL product was introduced by Honeywell in 1996. The VCSEL was offered in a traditional TO can, primarily for use in transceiver modules for Fibre Channel and Gigabit Ethernet. In these applications, transmission distance is less than 300 meters. Honeywell conducted early reliability studies that were very helpful in getting VCSELs accepted within the standards community and by customers. By accumulating a large amount of data, Honeywell showed that VCSELs could typically offer reliability improvements over CD lasers. Today Gigabit Ethernet and Fibre Channel transceivers are available from a large number of suppliers.

An important development in the last year or two has been the introduction of small form factor transceivers at approximately half the size of standard products, allowing the incorporation of more transceivers onto each network card.

The gigabit market is growing at a rapid rate, and will continue to do so for the next several years. Recent data from Electronicast states that the gigabit Ethernet transceiver market was worth about $600 million in 1999, and will grow to over $2 billion by 2004. Other projections suggest that this figure will be even higher.

There are a number of emerging opportunities for VCSELs, particularly in the area of multi-gigabit networking, where VCSELs are being developed for emerging standards such as 2X and 4X Fiber Channel and 10 Gigabit Ethernet. Parallel optical products for Very-Short-Reach (VSR) applications have reached the market and are gaining support in industry groups such as the Optical Internetworking Forum (OIF). For these applications, the development of oxide-confined devices and arrays is crucial. Long wavelength VCSELs operating at the single-mode telecom transmission wavelengths (1.3- and 1.55-ìm) will move the cost benefits of VCSELs to longer distance applications. Also, there are a number of emerging applications for VCSELs in areas relating to WDM.

Oxide-Confined VCSELs

First generation VCSELs, including many of today's standard products, are fabricated using ion implantation. This process creates a resistive area that funnels the current into the active region. The second generation of VCSELs use oxide confinement to provide both current guiding and index guiding for improved efficiency. Oxide VCSELs are grown with a high aluminum content buried layer, which is subsequently oxidized under conditions of high temperature and high humidity to form the device aperture. Oxide devices, which are now commercially available, operate at higher speeds and have lower threshold currents than the ion implant VCSELs. Oxide VCSELs also exhibit improved temperature performance. The advantages offered by oxide VCSELs enable low cost, higher speed LAN transceivers. These devices also allow the fabrication of highly uniform arrays, enabling multi-channel, parallel optics applications.

Figure 2 illustrates the performance of oxide-confined VCSELs manufactured by Gore. In an uncooled package, the devices can be directly modulated at a rate of 12.5 Gb/s with a very low modulation current. As the networking environment moves towards higher speeds, the transmission distance in standard multi-mode fiber shrinks dramatically (see Figure 3). To address this problem, new multimode fibers have been developed that are optimized for lasers emitting at 850 nm. Gore has been on the leading edge of working with companies including Lucent, Corning and Alcatel, who have been developing this fiber to create next generation multimode systems.

A demonstration combining Gore's 850 nm VCSELs and Lucent's high bandwidth fiber showed 10 Gb/s transmission across a distance of 400 meters. Subsequent experiments have shown that a 12.5 Gb/s signal can be transmitted across nearly a kilometer. These are laboratory-type experiments, but they indicate that robust transmission can be achieved through several hundred meters of multimode fiber using VCSELs.

Parallel Optical Interconnects

One of the major challenges in emerging high-speed networks is the ability to cost-effectively manage the huge amounts of data flowing through the central offices and Internet service providers (ISPs). Parallel optics is a very attractive solution for short-distance (<300 meters) links between routers, switches, and transport equipment. A number of companies have introduced multi-channel transmitter & receiver products, utilizing arrays of VCSELs and detectors connected by ribbon fiber. Early implementations of this technology have 12 channels with 1.25 Gb/s per channel, equating to an array with an aggregate data rate of about 15 Gb/s. It is important to note that this solution costs a few hundred dollars, compared to the thousands of dollars required to implement today's OC-192 (10 Gb/s) serial links.

Gore is in the process of commercializing its nLIGHTENTM parallel modules. The modules are only about three-quarters of an inch wide, allowing many modules to be incorporated onto a card edge (see Figure 1). Both Gore and Infineon produce parallel optical modules compliant with a multi-source agreement that provides for footprint-compatible modules in an effort to speed the market acceptance of these products.

Gore's nLIGHTENTM parallel modules were incorporated into a system displayed by Cisco Systems at the recent Optical Fiber Communications Conference (OFC). The Very-Short-Reach (VSR) parallel optical interface is being promoted within the Optical Internetworking Forum as a leading choice for OC-192 (10 Gb/s) transmission at distances up to 300 m. In this application, an ASIC converts the incoming 16 3 622 Mb/s signal into a 12 3 1.25 Gb/s signal for transmission across the 12-channel ribbon fiber.

A key part of making low cost parallel modules is the availability of a suitable VCSEL array technology. Gore VCSEL arrays are designed to have very stable performance over temperature at the actual operating current of ~5 mA (see Figure 4). This simplifies the system design, and eliminates the need for a monitor photodiode feedback loop. The VCSELs are scalable to high data rates so that in the future, low cost, hundred gigabit connections over very short distances is a viable concept.

Life test data for Gore VCSELs operating at 100°C with a drive current of 5 mA shows no failures after more than 2500 hours of test time. A large-scale reliability test is currently being performed, and the results will be available shortly.

Long Wavelength (1.3 ìm and 1.55 ìm) VCSELs

VCSELs operating at the transmission wavelengths of single-mode fiber (1.3 ìm and 1.55 ìm) will extend the cost benefits of VCSELs into the single-mode domain, enabling longer transmission distances and opening up new applications. This will move VCSELs from local area networks (LANs) with transmission distances of several hundred meters into access networks, campus backbones, metropolitan area networks (MANs) and wide area networks (WANs) with multi-kilometer transmission distances.

Despite strong demand for long wavelength VCSELs, the technology has developed relatively slowly due to fundamental materials challenges. Good mirrors can be grown on GaAs substrates, however it is very difficult to grow active regions emitting at 1.3 ìm on GaAs. Conversely, good active regions can be grown on InP substrates, however high reflectivity InP-based mirrors are extremely difficult to fabricate. There has been, and continues to be, a large effort to try to address these problems through fundamental materials research. Thus far, however, the best devices have used wafer fusion to combine materials of different lattice constants into one structure.

Gore has combined wafer fusion with an integrated, optical pump to produce the world's best long wavelength VCSELs (see Figure 5). The Gore structure has several major technical advantages resulting in higher single-mode powers and improved performance over temperature. Gore's 1310 nm VCSELs operate at temperatures approaching 100°C, and about a milliwatt of power can be coupled into fiber even at 70°C. The devices work very well in excess of 2.5 Gb/s and higher speed parts are in development.

Gore's VCSELs are now being sampled to a couple of customers, and will be more broadly available in the near future. At this year's OFC, Gore demonstrated 2.5 Gb/s transmission over a distance of 25 km using its 1300 nm device in a standard package (see Figure 6).

VCSELs are normally associated with multimode operation, however long wavelength devices can be designed such that single longitudinal mode and single transverse mode operation are obtained. The linewidth is similar to that of a DFB laser, so the devices are well suited to high speed, extended distance single-mode fiber applications. Long wavelength VCSELs also have major coupling advantages (see Figure 7). 1300 nm VCSELs have a low divergence, circular beam profile that enables more than 80% of the light to be coupled into single-mode fiber without lensing, creating very obvious advantages with regard to package design.

VCSELs for WDM

There is great potential for using VCSELs in wavelength division multiplexing (WDM) applications, and new concepts are emerging. VCSELs are being applied to coarse or wide WDM, and tunable VCSELs and VCSEL arrays emitting at multiple wavelengths have been demonstrated.

Coarse (or wide) WDM and dense WDM differ in the spacing between adjacent wavelengths. For a DWDM system in which the channels are placed on a 100 GHz ITU grid, the wavelengths are separated by only 0.8 nm. Although this allows 40 channels in the L-band (1530-1560 nm), cooling is typically is required to ensure that the wavelengths of the channels do not drift. Coarse WDM products are focused low cost, local area networking applications and use much wider channel spacings to simplify product design. Channel spacings of 10- to 25-nm are employed, and because VCSELs drift less than 0.1 nm/°C the products can effectively operate uncooled, from 0–70°C without channel overlap.

Blaze Network Products has developed an eight-channel coarse WDM transceiver using VCSELs with wavelengths ranging from 790 nm to 860 nm. Plastic, molded components combined with broad wavelength filters enable the optical mux and demux functions to be accomplished at low cost.

Tunable VCSELs have also been demonstrated. CoreTek (now part of Nortel) has combined a VCSEL technology with a MEMS mirror that can be moved in order to tune the cavity. About 40–50 nm tuning and 5 mW of coupled power has been achieved in this manner.

Finally, Gore has demonstrated VCSEL arrays containing four devices emitting at different wavelengths in the 1550 window. Using this technology, the possibility exists for fabricating low cost, integrated WDM sources with 4, 8 or 16 channels for high bandwidth single-mode links.

Employing VCSELs in WDM applications area should prove to be a fertile ground for innovation and future development.

Summary

For certain applications, VCSELs have fundamental advantages over other types of lasers. No longer lab curiosities, VCSELs are in volume production from multiple suppliers, and technological advances continue to enable new applications. Oxide-confined VCSELs have already enabled higher data rate products and parallel optical interconnects, while longer wavelength devices hold the promise of longer transmission distances over single-mode fiber. Finally, there is a great deal of potential to utilize VCSELs for WDM-related applications.

© 2000 Compound Semiconductor Magazine. All rights reserved.

This technology will bring the needed capability to mass produce optical components and alter the price structure for
the networks of the future.

'Gonna keep my VSCEL seats for a few good years.