Recollection of my Work Years at Bell Labs

(Atgofion o fy Mlynyddoedd Gweithio yn Bell Labs)

July 2023

It is now 23 years since I retired from Bell Labs after a 32 year career (4 years at Lucent Technologies). Those years were spent working on Fiber Optic subsystems in a technology that has revolutionized world communications. In Bell Labs I was part of a team that developed the new technology and brought it to fruition in a short period of time for AT&T. I couldn't have wished for a more exciting and satisfying job after leaving Wales for America as a young man in 1968. In early 2023 I made an effort to write a brief summary for my grandchildren on what I did during those Bell Lab years. It was a difficult task since my job was technically complex, not easy to explain in simple terms and my memory was not as acute as it once was. This is the brief that I wrote.

I indicated in my book(1) how pleased I was to be offered a Member of Technical Staff position in Bell Telephone Laboratories in Allentown, Pennsylvania after completing my PhD in Electronic Engineering at the University of Birmingham, England in 1967. I had been interested in electronics since I was a boy in Nefyn, Wales and Bell Labs was without doubt the most prestigious electronic research and development (R&D) organization in the world. Bell Labs was the R&D part of the American Telephone and Telegraph Company (AT&T nicknamed Ma Bell), the company responsible for providing telecommunication services throughout the United States. Over the years, Bell Labs won many awards including nine Nobel prizes but the most important of its inventions were the transistor, the laser and the charge-coupled devices currently used in phone cameras.

The job position I was offered was an entry-level MTS in Clare Barnes's Group in the Microwave Device Department headed by Louis Moose at the Bell Labs/Western Electric plant in Allentown, PA. I was familiar with most of the microwave devices used in the department from my research work at Birmingham. Other departments at Allentown were starting to work on silicon bipolar integrated circuits and metal-oxide-silicon (MOS) integrated circuits (IC's). I was also familiar with the basic devices in both those integrated circuit structures especially the MOS device because of my PhD work on the Space Charge Varactor (2). I thought myself well suited for the work at that location and that I could be an useful contributor. My mind was made up and I made plans to head for the States.

I started at Allentown on March 15, 1968 and the work I was assigned was focused on the development of millimeter wave devices for long haul communication systems. AT&T had coaxial cable transmission system corridors all over the United States but the busiest were those running along the East and West coasts. The traffic in those corridors was growing rapidly and more capacity was urgently required. A new WT-4 system had been researched and proposed as a potential upgrade where a high speed millimeter wave signal was transmitted through a special cylindrical waveguide tube buried underground. Transmitting the signal in the tube was to keep it from being scattered by rain or fog. Clare's group was already doing work on microwave components and was now starting development work on millimeter wave components, transmitter and receiver subsystems for WT-4.

I started by doing a detailed eigenvalue analysis on the waveguide Y-junction circulator, a passive device which was used to guide and isolate signals in microwave subsystems. The analysis(3)(4)(5) provided an improved understanding of the circulator and millimeter wave versions suitable for use in WT-4 transmitters and receivers were developed( 6)(7). The transmitters were incorporated into the WT-4 system together with a myriad of other components and the system was successfully field tested with an underground tube between Netcong and Long Valley, New Jersey in 1974. I remember being transported in a bus to Netcong with other engineers from Allentown to see the system in operation in late 1975. The WT-4 system test results were reported in 1977(8).

Although the WT-4 field test was successful, a decision was made shortly after to pursue an alternate system based on the propagation of light pulses along glass fibers. Research at the Standard Telephone Company in the UK by Charles Kao in 1966 had demonstrated that light signals could be transmitted over long distances if propagated on fiber made with high purity glass. Bell Labs had a lot of experience with semiconductor lasers and those type lasers would be required as small size light sources in the system. Bell Labs research confirmed that with further development such a fiber-optic system was feasible and that it would have many advantages over other potential systems. With WT-4 paused, Bell Labs Allentown was assigned the responsibility for the development of the laser transmitters and photo-detector receivers for the new system. It was a small effort in Louis Moose's department initially, with WT-4 holdovers Fridolin Bosch and myself tasked with developing the optical transmitters and the optical receivers respectively.

The optical receiver work started at a bit rate of 12.5Mb/s but was quickly increased to 45Mb/s The silicon avalanche photodetector (APD) was the detector of choice from the outset and its output was fed into a discrete Gallium Arsenide Field Effect Transistor (GaAsFET) transimpedance amplifier layed out as a thick film circuit. The GaAsFET was a bit of an overkill since it was capable of operating at much higher speeds and it was already starting to be used in microwave amplifiers. A lot of thought was given to its use at the time and in hindsight it was the best decision made since only minor changes were required in the transimpedance amplifier circuit design when the system speeds were later dramatically increased. The APD receiver provided a 7-8dB improvement in sensitivity at 45Mb/s when compared to a silicon P-type/Intrinsic/N-type (PIN) photo-detector which was more commonly used in optical applications at the time. However the APD needed ~300 volts to operate and in order to avoid electro-migration effects and pin-holes, the receiver package was hermetically sealed. This presented a problem at the fiber/package interface where the fiber was plated and soldered around its entire circumference and attached through a hole in the package wall. Coupling the light from the fiber into the detector itself was not a difficult problem. The APD had low capacitance per unit area and a large area allowed the light to be captured easily from the fiber while still providing an adequate bandwidth for the receiver to operate at the required speed. The same was not true in the transmitter where the coupling from the laser into the fiber was much more complex.

Even though there were still major challenges, a decision was made by AT&T to target the first commercial fiber optic system for deployment in United States metropolitan areas in 1980. This was the so-called FT3 (Fiber Transmission 3) system operating at 45Mb/s with multi-mode fiber, Gallium Arsenide (GaAs) laser transmitters at 0.82 micron wavelength and with silicon APD receivers. The system was for use in short span metropolitan areas where no regenerators (repeaters) were required. The 300 volts for the APD's was easily provided from the in-house terminal equipment.

The effort in Allentown had now been expanded to include two more engineers. Mel Dixon was partnered with Fridolin on the transmitter development and Isamu Tatsuguchi was partnered with myself on the receiver development. Western Electric, Reading agreed to manufacture the transmitter and receiver subsystems and after the development was completed, a lot of time was devoted to transferring the technology to the factory, designing test-sets, setting up manufacturing lines etc etc. The customer for the subsystems was the Western Electric Plant in Merrimack Valley, Massachusetts where engineers worked with the terrestrial network system Bell Lab designers in Holmdel, New Jersey on the system architecture and the design of the system level boards etc. It was a hectic but exhilarating period of time for all those involved with long workdays and many weekends, constructing transmitters and receivers, testing, establishing specifications and preparing reports for management. The receivers worked well with high yields but there continued to be an effort on improving the transmitter yields. With close co-operation between organizations and many last minute deliveries, the FT3 system was successfully deployed on time for AT&T.

By the late 1970's, research had shown that the use of single-mode fiber had significant advantages over multi-mode fiber in such systems, with much lower fiber loss at the longer wavelength windows of 1.3 and 1.55 micron. The fibers were manufactured in the Western Electric plant in Atlanta, GA and the network system designers in Holmdel were eager to take advantage of those fiber loss improvements. They also wanted to include regenerators (repeaters) in their portfolio of parts so as to design long haul terrestrial systems for the coastal corridors as mentioned earlier and also to explore the design of submarine systems. The regenerator would consist of an optical receiver subsystem, an electronic amplifier with a decision and retiming circuit followed by an optical transmitter subsystem. This resulted in a major expansion in the optical development effort at Bell Labs, Allentown and it was accompanied by a large influx of development engineers from Bell Labs, Murray Hill. Three optical subsystem departments were formed in Allentown focused on the three areas of interest for AT&T - long haul terrestrial subsystems, undersea subsystems and low cost data-link subsystems. The original optical transmitter and receiver group was included as part of the long haul terrestrial subsystem department.

Data was reported at the annual Optical Fiber Conference (OFC) in Williamsburg, Virginia that major progress was being made in fiber optics not only by AT&T Bell Labs but also by British Telecom (BT) and France Telcom (FT). The OFC was the prime symposium venue for this new technology and among other things, the BT and FT progress prompted high level so-called GoGo meetings to be held for the Bell Lab directors to monitor the progress of both the terrestrial and undersea subsystem developments. The meetings alternated between Allentown and Murray Hill and their main focus was on the laser and the laser transmitter. The transmitter work was complicated since the lasers often had non-linear characteristics making them difficult to control when switched on and off. Their long term reliability was also an issue due to dark line defects etc. and a key engineer on lasers defects, Bob Hartman in Murray Hill, was also coupled indirectly into the Allentown transmitter efforts.

All these developments indicated that fiber optics was starting to be recognized as a disruptive technology within the industry. Moving photons around instead of electrons was more advantageous and it was about to cause a major change in the design of telecommunication systems. The first AT&T terrestrial repeatered fiber optic system was the FT3 Series C system operating at 90Mb/s which was installed along the North East corridor between Boston and Washington in 1983. The installation started with multi-mode fiber, 0.82 micron GaAs laser transmitters and silicon APD receivers with a repeater spacing of 7km, a big improvement over the 1.5km spacing achievable in a coaxial system. Since it was now demonstrated that single-mode fiber had a loss of ~0.6dB/km at 1.3 micron and ~0.4dB/km at 1.55 micron, much lower than the ~4.0dB/km multi-mode fiber loss at 0.82 micron, then a mid course change was made to finish the FT3 Series C system with single-mode fiber at the new wavelength of 1.3 micron.

This was a major change for the optical device developers in Murray Hill and the subsystem developers in Allentown. There was an intense effort to complete the required changes as quickly as possible. To operate at the longer wavelengths, the transmitters had to use Indium Gallium Arsenide Phosphide (InGaAsP) lasers (normally referred to as InP lasers) and the laser coupling to the fiber became even more difficult since the single-mode fiber had a smaller core diameter. Also silicon photo-detectors could not be used at the longer wavelengths and they were changed to InGaAsPIN detectors. Germanium APD detectors were considered but they had large dark currents at higher temperature which resulted in high receiver sensitivity penalties. The InGaAsPIN detectors required further development work since they were not available at the time in a passivated format. I recall unpassivated samples with exposed junctions being made by Charlie Burrus in Bell Labs, Crawford Hill and installed in receivers at Allentown for an AT&T 1.3 micron system demonstration in the Los Angeles Olympic Games in 1984. Several receivers were built as back up to insure the system could continue to operate smoothly for the few weeks duration of the Games. Thankfully, there were no problems. Passivated InGaAsPIN's were later developed in Bell Labs, Murray Hill and successfully transferred into manufacture at Reading (9).

The 1.3 micron transmitter and receiver subsystems were introduced into manufacture in 1984. The receiver subsystem used the passivated InGaAsPIN detector with the GaAsFET transimpedance amplifier now on a thin film circuit and in a non-hermetic package since it required only 10volts to operate. A re-evaluation was made to show that a silicon bipolar or a fine-line NMOS transistor(10) could be used in the transimpedance amplifier but there was no objection to the continued use of the GaAsFET from the silicon community in Allentown. The new subsystems were used in an FTX180 overlay in the north east corridor and wavelength division multiplexed into the FT Series C system at 180Mb/s with single-mode fiber and with the same repeater spacings.

In 1984 I was promoted to supervise the receiver group and no sooner than I had started with those additional responsibilities, that the customer was asking for 1.3 micron subsystems operating at even higher speeds. Proposals were made for a new FT Series G 417Mb/s system and even an FT Series G 1.7Gb/s system with an aggressive schedule provided for subsystem deliveries. With the GaAsFET transimpedance amplifier, the receivers worked just as well at 417Mb/s as they did at 180Mb/s and the manufacturing line in Reading was able to accommodate the new requirements and continue to operate with high yields. As shown in the attached table, the FT Series G 417Mb/s terrestrial system was introduced in 1985. For the FT Series G 1.7Gb/s changes were made to a microstrip circuit design for the GaAsFET tranimpedance amplifier and the receivers were tested in the customer system board before being shipped to Merrimack Valley. The FT Series G 1.7Gb/s system was introduced in 1987.

The full advantage of switching to 1.3 micron was now evident since the regenerator spacing in the new system was ~27km. The system was deployed throughout the United States and in some other countries. It was a major achievement at the time. The 1.7Gb/s system was capable of carrying 24 thousand simultaneous telephone conversations on one glass fiber or expressed differently, it could transmit the complete contents of the Encyclopedia Britannica across the country from New York to Los Angeles in less than half a second! It was a huge advancement compared to the coaxial cable system that existed less than ten years earlier.

AT&T FIBER OPTIC SYSTEMS (1990)

System	Intro Year	Wavelength	Bit Rate	Voice Ccts	Regen Spac
Terrestrial
FT3	1980	0.82 micron	45Mb/s	600+	7km
FT Series C	1983	0.82 micron	90Mb/s	1200+	7km
FTX180	1984	1.3micron	180Mb/s	2400+
FTSeries G	1985	1.3micron	417Mb/s	5000+	27km
FTSeries G	1987	1.3micron	1.7Gb/s	24,000+	27km
Submarine
TAT8	1988	1.3micron	296Mb/s	40,000+	40Km

In the meantime as the above table shows, a combination of AT&T, BT and FT developed and deployed in 1988 the first undersea fiber optic system called the TAT8 system (Trans Atlantic Transmission) operating at 296Mb/s with regenerators every 40km. This was also a remarkable achievement since the system, conceived just in 1980, was fully deployed and under the ocean eight years later. There was little hardware differences between the components in the terrestrial and undersea subsystems, a reflection perhaps on the close co-operation between the respective departments. If I recall correctly, a silicon bipolar transistor was used in the undersea receiver instead of a GaAsFET and both undersea subsystems were housed in hermetic packages. A large number of the undersea transmitters and receivers were subjected to extensive stress tests at Allentown to ensure their long term reliability and they were also carefully screened before being shipped. The customer for these subsystems was the Western Electric Plant in Clark, NJ and they worked in conjunction with the undersea network system Bell Labs designers in Holmdel. Both those organizations had prior experiences with transatlantic coaxial cable systems and they were familiar with the harsh environment the fiber cable and regenerators would be subject to at the bottom of the ocean. The electronics for the regenerator plus decision and retiming circuits also used silicon bipolar transistors and they were installed in the undersea housing at Clark. The cable across the ocean had the regenerators spaced every 40km and they were buried beneath the sea bed. The American terminal was in Tuckerton, NJ and on the European continental shelf, the cable was split into two arms with one arm leading to a terminal at Widemouth Bay in the UK and the other to a terminal at Penmarch in France.

The transmission speeds in all these systems had advanced rapidly and concerns had been raised they might be approaching the limit of what the Western Electric silicon bipolar transistors were capable of providing. The speed issue was both alarming and potentially embarrassing since a large investment had already been made to certify the long term reliability of the silicon bipolar components for undersea systems. Fortunately this was not an an issue since TAT8 was successfully installed, albeit at a lower speed of 296Mb/s. It was a major achievement for AT&T as the leader of the consortium. The alarm however was a wake-up call on the status of both the silicon bipolar IC's and MOS IC's in Western Electric. Bell Labs was more focused on using the IC devices for switching applications and not for higher levels of integration like they were in Silicon Valley in California. It did however prompt a focus on the development of higher speed GaAs components at Reading and apparently the FT Series G 1.7Gb/s did include a high speed GaAs IC decision circuit manufactured at Reading in its regenerator electronics.

On a personal note, having a call on the TAT 8 Submarine System was a particularly pleasant experience for me, even though I was not aware of it at the time. I was able to have a normal telephone call with my mother in Wales without that awkward 'transmission delay' that occurred with a satellite trans-Atlantic connection. The delay was present because the satellite was in a stationary geosynchronous orbit 22,000+ miles above the earth and my poor mother, nervous in using a telephone at the best of times, could not learn to pause sufficiently long without blurting out in Welsh “Wyt ti yna o hyd?” (Are you still there?) and messing up the whole conversation. The TAT8 optical link on the other hand was directly across the Atlantic and the 'transmission delay' was so short as to be insignificant.

The advancement in photonics technology between 1980 and 1989 was nothing short of remarkable. The twenty one years of experience I had as an engineer at Bell Labs by 1989 had been more than fulfilling. I could not have wished for anything more exciting and I thank Louis Moose and Clare Barnes for giving me the opportunity to join their team in 1968. The future growth in photonic products and sales were all but guaranteed and improvements continued to be made. The designers could not take full advantage of the lowest fiber loss at 1.55 micron due to chromatic dispersion which smeared out the light pulse due to the lack of wavelength 'purity' in the laser sources. The chromatic dispersion penalty was a minimum at the 1.3 micron window and fiber designers in Atlanta had already found how to shift the minimum to 1.55 micron. This so-called dispersion-shifted fiber was now available and all future development work for long haul systems was focused at 1.55 micron.

By the late 1980's several changes within AT&T were starting to affect the development work at Bell Labs, Allentown. The government breakup of AT&T in 1984 was on top of the list and it was having an impact. Its intent was to end the monopoly AT&T had on the US telecommunications industry and to spur more competition. External suppliers could now sell equipment into Ma Bell while the internal suppliers within AT&T could sell to its competitors such as MCI and US Sprint. Several divisions were formed within AT&T in response to the breakup including the AT&T Microelectronics Division, the AT&T Network Systems Division and the AT&T Submarine Systems Division. Those divisions included the appropriate personnel from both Bell Labs and Western Electric. All the personnel at Allentown and Reading were included in the AT&T Microelectronics Division.

Around 1989, the AT&T Microelectronics Division itself was split into several smaller business units including Photonics, MOS IC's, Bipolar IC's, Power etc. and the leaders of those units had to meet challenging financial goals. The expectation was that each business unit would manage its own costs and eventually finance much of their own R&D. The IC businesses were not doing well because of focusing on switching applications only, as mentioned earlier. The Photonics business unit on the other hand was doing better because of the huge terrestrial system deployments made earlier and the undersea customer-supplier relationship was still not affected by all the break-up turmoil. The undersea lightwave components and subsystems continued to be sourced internally from the Photonics business unit within AT&T Microelectronics. This was assumed to be due to AT&T's leadership role in the consortium that deployed TAT8 and more undersea systems were planned for the Pacific Ocean.

A new AT&T building, the Solid State Technology Center (SSTC), was built in Breinigsville PA in closer proximity to Western Electric, Reading. It was constructed specifically for optical work and had been planned since 1983. It was naturally financially allocated to the Photonics business unit. The building was opened around 1987 and the Allentown development team, together with their new marketing team housed in Cedar Crest, were moved into the building by early 1989. While it was exciting to be in a new building, the conversations among some of the engineering personnel were more focused on whether the business unit could afford it. The building was nicknamed 'Bowers Towers' in honor of Klaus Bowers(11) who was a Bell Labs director in Allentown when I joined in 1968, and who had since been promoted to Vice President of the Bell Labs Electronics Technology Area. He was an excellent leader and familiar with all aspects of the optical and electronic technologies.

There was a sense that major changes were coming when Klaus appointed two ombudspersons to insure that personnel in his organization were treated fairly. Among many other things, the ombudspersons arranged for the new Photonics marketeers to be hired from volunteers within the engineering community. They had prior knowledge of the products and they became effective in the markets very quickly. We were all sent off on team-building exercises on the banks of the Rio Grande near San Antonio, Texas to unite, boost spirits and improve morale. AT&T Microelectronics also issued a new mission statement which stated that “Our Aim is to be the World's Best Development Organization for Electronic and Photonic Devices”. Such things helped in the short term, but the engineers were aware of the pitfalls ahead. Sadly, Klaus Bowers retired in 1990.

In the meantime, I had the opportunity to move within the subsystems department from supervising the receiver group to supervising an exploratory group. I took that opportunity. Also around that time Kinichiro Ogawa, who was my equivalent supervisor in the network system terrestrial design department in Holmdel was transferred to head the terrestrial subsystems department in Breinigsville. That was a commendable promotion since Kinichiro was a key person in the technical success of the FT Series G program and had a good understanding of system networks. I had interacted with him for many years ever since we were engineers evaluating the fine-line NMOS transistor for the receiver transimpedance amplifiers in 1980. We continued to interact at the supervisory level throughout the 1980's. He was without doubt the smartest individual out of the many that I had the pleasure of working with in Bell Labs.

Shortly after Kinichiro started in Breinigsville, he asked if I would take on additional responsibilities as a temporary Department Chief in Reading managing the optical receiver manufacturing line. The recently promoted supervisor on optical receivers was heavily involved in developing components for the FT2000 2.5Gb/s system and the new Next Generation Lightwave Network (NGLN) for ATT Network Systems. There was also a yield issue when using the Holmdel regenerator for testing the 1.7Gb/s receivers and Kinichiro wanted to insure Reading could ship sufficient receivers to meet the schedules at Merrimack Valley. I agreed and spent a few months splitting my time between Breinigsville and Reading. We fixed the receiver yield problem by using the latest customer board for the testing at Reading. My recollection from those hectic days was going a little too fast and colliding with a deer on the Mountain Road near Topton early one morning as I was commuting to Reading. Fortunately Kinichiro had allowed me the use of a rental car and although the damage to the side of the car was extensive, it was replaced quickly.

The exploratory group was originally formed by Mel Dixon to evaluate forward-looking optical subsystem opportunities. It was a group of bright, self-motivated young engineers with excellent future prospects. They had already done a lot of work with dispersion-shifted fiber at 1.55 micron, on improving the 'purity' and stability of lasers and on coherent optical systems where, if the purity and stability of the lasers were high enough, it would allow heterodyning the received signal with a local laser much like what is done with a local oscillator in a radio receiver. The group had also evaluated the alternate methods of modulating the signal laser with ASK or FSK (Amplitude or Frequency Shift Keying) techniques and concluded ASK (on/off) was adequate for the time being. They had also demonstrated the basic principles of dense wavelength division multiplexed systems (DWDM) with an even larger increase in the number of signal channels transmitted per fiber (12). This was done in co-operation with an exploratory optical passives group at Breinigsville.

Bell Labs research in Murray Hill had also been investigating the use of a semiconductor laser without the facet reflectors at each end as a single-pass optical amplifier. It was a novel idea but not particularly useful since the amplifying path was too short to realize significant gain. But it had kindled interest in an alternate option where the single-mode fiber itself was used as an amplifying element and hence the introduction of the Erbium Doped Fiber Amplifier (EDFA). Researchers at Southampton University in the UK had found that by doping the fiber core with the rare-earth element Erbium and flooding (pumping) the fiber with light at a wavelength of 1.48 micron, then the fiber could be converted into a gain medium for signals in the 1.55 micron window. With an adequate amount of pump power, there was sufficient confinement and length in the fiber to generate a large mount of gain. It was also found that it would faithfully reproduce the signal irrespective of how the signal was modulated. As the signal propagated along the Erbium doped fiber, photons were transferred from the pump at 1.48 micron into photons in the signal at 1.55 micron with the pump level decreasing and the signal level increasing as they progressed along the fiber. The EDFA was an important disruptive technological device in itself, offering essentially zero-loss fiber with unlimited bandwidth and thereby eliminating the need for high-speed electronic regenerators in long haul fiber systems. It moved to the head of the forward-looking work queue and the group showed how the EDFA not only could be used as an optical regenerator, but also as an optical power booster at the output of a transmitter or as an optical low-noise preamplifier at the input of a receiver.

Meanwhile another financial blow occurred to the Photonics business unit when the optical data-link development work completely collapsed. The optical data-link subsystems (ODL200) were focused on low cost, short distance applications <2km with multimode fiber, LED transmitters, silicon PIN detectors and silicon IC receivers at 200Mb/s for interconnections between computers or between adjacent buildings. There was an internal AT&T customer associated with the #5ESS switch for the ODL200 product but the main customer was a large US computer developer. The product was touted as a significant revenue generator for the Photonic business unit going forward. There were several delays and specification changes during the course of the ODL200 development but large volume production had started in Reading. Then apparently the business totally collapsed with the customer withdrawing from any further purchases. We were told it was not a technical issue, more perhaps of a business issue, but we never got a clear explanation of what happened.

The exploratory group continued to move forward evaluating EDFA gain control techniques and methodologies for flattening the gain over the band of interest. The EDFA provided gain from 1.53 to 1.57 micron and it was important to flatten that gain so that when the EDFA's were cascaded in a system, any signal within that band would be amplified by the same amount. Again, in co-operation with the passives group, optical filters were developed to flatten the gain. Kinichiro arranged for the group to conduct field tests on these new technologies in a month long string of tests over 35km of installed fiber cables between Sunbury and the Roaring Creek Station in Pennsylvania (13). The tests were successful in establishing the feasibility of the EDFA as a 'ready to be deployed' component and heterodyne detection coherent systems as 'practical when required'.

A model shop was created in Breinigsville for AT&T Microelectronics to fabricate and test EDFA's in small numbers for select customers. Those included a Fiber Amplifier Booster Module (FABM) for an AT&T terrestrial feeder linking an undersea cable in California to the US terrestrial network. Longspan was an 'island hopping' repeaterless submarine system with many potential future customers and a link between Israel and Cyprus in the Mediterranean Sea was its first application. All that system required was a high power booster EDFA at the output of the transmitter to provide a direct 'hop' between Cyprus and Israel. Fiber amplifiers models were also provided for the Wavestar project on an OC-192 self-healing ring network done in co-ordination with Bellcore and ATT SBC. There were also a myriad EDFA's provided for several customers for evaluation purposes. The EDFA certainly had an encouraging growth potential.

The ODL-200 debacle had now placed the business in a more precarious financial position and lunchtime discussions became more dire when retirement packages were offered by AT&T Microelectronics to senior members of the Bell Labs technical staff. Many took those offers since they were of retirement age and pension eligible. There were other engineers and technical managers like myself hired in the late 1960's, who were too young by a few years to qualify for a company pension. Most of those decided to stay put and weather the storm. Ma-Bell was still a highly regarded company with excellent benefits, a great health insurance program and an employee savings plan that had just been been rolled over into a matching 401K. It was not a company to just get up and leave without a lot of measured thought. Then in a period from around 1991 through 1993 there was a high level departure of Bell Labs leaders from throughout the Photonics business unit leaving hardly any leaders in charge with knowledge of the technology.

This was followed by a deeply disappointing period. I am not a judge of what qualifies for good business leadership but a few persons emerged from the post-1984 Western Electric marketing and sales (M&S) organization and became the new leaders of the Photonics business unit. They inherited assets that should have made the business immensely successful - an established portfolio of products with a huge growth potential, a world class supportive engineering team and at a time when there were hardly any competing businesses. But they blew it! They had little technical knowledge and worst of all they made little effort to learn. They proceeded to treat the engineering staff as second class citizens and it was disheartening to make a presentation to them when it was obvious they had no idea what you were talking about. Their only focus was sales and cost reduction, cost reduction, cost reduction. It was in vogue at the time to send engineers abroad to train foreign personnel how to manufacture product so those same engineers could afterwards be made redundant – a great cost reduction strategy! This was precisely what the earlier ombudspersons were supposed to monitor for - fair treatment for all! I am not sure if the practice ever happened in AT&T Microelectronics but there was shock, awe and frankly disgust associated with it. Manufacturing facilities were opened in Mexico, Singapore and China but it was a different time period then and transferring such critical technologies abroad most certainly would not be tolerated nowadays.

There were quite a few after-work meetings with Kinichiro and his supervisors in the Breinigsville Diner to discuss how to navigate interactions with those business leaders and to talk about personnel issues within our groups. Kinichiro was a caring individual and very devoted to the people working for him. He was aware that the younger Bell Labs engineers were looking for opportunities elsewhere and it bothered him a great deal. On the bigger scale, he was also aware that the internal cohesion that existed between Bell Labs organizations was disintegrating and meetings between divisions were now 'difficult'. In 1993, Kinichiro and his family decided to move back to live in New Jersey. He remained the department head for the exploratory group in Breinigsville and assumed more responsibilities for Network System research in Bell Labs, Murray Hill. He insured that key members of the exploratory group remained within the AT&T Bell Labs research organization. He was quite cool at handling the business people and the sparks would fly after he participated in their meetings and then slipped quietly out the door. He confided that he was all done with AT&T Microelectronics. I realized then that I had reached the peak in my own career within Bell Labs and that the enjoyment of coming to work every day was going downhill rapidly. I had no complaints since it had already been a very satisfying experience but I was certainly glad that I was now fully pension eligible.

In 1996, Lucent Technologies was formed from a combination of AT&T Microelectronics, Bell Labs and Western Electric. There was some initial euphoria about being part of a new company especially when stock options were issued to key people in an attempt to boost morale and counter the pressure from headhunters. The price of Lucent stock on Wall Street became the key barometer of progress and with it the perpetual short-term focus. The stock price was volatile and reached a peak in late 1998 and then started sliding. Thankfully most people had isolated themselves from investments in the company stock and had focused their savings etc on more conservative options. A few people however did go all-in with Lucent stock and were continuing to buy as the slide continued. They paid a hefty penalty later. The stock was sliding because the Lucent silicon IC business was faltering having been focused for too long on switching applications. I believe they had tried to shift their MOS strategy more in line with those LSI (large scale integration) strategies in Silicon Valley but had left it too late to catch up. They were nowhere near as advanced as those companies setting the trend and dominating Moore's Law by doubling the number of transistors integrated on a microchip per year. The Lucent optical business in comparison was doing better but external suppliers were making steady in-roads with the network systems customers.

In early 1998, there was sudden volatility among the leadership ranks of the optical business unit and the lead person together with the first lieutenant departed. The inside joke was that they left because of the poor prospects on their Lucent stock options. There was hope that the lead replacement might prove better but alas it was not to be. On top of that came the tragic news from New Jersey in December 1998 that Kinichiro had suffered an incapacitating stroke. He passed away six months later in June 1999. He was only 56 years old.

Then in early 2000 a major disintegration occurred in Lucent Technologies with the voluntary departure of seven key lead managers from the optical business unit. Chaos ensued and it was reported in the Bloomsberg Technical news that the departure could seriously hurt the business (14). Three of those managers opened a start up nearby to manufacture InP lasers financed with venture capital from Silicon Valley. A second startup followed later, financed locally in the Lehigh Valley, with plans to manufacture subsystems. There was a wealth of disillusioned engineering talent available in the area and Lucent was now facing competition on the optical front not only with Nortel and JDSU but also with those local start-ups. It was sad and humbling to see all this happen especially to Bell Labs. But its prestigious intellectual portfolio was still considered valuable and its research division lived on at the headquarters in Murray Hill. I resigned/retired from Lucent Technologies in December 2000.

This was not the end of my career in the optical technology business. I joined the second start-up as Director of Engineering in early 2001. The company was called OptronX. There followed for me three years of exciting work with a small team of focused engineers, marketeers and sales personnel. It was an enjoyment to go to work again but that is another story.

Looking back on those earlier Bell Labs years, there were plenty of people available within AT&T Microelectronics who had much more leadership skills, business acumen and technical know-how than those who were put in charge in the early 1990's. It was a sad story and perhaps it was bound to happen after the break up of AT&T in 1984. It was somewhat bemusing and pathetic however to read the embellished write-ups by those people about the financial growth they achieved during their tenure as leaders of the optical business unit ...etc. etc. The financial growth achieved in the 1990's was preordained by the work done by Bell Labs and Western Electric engineers during the 1980's, long before they arrived on the scene. The growth was handed to them on a plate and all they did was trade technical excellence for mediocre business management. They most certainly contributed to the eventual collapse of the photonics/microelectronics part of the business and the dismemberment of Bell Labs. Lucent Technologies was bought out by Alcatel in 2006 thereby terminating the telecom equipment industry as we knew it in the United States.

Bibliography
1 “Nefyn, Wales. Recollections from America”, Brian Owen, 2012
2 The Space Charge Varactor Solid State Electronics, Vol 8, 1965
3 The Compact Turnstile Circulator IEEE Trans MTT-18, No 12, 1970
4 Identification of Modal Resonances.... BSTJ, Vol 51, No 3, 1972
5 A Narrow Band Millimeter Wave Y-junction Circulator with Wideband Tuning Capability MTT-Symposium, Atlanta, GA 1974
6 Mechanically Tunable, Cavity Stabilized, Millimeter Wave Impatt Oscillators MTT Symposium, San Diego, CA 1977
7 An Experimental MM-Wave Path Length Modulator, BSTJ, Vol 50, No. 9, 1970
8 WT4 Millimeter Waveguide System.... BSTJ, Vol 56, No 10, Dec 1977
9 PIN GaAsFET Optical Receiver with a Wide Dynamic Range, Electronic letters, Vol 18, No 14, July 1982
10 Long Wavelength Optical Receiver using a Short Channel Si-MOSFET, Conf on Lasers and Electro Optics, Washington, 1981.
11 “Non Frangimur, My First Six Decades”, Klaus Bowers, 2015.
12 Coherent Communication Systems R&D at AT&T Bell Laboratories, Proc. SPIE, Jan 1991
13 1.2Tb/s WDM Transmission Experiment over 85km Fiber ...OFC 1998
14 Bloomberg Technology News, April 2000.

Dr. Brian Owen
Emmaus, PA, USA

Top of Page