Repeaters are a fascinating topic as they have been as instrumental to the early success of open wire as well as cable installation and operations growth, just as they also would accomplish for future 20th Century fiber optical systems. Through the latter development, a new moniker was introduced to telephony: a term by which “regenerator” or “optical line repeater,” or otherwise known as the OLR is used. But, let’s not get beyond our present introductory discussion. We need to define the device for our open wire fans. The repeater evolved comparatively as a worm to a fully developed butterfly, by unique features of its improvements in construction and operation over the many years. Let’s get started!
If you read through the chapter on Open Wire Loading within the Song website, you’ll notice a late mention of the “repeater” as the load coil evolved. Both work together.
Early Telegraph Repeaters
Telegraphy, having been born in 1843, advanced to a state of the art where nearly the entire North American continent was connected by wire. Each digital signal (the off and on of the key to the contact) and a battery allowed transmission to a distant sounder. This receiver could then deduce the code transmitted, transcribe it and answer accordingly.
These early transmissions were remarkable, but the signals could only travel only so far. Because of this limitation, telegraph repeaters were developed. They were semi-automatic. Essentially, a telegrapher at one office would send out a coded message. Instead of a physically non-connected “stop” at the receiver, a switch was moved. A battery at the location furnished additional power strength which compounded the voltage and sent a further improved signal to the next station. Telegraph repeaters operated very simply. There was only a need to understand the clicks and clacks of the sounder and send similar limited information forward. Repeaters were simple for telegraphy.
Launching the Search for a Workable Telephone Repeater
Telephony, on the other hand, opened a great challenge to engineers, because as we know the human voice has a great many differences in frequency and vibration. Many people strived to build a device whose characteristics in sound could be reproduced and re-conveyed with quality regeneration.
Herbert E. Shreeve of the predecessor of Bell Labs–the Western Electric Engineering Department–also was frustrated, until one day when he came upon a mechanical solution to the problem. He constructed an experimental device–or devices–and soon they were successfully tested on lines between Amesbury, Massachusetts and Boston. The line had previously suffered from poor intelligibility in speech and the equipment improved this fault remarkably in 1904.
Shreeve, and his colleagues, notably R. L. Jones, put the device through rigorous testing to improve the repeater and prepare it for practical production. It was good but not a quality instrument in their opinion. Before it could see practical application and production, some defects had to be cured. Major among them was the device’s lack of sensitivity. Loud speech was necessary to make it function and a whisper or lighter intensity of voice tone would make it useless. Voice is a function of current.
Furthermore, the device could not audibly communicate a person’s voice with any fidelity. If some pitches could be heard, others could not. If additional repeaters were on the pair in series to the conversing persons, the speech was rendered “unintelligible.” If the line were loaded, then the repeater failed to work . . . at all! Loading was crucial to long distance telephony and this restriction couldn’t be removed where the new repeater might be applied. While work continued to perfect this version of the repeater, and a bunch of newer models underwent testing, still other technicians pursued other divergent ideas for a workable device.
Chief Engineer for AT&T, J. J. Carty, wrote his company’s President, Theodore N. Vail, a long, detailed letter begging to apply further budgetary dollars towards a workable repeater for open wire and cable. Enumerating the various trial efforts disproportionate near-successes and failures, Carty insisted that with the development of a reliable repeating instrument for physical lines, such features might be applied to wireless as well. Radio was in its infancy, and similar non-wire issues of propagation were rearing their heads as well. Money to develop a successful repeater would have significant aftereffects on other technical developments, too.
Carty’s arguments prevailed and Vail released funds for a select group of engineers and scientists under Dr. F. B. Jewett to investigate the problem and solve it. Jewett, a Transmission & Protection Engineer at AT&T, gambled his time, effort and company funds to arrive at a satisfactory solution where repeaters of the early type might function with a loaded pair. With Jewett, Dr. E. H. Colpitts of the Engineering Department of Western Electric, combined his staff to fight this challenge. In doing so, the genus of this group evolved into the Research Department of Bell Telephone Laboratories, much later. This was during December-March 1910-1911. In 1912, Jewett became the Assistant Chief Engineer of Western Electric. Here, he and his fellows formally convened a major coordinated attack on not only the repeater but the design and completion of the first U. S. Transcontinental Toll Lead by 1915.
With an eager intellectual spirit and gung-ho attitude was Dr. H. D. Arnold, whom had been recommended by a Professor at the University of Chicago, R. A. Millikan, for his interest in the developing science of atomic research. He promptly joined the team tackling the problem of repeaters.
What would be considered a major problem in electronics versus electro-mechanical devices was “time base correction.” Although at the time, it was not called this. By creating a mechanical repeater, there was the need to allow for inertial energy, such as in the telegraph repeater and sounder, and deal with these effects in its operation. The team sought a solution in locating lightweight vibrating materials, where diaphragms were used. For example, these movable surfaces of a telephone receiver had to receive all frequencies within speech. In turn, the movable electrode of the carbon button transmitter then in use, had to be identically driven.
Arnold studied the problem from a non-mechanical perspective, inviting others at his offices to consider a molecular gas solution to repeaters. One day, he set up a glass tube with mercury vapor. Connected to terminals at each end, an electrical charge entered the glass tube internally where a stream of ionized mercury particles were carried. His onlookers watched as he tipped the stream sideways when the magnetic device acting as a telephone receiver, interacted. Within the tube were electrode plates battery-energized, also connected to a transformer. These together acted as a transmitter. Ionized particles of mercury were akin to carbon grains and acted similarly in the experiment.
Mercury Arc Rectifier Tube, on display at the Museum of Independent Telephony, Abilene, Kansas.
What happened was voice energy would activate current; the current would then cause a deflection in the energized gas stream. By the use of the battery and transformer, a substantial magnitude of voice receiving strength could be increased and reproduced by this method. Where no receiver current entered, the gas stream remained undeterred from its original anodes and cathodes.
Everyone was very impressed. As brilliant as this innovative technique was it still had drawbacks. Amplification was very, very splendid. However, as with the early carbon transmitters of Bell’s early experimental phone, it rendered speech discordantly, with a raucus tone. Several onlookers termed it “noisy.” While attempted on supervised lines it’s practical use was largely prevented by its “noisy” nature. And while disappointing as this revolutionary approach was, in the meantime, other remarkable developments were underway.
This came in the form of what we might consider the first electronic device, Dr. Lee DeForest’s audion. Grasping the full implications of the Council Bluffs, Iowa native’s remarkable experimental work, Jewett, Colpitts and Arnold leaped upon its unmistakable potential for regenerating distant human speech. As a telephone amplifier this technology was the key by 1908.
Developing the High-Vacuum Repeater for Practical Application
What was the audion? Similar to an early incandescent light bulb, this clear glass tube looked similar to a high pressure mercury vapor bulb, except four sealed wires extruded through the top of the device and within was a lamp filament. When energized, the electrons rush into the filament, some succeeding in entering into the near vacuum of space within the bulb. The electrically charged particles then frequently return to the filament by striking it. Generally as many electrons return and leave making the equilibrium constant. However, inside is also a metal plate. By connecting a battery creating a positive charge on the plate, the equal nature of the filament’s particles are drawn towards the plate. What then happens is an electron stream from filament to the plate and back to the filament through the connected battery’s conductors. No current in the filament plate circuit occurs when the battery is reversed, the plate repels the electrons. This “one-way conductivity” was not entirely unknown. Edison and Fleming both recognized the natural properties of this “thermionic emission of electrons.” In radio, Fleming had used the process like a “valve”. He realized the high frequency currents of a radio receiver would be conducted upon the filament-plate, offering alternative portions of the waves to be conducted. This “rectifying” of the alternating current made possible telephone receivers whose signal-bearing strength could act as a radio detector.
What DeForest accomplished, and which he could not adequately fully explain in his early years of its development, was the physics basis for the device’s actions. While mystified to some extent over his invention, his inspired tinkering created a device quite different than any other up to this time. He placed the wire grid within the tube. This influenced the electron stream on the plate circuit. When the device was used experimentally at Western Electric and with the attending experts, the modulating radio waves received were found to simultaneously reproduce within the receiver the same variations of speech coherently and accurately. The same was true when they applied the device to telegraphy.
Upon inspection of the device, and with patents in hand, DeForest offered the device to AT&T in 1912. But there was still work to be done before the device could be released for actual application.
One of the problems was that however astounding the reproduction qualities were of the experimental device over a closed, laboratory telephone circuit, in reality, all telephone circuits experience much interference and intruding currents of all nature. If voice production is too low or too high, it can be cancelled by noise.
The “Blue Haze” Effect
And . . . there were other problematic issues. One was the “blue haze” effect. Arnold knew the device had considerable potential for development, however one strange effect had been encountered in its use at Western Electric. Because the device had been used in radio previously and those transmission currents are remarkably low, the device had no challenge in reproducing them efficiently. But telephone currents were a different matter. Since this device would be, at its successful development, installed in many repeater huts along the Transcontinental as well as other long toll leads, telephone currents would be much higher than radio.
So, they tested the device where it would be expected to perform at a repeater hut with voltages and current most likely to occur. When applied, the device’s healthy glow electrically shuddered, appeared to dim and gasp, then produce an eerie blue smoke or haze. Transmission ceased at higher currents. However, speech transmission would healthily re-emerge after the transmitting current was reduced.
Arnold was quick to grasp at what was causing the “blue haze” effect. Because the stream of positive charges conducted within the tube increased when higher EMF was applied, he promptly came to a conclusion. The grid within the tube was a controller and had values which could not be exceeded. The current between the filament and plate would be effective only if kept within a specific range. The stream of electrons wildly meandered and lost control in these higher charged situations. The grid had no control over them. The haze was the result of this lost interaction. Writing later, Arnold reviewed the problem: “When the stream was controllable, it was a stream of electrons; but when it disallowed the authority of the grid, it was because ionized molecules of gas were taking part in the performance.” Continuing, Arnold surmised, “To permit stability of operation and accurate control, the current through the tube must comprise only electrons.”
Another issue clearly reared its head: the glass tube was insufficiently evacuated of air. It had a “near vacuum,” not an efficient vacuum. Much of what the tube contained was what we breathe. Now, it appeared the problem would reveal a need for industrial vacuum abilities to “clear the air” so to speak, within the tube.
Clearing the Air and the Legal Struggles Which Followed
The decrease of air within the tube and causing a near-perfect vacuum resolved the blue haze issue since it was evidence of spontaneous ionization, where electrons were forced to move in opposition to the electron stream. These collisions if decreased, would cause the filament-plate stream to be purely electronic. Arnold, with further contemplation, felt the device required re-design: a three-tube high-vacuum thermionic tube design was advanced to his colleagues which then became the first major step towards an amplifier and its use inaugurated as the U. S. Transcontinental lead was successfully built.
There were others who viewed this development at Bell with trepidation as Dr. Irving Langmuir of General Electric had designed a similarly functioning device and patented it at the same time. Arnold felt the device lacked the substantiality of an “invention,” rather “innovation” of existing know-how. A court case emerged, whose complications were expedited to the U. S. Supreme Court for a decision. This judiciary clearly determined, that for the device, calling it an “invention” was improper; but . . . had it been an “invention” Arnold would have been the father.
“Singing” Wires Seek A Solution
The team at Western Electric, proceeded to advance the device beyond its initial stages and perfect it. By 1918, 700 of these audion-type electronic repeaters were in use; half on open wire; the remainder on limited cable routes. Their success dictated further production and experimentation on improved units.
The mechanical repeater’s problems were well known and recognized. When applied to loaded toll circuits, there wasn’t enough amplification. If tweaked beyond a minimal amount of transmission gain, it squealed, sang and screeched.
Colpitts, Jewett and his colleagues agreed this problem resulted from how the repeater was physically inserted in the midst of the circuit. Since the repeater was amplifying from both directions where speech conversation was being produced, it couldn’t tell from which direction speech came? The repeater responded to this conflict by distributing amplification both ways!
It was a contest between two speakers of a conversation: if talk loudness was simultaneously identical in intensity, there was no problem. However, if one was unbalanced–in excessive loudness–it naturally amplified this speech current and in doing so was not equally divided. Instead the receiver got an earful of excessive noise.
Not only this problem arose, but the static created by the actual device, created noise even on a quiet line with no speech. The “singing” or “screeching” was due to its own inherent mechanical vibration preference rate. By singing at its own natural frequency, it didn’t matter whether it could be made more efficient, as increased efficiency caused further sonic effects. The singing could be deterred if the output current was divided evenly between the circuits on either side of its mid-line placement. However, such lines also were typically loaded, too.
Repeatered lines with loading coils appeared to have negative transmission effects when measurements were taken. The impedance-frequency characteristics appeared to be wildly different on the cables which were undergoing tests. Open wire was different, however and recognizing how repeaters might improve operation on them, fell to a gentleman by the name of R. S. Hoyt. Hoyt, along with two of his fellows, Otto B. Blackwell and C. A. Robinson, relocated their night-time experiments to a massive junction of open wire toll at Morrell Park (near Chicago), Illinois. This was a major test board operating center. These lines, for the most part contained considerable loading and obviously because of their varying lengths and styles of construction, no two were alike in impedance qualifications. Measurements were taken on impedance of many open wire lines.
Reviewing the charts of bell-shaped differences in electrical characteristics on open wire, one particularly striking consistency stood out to Hoyt: an open wire line with installed loading did not, within several nights, have the same characteristics, however, C. A. Robinson had found one unique feature common to all of these lines. These suspicions called for an investigation by all concerned. Henry S. Osborne, one of the pioneers of patented telephone open wire transposition design calculations, was also intrigued. Robinson’s question to Hoyt, who originally had created the charting, inquired, “Had an office loading coil been included in the circuit?”
What had occurred was this. This particular open wire lead was designed where typically a load coil would have been installed–instead, a termination office was located there. If a placed telephone call originated beyond that office and simply went through the office instead of ending there, a coil would normally have been cut in.
Osborne grabbed a pencil and began charting curves on a theoretical circuit where loading coils were added or removed from the circuit. Substantial irregularities ensued in the circuit without the typical placement of a coil. Osborne derived an equation to estimate loading irregularities and asked his associates to test lines using his impedance formula. With this new information in hand, Osborne’s colleagues hastened to derive new figures from similar open wire lines’ testing.
Could loaded open wire be smoothed out relative to impedance-frequency? Closed tests were made on a physical but artificially-lengthened loaded cable. Open wire, of course was different. Inductance would jump within two or three amperes of D.C. current. In fact, each open wire lead had constant differences in electrical characteristics when measured. The Newtown Square-Pittsburgh open wire toll lead underwent considerable testing by portable measuring equipment of its pairs under loaded conditions.
Other possibilities creating the likelihood of such disturbing irregularities were checked against lines in California and distant areas. Essentially, the problem was that loading coils were less efficient when composed of iron dust and segmented rings placed upon each other, were far more stable magnetically. This made design limits more predictable. Improved loading coils were manufactured and installed. Furthermore, advanced calculations specified far stricter installation along a line, with a less margin than before. Half of the span from one pole to another was the appreciable margin possible with these new coils.
In combination with these highly efficient coils came the advancement of the two-way repeater, known as the Twenty-one Type followed by a Twenty-two style. By the late summer of 1914, repeaters were being installed of the 22-Type along the Transcontinental U. S. lead which led to their adoption on many other major open wire stretches in North America.