History | IEEE Power & Energy Magazine
IEEE Power & Energy Society


The Rotary Era, Part 1

Early AC-to-DC Power Conversion

The early years of the development and expansion of electric power for residential, commercial, and industrial use was marked by intense and well-documented competition between the proponents of dc power and those supporting ac power. As ac emerged by the early 20th century as the preferred system and dc power began a long, gradual decline, it remained necessary to continue to supply existing dc distribution systems and provide power to such uses as dc streetcar and subway systems.

One important early means to convert ac to dc power was a complex machine known as a synchronous or rotary converter or simply a rotary. This two-part article, written by Thomas J. Blalock, a frequent contributor to these pages, investigates these now long-gone machines that allowed ac power systems to actually support their old rival systems during the decline and eventual extinction of dc power systems.

This is the 15th “History” article authored by Tom Blalock for publication in IEEE Power & Energy Magazine since the magazine’s inception in January 2003. Tom earned a B.S.E.E. degree from Lafayette College and an M.E.E.E. degree from Rensselaer Polytechnic Institute. His duties as a development engineer at the former General Electric High-Voltage Engineering Laboratory and later as a test engineer in the Transformer Test Department, both in Pittsfield, Massachusetts, included a broad range of duties, including lightning protection and high-voltage switching surge studies. Since retiring from General Electric, Tom has actively pursued his hobby of “industrial archaeology,” with particular emphasis on the exploration, preservation, and careful documentation of historically important and interesting electric power projects and equipment.

We are honored to have Tom back as our guest history author for this and the next issue of IEEE Power & Energy Magazine.

—Carl Sulzberger
Associate Editor, History

During the early 20th century, machines known own as synchronous converters, or rotary converters, were mainly responsible for the successful transition from early Edison direct current (dc) power distribution to alternating current (ac) distribution. Rotary converters were the most efficient means of converting ac power into dc power so as to be able to continue supplying established dc motor loads (see Figure 1).

figure 1. A typical traction rotary converter substation of Old Colony Street Railway Company, Brockton, Massachusetts, with three 750-kW rotaries at left (from Cyclopedia of Applied Electricity, 1913, vol. 1, p. 331).

figure 1. A typical traction rotary converter substation of Old Colony Street Railway Company, Brockton, Massachusetts, with three 750-kW rotaries at left (from Cyclopedia of Applied Electricity, 1913, vol. 1, p. 331).

Eventually, however, these unique machines were replaced, first by devices known as mercury arc rectifiers and, later, by solid-state rectifying equipment such as is still in use today for ac-to-dc power conversion.

Regardless, these “rotaries” were essential for several decades as coupling devices between the old dc power installations and the new ac power generating stations.

Early Developments

Thomas Alva Edison used dc for his early electric power installations because, at that time, it was the only viable alternative. In fact, even the respected scientists of that era labored under the mistaken belief that the ac power discovered by Michael Faraday during his early induction experiments would be useless for electric power distribution because any power flow during one half cycle would be negated during the subsequent half cycle.

Consequently, virtually all electrical development during the early and mid-19th century concentrated on dc equipment. Edison’s early success in the realm of electric power distribution meant that the use of dc became firmly entrenched, particularly in the downtown areas of large cities.

During the late 19th century, the inefficiency of dc power distribution became apparent, and the early erroneous opinions regarding ac had been dispelled. The work of, primarily, William Stanley and Nikola Tesla had led to the availability of both a practical transformer and a functional ac motor so that the use of ac distribution became superior to dc.

An inventor named Charles S. Bradley is generally credited with the invention of the rotary converter as early as 1888. However, during the early 1890s, Benjamin G. Lamme, of the Westinghouse Electric & Manufacturing Company in Pittsburgh, Pennsylvania, also laid claim to its invention. It is certainly likely that the Westinghouse Company was responsible for improvements to the basic design of these machines during this period.

Bradley had worked with Edison in Menlo Park, New Jersey, but subsequently formed his own Bradley Electric Power Company in Yonkers, New York. However, in 1893, the recently incorporated General Electric Company bought Bradley’s small company to acquire his rotary converter patent. By then, Westinghouse was manufacturing rotary converters, and, in 1896, the inevitable squabbling between the two companies regarding patent rights was resolved as part of an extensive cross-licensing agreement that covered all forms of electrical equipment.

figure 2. An operator lubricating slip rings on a large rotary converter (photo courtesy of Robert W. Lobenstein).

figure 2. An operator lubricating slip rings on a large rotary converter (photo courtesy of Robert W. Lobenstein).

It was (and still is) possible to convert ac into dc by means of a straightforward motor-generator set. However, the overall efficiency of such a set is the product of the efficiencies of the individual machines. The rotary converter was a single machine that combined the functions of motor and generator and, so, inherently exhibited a significantly higher efficiency than a motor-generator set.

Regarding its function as a motor, a rotary converter was essentially an inside-out synchronous motor. That is, its rotor (armature) contained a multiphase winding that was fed via a set of slip rings on the ac end of the machine, and the stator then contained a dc field winding. The generator function was provided by a commutator and brushes on the dc end of the armature. Thus, from that end, it resembled a conventional dc generator. In fact, the earliest experiments at Westinghouse involved the addition of slip rings to a standard dc railway generator (see Figure 2).

Since the slip rings and the commutator connected to the same armature winding, there was a direct electrical connection through the machine. This meant that there was a fixed relationship between the ac voltage applied to the slip rings and the dc output voltage, with the latter being slightly less than the peak value of the former (on a sine-wave basis). The two major (but by no means the only) uses for rotary converters were to supply Edison dc distribution networks at 250 V and streetcar trolley wires at 500 V. For 500-V dc, the ac voltage applied to the slip rings had to be about 370 V and, similarly, for 250-V dc, the ac voltage had to be about 185 V.

figure 3. A rotary converter installation with air-blast step-down transformers and a blower for cooling at left (from Cyclopedia of Applied Electricity, 1913, vol. 2, Forward page).

figure 3. A rotary converter installation with air-blast step-down transformers and a blower for cooling at left (from Cyclopedia of Applied Electricity, 1913, vol. 2, Forward page).

Because these are decidedly nonstandard voltages, a bank of step-down transformers was always required for use with a rotary converter. Even at the turn of the 20th century, however, transformer efficiencies were in the high 90% range. Thus, the need for transformers did not adversely affect the overall efficiency of a rotary converter installation. Most often, the transformers were located inside the converter substation so as to minimize the lengths of the low-voltage, high-current connections between the transformers and the converter. It was not good practice to use oil-filled transformers inside due to the fire hazard, so dry air-blast transformers were normally used, with the cooling air supplied from an electric motor-driven blower (see Figure 3).

The “Hunting” Problem

The 1893 World’s Columbian Exposition in Chicago, Illinois, was a very important world’s fair. It was conceived as a celebration of the achievement of Christopher Columbus 400 years earlier, but it opened one year late due to construction delays.

The Exposition was to be entirely illuminated by electric lights, and the electrical industry was rather startled when Westinghouse managed to wrestle the contract for this endeavor away from General Electric, which had in its possession all of Thomas Edison’s electric light patents. In fact, Westinghouse had to develop a short-lived type of electric lamp (the so-called “stopper lamp”) to avoid infringing on the patents held by General Electric.

Westinghouse created an immense display in the Electricity Building of the Exposition, which included steam engine-driven ac generators that provided the power to light the fair. As a part of this exhibit, Westinghouse included rotary converters to demonstrate the ability to change ac into dc for use with conventional dc motors.

It was noticed that one of these machines exhibited a peculiar “beating” sound with accompanying sparking at the commutator under certain load conditions. The operators managed to deal with this matter so not much attention was paid to it at the time. That was, however, the first indication of the problem of what is termed “hunting” in rotary converters that eventually proved to be very problematic in larger machines.

Then, in 1895, Westinghouse provided two rotary converters to supply the excitation for the ac generators that the company had constructed for the first significant power station at Niagara Falls in New York. The hunting problem manifested itself rather severely in these machines, and, consequently, the machines proved to be unusable.

It turned out that other sources of excitation would have had to be provided anyway because the rotary converters operated from the ac output of the generators that they were to excite. Thus, there would have been no way to start up the station. When this basic fact was first realized, provision was made for the temporary use of a dc generator in a nearby construction shack to provide excitation needed to start up the station for the first time.

As a result of the problems with the Niagara exciters as well as with rotary converters that Westinghouse had built for chemical factories at Niagara Falls, Benjamin Lamme and Charles Scott (both Westinghouse engineers) embarked on an extensive study of the hunting problem.

They determined that the cause of the phenomenon was a repetitive shifting of the magnetic flux distribution in the salient (projecting) magnetic poles of the field winding on the stator of the machine. In effect, this was causing the machine to cycle back and forth between generator and motor operation with consequent small variations in speed that manifested themselves as the audible hunting behavior.

At this same time, General Electric built (using Bradley’s patents) several large rotary converters for the Pittsburgh Reduction Company at Niagara Falls for the production of aluminum. These converters did not suffer from the hunting problem. The only major difference between the design of the General Electric and Westinghouse machines, as far as their field structures were concerned, was that Westinghouse used laminated field poles while General Electric used solid poles (following standard dc generator practice).

The solution to the hunting problem, pioneered by Lamme and Scott, turned out to be the installation of copper damping coils on the salient field poles. These coils effectively short-circuited the periodic magnetic field variations causing the hunting behavior. It so happened that the solid field poles used by General Electric in their converters provided this same damping action. In later years, however, when General Electric shifted to laminated field poles to reduce the eddy-current losses in larger machines, they then had to resort to the use of copper dampers on the field poles as well.

The concept of such a damper was patented in 1892 by two French engineers, Maurice Hutin and Maurice LeBlanc, both of Paris. They also held several patents in the fields of telephony and electrical resonance (there was another Maurice LeBlanc, who was a noted French novelist). As a result of this French origin, the damper winding used in rotating electrical machines was known as an “amortisseur” winding throughout the 20th century. Amortisseur translates roughly as “to deaden,” which is in keeping with its function to damp out or deaden various forms of magnetic oscillations.

During early investigations using a standard dc generator modified by the addition of slip rings to become a rotary converter, Lamme and his associates noticed another peculiar result. The same machine could handle a much larger current when operating as a rotary converter than it could when operating as a dc generator; in one case, 400 A versus 280 A. These limits were referred to by Lamme as the “smoking currents,” reflecting an apparent standard practice of the time wherein the maximum current a machine could handle was a current slightly less than what caused noticeable smoke to issue from the windings.

Lamme claimed that a detailed mathematical analysis of this phenomenon was carried out by one of his associates but that they never received credit for this effort. He claimed that exactly the same analysis was carried out years later by others, most probably by General Electric engineers, and that this later effort received “very considerable” credit.

Rotary Converters at Niagra Falls

A world’s fair, the Pan-American Exposition, was held in Buffalo, New York, in 1901. By that time, the electric power development of Niagara Falls had become quite extensive, and a presentation on this topic was given by Philip Barton, a representative of the Niagara Falls Power Company.

Barton’s presentation included an elaborate schematic drawing showing in detail all of the various electrical loads supplied by power generated at Niagara Falls. A majority of these loads used dc supplied by rotary converters.

In that era, it was common practice to define the capacities of electrical machinery in terms of horsepower (hp), rather than kilowatts (kW), with 1 hp being equivalent to 0.746 kW. Barton’s schematic indicates that a total of 5,000 hp in rotary converters were installed at the Pittsburgh Reduction Company in Niagara Falls for the production of aluminum (this company later became Alcoa).

Also, locally in Niagara Falls, three chemical companies used a total of 3,000 hp in rotary converters to supply dc for electrolysis processes that produced chemicals such as potash (potassium hydroxide), sodium, and caustic soda (sodium hydroxide).

The Natural Food Company used a total of 2,500 hp in rotary converters in its factory that produced its famous shredded wheat breakfast cereal. Perhaps adjustable speed dc motors were needed for mixers or for conveyors through continuously fed ovens.

The ac from the Westinghouse generators at Niagara Falls was stepped up to 22,000 V and sent to the city of Buffalo, a distance of about 25 mi (40.2 km). There, rotary converters were used to supply dc for the operation of street railways. The same was true in the nearby towns of Lockport and Tonawanda, as well as on the Canadian side of the Niagara River. The total streetcar load fed by rotary converters at all locations was nearly 10,000 hp (about 7.5 MW), which included a streetcar line in the Niagara River gorge itself.

In addition, the Lockport Gas and Electric Company used a rotary converter to supply the Edison dc distribution for lighting. In Buffalo, however, a conventional motor-generator set was used for this purpose.

Eventually, the rotary converter exciters at the Niagara Falls power station were fitted with Lamme’s amortisseur windings and, so, became usable. Also, a 500-hp rotary converter from the 1893 Chicago Exposition was installed in this power station, presumably to supply dc for the local streetcar line.

Six Phase and Three Wire

During Lamme’s investigation into the fact that a rotary converter could handle more current without overheating than the same machine used as a conventional dc generator, it was learned that the greatest armature heating occurred near the points where the three-phase slip rings connected to the armature winding.

This naturally led to the idea of supplying current to the armature at more than three points. Since step-down transformers were always needed to operate rotary converters, the secondary connections of these transformers could be used to provide six-phase power from the available threephase power. The feeding point heating would then be more evenly distributed around the armature.

The simplest means of accomplishing this was to connect the six transformer secondary leads, two from each phase, to six slip rings on the converter in the proper sequence.

Each phase would connect to points diametrically opposite each other on the armature winding. Logically, this was referred to as a “diametrical” connection. It effectively supplied six voltages to the armature, which differed from each other by 60°.

If the step-down transformer primaries needed to be wye-connected, then a so-called “double-delta” secondary configuration could be used to provide closed paths to circulate third-harmonic currents. However, this connection required two isolated secondary windings per phase. Each delta, then, fed three of the six slip rings.

The complication and expense of providing six rings and six brush assemblies meant that six-phase operation usually was resorted to only on the largest capacity machines where every means of reducing their size and weight was of value. The more even distribution of heating in the armature meant that less core material was required to dissipate that heat. In addition, a smoother dc voltage waveform would be obtained.

Rotary converters used to supply Edison 110/220-V, three-wire dc networks required a neutral connection in addition to the positive and negative brush connections. Such a connection could be provided via the step-down transformer secondaries so as to provide a path for neutral current to flow into or out of the armature winding.

For three-phase rotaries, simple wye-connected secondaries could not be used to provide a neutral connection because the resultant flow of dc neutral current in the transformer windings would tend to saturate the magnetic cores of the transformers. Accordingly, a zig-zag, or interconnected, type of wye winding had to be used to prevent this.

For six-phase diametrical rotaries, the center points of each transformer secondary could be tapped and connected to provide a neutral connection. The dc current flow in one winding half was opposite that in the other, so the same effect was obtained as with a three-phase zig-zag connection.

Starting Rotaries

In the early 20th century, electric power systems often were not yet sufficiently robust to allow for the full voltage starting of large machines such as rotary converters. Since rotary converters were equipped with amortisseur windings on their stators to prevent hunting, as discussed previously, they were capable of being started as induction motors just as was done with large synchronous motors.

figure 4. Chicago City Railway Company rotary converter substation with pedestal at right for two three-pole, double-throw starting switches (from Cyclopedia of Applied Electricity, 1913, vol. 6, p. 379).

figure 4. Chicago City Railway Company rotary converter substation with pedestal at right for two three-pole, double-throw starting switches (from Cyclopedia of Applied Electricity, 1913, vol. 6, p. 379).

However, reduced voltage starting had to be used to minimize the disruption to the ac power system supplying the machine. Since step-down transformers were always required with rotaries, taps on the secondaries of these transformers were used to provide the reduced starting voltage. Only a single reduced voltage was necessary for smaller rotary converters, but two reduced voltage steps (about 1/3 and 2/3 running voltage) were needed for larger machines (see Figure 4).

Huge three-pole, double-throw knife switches, mounted on vertical slate panels, were usually used to provide the switching from reduced to full running voltage. These switches were located between the adjacent transformers and the rotary to minimize the lengths of high current connections. A 1,000-kW rotary converter, for example, that supplied a streetcar line at 500-V, 2,000-A dc would require about 1,500-A, threephase, ac for its operation.

One problem encountered during the ac starting of a rotary converter was that a high voltage would be induced into the unenergized field winding due to its many turns. Accordingly, the field winding was temporarily broken into several sections during starting to minimize this induced voltage and avoid puncturing the field winding insulation. A small multipole knife switch, called a field break-up switch, was mounted on the frame of the machine to provide this function.

figure 5. Three rotary converters at the New York City Transit Authority’s Nostrand Avenue, Brooklyn, substation with a starting motor-generator set and two battery-charging motor-generator sets in the foreground (photo courtesy of Thomas J. Blalock).

figure 5. Three rotary converters at the New York City Transit Authority’s Nostrand Avenue, Brooklyn, substation with a starting motor-generator set and two battery-charging motor-generator sets in the foreground (photo courtesy of Thomas J. Blalock).

In large substations housing several rotary converters, any rotary could be started from the dc end using the power available on the station’s dc bus bars, supplied by the rotaries that were already running. A multistep starting resistance was provided on the station switchboard, which could be connected to any rotary converter to enable it to be gradually brought up to speed while operating as a dc shunt motor.

When close to running speed, the rotary would be generating an ac voltage at its slip rings, and it would then have to be connected to the ac power system with the aid of a conventional synchroscope. When synchronism was indicated, the remotely controlled circuit breaker in the circuit supplying the step-down transformer primaries would be closed.

A complication with the dc starting of rotary converters supplying streetcar or subway lines was that significant dc voltage fluctuations often appeared on the station bus bars due to the constant starting and stopping of cars or trains. This fluctuation would make it difficult to synchronize a rotary to the ac line when starting.

Accordingly, in such substations, a small (that is, relative to the size of the rotary converters) motor-generator set, called a starting set, was sometimes provided. This consisted of a dc generator driven by an ac motor. It supplied a more constant voltage for the startup of the rotaries (see Figure 5).

figure 6. An induction starting motor on the ac end of a 2,000-kW rotary converter at the New York City Transit Authority’s Prospect Park, Brooklyn, substation (photo courtesy of Thomas J. Blalock).

figure 6. An induction starting motor on the ac end of a 2,000-kW rotary converter at the New York City Transit Authority’s Prospect Park, Brooklyn, substation (photo courtesy of Thomas J. Blalock).

Other small motor-generator sets often were required in such substations to charge the station battery that provided 125-V dc for the operation of circuit breakers and similar control functions.

A large rotary converter could also be brought up to speed by means of a dedicated starting motor on the end of its shaft. Usually, this would be an ac induction motor having windings with two fewer poles than the rotary converter to assure that the latter could be brought up to, or slightly above, its synchronous running speed. It could then be synchronized to the ac power system as with dc starting (see Figure 6).

Part 2

The further development and use of rotary ac-to-dc converters will be discussed in the second and concluding part of this article in the November/December 2013 issue of IEEE Power & Energy Magazine.

For Further Reading

D. O. Woodbury, A Measure for Greatness. New York: McGraw-Hill, 1949.

H. C. Passer, The Electrical Manufacturers: 1875–1900. Cambridge, MA: Harvard Univ. Press, 1953.

B. G. Lamme, “Sixty cycle rotary converters,” Elect. J., vol. 10, no. 11, pp. 1124–1138, Nov. 1913.

F. D. Newbury, “The engineering evolution of electrical apparatus—Part IX,” Elect. J., vol. 12, no. 1, pp. 27–38, Jan. 1915.

B. G. Lamme, “The technical story of the synchronous converter,” Elect. J., vol. 17, no. 2, pp. 55–60, Feb. 1920, and vol. 17, no. 3, pp. 91–99, Mar. 1920.

“The Waterside station of the New York Edison Company—Part IV,” Elect. World, vol. 39, no. 5, pp. 191–194, Feb. 1902.

C. Payne, New York’s Forgotten Substations. New York: Princeton Architectural Press, 2002.

D. Ellison, “Hail and farewell to two historic stations,” Around the System (Consolidated Edison Company newsletter), Nov. 1977, pp. 7–9.

Cyclopedia of Applied Electricity, 7 vols., 6th ed. Chicago, IL: American Technical Soc., 1913.

In This Issue

Feature Articles

Departments & Columns

Upcoming Issue Themes