Pittman Model DC81 Per-Mag Motor
This paper will attempt to describe for the non-technical reader the characteristics of common miniature DC motors with particular attention to their application to model railroad locomotives, to describe the differing types of signals used to control them, and to explain the widely accepted need for AC modulation of the "DC" control signal. The treatment will begin with simplified theoretical foundations, and proceed through historical applications to the tradeoffs inherent in today's competitive commercial marketplace.
The modern "DC" permanent magnet field, wound armature design of motors used in contemporary model railroad manufacture is a peculiar device with a number of problematic design features that would overcome its few fortuitous features if only a more efficacious or economic design could be found. As it is, we have a device that will produce propulsion power for miniature locomotives by the application of relatively small amounts of electrical power, whose speed can be controlled by control of the electrical current, and whose direction can be reversed by reversing the polarity of the applied current.
The drawbacks of the design are a high operating speed, necessitating high gear ratios and their consequent noise, friction, and expense; their generation of waste heat within the often narrow confines of the locomotive; their often non-linear operating characteristics, including cogging; and their relatively low efficiency.
They are however lauded for their improvement over preceding designs due to both evolutionary and revolutionary improvements. Compared to their immediate predecessor designs, modern motors have more poles (five, and even seven now being commonly found), skewed armatures to reduce cogging, improved magnetic materials, long-life brushes, near-permanent lubrication, and enclosed cases to exclude dust.
While these incremental improvements are significant, the largest improvement in model railroad motors occurred with the adoption of the permanent magnet field to replace the wound field, or "universal" designs that preceded them. The wound field motors were larger, dissipated more heat, and required either an on-board rectifier or a sequencing relay mechanism to provide controlled reversing. While this last could be done more easily today with solid state electronics, the bulk and cost would still doom the universal motor to obsolescence, despite their ability to operate from either AC or DC with impunity.
While there are a small number of other miniature motor designs possible, few have seen commercial application in the hobby. The "pancake" design has never been given a fair trial, as torque requirements generally require more windings than can reasonably be provided in a single (or double) layer, and all known examples of this design were but oddly shaped conventional designs. The "coreless" motor, where the armature consists of only the windings without any supporting magnetic material, has been tried with generally poor results due most often to the low thermal mass of the armature that makes them more prone to burnout under stall conditions--which are not infrequently encountered in this hobby.
While the nominal theoretical application of per-mag motors is for operation at a fixed design speed on pure DC current, the application of such motors to model railroad applications requires varying speeds, with particular attention to low speeds (compared to the nominal design value) and smooth starts. Further complicating matters is a concern with low cost in both the motor and the power source.
These economic concerns, especially, have lead to a number of economies (one might even say liberties) being taken in motor design. The most frequent design met, particularly in units produced for the low-end market, are three-pole, non-skewed motors with poorly shaped commutators and unshaped brushes. These same motors are expected by many naïve buyers to run alongside and to provide performance equivalent to better designs, and to operate from the same power source with equal satisfaction.
While consumers really should expect to get what they pay for in most cases, their expectations are a reality of the marketplace, and must be taken into serious consideration by manufacturers who do not have sufficient market domination to impose new expectations on their customer base. What this means is that new locomotives must be designed with due consideration for the installed base of power supplies, and that new power supplies must likewise be designed with due consideration for the installed base of locomotives--for better or worse. Standards, such as those promulgated by hobby organizations like the National Model Railroad Association (NMRA), standards groups like UL and CSA, and national standards agencies must also be observed in order to obtain market and governmental acceptance.
A number of characteristics of the motor are of concern in the design and operation of not only a successful model locomotive, but also in the design and operation of a successful power supply. Let us examine these characteristics and their importance to performance.
The torque, or ability to do work, produced by a per-mag motor is a function of the instantaneous current less electrical, magnetic, and frictional losses.
The speed produced by a permanent magnet motor is a function of the average current less electrical, magnetic, and frictional losses; and the load applied. While a theoretical motor would be perfectly linear in its speed response, starting and low-speed non-linearities in friction and armature cogging are the main causes of speed response being non-linear to the current applied. As many power supply designs control voltage rather than current (and that at the power pack rather than the motor), further non-linearities in response are introduced unless compensated for by the power supply design (such as "taper-wound" rheostats).
Note particularly the phrase in the above paragraph that says "average" current. The time period over which the speed-controlling current is averaged is equivalent to the mechanical time-constant of the armature and attached load, and this feature will be put to good use later in this paper.
Electrical losses are almost entirely resistive. They are thus a linear function of current, and may also generally be neglected, except as they are a prime source of heating, discussed below.
The magnetic losses in per-mag motors are of two types: hysterisis losses, which are a function of changes in current and can largely be neglected at normal frequencies; and cogging, the degree of which is a function of design (number of poles and amount of skewing) and the effect of which is felt most strongly at low speeds.
"Magnetic losses" as it is used here should not be confused with a loss of magnetic force in a motor's permanent magnet. While all magnets are subject to loss of magnetization with time, and while the application of external magnetic fields (such as those from a motor's armature) will hasten this loss, modern magnetic materials do not show measurable loss of magnetization under normal service. Any motor that did suffer such a loss during normal service would have to be regarded as having been built with inappropriate materials. (For a definition of "normal service", see remarks below about NMRA Standards.)
Frictional losses are often the most significant. While some frictional losses are linear functions of speed, with perhaps a temperature factor, many frictional losses are non-linear. Starting friction in particular is different from and often much larger than running friction, and gear loads can also vary with the angle of the gears, with this variation being most significant at lower speeds. These non-linearities, while of only minor concern for constant-speed applications, make starting friction a serious factor in obtaining satisfactory operation in the variable speed application of model railroad propulsion and especially the low speed operation often encountered.
Note that both magnetic and frictional losses are worst at low speeds, with some frictional losses incurred only on starting (not on slowing). It is these low-speed losses that are of greatest concern to model railroad engineers and designers as they are the source of most complaints about poor operation (excepting only electrical pickup problems, which are outside the scope of this paper). Even the most poorly designed motor and power supply will generally give satisfactory performance at maximum speed.
Heating is of particular concern in model railroad applications due to confined spaces within locomotives and the low-melting point plastics used in some models. Heating is produced by frictional, resistive, and magnetic losses. Generally speaking, friction heating is a function of speed; resistive heating is a function of the RMS current (nearly, but not quite equal to the average current defined above as controlling speed); and magnetic hysterisis heating is a function of changes in current, or the AC component of the drive signal.
Note that there is a difference in heating between a filtered DC signal that produces only RMS heating, and an AC signal of whatever form with the same average value--not only is the RMS value different from the average (though not always larger), but the magnetic losses caused by the AC component are also added. This is why concern for the "excess" heating effects of AC is a legitimate--though generally insignificant--point of interest.
Cooling must also be taken into consideration, and the design of the motor, its mounting arrangement, and the ability of the locomotive's mechanical design to dissipate heat by convection and radiation all contribute to the determination of the maximum dissipation that can be tolerated before the temperature of either a motor or locomotive part exceeds its maximum safe value. While the writer knows of no small-scale locomotive designs that include active cooling systems (such as fans), the earlier trend toward resilient motor mounts and low melting point plastic locomotive superstructures now seems to be turning in the direction of higher thermal conductivity and higher temperature materials.
Thermal time constants must also be taken into account, as should changes in common operating patterns. A design with a long thermal time constant that might endure an hour of intermittent, heavy-load operation on a conventional home layout could well fail during ten hours of continuous, medium-load operation on a modular layout at a public show.
Any mechanical object when mechanically excited by a periodic disturbance (such as a motor subject to intermittent pulses of current) will vibrate, and when the frequency of that vibration is within the range of human hearing and of sufficient volume, the result is noise. This can be a good thing, as with the speaker cone of a radio; or a bad thing, as when a singer shatters a glass with a high note. The difference being the intent of the application and the presence of damage.
Obviously, we are not interested in our miniature motors acting as speakers, particularly for a power line frequency hum. Some such noise will be the inevitable result of applying any varying signal to the motors, but such noise is not in itself harmful, and while occasionally objectionable, it is one of the tradeoffs to be made in locomotive design.
Noise and vibration can be harmful when they are so extreme as to cause physical damage to the locomotive (or the operator), but is this in itself likely? Usually it is necessary for the frequency of excitation to match a resonance of the mechanical system (or one a harmonic of the other) for simple "noise" to be amplified to damaging levels. One then need ask if the normal resonances of model locomotives or motors are near (harmonics of) the frequency of excitation. As most power supplies, for economic reasons, use AC signals that are equal to or twice the power line frequency of 50 or 60 Hz, reputable model manufacturers will test their motors and locomotives for resonances at these four frequencies and make suitable design changes if a potentially harmful resonance is found. Most model locomotives have much higher frequencies of resonance, and thus this becomes of concern only when high-frequency AC signals are applied.
The early days of the hobby of model railroading saw most locomotives powered by pure DC from storage batteries, as shown in Figure 1: Pure (Filtered) DC. Indeed, it was the change of the American automotive battery standard from 6 volts to 12 volts that incited a corresponding change in model voltage standards in the 1940s. With the passage of time, widespread availability of commercial electric power, and increasing technical sophistication in the hobby (not to mention the hazards inherent in the use of lead-acid storage batteries in the home), a gradual change was made to power supply designs based on transformers and rheostats with an output waveform similar to that shown in Figure 2: Rectified AC. |
As early hobby motors were universal, or wound field designs, there was little difference between AC and DC performance, particularly as the hobby was still in a state of amazement that anything would run at all. With the post-war development of improved per-mag (DC) motors, they were rapidly adopted as the de-facto standard largely because they could be reversed without use of the bulky and troublesome relays or rectifiers of the day. |
The earliest DC power supplies simply added rectifiers and reversing switches to the basic transformer/rheostat power pack so that it would produce DC. While some of the more careful modelers did install filter capacitors to smooth the output, as shown in Figure 3: Filtered AC, others tried doing without for reasons ranging from economy to experimentation to ignorance. |
It was soon discovered that the performance of most locomotives--particularly at low speed--was better when the full-wave DC was not filtered. The resulting waveform remained of one polarity, and provided an average DC voltage that produced unidirectional rotation of the motor. Because of the non-linear characteristics of the motor, particularly at low speed, this waveform provided better control, a lower overall minimum speed, and a starting speed closer to the stopping speed (which had previously be much higher due to starting friction).
The reason such a varying signal produced better performance is that given the same average voltage as a corresponding DC signal, and consequently the same nominal speed, the maximum torque produced by the peaks of the AC signal were sufficient to overcome starting friction where the torque produced by the constant DC voltage would not. By pulsing the signal to a value producing high torque, the starting friction would be overcome, and the motor started to turn. By then lowering the voltage below the equivalent DC value, the time-weighted average current was kept low, and with it the resultant speed.
In addition, the difficulty of economically obtaining pure DC from conventional AC power sources offers further incentive to deviate from pure DC current. As the time constant of most motor/gearing arrangements is sufficiently long that 50 to 120 Hz signals can conveniently be used, the opportunity for economic applications is significant.
|
|
Many power packs have been built that incorporate both of these waveforms, selected by a switch, and a number of innovative designs gradually switch from half-wave at low speed to full-wave at mid-range to take advantage of the best of both, as suggested in Figure 5: Gradual Pulsed Power.
Since Linn Westcott's pioneering work in this field in the early 1960s, much attention has been given to designing waveforms and frequencies that will give superior performance, without incurring unreasonable costs in motor heating, noise, or power supply complexity.
Some throttles were designed and marketed that generated square-wave signals, and controlled speed by adjusting the duty cycle, and thus the average DC value, as shown in Figure 6: Square Wave Power.
These proved to be unsatisfactory, because at medium to high speeds, the DC heating value was increased by the RMS value of a waveform with a peak-to-peak value equal to the maximum voltage. The resulting heating exceeded not only the DC heating effect at full throttle, but the full-wave heating effect at an equivalent throttle setting. Few such models remained long on the market when first introduced.
Today, most if not all DCC throttle receivers use such pulse-width modulation for throttle control. The direct performance is no better with them that it had been with conventional throttles of this design, but the operating superiority of command control has to the largest extend hidden this problem from view. DCC systems compete on a wide variety of features, but I have yet to see one advertise NMRA Standard-compliant motor drive waveforms (but I would love to hear from any manufacturer who does meet those Standards).
Are there then other AC signals that could be put to practical use in model railroad propulsion? In his seminal 1962 work on third generation power supply design (the so-called "True Action Throttle" or TAT), Linn Westcott described a throttle that produced a train of short pulses at near power-line frequency (he recommended 40 Hz) from starting voltage to mid-range, at which point the AC pulses would submerge under a rising DC value. This produced high torque at minimum speed for smooth starts and reliable low-speed operation, a gradual removal of pulses so that medium and high-speed operation is not burdened with additional (and unnecessary) AC heating, and pure DC for full-speed operation, as shown in Figure 7: TAT Output.
However, these improvements came at the cost of much greater complexity in the power supply. While evolution in motor design and electronic control circuits have brought about changes in the TAT circuit, no one has yet credibly challenged the basic principles or the preferred waveforms that Westcott specified.
It is reasonable to suspect that these principles are due for revision due to dramatic developments in the state of the art of miniature motor design in the subsequenty thirty years. The photo at th ebeginning of this paper, showing a Pittman #DC81, was chosen to represent the class of motors on which Westcott's research was necessarily based.
All of the waveforms discussed above are unipolar--while they vary in amplitude, current flow is never reversed. Not all waveforms suitable for DC motor control are unipolar, however. Outside the field of model railroading, there are a number of examples of the use of bipolar AC waveforms to control "DC" motors. One common technique is used where it is desired to have a motor hold a fixed position or a fixed, low speed against a varying load. The impossibility of using pure DC for such purposes can easily be seen from the fact that for zero speed, DC control signals have zero voltage, thus zero power, and cannot therefore hold a fixed position against a load. Controlled low-speed operation also suffers from a lack of power available from a low DC voltage.
In these applications, it is common to add a relatively large AC holding current to the DC control signal. In some applications, the AC signal can have a peak-to-peak value many times the average DC signal. A suggested waveform for such a model railroad throttle is shown in Figure 8: Bipolar Duty-Cycle Control Signal. In such circuits, when the average DC signal is zero, the motor is stopped, regardless the amplitude of the AC signal. If a live load attempts to disturb the motor, however, it now has to work against not a zero-volt DC signal, but a high amplitude AC pulse of the opposite polarity. Low speed operation is created by setting either the peak voltage of one polarity slightly larger than the other, or altering the duty cycle of the positive and negative phases slightly. Because of the high peak voltages present, starting friction is easily overcome, and as with zero motion, the high but balanced direction signals cause the speed to be far less affected by load than would an equivalent pure DC signal. Such applications pay an even higher penalty than the pulse-addition circuits like the TAT in motor heating and noise. Motors for such service need to be selected and mounted with such heating in mind, and appropriate cooling provided--bearing in mind that resistive and magnetic heating at zero RPM is as great as at maximum speed. Because of this high heat load and the difficulty of cooling model locomotives, such designs are not common in model railroading, but they do show that applying AC signals to "DC" motors is a common and practical industrial design that does not of itself damage the motor. |
The improvements in low-speed control to be had from the addition of various forms of AC are not, however, without their drawbacks. Because there is now an AC signal applied to the motor, there will be resistive and magnetic heating that would not occur in a pure DC environment. The practical effect of this is small at low to medium speeds, and only becomes serious when sustained high-speed operation is attempted. For this reason, many power supply designs either limit the maximum AC signal or remove the AC component as the DC value rises. While there are many ways of accomplishing these effects, some are quite simple and involve only a single resistor or capacitor in addition to the conventional circuit.
In some motors or motor/drive systems, there will be mechanical resonances at some harmonic (or sub-harmonic) of the AC signal frequency. If such a resonance is present, it can lead to audible noise from the locomotive, which is often amplified by the track structure commonly used, even when such widely touted "sound deadening materials" such as Homasote® or VinylBed® are used. Motor noise at the drive frequency will also often be heard where no mechanical resonance is present, and such sound is not generally harmful (except for esthetic reasons). To the extent that sound is produced by vibrating material, it will increase heating in the motor and drive, but only to the extent that any audio speaker warms as it produces sound--a negligible effect in this application.
Resonance effects can be profound in some rare cases. As most commercial power supplies have AC signals of 60 or 120Hz (50 or 100Hz where 50Hz power is used), due to the use of power-line frequencies, resonance at these frequencies can be a serious defect in motor and drive designs. As all companies manufacturing motors for the model railroad hobby are aware of the prevalence of these frequencies in commercial power supplies, it is rare to find problems of this sort. Where power supplies use different pulse rates, good designs will provide for varying the frequency so that any such resonance can be avoided by de-tuning the supply from the motor.
The only serious penalty to AC signals of reasonable frequencies in model railroad applications is the introduction of additional heating. For sophisticated throttle designs that reduce the amount of pulse as the DC voltage increases, the maximum heating from AC is kept below the maximum heating from full-wave DC at the maximum specified voltage. Some lesser designs, particularly those that use half-wave DC, can generate greater heating if their peak voltage is allowed to exceed the DC maximum, as they must in order to reach full average voltage and full nominal speed.
The above notwithstanding, there may be present some level of noise from the motor and gear train under AC propulsion that some modelers may find unpalatable. Such side effects as noise are the natural consequence of trying to get improved performance out of low-cost motors, gear trains, and power supplies, and are not to be avoided--though they can be minimized by the various sound attenuating techniques described in the literature. If the noise level remains objectionable even after all reasonable steps to quiet the locomotive and track structure have been taken, one must conclude that the modeler has unrealistic expectations for the quality and price of locomotive and power supply used.
The National Model Railroad Association has adopted Standards for Electrical requirements of model railroad propulsion in NMRA Standard S-9, which are adhered to by all reputable model railroad power supply manufacturers worldwide (except DCC receiver manufacturers, as noted above). The NMRA Standard implicitly states that full-wave rectified sine waves are an acceptable propulsion signal, and that locomotives must be designed to operate safely under such a signal at any voltage up to 12 volts RMS. It also states that power supplies using different waveforms must be constructed so as to prevent greater overheating than that produced by such a signal.
Thus any locomotive that fails to provide safe and satisfactory operation under full-wave DC at power line frequencies would not be in compliance with NMRA Standards, and should neither be marketed nor used. Neither should any power supply that caused greater heating than full-wave DC be in compliance with NMRA Standards or suitable for sale and use.
The realities of economics and physics dictate that satisfactory operation of modern model railroad locomotives is seldom to be found in the exclusive use of filtered or "pure" DC control signals. The addition of a varying or AC component to the drive signal is both economically and technically advisable. Modern, well made DC per-mag motors are neither degraded nor damaged by being driven by signals with an AC component, either unipolar or bipolar, so long as heating is kept within acceptable limits.
Extensive research has shown that the best possible drive signal for control of modern DC per-mag model railroad motors is a fixed DC signal with short pulses added at low speeds to overcome starting friction and low-speed non-linearities. When economic considerations take precedence over optimum operation, the lowest cost practical signal is full-wave rectified AC. Compromises between these extremes, such as half-wave rectified AC at low speeds blending to full-wave at high speeds are also worthy of serious consideration. Further research, especially into cooling techniques and dynamic characteristics of contemporary motors, should be conducted before bipolar drive signals should be contemplated for commercial sale.
The improved DC per-mag motors now being produced are reliable and compact sources of motive power for model railroading and similar applications. The design of equipment to produce their control signals deserves more attention that it usually receives--not to prevent damage to motors or locomotives, but to ensure that optimum performance is provided to the consumer at a reasonable price. The inclusion of an AC component in control signal waveforms is not only safe, but necessary to obtain such optimum performance.
National Model Railroad Association, Inc.
NMRA Standard S-9: Electrical.
NMRA, Chattanooga, revised August 1984.
Westcott, Linn.
"Getting Closer to Realistic Performance."
Model Railroader, February 1962, p. 56.
|
|
Original edition published in The Portland Clinics, (National Model Railroad Association Convention Proceedings), Bruce A. Metcalf, editor. Kalmbach Memorial Library, 1994, pp. 232-245. This revised edition copyright © 1994, 1997, 1998, 2000 Bruce A. Metcalf. Updated 11 March 2006 by Bruce A. Metcalf, who will appreciate your comments. |
The full-wave rectified DC waveform is not the only AC waveform, nor necessarily the best possible waveform available for improving the low-speed performance of per-mag motors. Another that is even easier to produce is half-wave rectified DC. Using a single rectifier, rather than a full-wave bridge, this waveform has half the frequency, and twice the peak-to-average voltage ratio of the full-wave waveform, as shown in Figure 4: Pulsed Power.