Model Railroad Power Supplies:

The output of any DC power supply will be limited by at least two, and sometimes three factors. These are Voltage, Current, and Power.

Voltage is limited by the design of the supply, and is the value most often cited when describing a supply. For model railroad propulsion, 12 volts is specified by NMRA Standards, although some scales, such as G and Z typically use higher or lower voltages, and most tinplate equipment uses AC power. For most purposes, it is the voltage that determines the speed of the train.

Any given supply will have a design maximum voltage determined by the components used to create the supply. Throttles will allow the actual voltage to be adjusted to control train speed, either by changing circuit values to produce a different maximum voltage (as with a potentiometer) or by controlling current, thus producing a different voltage indirectly.

In a perfect supply, the desired voltage would be produced at any desired current without change. Real supplies will show some decrease in voltage as the current increases due to resistive and other losses in the components. This drop can be minimized by regulation, but as the drop is generally small, such regulation is rarely used in model supplies. Following this to its logical conclusion, one can foresee that at some large amount of current, the output voltage would drop to zero, but this limit will not be reached in practice because of two other limits that are reached first. Figure 1 shows the limits of a theoretical power supply's output, with a realistic voltage limit shown as line E.

The current output of a power supply is generally limited deliberately, both to prevent damage to the power supply and to prevent damage to the locomotive (and not incidentally to avoid hazard to the modeler). There are several popular types of current limiting:

Fuses are probably the most common current limiting device, but as they work only by self-destructing, and as overloads are common occurrences for those modelers who occasionally leave a screwdriver on the track, they are not appropriate for modeling.

Light bulbs rated at the desired maximum current at the design maximum voltage are very efficient, fast enough to protect the other components, and the side effect of glowing can be used as a warning device. Unfortunately, as they sometimes burn out and need replacement, it is difficult to receive regulatory approval (UL, CSA, etc.) for their use in a commercial design.

Electronic current limiting is an elegant approach, as it allows the designer to carefully set any value desired, or to leave the value adjustable by the operator, but the additional component costs generally remove this from consideration in commercial applications except for the most sophisticated throttle designs.

This leaves two types of thermal circuit breakers. Both work on the principle of a bimetallic strip that deforms when heated by current passing through it, breaking the circuit. Open designs often have a mechanical latch that requires the operator to push a button and reset it manually. Again, safety regulations now make this difficult to include in a commercial design.

Sealed thermal circuit breakers reset themselves after a period. This can be beneficial, as it requires no operator intervention, but entails the new hazard of re-energizing the circuit without warning, possibly while the operator is trying to remove a short. Sealed thermal breakers also give no indication of their condition, and additional circuitry to give a warning to the operator brings us back to the cost problem with electronic current limiting.

In a perfect supply, the current limiter would engage at exactly the same current, regardless of the voltage or any other variable. In a real supply it may vary according to the voltage, and in the case of thermal circuit breakers, they will vary according to the temperature in the room and the amount of heat generated by the power supply itself. In Figure 1, a realistic current limit for a range of power outputs is shown as line I.

There is a third limiting factor at work in most power supplies, and that is their maximum power rating. Power, measured in either Watts or Volt-Amps (which for the sake of this discussion are the same thing), represents the amount of work being done. It is a function of the load drawn by the train and the efficiency of the power supply and its components (and to a lesser degree, the quality of the input power).

Power is calculated by multiplying the voltage times the current at any given point in time. The power output to the train is of little concern, as supplying power to the train is what we are trying to do, but because no power supply is 100% efficient, some power is always lost in the supply in the form of heat. Heat is a most troublesome thing to deal with, because if it is not dissipated, the temperature will rise until something malfunctions, melts, or catches fire--none of them desirable outcomes.

Figure 1 shows a realistic power limit as a pair of lines marked P. The reason there is a range of power limits is that it takes a while for heat to build up to hazardous temperatures, and you can get away with short overloads that could not be sustained for longer periods of time.

Heat is dissipated by radiation, by conduction, or by convection, and power supplies use all three. A component, say the transformer, conducts heat to the case to which it is attached. It also heats the air around it, removing heat by convection. Finally, it radiates infrared--not bright enough to see (one hopes) but still transferring heat to something else. In a perfectly sealed box, the heat would be contained, and the temperature would rise, until something failed. In a real power supply case, heat must be removed, and in the same three ways.

In commercial applications, safety regulations now generally proscribe the use of ventilation. This means no holes in the case, and no convection of heat from inside components--only from the case itself. This also means that radiation from internal components is not possible (without transparent cases)--but again, the case can still radiate. Finally, cases are often set on wood or other surfaces that do not conduct heat well, and this pretty well eliminates conduction as a way to remove heat.

One problem with heat dissipation by a power supply case is that the temperature of the case has to be kept low enough that an operator is not burned when he touches it. On the other side of this coin is the fact that the greater the temperature difference, the faster heat transfer takes place. What does this mean to power supply case design?

This means that the surface of the case, both total area and the type of surface, create a limit to how much heat can be dissipated from the box without raising surface temperatures above acceptable limits. Effective dissipation will also be limited by the ability of the internal heat to be conducted to all surfaces of the case--maybe the parts are attached to the back and bottom and heat these surfaces to the limit, but the front and sides are much cooler, thus dissipate less than they could.

In order to increase heat dissipation, a number of steps can be taken. A black finish can be applied, which increases slightly the efficiency of radiation, but as many power supplies are installed in panels or enclosed areas, this is a minor consideration at best. The total surface area of the case can be increased, either by making the whole box bigger, or by adding fins like a heat sink or even just wrinkles. This is probably the most effective, but may cause serious cosmetic changes. All of this works to help remove heat from the outside surface of the case.

Another alternative is to make changes inside. The heat path from hot components to the case must be made short and efficient. This means increasing the surface area of a component in contact with the case to promote conduction, or even the use of heat conductive grease or plastic fittings between case and component. Heat sinks inside the case can aid convection to the air inside the case, which helps heat the entire inside surface of the case. In extreme cases, thermally conductive potting compound is used, although this generally makes repairs impossible (and manufacturing expensive). These steps help to effectively transfer the heat to the inside surface of the case.

The last step of the problem now becomes transferring heat from the inside surface of the case to the outside. The only factors that have a big impact on this are the thickness and material the case is made from. Metals like aluminum conduct heat very well, but complicate the problem of providing insulation for electrical safety. Plastics are often good electrical insulators, but frequently conduct heat poorly, and must often be thicker to provide the same strength. The best choice is a material that conducts heat, but not electricity, is very strong in thin sections, and is easy to fabricate. Well, now you're talking money again. As ever, compromises must be made.

The only other choice is to make the internal components more efficient. While some gain can be had from higher quality parts, this usually means a more sophisticated electronic design, which can mean anything from simple autotransformers to high frequency chopper-driven power oscillators, fast recovery diodes, and electronic regulation. Brother! We've come a long way from taking the batteries out of our automobiles, haven't we?

What does all this nonsense mean to the average modeler who just wants to see his train go? The amount of power available from even a very small and inefficient power supply is enough to move light trains at some speed on a small layout, so obviously the problem arises when the modeler wants to run heavy trains, such as six or eight old diesels pulling 100 freight cars; when a full 12 volts are needed to achieve desired train speed; when losses in wiring on a large (or poorly built) layout require excess output for compensation; or when the modeler wants a hand-held unit of very small size.

The modeler sees only a few factors as influencing his operation, and sadly these are not easy for a designer to determine in advance. Case design and features are obvious--size, placement of controls, special features like brakes or walkaround design are relatively easy to figure. Current ratings and power specifications are often hawked in advertising, but these relate only indirectly to realistic operating conditions. Why is "performance" so difficult to quantify?

One problem is locomotive speeds. While NMRA Standards say they shall operate at a realistic maximum speed at 12 volts, very few do. Even if they did, not every modeler wants to operate every train at full speed. Finally, the current demands of locomotives differ because of motors (ranging from multiple open-frame designs of the 1950s to single efficient "can" designs of the 1990s), train lengths (trolleys to long drags), and lighting, signaling systems, and other loads.

What determines the ability of a power pack to provide acceptable performance is its capacity for current at the operating voltage that produces the desired speed on a given layout. This voltage will vary with the number and type of engines, train resistance, grades, layout wiring, and personal taste, but it is the one essential operating variable that is unknown to the designer of power packs. Three examples:

In Figure 1, the point marked 1 is the maximum current available at 12 volts. This is the effective rating available to a modeler who operates at full speed. The point marked 2 is the maximum current available at 8 volts. A modeler with some older locos who hauls long trains might find this his effective rating. Finally, point 3 shows the effective rating of this supply for a modeler with nothing but efficient can motors who operates a switching layout at low speeds.

None of these ratings are the same as the current rating of the supply. Point 1 is limited by output voltage, point 2 by power dissipation, and point three by the current limit less the heat generated in producing the power. Is the capacity of this power supply adequate to each of these modelers? It depends on how much current they need for their own type of operation.

A manufacturer of power supplies has to sell his product to turn a profit. How should a power supply be defined, and how should it be marketed? In the best of all worlds, a manufacturer would publish a chart such as shown in Figure 1, giving thereby the voltage, current, and power limits of his supply, and the wise modeler would understand these charts and choose the least expensive model with all the desired features that would provide him with sufficient current at his normal operating voltage.

In the worst case, manufacturers publish whichever of the three specifications make them look best when compared to the market leaders, and modelers buy the pack their dealers stock because of the discount offered by the distributor.

A new power pack that significantly outshines the competition has no problems, but in a real and competitive world, with relatively equal access to components and design ideas, and minimal control of distribution channels, it can be difficult to find a niche for new products. For any given item, it will likely outshine the competition in at least one area, and it is this that modern advertising wisdom pursues.

Education of modelers to their own real needs, preferably through side-by-side, hands-on comparisons of different models in realistic service on the same layout would provide the modeler with the best basis for informed purchasing. Short of that--and the power supply demonstrations of most hobby shops (and all mail order shops) generally fall far short--the best choice would be to conduct such comparisons on real, well known layouts, and advertise the comments of real, well respected modelers. While even these results can be slanted by careful selection of layouts and modelers, this is probably a better approach than trying to explain electronics to hobbyists who don't want to know what's in their little black throttle anyway.

Designing and marketing a power supply for model railroaders is a complex and difficult task, with decisions and compromises to be made in physical design, cost, features, ratings, efficiency, and marketing. It's a dirty job, but somebody's got to do it--otherwise we would all have to go back to automobile batteries and hand cranking our cars after each night of railroading.

• Top of Page
• Back to Publications List
• Back to Home Page

Originally published privately for the National Model Railroad Association, Inc., 1987. Copyright © 1987, 1997, Bruce A. Metcalf. Updated 14 March 2006 by Bruce A. Metcalf, who will appreciate your comments.