The Core Issues: Choosing A Power Transformer
"The Core Issues" will examine the function, construction, design, and operation of magnetics for audio applications. This series will combine practical advice in the form of useful "rules of thumb" with enough pure theory to give us some concept of how these magnificent devices work. This column is not an academic engineering tutorial but rather a guide to becoming an informed consumer.
This series begins with a look at the theory and operation of the power transformer. Power transformers are relatively uncomplex devices compared with output transformers and are often given only the most cursory attention by audiophiles. More of- ten than not, they are treated generically. "Oh, all you need is a transformer that will supply, say, 360-0-360 VRMS @ 200 mils for the plates, a 6.3 VCT @ 4A filament winding, and 5 volts at 3A for the rectifier tube." Beyond this point, the consumer is often at a loss as to how to properly select a good unit or is misguided by myths about magnetics. Where do we go from here?
In the ideal world, power transformers are conceptually pretty simple. We want the transformer to either step up (increase) or step down (decrease) the voltage supplied to the primary. Power transformers, like all transformers, are simply ratio devices. If you have 100 turns on the primary connected to a 120 VAC supply and you want 360 volts on the secondary, then there must be 300 turns of wire on the secondary -- i.e. a 1 to 3 ratio. To obtain our 5V winding, the primary voltage of 120V must be stepped down by putting fewer turns on the secondary. The exact number needed is 4.1666 turns. In the ideal model, turns ratio precisely equals voltage ratio.
The ideal model assumes a perfect conductor with zero resistance. Therefore, the size of the conductor is wholly unimportant and can be infinitely small for convenience. But, and this is a big but, we do not live in an ideal world. In the real world, we must contend with a variety of electrical and magnetic phenomena that blow our ideal models to hell and back. There are always losses when using real materials, e.g. copper is one of the better conductors, but it significantly inhibits the flow of electrons. Similarly, losses occur due to the properties of core materials.
Due to the resistance of copper, not all of the 120 volts applied to the primary will be available to be stepped up or stepped down. Some of the voltage will be burned up (gone forever) in traveling from one end of the copper to the other. The telltale sign of this inefficiency is heat. The lost voltage is reincarnated as a temperature rise. The steel laminations also introduce losses that are not accounted for by the ideal model. Our ideal model assumes infinite permeability and proposes that the laminations will not, in fact cannot, be saturated ("overloaded"). Real transformers aren't like this, unfortunately.
The value of this ideal model is to demonstrate the underlying simplicity of the device and to reinform us as to where deviations from the ideal occur. But you can't buy an ideal transformer.
If it's not possible to have an absolutely perfect transformer, "one that simply transforms", then what should we realistically seek? First and foremost, we want to obtain our specified secondary voltages -- using our earlier example, we need 360-0-360 while the transformer is delivering its fully rated 200 mA current. Also, we would like to minimize the variation in this voltage even when the Circuit is pulling consider- ably less than the full current for which it is designed. In other words, we want the voltage to remain relatively constant from no-load to its maximum current capability. This is called "voltage regulation". Many power transformers (notably stock Dyna iron) exhibit pronounced voltage sags at maximum rated current. A well-designed and properly built power transformer can achieve a remarkably high degree of voltage regulation. Next, we would like it to run relatively cool. And, we would like it to be as physically small as possible without sacrificing performance. And we would like it to be as inexpensive as possible without trading off real performance for dollars.
Let's take a look at a few dubious claims about power transformers you're likely to encounter in the audiophile press. In philosopher's parlance, these are "fallacious arguments". By examining these fallacies, we can develop a better perspective on important performance parameters that must be considered in choosing transformers.
One of my favorites was an ad that advised hobbyists to buy the house brand of power transformers because they were heavy -- the assertion being the more weight, the better. Another good one was the high end manufacturer who claimed that to get your money's worth you should buy a transformer that runs hot. A widely held and popular belief is that you should buy a mil-spec trans. The supporting logic is that because Uncle Sam has an unlimited budget and only buys the best components, the smart consumer benefits by buying them on the surplus market for pocket change and makes a real killing because its "mil-spec" and, therefore, a high-performance unit. But by far the most common misconception is that a power transformer with double or triple the current capacity actually required will yield a better performing transformer than one that is rated at the maximum current honestly necessary for proper operation of the circuit at hand. This is a potentially dangerous assumption.
Let's start with the weighty issue first. The assumption is that you will get more for your money. But more of what? Often, but not always, a transformer is beefy because it utilizes a low grade of electrical steel. Transformer laminations come in a multitude of grades ranging from M6 to M55. The lower the number, the lower the core losses as measured in watts loss per pound of material. This is because the reluctance (the magnetic equivalent of resistance) of M6 is lower than the lesser grades with a higher number behind the "M". The higher the number, the higher the losses; there- fore, you must use more (all other things being equal) of MI 9 for a given VA (power) rating than if you use M6. As the number increases, the saturation point falls lower and lower. Even if you run Ml 9 at lower levels of flux density, magnetic distortion is still greater than for the more premium grades of laminations at the same operating level. Simply put, you must use more MI 9-to build our example transformer than you would need using M6 and your iron losses will be greater if all the other de- sign considerations are equal.
This is not meant to say that you should always reject a power transformer if it is heavy. Some power transformers built with M6 are heavy, and some transformers built with M19, M22, or M27 are designed to provide reasonable performance levels. The big mistake is to make decisions based solely on weight without knowing anything about the composite performance capabilities. We must know what the voltage regulation of the unit is, what temperature rise to expect, how current capacity was calculated, what flux density level the unit runs at, etc. Just don't buy on weight alone since it might indicate cost constraints imposed by the manufacturer (Ml 9 and M27 cost 40% to 70% less than M6) and a heavy unit may have poor overall performance characteristics.
Another misconception is that a transformer should run hot if you're getting your money's worth. Actually there is a small grain of truth in this assertion, but only if you don't particularly value characteristics like good voltage regulation, low magnetic distortion, and the like. The aviation industry, for instance, will tolerate extremely high temperature ratings in order to gain small size and weight. But for audio projects, especially quality projects like building 300B amps, this would be a very unwise tradeoff. To get a transformer to run hot, we can do any number of things: use a poorer grade of lamination, keep the size of the lamination small, use small diameter wire in the windings, or operate a good grade of lamination at very high flux density levels. Presto, it'll run hot. Sort of like a New York City Saturday Night Special. Will it survive? If it is built with class B insulating materials that are rated for 1050 C operation and it truly runs hot (say it has a 750C or higher temperature rise when pulling a full load) on the surface, then we are probably flirting with disaster. But even under these conditions, more than likely, it will survive. Forced air cooling would provide an added measure of safety.
Remember that inside the transformer there will be localized hot spots that operate at 15 to 20 C hotter than the surface. The aviation industry gets away with running iron hot because they specify insulation materials rated for 155, 180, or 220 C environments and they're willing to sacrifice some performance factors for small size/weight. Most consumer and industrial power transformers use 105 or 130 C materials which are adequate for a transformer without serious design flaws. If your manufacturer specifies temperature rise, look for a unit with a 35 to 55 C rise. Transformers with a 55 to 75 C rise are less desirable but worthy of consideration if all other performance factors are acceptable. Pass it by if rated at 75 C or higher rise. Again, don't make decisions based on any single performance criterion -- look at the composite characteristics.
Let's examine the most common misconception: that radical de-rating of transformers leads to better performance. In our ideal transformer model, the voltage ratio is equal to the turns ratio. If you built an actual transformer according to the model, the actual voltage increase will be smaller than anticipated. Why? Because there are always resistance losses in the copper and iron losses in the laminations. How much less depends on the grade and size of lamination, the size and length of the wire, the varnish or impregnate used, the type and amount of insulation employed, type of housing, and yet other factors. The principle determinants are the copper and iron losses mentioned above.
In practice, we can compensate for voltage losses by adjusting the turns ratio. To achieve a given voltage ratio, we need a slightly higher turns ratio. The adjusted turns ratio is calculated by taking into ac- count a number of factors which are related to current flow: copper and iron losses and temperature rise (resistance and core losses go up with temperature). We need to calculate losses at full load so our transformer will deliver the target voltages with good regulation. The current values plugged in to the formulae must be the ACTUAL maximum conditions which the circuit will demand of the transformer. Otherwise, the secondary voltage will be a surprise.
Suppose a customer specifies a 400 mu transformer instead of a 200 mu transformer. OK, this requires a significantly different design. It will be larger and it will need heavier conductors in the primary and plate windings to offset increased I2R losses. The voltage regulation of this transformer will be about the same with 400 mA as the smaller transformer would have with a 200 mA load. The customer is right in one respect -- if he only draws 200 mA from a 400 mA transformer, there shouldn't be any problem with voltage sags as with the Dyna iron. But is he home free?
By overstating current requirements, our friend will end up with a transformer that delivers a voltage higher than optimum for a given circuit. The compensated turns ratio for 400 mA is higher than for 200 mA. Therefore, he might get better voltage regulation and a cool running transformer, but at a voltage which is assuredly higher than bargained for. Not only the plate voltages but all secondary voltages will be too high since copper losses in the primary will be less than anticipated. The ultimate costs of this error could include decreased reliability and longevity of tubes, caps, and other parts or even improper operation of the circuit. The wise move is to get a competently designed unit at the desired voltage and realistic current levels.
A few more notes on current capacity: The current rating of transformers is what I would call a "soft" number. If a manufacturer says his unit is rated at 300 mA but provides no further info, we have not learned enough to evaluate the import of his claim. Perhaps it will deliver 300 mA, but only at a 90 C rise or it might drop so much voltage at 300 mA that the initial design target voltage is inaccessible. Voltage regulation might be poor indeed at 300 mA. To really judge this claim you would need to know at least the temperature ripe and voltage regulation from no-load to full load. Current capacity specifications can vary between very conservative and wildly optimistic. Be sure you're comparing apples with apples. If two units give the same voltage and current specs, try to find out more about how it was designed with respect to voltage regulation, temperature rise, flux density level, etc. Don't treat the two units as "generic" equivalents performance-wise without first getting as much information as possible.
Here are a few characteristics to look for when choosing a power transformer.
- A 4Q0 to 550 C temperature rise
- M6 material
- A copper current density level of 600 to 1,000 circular miis per amp (the higher the better)
- Between 3% and 6% voltage regulation
- Insulated hardware - A flux density level under 15 kilogauss
- Mechanically quiet
- Electrostatic shielding between primary and secondary windings
- Designed for the current that your application requires
Don't use this as a rigid checklist since there are many good designs out there that will fail to meet one or more of these criteria. In general, these are indicators of a quality transformer.
Lastly, a few comments on mu-spec transformers. Many audiophiles are under the misconception that mu-spec transformers are somehow superior to other, commercially available units. In reality, Mil T-27 addresses issues relating to physical packaging of transformers almost exclusively. Specific requirements are laid out to ensure a high degree of mechanical integrity for phenomena such as vibration and shock, salt spray, flammability, fungus proofing, altitude testing, etc. On electrical matters such as voltage regulation, current capacity, temperature rise, core material and grade, or permissible flux density, Mil T-27 is wholly silent. In fact, unless the government agency or the approved contracting agent specifically lists particular performance parameters, the units supplied must only meet longevity (as measured by insulation life), corona standards, and mini- mum standards for insulation resistance.
Unless you're privy to the actual contracting order, you cannot be sure of the actual performance parameters (if any were specified) to which the transformer was de- signed. You could wind up buying a unit that offers no real advantages over other designs. But, of course, those surplus items are available cheaply and have great appeal to constructors on a tight budget. But it really is taking a shot in the dark from a performance point of view.
When shopping for transformers for your next project or design, it may be useful to review some of the points in this article -- especially the indicators of quality' outlined above. In certain cases even more information will be necessary. For instance, this article does not address very important issues like rectification and filtering on AC voltages and currents, a topic of great importance in choosing your power components... More on this important topic later. In the meantime...
Get the entire collection of Sound