By John I. Brown [Chief Development Engineer, Avel-Lindberg Ltd., South Ockendon,
Essex, England]
A problem common to all branches of electronics is the hum, both electrical and acoustic, which is generated by power transformers. Electrical hum is caused by stray magnetic flux leaking from the core and windings, and interacting with nearby conductors and components. Acoustic noise is caused by the laminations in the core literally being "flexed" by the magnetostrictive effect of the currents passing through the windings. An accident of design, because its inventor never had hum in mind, gives the toroidal transformer very considerable inherent advantages over other types of transformers when it comes to the suppression of hum despite the influences on audio design which seem to conspire to create a hum problem.
The introduction of veneer type laminates, synthetic surfaces, and constant changes in home furnishing fashions have meant that the industrial de signer involved in the audio field had to produce equipment with less "battleship" and more "furniture" if his product was to be readily accepted as an integral part of the home. A typical example is the "slimline" trend where the height of the case has been considerably reduced, and trim good looks have become a marketing necessity. Any physical limitation, imposed by dimensional constraints, makes it that much more difficult for the de signer to select a transformer and choose a site where it will not interfere with other circuits. The user has also demanded higher power from stereo amplifiers, with negligible noise, of course, and user requirements have also led to reduction in loudspeaker size with consequent lowering of efficiency so that higher power input is now needed to produce an equivalent sound level. This in turn requires a higher output from the amplifiers. The introduction of solid-state rectifier technology has also caused problems, as has the higher component packaging density brought about by component miniaturization and printed circuit techniques. The circuit engineer therefore faces a mounting difficulty, sooner or later, of what to do about the power transformer-that well known generator of unwanted hum.
Hum and Leaking Flux
The trouble with a transformer is, simply, that it depends on the creation and decay of an alternating magnetic flux field through its windings to per form its function in life. Any stray flux, leaking from the transformer, will introduce an emf in any adjacent wiring or susceptible components and will cause hum at the a.c. line frequency.
High-gain, high-impedance, low-signal-level circuitry is at greatest risk, which is why the old unwritten law of keeping the input circuitry as far away as possible from the power supply components always made good sense.
In the days of tubes, if trouble from power transformer hum developed, then mu-metal screens (which attenuate magnetic flux), twisted pairs, screened leads, and hum-bucking windings could be employed if the transformer couldn't physically be moved to a position where it stopped being a nuisance. The smaller the equipment, though, the more difficult it becomes to find a satisfactory position. In certain instances, it is also necessary to use multiple mu-metal screens, which adds to the cost of the unit.
The Gap
In the traditional wound-bobbin transformer, with E and I laminations stacked together to form a core, it is the air gap at the junction of the I across the three legs of the E that causes most of the trouble. The air in the gap has a high reluctance, or magnetic resistance, compared with the metal and the concentration of flux which results and radiates out into its surroundings.
The same effect, though to a lesser extent, occurs if the laminations have mounting holes or notches punched through them, because this imposes a localized concentration of the magnetic field and causes some of the flux to be spilled out. The answer to this problem is, obviously, "get rid of the air gap," and to this end, the C core type of construction is a definite improvement. However, even with the butting faces lapped, ground, and polished, a residual air gap remains, and so does some unwanted stray flux.
Torus Concept
Theoretically, the ideal answer would appear to be to have a magnetic circuit without an air gap, and this can be achieved in practice using a ring or torus, wound from strip steel material rather like a tightly wound clock-spring. Granted, with this system you can only wind one toroid at a time whereas you can multiple-wind bob bins, but a fair trade-off can be achieved, however, because you have to assemble the conventional core stack by hand from separate laminations. The mechanical construction of the toroidal core also has the inherent advantage that once it is wound, its magnetic properties can be measured before the windings are applied in the knowledge that they will remain constant throughout all subsequent operations. The windings are put on the core, using high speed machines, across the three outside faces and through the hole in the center so that they encompass the core. Impregnation, potting, and packaging in thermoplastic cases provide good protection from environmental hazards.
Electrically Induced Hum
The pulsing magnetic field developed by an unscreened stacked-lamination type transformer can generate a flux leakage which extends completely through the entire spatial volume of a typical modern 100-watt amplifier.
Where multiple channel amplifiers are involved, there is an additional problem of unequal demands for current which affect the regulation characteristics as the currents in the secondaries vary in sympathy with the different demands of each channel on the power supply. The use of separate secondary windings and rectifiers for each channel is desirable and, for the purist, two completely separate toroids can be mounted on top of each other and still be lower than a single stacked-lamination type.
The hum problem is not limited to low-level, high-gain stages, however, because it can also make itself a nuisance in the negative-feedback circuits of power amplifier stages. The combi nation of a "gapless" continuous magnetic circuit and the natural' screening effect of the copper windings, which completely enclose the core of a toroid, give an 8:1 reduction in radiated field when compared with a stacked-lamination type as shown in the polar diagram (Fig. 1).
Where radiated magnetic fields are a problem, then the toroidal transformer is more likely to provide a solution without resorting to expensive mu-metal screening or completely reorganizing the circuit layout. Changing a layout might not have presented too many difficulties in a unit constructed with tagboards and discrete wiring be cause it was a relatively simple matter to reroute a couple of wires or move some components around. If the circuitry is based on a printed circuit board, however, and hum is only detected at a late stage in the design, it can be an expensive business to modify a complex board layout. In this situation, toroids have often been used as a "last resort," before calling for radical design changes, and have proved entirely satisfactory.
Acoustic Noise
The best known example of the noise generated by magnetostriction is probably the characteristic "ping" sent out from sonar devices used aboard ships for detecting submarines, fish shoals, or the depth of the sea bed.
The current through the transducer causes the laminations to move and displace the medium in which it is mounted--water in the sonar example and air in the transformer example.
In sonar it is a necessity; in audio equipment, however, it's the last thing you want from a transformer.
Advances or changes in circuit technology sometimes aggravate the problem of noise while trying to improve other features, e.g. high-voltage, low-current tube circuits with low-value filter capacitors (8 to 16 pF) and high-resistance, high-tension windings operated at very low peak currents. The advent of low-voltage, high-current transistor circuitry, however, meant that very much larger (1,000 uF) filter capacitors were needed for ripple reduction (smoothing), and these capacitors caused high peak currents to be developed. The general use of low-impedance silicon rectifiers, with large filter capacitors to smooth their d.c. supplies, resulted in the transformer current being in the form of large pulses with steep edges lasting only a fraction of each half-cycle of the a.c. supply. Typically, the secondary winding feeding the rectifiers of a power supply for a 100-watt audio amplifier delivers pulses in the order of 15-A peak and 2-mS duration at a pulse rate frequency of twice the supply frequency. The harmonics resulting from the sharp pulses begin in and extend well up into the audio range. In the case of a stacked and laminated transformer, the core can be clamped (but, of course, screw holes distort the field), and heavy impregnation helps a little to damp down the noise. The toroidal transformer, however, has a much higher core packing density and is almost a solid ring to start with be cause it is wound from strip under constant tension. It is capillary impregnated and then the copper windings are wound round the circumference, which in itself is a clamping operation, and also has a damping effect on any sound that is generated.
Fig. 1--A polar diagram comparing the radiated field strengths of a toroidal
transformer (at 1 mV) and a laminated transformer (at 8 mV) measured with
a standard 1000-turn pick-up coil.
Fig. 2--The physical advantages of toroid transformers are demonstrated
in this photograph of a 90-VA toroid transformer compared with a similarly
rated conventional transformer. The toroidal transformer is 1.8 in. H and
4.2 in. diameter, with a weight of 3 lbs.; the conventional transformer is
33 1/4 in. H x 3 1/2 in. L x 3 3/8 in. W.
Fig. 3--Magnetic performance of grain-oriented steels is very dependent
on grain direction.
Core Material
As a general rule the toroidal transformer is smaller and lighter than a stacked laminated type--of equivalent VA rating and function--mainly because the core material is used more effectively (Fig. 2). The strip is manufactured from grain-oriented silicon steel (GOSS) and wound so that all the molecules in the metal point in the same direction as the flux. Any molecules out of this alignment increase the reluctance (magnetic resistance) and therefore progressively degrade the performance until, at 90 degrees out of phase, they reduce the effectiveness to that of ordinary mild steel.
With a stacked-lamination core, it is possible that at least 40 percent of the total core area will be at 90 degrees to the required grain direction and another 40 percent will be effective only as a return flux path. A smaller transformer--and the savings can be in the region of 50 percent in volume and weight--means greater flexibility in the choice of mounting position and a greater probability of being able to site the transformer where it will not cause interference.
Core Geometry
The use of strip steel as the core medium enables the transformer designer to produce a large number of variations in core sizes from a single strip width, and this gives more flexibility to the circuit designer who may require a different width-to-height geometry from the optimum two to one. If there is plenty of room for diameter and little height, then a three-to-one ratio could be used, and if space is at a premium then a ratio of 1.5 to 1 could be supplied. In the slimline-styled equipment, the toroid has the obvious advantage of presenting a very low profile compared with a stacked type, but this is not the only area where the toroid scores because, in any equipment, single point fixing, coupled with low weight and a low center of gravity, helps the engineering.
Efficiency
Fig. 4--Even distribution of the primary over the secondary in, a, ensures
that the magnetic fields generated in the windings cancel, b.
Table I--Comparison of Two Toroid vs. Two Laminated Transformers.
Fig. 5--Mounting methods for toroidal transformers.
With full advantage being taken of all the grain orientation being in the preferred direction and having no air gap, the toroidal core can be operated at a flux density of 1.6 Tesla (16,000 Gauss) to 1.8 Tesla (18,000 Gauss), while a stacked-lamination transformer would be limited to the 1.2 Tesla (12,000 Gauss) to 1.4 Tesla (14,000 Gauss). This higher efficiency means that the alternatives of using less magnetic material or fewer turns are avail able to the designer. The iron losses are much less significant in a toroid because the reluctance is so low, and therefore it is often the required physical size that determines the core material, not the losses. The higher efficiency also means that there is less heat generated, and a value of 0.4 watts per square inch of surface area is a reason able midrange target. The toroidal transformer designer can therefore trade off size against efficiency.
The toroid also saves power, and comparison between the low magnetizing current (which is being drawn the whole time the transformer is connected to the power supply, no matter if secondary current is being drawn or not) of a toroid and that of a laminated transformer amply demonstrates this point. Comparison between the other losses, which also consume power to no good effect, make quite interesting reading (see Table I).
Mounting
Fig. 6--Comparison between toroid core and laminated E and I core showing
differences in grain direction.
Fig. 7--The winding process.
The toroid might well have been de signed originally with printed circuits in mind because the center hole fixing, the low center of gravity, and the PC board compatible pins enable it to be mounted at the same time as the other major components and flow-soldered using standard production techniques.
The availability of a library of entirely "standard" designs also gives the circuit designer the facility of having the "standard range" dimensions--pin layout, fixing drilling ordinates, and track constraints stored as macros in PC board computer-aided drafting systems. In this way the transformer de tails can be laid down simply by pressing a single key; the computer can also be programmed to take into account the various constraints regarding track widths and spacings to take the cur rent involved.
Summary
When presenting a summary of the technical advantages of toroidal transformers, there is a danger of establishing a credibility gap by overkill.
There is a tendency for the uninitiated circuit designer to speculate "if they are that good, they must be expensive." The simple fact is that it is harder to make a comeback than to start from scratch.
The eight-to-one lower radiated field has been demonstrated, and. the acoustic quietness is self-evident when the method of construction is investigated. A working flux density of 1.7 Tesla (17,000 Gauss) for the all-in-line grain-oriented toroid, as against 1.3 Tesla (13,000 Gauss) for a conventional transformer, is inherent in the design; as are iron losses of typically only 0.46 W/lb. (Fe) as against 1.25 W/ lb. (Fe). The absence of the air gap means the toroid only requires a magnetizing current of one-tenth of that needed by a transformer with a gap.
Center-hole, single-point mounting, coupled with the ability to mount the toroid directly onto a PC board, make the production engineer's life easier.
The typically 50 percent lower weight and volume, with the lower height profile, must also contribute to easing the designer's ulcer when space is at a premium in high-component density equipment.
The toroid is certainly the oldest type of transformer; in fact, Faraday wound the first ever. But modern winding techniques coupled with advances in wire technology, especially in insulation, have enabled its manufacturers to produce a product of considerable benefit to contemporary circuit designers. The advantages, which stem largely from characteristics inherent in the toroidal construction, will commend the toroid to designers with transformer problems, and in their next project they are likely to specify a toroid as the prime choice as a matter of course. In the past, the toroid has suffered from its own negative feed back effect, where small quantity production kept costs up and the small numbers in use meant that the toroid did not warrant a very prominent place in the educational syllabus for electronic engineers. This educational gap meant that only the "enlightened" specified toroids and, because of the cost penalty, only when any other type would not do the job.
(Adapted from: Audio magazine, Dec. 1979)
Also see: The Importance of Dynamic Range (Jan. 1980)
= = = =