Power Transformer Design and Construction

Radio Record, Volume III, Issue 23, 20 December 1929, Page 28

Power Transformer Design and

Construction

Rules for Calculation

(By

CATHODE

ISHING to construct ‘a ‘transformer that will supply’ his needs for some’ ‘time, and yet simplify the process, a Wellington ‘'’ reader '* has raised somé questiozis that, in view of their general interest, the writer has expanded into a special article. The design called for requires windings for the following :- (a) Anode current supply up to 200 milliamps at 400 volts, with tappings at 100 volts, 170 volts, 200 volts, 300 volts, 350 volts, and 400 volts, or as required, with full-wave rectification. (b) Grid bias up to.100 volts with separate smoothing apparatus. (c) Windings for full-wave. accumulator charger. (ad) Windings tor filaments of a.c, valves. (e) Windings for filaments. of rectifying valves.° The first operation in designing a transformer is to culculate the total load, assuming a certain efficiency, We will assume an efficiency of 80 per cent., a figure which should be fairly reached. In order to obtain an anode current supply of 400 volts on load, it is safe to assume that a transformer winding providing at least 500 volts each side of the centre-tap will be called for. Strictly speaking, we should use the "average" current for calculating the power drawn from this winding, but, as no great accuracy is necessary, it will be less confusing to readers if we use the "effective" current. Thus we may say that 200 milliamps at 500 volts must be supplied by this -winding, or, in other words, 100 watts. No power is drawn from the gridbias winding except such as is expended in the potential-dividing résistance which wil] be connected actoss the out-

put of the grid-bias rectifier. Assuming that this resistance has a value of 20,000 ohms (the voltage across it being 100), a current of 5 milliamps will be drawn from the trinsformer winding. To maintain a voltage of 100 with adequate smoothing apparatus and the loud imposed by the resistance will necessitate a transformer winding supplying about 140 volts, or, in the case of a full-wave rectifier, 140 volts each side of the centre-tup. Thus the power drawn from this winding is .005 x 140 equals 0.7 watts-practically hegligible. Whether half or full-wave rectification is employed is a matter for individual preference, remembering that a fullwave rectifier has a slight theoretica! advantage, in that the variations arising. from a partly-smoothed full-wave rectified current for anode supply are to some extent cancelled out by corresponding variations of the same phase and frequency in the bias voltage. The cost of+the full-wave rectifier need be little greater, since cheap receiving valves with grid and plate terminals joined are quite adequate as rectifiers for this purpose. If desired, the secondary of a disused audio-frequency transformer may be utilised for a smoothing: choke for a bias rectifier, although ts life will probably not be loug and the voltage drop across it will, by reason of its high resistance, be a trifle excessive;

this could be compensated by making the voltage of the a.c. input to the rectifier rather higher. ~~ The windings for the charger, assuming a full-wave miereury vapour rectifier is used, will comprise two 16volt anode windings carrying between them 1.8 amps. and a 1.75 volt filament winding carrying 3.5 amps, Power

drawn from windings will be about 27 watts, Our correspondent does not specify precisely what filamént windings he requires. However, he wil] ‘no’ doubt wish to provide for filament windings to feed about four ae. amplifying valves at either 1.5, 2.5, or 4 volts, and two power amplifying valves at 7.5 volts, the total consumption being per-

haps 40 watts.. In additibn, the filament windings for the rectifying valves will absorb another 40 watts or so. We are now in a position to ascertain the power rating the titansformer must have, Adding the 100 watts for © the plate supply, 1 watt or less for bias, 27 watts for charger, 40 watts for receiver filament, and 40 watts for rectifier filaménts, a grand total ‘of 208 is\arrived at. An efficiency of 80 per cent. Was assumed so that it is safe to say that a 250 watt core and primary will be necessary. . ‘Our correspondent is insistem’, that the windings shall be carried on Hound bobbins, holding that winding -*on a square bobbin is too difficult for the amateur. The writer does not altogether agree with this view, since a bobbin to accommodate a square or rectangular core can be readily constructed. Two types are suggested. The one fills the circular spacé of the former. by a series of "steps and stairs," the other by placing a square core in a round former, Both these have serious disadvantages, the former on the ground of difficulty in cutting the iron to the many different sizes required, and the latter on the ground of the probability of introducing too high a leakage reactance and the certainty of making the iton path ‘ unnecessarily long, The compromise shown at Fig: 1 is best, the core being s‘4p-cut to\£U. most of the bobbin, yet requiring oily three, different sizes of strips. The stalloy must be bought cut to size, as it cannot be cut with snips. The crosssection of the core shown, which just fits inside a piece of 2}-inch outside diameter formica tubing, is 8.86 square inches, so that, in order that the yokes may fill the requirement of having approximately the same cross-section as the core, the width of the laminations forming them must be 1% inches. Calculating the Turns. WE have a core cross-section of 3.36 inches, of which we can safely assume that 80 per cent. will be iron. Stalloy can be worked economically at a flux density of 60,000 lines per square inch, so that the total flux will be 60,000 X 3.36 * .8 = 161,280 lines. Applying the formula ‘ . E x 108 a N= 4,44.Kx £«'S

where N is the number of primary turns, E the applied aa voltage which will no doubt be the standard 230 volts; f the frequency of the applied a.c., also presumably the standard 50 eycles per second, and S the total flux, it is found that the required number of primary turns is x = 645 primary turns 4,44 * 50 * 161280 It is suggested that, partly to reduce the magnetising current, partly to simplify the calculation of the secondary turns, the number of primary turns ra increased to about 670, which will b equivalent to 2.918 turns per. volt, Then, by calculating the secondary windings on the basis of 3 turns per volt, an allowance of about 8 per cent. ta compensate for voltage drop due to leakage reactance and resistance of windings betas automatically be made, >

E are now in a position to figure out the various windings and detetmine what window space will be needed to accommodate them, A current density of 1200 amperes per sq. in. would be sufficient: for the primary in view of the high ratio of copper to insulation; this winding, however, has to carry the magnetising or "wattless" current as well as the equivalent of the secondary load, so that it may be well to reduce the current density to 1000 amperes per sq. in. or less. No. 18 ‘S.W.G. has a sectional area of 0.00181 &q in, and may therefore be relied on to carry 1.81 amperes; apart from the magnetising current, the primary will carry only a little over an ampere, so that this wire is entirely ‘Suitable. No. 18 D.O.C. winds 17 turns per inth, so that 670 turns will require +--\sq. in., or 2.82 sq. inches, 172 . -Hach half of the high-voltage winding carries current only half the time, So that the wire may be smaller than would otherwise be the case. Actually 28 S.W.G. proves sufficient, and it is recommended that either double eotton covered or single cotton enamelled wire’ be used. Even so, for absolute safety it is desirable to put a layer of good paper (varnished when in position) or oiled silk between each two layers of the winding. At 8 turns per volt, the 1000 volts right across the winding will require 3000 turns which, since 28 D.C.C. winds 39 turns to the inch, will take up approximate 2 square inches. The grid bias winding may, be of No. 86 D.0.0. and will take ugg About .25 of a square inch. > 5A ree =

The filaments of the rectifier valves, which may be either one or two Mareconi or Osram U8 full-wave rectifiers, or two or four UX 281 half-wave rectifiers, may be fed from 22 turns of No. 14 D.C.C., which will take up another .25 of a square inch. The recti‘fiers for the grid-bias will no doubt be 5-6 volt 2012 type valves, and. may have their filaments fed from 16 turns of No. .22 D.C.C. or larger wire. The power-valve filaments will re: quire, if of the 250 or 210 type, 22 TITTLE

turns. of No. 14 D.C.C. taking up .25 of a square inch. For feeding several a.c. valves in the earlier stages No. 12 D.C.C. will be needed, the number of turns depending on thé valves employed; if 4 volt a.c. valves are employed 12 turns will be needed. Some constructors would prefer, perhaps wisely, to make provision for all three types of a.c. valves, so that it will be —

wise to allow .5 of a square inch for these windings. The total cross-sectional area of the windings is, then, 2.82 + 2 + 25 + 25 + 25 + :5 square inches, or a total of 5.57 square inches. The winding bobbin, the air-space between it and the core, the insulation between windings and the insulation between the layers of the high-voltage windings, if any is used, will all take up space, and a "window" area of 8 square inches will be none too great. Efficient design de-

mands that the "yokes" should bé appreciably shorter than the legs on which the windings are mounted, so that suitable dimensions ‘for the window will be 4 inches by 2 inches. The Iron Required. A LITTLE figuring shows that the iron required will be as follows: If .014 stalloy (handled by Johns, al —

Ltd., Auckland) is used, theré will he needed 160 pieces 5%#in. by Vin., 268 pieces +n. by 1fin., and 108 pieces 54in. by 1 8-16in., or, if the suppliers object to the odd size, 53in, by ldin. If .018 Stalloy (handled by National BPlectrical and Engineer Co., Wellington) is used, there will be required 126 pieces 5zin. by 2in., 212 pieces 4in. by 12in, and 86 pieces 5%in. by 1 8-16in., or 14in, It must not be forgotten, when assembling the core, that the joins in the laminations must be staggered. The mean length of the iron path is about 19 inches, and since it takes 10 ampere turns per inch to magnetise Stalloy to a.flux density of 60,000 lines per square inch, 190 ampere turns will be called for for magnetising the core, The primary turns are 670 in number, so that the magnetising current ig 190 /-- = 0.28 of an ampere. 670 ; Fortunately the 18 s.w.g. used in the primary will carry this "wattless" cure rent in addition to the useful current, so that no increase in wire diameter is called for. The core contains about 16lb: of iron (1 cubic inch weighs 4oz.) and the loss in Stalloy at the flux density mentioned is 0.7 watts per Ib. Thus the iron loss will be about 10 watts. Assuming that the secondaries are to be wound over the primary, the mean | primary turn will be about 9 inches. Thus the primary will need nearly 4 lb. of No. 18 d.ec.c.,; which will have a resistance of 2.2 ohms. The copper or 12 R loss in this winding will be, assuming the primary current to ne -_

13 amps., 1.3 X 1.3 X 2.2 = 3.72 watts. ' he high-voltage secondary,. having a mean turn of 12in., assuming it is wound on next over three or four layers of Bmpire cloth, will require about 241b. of No. 28 d.ee. Its resistance will be 140 ohms, and its 12 R loss at full load 1.4 watts. The grid-bias winding, wound on next over adequate insulation, needs only about {lb., the losses in it being negligible. The two 16-volt plate windings for the charger, each of 48 turns, need altogether about 4lb. of No. 20 d.c.¢c., while the filament winding will require 5 turns of No. 14 d.ae. The losses in these windings will bé only about 0.6 of a watt. The high-voltage rectifier filament winding, 22 turns of No. 14 dec. will need lb. of wire, the losses amount-

ing to 1.25 watts at full Ioad. The filament winding for the bias rectifier may be wound from No. 18 d.c.c. left over from the primary, the losses in it being negligible. +The power valve filament winding is the game as the high-voltage rectifier filament winding, and, running four 250’s and 210’s, contributes a further 1.25 watts loss. The other filament windings, wound with 12 D.C.C., will probably take a little over one pound, depending on just what windings are included, and the loss will not be more than about 1 watt. [THE copper losses total 9.22 watts, ._ while the iron loss, it will be remembered, was about 10 watts. Thus, not only does the design fulfil the wejlknown condition that for maximum efficiency copper and iron losses must be approximately equal, but the efficiency is round about 90 per cent. (even allowing for regulation losses) instead of the 80 per cent, which it was thought necessary to attain. The cooling surface is ample to dissipate the heat generated by the very modest losses, and: the transformer will run practically cold. In insulating the windings one from the other (Hmpire cloth, obtainable from Johns’ or Ballingers’ is probably the best) it must be remembered that in addition to insulating the high-

yoltage windings well, the filame windings for the high-voltage rectifi and the grid-bias rectifier must be jinsulated from each other and from the receiver filaments with extreme carey the voltage between these ‘windings being very high. The power valve filament winding should be also reasonably well insulated. . The tappings on the high-voltage winding will present a little difficulty, and should be brought out between two layers of Empire cloth or oiled silk. In joining the wires from the two bobbins to form the centre tap for each winding, it must be clearly understood that the wires must cross over through the windows in the manner illustrated in Fig. 2. Otherwise the transformer will blow the house fuses as soon as it is connected to the mins. ad

If both bobbins are wound and mount. ed in the same’ direction, the starting, or inner ends, should be joined to form the centre tap; if this instruction is followed there will be no possibility of error. It is probable that our original correspondent has already worked out the details of the remainder of the power supply and charger. However, for the benefit of anyone else wishing to copy his very ambitious ideas, a cpmplete circuit diagram is reproduced in Fig. 3, aiming at an ideal rather than at economy. Suitable chokes have been previously described in the "Radio Record" (see, for. example, the issue of June 14 last). The rather detailed exposition of the process of transformer design given here has, of course, been given with the idea of enabling readers so inclined to. copy the procedure and design their own transformers for their individual requirements. Nevertheless, any reader mistrusting his ability in the directioh of transformer design is cordially invited to indicate his requirements to the "Radio Record" information. ser- / vice, when a suitable design will b¢ briefly outlined in the, "Questions and Answers" page. ¢