Cyclically Operating Transistorized Power Switching Circuit System

The present invention relates to a transistorized switching circuit system which is cyclically operating, and more particularly to such a system in which a driver transistor controls a switching power transistor, and the duty cycle, that is, the ON-OFF relationship of cyclical switching is variable.

Transistorized switching circuits have been proposed in which the collector of a power transistor is connected to the collector of a driver transistor, and the base of the power transistor is connected to the emitter of the driver transistor. This circuit, known as a Darlington circuit, has many industrial applications, for example as voltage regulator control switches connected to automotive generators, such as alternators. The control currents required are comparatively low. The minimum saturation voltage which is obtained is the sum of the base-emitter voltage of the power transistor and the collector-emitter voltage of the driver transistor. Both voltages, together, are ordinarily in the order of somewhat over 1 volt; the power transistor alone may have saturation voltages of less than 0.5 V.

Transistorized power switches connected as a Darlington circuit have a comparatively high power dissipation level. When the switch is disconnected, that is, when the power transistor is blocked or in OFF state, then the power dissipated is zero. If such a Darlington circuit is operated with variable duty cycles, which occurs, for example, when such a switch is used in the voltage regulator of an automotive alternator system, then the average power loss due to dissipation varies between a minimum value in the order of zero and a maximum value which corresponds to continuous connection. In the Darlington circuit used in automotive voltage regulators, the variation in dissipated power is between 0.6 W and 6 W. If the circuit to which the transistorized circuit is connected has temperature dependent components, then operation thereof will become dependent on the duty cycle, unless the switching circuit is thermally decoupled from the remainder of the circuit. Such thermal decoupling, however, leads to difficulties even if conventional construction of the control circuit is used, employing discrete components; a comparatively large number of additional circuit components are required. Using hybrid integrated construction adds to thermal complication difficulties and, upon monolithic integration of the entire circuit, thermal decoupling becomes almost impossible.

The term "duty cycle" used herein may be defined as the ratio of the duration of conduction, or ON condition of the switch to the duration of the blocked, or OFF condition of the switch this; is also at times referred to as the "mark-space" ratio.

It is an object of the present invention to construct a transistorized switching system which, from a thermal point of view, is independent of duty cycle.

Briefly, the dimensions and values of the internal resistances of the components of the transistorized switch are so selected that the power dissipated -- considering normal manufacturing tolerances of the components -- is, in an initial approximation at least, independent of the duty cycle of the switch.

In a typical circuit, the collector-emitter path of the driver transistor is connected by means of a first resistor across a source of operating potential; the base of the power transistor is connected to the junction of the resistor with the respective electrode of the driver transistor over a second resistor. In accordance with a feature of the invention, the resistance values of the resistors are selected to meet the following conditions:

a. When the driver transistor is blocked, the sum of the resistance values of the two resistors is selected such that the current flowing to the base of the power transistor holds the power transistor, under all operating conditions, in saturated conduction; and

b. the resistor connected in the emitter-collector path of the driver transistor is so dimensioned that when the driver transistor is in conductive, or ON state, the sum of the power dissipated in the first resistor and in the collector-emitter path of the driver transistor is approximately equal to the sum of the power being dissipated in the two resistors and in the power transistor when the driver transistor is blocked or in OFF state.

As can be seen, the power dissipated in the switching circuit then becomes independent of the state of the switch, as such -- at least in a first approximation, considering tolerances of manufacturing components - and thermal effects become independent of duty cycle.

If the load being switched by the transistorized power switch is a pure resistive load, then the dissipation losses in the transistorized power switch become entirely independent of the duty cycle of the switch -- or nearly so, considering usual manufacturing tolerances. If the connected load, however, is a reactive load, for example an inductive load which has a free-wheeling diode connected thereacross, then the dissipation losses in the transistorized power switch only become independent of the duty cycle in an initial approximation by dimensioning the resistance as above described, since the amplitude of the interrupted current depends on the mark-space ratio thereof.

In accordance with a feature of the invention, the initial approximation may be extended to a first approximation by including the power dissipated in the free-wheeling diode in the energy balance, that is, in the equality of (b) above.

The driver transistor may include a transistor which is complementary thereto. The power dissipated in the control circuit of the driver transistor then can be also included in the energy balance -- equality (b) above -- in such a manner that the collector resistor of the driver transistor is so dimensioned that, when the driver transistor is conductive, the sum of the power being dissipated in the first resistor and in the collector-emitter circuit of the driver transistor, including the dissipated power in the control circuit of the driver transistor, is equal, or approximately equal to the power dissipated in the two resistors and in the power transistor, when the driver transistor is blocked.

The power transistor may be constructed as a Darlington transistor; this is of particular advantage in circuits which are designed for relatively high operating voltages.

The invention will be described by way of example with reference to the accompanying drawings, wherein:

FIG. 1 is a schematic circuit diagram of a transistorized switching circuit system in accordance with the invention, showing currents and voltages arising in the circuit;

FIG. 2 is a schematic cross-sectional view through a monolithic integrated structure of a circuit in accordance with FIG. 1;

FIG. 3 is a top view of the structure of FIG. 2;

FIG. 4 is a schematic circuit of the power transistor in multi-cellular arrangement;

FIG. 5 is a cross section through a monolithic integrated structure of the circuit of FIG. 1, in which the second resistor is formed by conductive layer diffusion;

FIG. 6 is a top view of the arrangement of FIG. 5;

FIG. 7 is a schematic cross sectional view through a monolithic integrated structure of the circuit of FIG. 1 in which the first resistor is formed by conductive layer diffusion and the second resistor by base diffusion;

FIG. 8 is a top view of the structure of FIG. 7;

FIG. 9 is a top view of a monolithic integrated structure of the circuit of FIG. 1 in which the driver transistor is arranged transversely to the power transistor, and in which the power transistor is a multi-cellular structure;

FIG. 10 is a fragmentary circuit diagram showing a cascaded driver circuit, utilizing two transistors;

FIG. 11 is a fragmentary circuit diagram of another embodiment of a cascaded driver circuit;

FIG. 12 is a schematic transverse sectional view of a chip, on which the transistor switching system is formed, mounted on a header;

FIG. 13 is a schematic part-sectional view of a voltage regulator assembled to part of the housing structure of an automotive alternator, and having a discrete free-wheeling diode;

FIG. 14 is a view similar to FIG. 13, utilizing a monolithically integrated free-wheeling diode;

FIG. 15 is a perspective, part-sectional view of the structure of a monolithic integrated free-wheeling diode. FIG. 16 is a schematic top view of a monolithic integrated structure in which the power transistor is subdivided into cells and cell groups, symmetrically located on either side with respect to the driver transistor, with symmetrical parallel connecting resistors extending centrally from the driver transistor to the power transistor, with respect to the central axis of the driver transistor;

and FIG. 17 is a perspective, schematic view illustrating connection of a connecting strip, connecting the various transmitters of the power transistor of a cell group to the edge of the crystal chip, for adhesive connection to a base.

The invention will be described in connection with npn transistors; of course, it is equally applicable to complementary pnp transistors. For simplicity of language, the claims are likewise directed to the npn transistor structure and are deemed to include, likewise, the complementary pnp structures.

A representative circuit illustrating the concept of the present invention is shown in FIG. 1 in which a driver transistor 1 is connected to a load transistor 2. The driver transistor, shown as an npn transistor, has its collector resistor R.sub.1 connected to one terminal of a power supply. The base of the switching transistor 2 is connected over coupling resistor R.sub.2 to the junction between the collector of driver transistor 1 and the driver collector resistor R.sub.1. The load is generally shown as R, L. The load, in any event, contains a resistive component R and may (but need not) contain a reactive component which, in the present example, is inductive. R.sub.4 is the equivalent collector resistor of switching power transistor 2; u.sub.2 is the inherent collector-emitter voltage of transistor 2. A free-wheeling diode D is connected in parallel to the load R, L. It has an inherent voltage u.sub.3 and its inherent equivalent resistance is shown by resistor R.sub.3. The current through R.sub.1 and R.sub.2 is shown as I.sub.11 when the driver transistor 1 is blocked. Current I.sub.12 is the current through resistor R.sub.1 when the driver transistor 1 is conductive, or ON. I.sub.2 is the collector current of the power transistor 2 when continuously conductive; I.sub.3 is the current through the free-wheeling diode D. The operating voltage is shown as U.sub.1.

If the load is essentially inductive, free-wheeling diode D accepts current I.sub.2 as soon as the power transistor 2 is blocked or OFF. Let the duty cycle (ratio of ON to OFF duration of the power transistor) be m. The current I.sub.3 through free-wheeling diode D is then given by relationship (1).

The average power dissipated in the control circuit N.sub.St is given in relationship (2) and the power dissipated N.sub.L is given by relationship (3).

If the load in the collector circuit of the power transistor 2 is purely resistive, that is, is a pure ohmic load, then the power dissipated is given in relationship (4).

If the free-wheeling D is disregarded, then U.sub.3 and R.sub.3 are to be equated to zero in relationship (3). Substituting, one obtains relationship (3a).

The parameters now are to be selected in such a manner that the sum of the power dissipated in the control loop and in the power or switching loop is as constant as possible, that is, is in accordance with relationships (5) or, even better, (5a).

The inherent voltages u.sub.2 and u.sub.3 of the power transistor 2 and of the diode D - if used - can be considered as independent from current I.sub.2 for this consideration in an initial approximation.

The results obtained by the circuit in accordance with the present invention, in comparison to a customary Darlington circuit, are illustrated in an example in which the respective values are given in Example A, and in which the effects of the free-wheeling diode are also shown, by illustrating the example in connection with three free-wheeling diodes D.sub.1, D.sub.2, D.sub.3 of characteristics as given in the example.

Table 1 shows the average dissipated power in dependence on the duty cycle m for a customary Darlington circuit, and four examples in accordance with the invention, using an inductive load as shown in FIG. 1. In the first example, in accordance with the invention, the diode D has been omitted; in the last three columns, the respective diodes D.sub.1, D.sub.2, D.sub.3 of Example A have been used. Table 2 illustrates a maximum difference in power dissipation, which may cause temperature difficulties. As seen, the variation in the Darlington circuit of 6 W is reduced by using the present invention, even without considering the characteristics of the free-wheeling diode already to 0.9 W, that is, to 15% of the previous value. Considering the free-wheeling diode, the variation in dissipated power can be reduced to between 0.25 W and 0.45 W. The table illustrates the maximum change of the average power dissipated between duty cycles of m = 0 and m = 1, for a Darlington circuit, and for circuits in accordance with the present invention, and utilizing an inductive load.

It is practically unavoidable that the various circuit components have manufacturing tolerances. The temperature differences, which may be termed the error dissipation, may increase thereby. Table 3 illustrates the error for three different pairs of currents I.sub.11 and I.sub.12. As can be seen, the error doubles when the resistors or, rather, the currents through the resistors vary by .+-. 15%. Still, the advantage over the Darlington circuit without the present invention is apparent. If the circuits are made by thick-layer, or thin-layer technology, that is, conventionally, with printed circuit boards, then it is possible to utilize components of low tolerances, that is, to at least approximately maintain design parameters. In such a case, the power dissipated in the control circuit for the driver transistor 1 can also be considered. Inclusion of the power dissipated in the control circuit for the driver 1 as an additional factor in relationships (5) and (5a) will be obvious.

A structure, as a monolithic, integrated circuit for the transistorized switch of FIG. 1 is shown in FIGS. 2 and 3. The driver transistor 1 is at the left side and the power transistor component 2 is on the right side. The axes of the strip-shaped electrodes are located at right angles to each other which, in a multi-cellular structure of the power transistor 2 forms a particularly desirable and space-saving arrangement. An epitaxial layer 91 is applied to a substrate 90. An isolating diffusion is shown at 92, and 93 is the insulating convering layer. 94 illustrates the collector connection diffusion which, in the example shown, is a low-resistance connection formed between conductive layer L.sub.1 and the connecting terminal 96 of the collector of driver transistor 1, formed of a metal such as aluminum; the connecting line 95 connects the resistor R.sub.2 formed, for example by base diffusion to the base B.sub.2, of the power transistor 2. The partial collectors C.sub.21, C.sub.22 . . . C.sub.2n are connected to a connection line 97. The connecting windows of the resistor R.sub.2 are shown at 98, 99, respectively.

The power transistor 2 is split up into partial transistor elements 21, 22, 23 . . . 2n, as seen in FIG. 4, which shows the multi-cellular structure. Resistor R.sub.2 likewise is a multi-cellular component, split up into elements 41, 42, 43 . . . 4n; each one of the partial resistance elements has a resistance value of n-times the overall resistance of resistor R.sub.2. The voltage drop, individually, in the resistance elements, when using the present invention, is so great that the current distribution to the various cells remains stable, if the power transistor 2, with its partial transistors 21, 22 . . . 2n is operated in the region of an emitter current density in which the current amplification decreases with increasing collector current. Equalizing and stabilization resistors for the various connections to the emitters of the transistor elements 21, 22 . . . 2n are thus not necessary.

Embodiment of FIGS. 5 and 6: The resistor R.sub.2, or the resistor elements 41, 42 . . . 4n, are made by conductive layer diffusion. The terminal between the resistor R.sub.2 at the base side of the power transistor and the connecting line 95 to the base B2 of the power transistor is shown at 100, which is made by means of collector contact diffusion. All other elements are similar to those described in connection with FIGS. 2 and 3 and have been given the same reference numerals and will not be described again.

Embodiment of FIGS. 7 and 8: The resistors R.sub.1 and R.sub.2 are located in two superimposed planes, thus saving space. Resistor R.sub.1 is made by conductive layer diffusion and resistor R.sub.2 is made by base diffusion.

The resistor R.sub.1 has a higher power density. In order to better utilize the surface of the crystal, resistor component R.sub.1 may also be split up into two or more resistor elements. The partial resistor elements 31, 32, 33 . . . 3n may be connected in series, or in parallel, or in series/parallel.

Under certain circumstances it may be desirable to form the partial resistance elements in different diffusion zones.

The multi-cellular structure of the power transistor 2 is also shown. The connection between the end of the resistor R.sub.1 which connects to the power supply and the bus or line 102 to which the operating voltage is connected is formed by the element 101, made by collector connection diffusion.

Embodiment of FIGS. 9 and 16: This is particularly suitable for higher currents. A second multi-cellular, or multi-element group of the power transistor 2 is located, symmetrically, and as a mirror image of the first with respect to the driver transistor 1.

The power transistor 2 must be reliably blocked even if the temperature of the underlying crystal chip rises somewhat. To ensure reliable blocking of the power transistor, the driver transistor 1 must have a sufficiently low saturation voltage. This condition can be obtained, simply, by forming the driver transistor 1 in the form of a cascaded transistor array. Suitable circuits are illustrated in FIGS. 10 and 11.

Referring first to FIG. 10: The two transistors 11, 12, together, form the active components of the driver transistor circuit. The collector resistor R.sub.1 of the driver transistor 1 is now split up into two partial resistance elements which have a high resistance portion 3a and a low resistance portion 3b. The resistors 61, 62 connected between the bases B.sub.11 and B.sub.12 are provided in order to reliably control the base currents I.sub.B11 and I.sub.B12.

FIG. 12 illustrates another possible circuit for a cascaded driver transistor network. Driver transistor elements 11, 12 are connected in emitter-follower circuit. Emitter E11 of transistor 11 is connected to the base B12 of transistor 12. Resistor 63 forms the base resistor for transistor 12.

A suitable mounting and holding structure for a chip on which the entire circuit can be formed is shown in FIG. 12. A large-area chip is adhered to the header by means of a conductive adhesive. In this way the connection between the header and those components which are connected to ground or chassis voltage of the integrated circuit, such as E.sub.1, E.sub.2 and components not shown, can be made simultaneously. Chip 82 is adhered to header 80 by conductive adhesive 83. For reliable electrical connection, the portions of the circuit at ground or chassis potential are brought out to a metallized edging 84 which, by direct contact with the conductive adhesive 83 connects the metallized connections 84 to the metal header 80.

FIG. 17 illustrates how a conductive strip 111 is brought to the edge of the chip, and the adhesive attachment thereof.

The circuit of the present invention is particularly suitable for use in voltage regulators of automotive electrical supply systems. FIG. 13 illustrates the mechanical construction of a semiconductor voltage regulator switch for an automotive alternator, in which the entire voltage regulator switch is secured to the brush holder for the slip rings of the alternator, or otherwise to the alternator housing. The header 80 (FIG. 12) simultaneously forms the support carrier for the entire assembly. Chip 82, as before, is adhered to header 80 by means of conductive adhesive 83. The chip 82 has formed thereon, in addition to the switch, the entire control circuit for the switch in accordance with the present invention. Terminals 74 for the integrated circuit are melted, vacuum-tight, into the header 80. They are connected by means of bonding wires 75 with suitable connection dots (not shown) of the chip 82. The brushes (or brush connections, respectively) for the slip ring of the alternator field are shown at 76. The brushes 76 are secured in brush holder 77 which is connected by means of a river 71 to a portion of the metallic housing or support structure 80' of the alternator forming, electrically, the equivalent of the header 80 (FIG. 12). Interposed insulating washers 72 separate a contact spring 73, secured by a river 71', to form one terminal connection of the supply voltage. Other electrical elements may be included in the electrical circuit; the free-wheeling diode D, in this construction, is formed as a discrete element 50, connected in parallel to the winding of the field. A bypass capacitor 51 is likewise provided. The chip is protected by a cover housing 85. Lead-through connections 74 are shown with exaggerated clearance from the header structure 80 although, of course, they can be melted in, to be vacuum-tight. The housing structure portion 80' is secured to the alternator structure, and hence electrically and mechanically to the chassis of the automotive vehicle by suitable connection means such as, for example, screws 70 shown schematically only and with exaggerated clearance, for clarity of illustration.

While the free-wheeling diode 50 may be connected as a separate discrete element, it is also possible to form the free-wheeling diode 50 on the chip as an integrated circuit. The mechanical structure is illustrated in FIG. 14 which, except for the diode 50 and the connections to the chip, is otherwise identical to the structure of FIG. 13. The diode structure itself, formed on a monolithic integrated chip, is illustrated in FIG. 15.

Substrate 105 (FIG. 15) has a conductive zone 106 buried therein, by diffusion. An epitaxial layer 107 is deposited thereover. Thereafter, isolating zone 108 is diffused into the chip. Together therewith, a deep connection 109 is diffused therein. The diffused zone 110 forms, normally, the base of the transistors. In this construction, however, it forms one of the electrodes of the free-wheeling diode D. The other electrode is formed by the epitaxial layer 107, the conductive layer 106, and the connection diffusion 109. It is important for the operation of the free-wheeling diode that the deep connection diffusion 109 surround the diode, in ring form, and is reliably connected to the buried conductive layer 106. If the diode is operated in forward direction, the electrode 110 forms the emitter. The other electrode 106, 107, and 109 forms the base, and the substrate 105 the collector of a parasitic transistor. This parasitic transistor has full operating voltage applied thereto. Large dissipated power losses would arise even at small collector currents. It is therefore important that migration of minority charge carriers from the epitaxial layer 106 into the collector space formed by the substrate 105 is inhibited. This is obtained by the highly doped connection diffusion layer 109 if this layer, as illustrated in the drawing, entirely surrounds the other electrode 110 in form of a ring, since the minority charge carriers recombine in the highly doped zones 106, 108.

The specification has been written based on the illustration of an npn transistor; the switch can be constructed, similarly, with pnp transistors. When using a monolithic integrated construction, then the structures are complementary. The voltages and currents have reversed polarity or direction, respectively.

The control circuitry for the transistorized switch may likewise be formed on the chip; control circuits for transistorized switches of semiconductor voltage regulators for automotive vehicles are well known and shown, for example, in the cross-referenced patents. Additional references may be made to "Automotive Electronics," publication jointly under the auspices of the Automotive Electronics Committee, IEEE, and the Electrical Systems Committee, SAE, copyright SAE, 1973. Such other circuitry -- for example control circuitry for the switch -- is preferably located at one of the free ends of the chip, and at a position transverse to the axes of the driver transistor 1, and parallel to the axes of the power transistor 2, or to the cells, or elements thereof. Any desired circuit may be placed on the chip, and the header construction of FIG. 12 may be used to provide, simultaneously, a ground or chassis connection.

To provide for good thermal capacity, the header preferably is made of materials which have good heat conductivity, or at least is formed with a layer of such materials. The metals silver, copper, aluminum or steel are suitable; if an insulating header structure, preferably with a metallized coating is used, or with a connection to the conductive adhesive 83, then ceramics such beryllium oxide, aluminum oxide, or other insulating materials, or a combination of any of the foregoing materials are suitable. The carrier or substrate for the crystal preferably is heat-dissipating. Thus, the carrier may have an expanded surface to act as a heat sink, and heat radiator, by having its surface increased to have a profile which is, for example, U-shaped, ribbed, or the like. The attachment fins for the brush holder 77, FIGS. 13, 14, and for the power connection 73 form such an extended, heat-dissipating profile, this profile simultaneously providing attachment surfaces for electrical or mechanical elements with which the transistorized switch cooperates.

Various changes and modifications may be made and features described in connection with any one of the embodiments may be used with any of the others, within the inventive concept.