A METHOD AND APPARATUS FOR PARALLEL OPERATION OF ELECTRONIC POWER SWITCHES
BACKGROUND OF THE INVENTION Field of the Invention
The present invention relates to systems, methods and apparatus for the parallel operation of electronic power switches with negative temperature coefficient of the forward voltage or wide tolerance of the switching threshold voltages.
Description of the Prior Art
Valves of power electronic systems of high power are realized frequently by direct parallel connection of single switches. A substantial limitation in this type of connection, however, is that stationary, as well as, dynamic symmetrical current division cannot be guaranteed.
In particular, when single switches show a negative temperature coefficient of the forward voltage, then, because of the spread of characteristic values of the components, a small difference in current division inevitably follows, which causes a relatively higher junction temperature for one switch. The consequence of this temperature rise is that the forward voltage decreases and the current loading increases for that switch. This positive feedback results in the switch taking over the total current for the connection and thus overloading.
Furthermore, for voltage-controlled power-semiconductors, different switching threshold voltages produce an asymmetrical division of the switching losses between the single switches. Traditionally, this asymmetry is overcome by increasingly complex current controllers and measurement of the single switch currents.
Thus, there is a need for a simple (i.e., not complex) method, system and apparatus to provide control for the direct parallel single switches making up a valve, which creates a symmetrical division of the conduction and switching losses between the single switches.
SUMMARY OF THE INVENTION
The present invention relates to non-complex methods, systems and apparatus for controlling direct parallel single switches making up a valve, which create a symmetrical division of the conduction and switching losses between the single switches.
The basic idea of the invention is to create a stationary or dynamic symmetrical current division. This is accomplished by not impressing current simultaneously to the parallel single switches making up a valve, i.e. to turn them on in an alternating fashion rather than at the same time. If a control signal for one valve, e.g. consisting of two single switches, exists, only every other turn-on signal is accepted from one single switch, during the conduction time, the other single switch is blocked. In fact, one switch is stressed with double current in comparison with ideal parallel operation, however, it is only driven with the half relative on time duration, which causes no difference in the conduction losses at current-independent constant forward voltage (as given, e.g., for Isolated Gate Bipolar Transistors, "IGBTs") in comparison with ideal symmetrical simultaneous parallel operation. There is also no difference concerning the switching losses, based on the assumption that the switching energy loss is directly proportional to the switched current, which is achieved in practice with sufficiently good accuracy: every single switch is only switched with half pulse frequency and the permissible peak current load of a single switch is substantially higher than the current value occurring during symmetrical simultaneous parallel operation. An analogous expansion of the method to a higher number of parallel switches is possible in a simple way. One embodiment of a method according to the invention to guarantee a symmetrical load of valves with forward voltage drop proportional to the current and positive temperature coefficient of the forward voltage (e.g., power Metal Oxide
Semiconductor Field-Effect Transistors, "power-MOSFETs") comprises displacing the instants of switching time of the single switches relative to each other, but at least by the switching delay time and by the signal traveling time if a driver stage is provided. With that, the single switches take over the turn-on and the turn-off command alternately and they are stressed by turn-on switching power loss and turn- off switching power loss alternately. A simultaneous conduction of both switches does not cause any problems because of the positive temperature coefficient and the given self- stability of the current division.
If a control signal for the valve is given, e.g., the first switch takes over the turn-on command directly and the second switch is turned on with a small time lag, and therefore without switching losses. The turn-off command is applied directly to the first switch, too, and to the second switch with a small time lag, so there are no turn-off switching losses for the first switch, the breaking and the turn-off switching losses are taken over by the second switch, respectively. At the following control signal the sequence of the control commands of both switches is reversed, i.e. the second switch is turned on directly according to the given turn-on command while the first switch is turned on with a small time lag, so the turn-on switching losses are carried by the second switch and the turn-off switching losses are carried by the first switch. Therefore, a symmetrical division of the turn-on and turn-off switching power losses on both switches is obtained. A symmetrical division of the conduction losses on both switches is guaranteed, because the time lag of the switching commands is usually substantially smaller than the on time duration of the switches and therefore basically both switches are in conducting state simultaneously. Because of the complete utilization of the silicon chip area available for current conduction the forward power losses are minimized. An analogous expansion of the method to a higher number of parallel switches is possible in a simple way, keeping the advantages with regard to the state of the art.
A further embodiment of the present invention comprises an apparatus for two parallel single switches, wherein the control signal of the valve is applied to a frequency divider (fed back master-slave flip-flop and toggle flip-flop, respectively), to the first input of a first AND-gate, assigned to the first switch, and to the first input of a second AND-gate, assigned to the second switch. For the following
considerations the "high"-level of the control signal shall be assigned to the turn-on state, and the "low"-level shall be assigned to the turn-off state. There are two outputs of the toggle flip-flop available always showing inverse logic levels. That output changing with a negative edge of the control signal from "low" to "high" remaining at this level for a whole pulse period of the control signal and changing with the following negative edge of the control signal to the "low"-level (halving of the frequency) is connected with the second input of the first AND-gate. The inverted output of the toggle flip-flop is connected with the second input of the second AND- gate. The output of the first AND-gate is used for controlling the first switch, the output of the second AND-gate is used for controlling the second switch, interposing a driver stage if necessary. Due to the AND-comparison with a signal with half pulse frequency for the first switch every second control pulse is gated out (the assigned output of the toggle flip-flop shows a "low"-level). Those gated out control pulses are executed by the second switch, its AND-gate is controlled by the inverted output of the toggle flip-flop, which is in conduction state while the first AND-gate is blocking. There results, that only every second control pulse is transmitted, the control pulses lying between the control pulses of the first switch are applied to the second switch. Both switches are alternating in on-state and in off-state, shifted by a half pulse period. For both switches there result equal conduction losses and equal switching losses, but problems according to simultaneous conducting and switching of direct parallel valves are avoided.
For a further embodiment of the present invention, one toggle flip-flop and one AND-gate for each switch is provided and the control information is transmitted alternating to a switch by appropriate gating of the control signal with the outputs of the toggle flip-flop, respectively. Furthermore, the control information is connected with a time-delay element with a delay time resulting from the switching lag time of the driver stage and from the lag time of the switches, the resulting slightly time- delayed control pulses are transmitted alternating to the first and second switch. AND-gates gate out every second time delayed control pulse again, the second inputs of the gates are connected with the outputs of the toggle flip-flop. The resulting control information for the first and second switch is each generated by an OR-gate.
The inputs of the first OR-gate, assigned to the first switch, are connected to the output of the first AND-gate and to the output of the additional provided AND- gate. The inputs of this AND-gate are connected to the time delayed control information and to the output of the toggle flip-flop, which controls the gate in such a manner that time delayed control pulses are placed between the control information of the first switch occurring with half pulse frequency. The generation of the control information of the second switch results analogous, by which every control pulse causes an immediately turn-on of the first switch and a slightly time-delayed turn-on of the second switch, the first switch is turned off immediately by a control pulse, whereas the second switch's turn off is time-delayed.
For this control pulse, the first switch carries the turn-on losses, the second switch carries the turn-off losses, and both switches take part at the current conduction simultaneously, with the exception of the time delay. Both switches change their part at the following control pulse, by which a symmetrical division of conduction losses and switching losses for both switches is guaranteed, and always the whole silicon chip area is available for conducting the valve current. When the time-delay element is left out and the AND-gate inputs, connect to the output of the time-delay element, are connected to ground ("low"), the apparatus may be programmable for applications wherein the switches have a positive temperature coefficient of the forward voltage or for applications with the need for slightly time-delayed switching. In both cases minimal conduction losses at resistive conducting-state characteristic of the switches shall be achieved.
It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the invention, as claimed.
BRIEF DESCRIPTION OF THE DRAWINGS
The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments of the invention and together with the description, serve to explain the principles of the invention.
Figure 1 shows the current configuration of a particular embodiment of the present invention.
Figure 2 shows the signal behavior over time for a particular embodiment of the present invention.
DESCRIPTION OF THE PREFERRED EMBODIMENTS Reference will now be made in detail to implementations of the present invention as illustrated in the accompanying drawings. Wherever possible, the same reference numbers will be used through the drawings and the following description to refer to the same or like parts.
As illustrated in Figure 1 , the invention concerns methods and apparatuses for parallel operation of electronic power switches (3) and (4) with negative temperature coefficient of the forward voltage and/or wide tolerance of the switching threshold voltages. With the apparatus (1) according to the invention a control signal (7) applied by a preceding control unit is divided up into control signals (16) and (17) for the electronic switches (3) and (4). The clock input (18) of a toggle flip-flop (8) is connected with the inverted control information, and a signal (22) and an inverted signal (23), respectively, with half control frequency is generated. AND-operations (11), (13) and (10), (12) of one of the flip-flop-outputs with the control pulse (7) and the slight time-delayed control pulse (19), tapped off from the output of the time-delay element (9), respectively, gate out the even or the odd control pulses (the pulses shall be continuously and increasingly numbered with increasing time). The resulting auxiliary signals with half clock frequency are assembled by OR-gates (14) and (15), in such a manner that, e.g., switch (4) is turned on at the beginning of an even control signal, and switch (3) turns on slightly time-delayed and loss-free, respectively, by the assigned pulse of the control pulse (19). On the other hand no turn-off losses are caused while turning off switch (4) at the end of the even control pulse, switch (3) takes over the valve current and turns it off slightly time-delayed. For the following and every odd control pulse, respectively, the facts are inverted: switch (3) is stressed by turn-on losses, and switch (4) is stressed by turn-off losses. Therefore, both switches are stressed symmetrically with turn-on and turn-off energy losses, and with sufficiently good approximation the silicon chip area of both switches (3) and (4) is available for current conduction during a full conduction interval of the valve (2), by which minimal conduction losses result.
In the following, the invention is explained with Figs. 1 and 2 in more detail. Fig. 1 shows the circuit configuration of the apparatus (1) for the control according to the invention of a valve (2) comprising direct parallel connection of two electronic switches (3) and (4) with control inputs (5) and (6). The control signal (7) generated by a preceding control unit is divided by the apparatus (1), which consists of a toggle flip-flop (8), a time-delay element (9) (in a simple way, the time-delay can be realized-by a RC-low pass in connection with the switching thresholds of subsequent gates), two AND-gates (10, 11) and (12, 13), assigned to the switches (3) and (4), an OR-gate (14), assigned to the switch (3), and an OR-gate (15), assigned to the switch (4). The divided up signal (16) and (17), occurring at the outputs of the OR-gates (14) and (15), is connected with the control inputs (5) and (6) of the electronic switches (3) and (4). The inverted control information (7) is connected with the clock input (18) of the flip-flop (8), e.g., realized as known by NAND-gates, and connected by the time- delay element (9) with output (19) and directly with the inputs (20) and (21) of the AND-gates (11) and (13). The time-delayed control information (19) is connected with the inputs of the AND-gates (10) and (12). Furthermore, the output (22) (designated as "Q" in usual technical nomenclature) of flip-flop (8) is connected with the second inputs of the AND-gates (10) and (13) and the inverse output (23) ("(?") of flip-flop (8) is connected with the second inputs of the AND-gates (11) and (12). The outputs (24) and (25) of the AND-gates (10) and (11) are connected with the inputs of the OR-gate (14), and the outputs (26) and (27) of the AND-gates (12) and (13) are connected with the inputs of the OR-gate (15).
The function of the apparatus according to the invention is explained by the signal time behaviors shown in Fig. 2. The signal behavior is designated by the related inputs and outputs of the components in Fig. 1. The control signal is generated by pulses (28, 29, 30, 31 etc.); the time-delayed control information (19) is also generated by pulses (32, 33, 34, 35, etc.).
Due to a negative edge (36) of the inverted control signal applied at the clock input (18) of flip-flop (8) the logic state of the flip-flop-output (22) changes from "low" to "high", and at the following negative edge (37) of signal (18) the state changes from "high" to "low", by which the generated signals at output (22) and at output (23) (inverse to output (22)) of flip-flop (8) occur with half frequency
compared with the control signal (7). Due to the AND-operation (10) of the time- delayed control information (19) and of the flip-flop-output (22) there results an auxiliary signal (24), where only every second pulse (32, 34, etc.) of the time-delayed control information (19) occurs, the odd pulses (33, 35 etc.) are gated out. Due to the AND-operation (11) of the flip-flop-output (23) and the not-time-delayed control information (7), the undelayed odd control pulses (29, 31 etc.) are transmitted to the output (25) and the even control pulses (28, 30 etc.) are gated out. With the OR- operation (14) of the outputs (24) and (25) of the AND-gates (10) and (11) there results a control signal (16) for the electronic switch (3), composed of even time- delayed control pulses (32, 34 etc.) and of odd undelayed control pulses (29, 31 etc.) lying in between the time-delayed pulses. The control signal (17) of the electronic switch (4) is generated in an analogous way by AND-gates (12) and (13) and the subsequent OR-operation (17) of the outputs (26) and (27) of the AND-gates (12) and (13), using the even undelayed control pulses (28, 30 etc.) and the odd time-delayed control pulses (29, 31 etc.) lying in between. Therefore, switch (4) is turned on with edge (36) (at the beginning of an even control pulse, in general), by which the turn-on of switch (3) at instant of time (38) is loss-free at sufficient time-delay. On the other hand, the turn-off of switch (4) at instant of time (39) is loss-free, the valve current is taken over by switch (3) and turned off at instant of time (40) with a slight time-delay. For the next control pulse (29) (and every odd pulse, in general) the facts are inverted: switch (3) shows turn-on losses and switch (4) turn-off losses. Both switches are stressed symmetrically with turn-on and turn-off energy losses, with sufficiently good approximation during every conducting interval of the valve (2) the silicon chip area of both switches (3) and (4) is available for current conduction, by which the minimal conduction losses are resulting.
If inputs (41) and (42) of the AND-gates (10) and (12) are not connected with output (19) of the time-delay element (9) but with ground, the time-delayed control signals (32, 34 etc.) contained in control signal (16) of switch (3) and the time-delayed control signals (33, 35 etc.) contained in control signal (17) of switch (4) are suppressed. Therefore, both switches are conducting alternately.
The present invention is not limited in scope by the specific embodiments described which are intended as single illustrations of individual aspects of the
invention. Other embodiments of the invention will be apparent to those skilled in the art from consideration of the specification and practice of the invention disclosed herein. It is intended that the specification and examples be considered as exemplary only, with a true scope and spirit of the invention being indicated by the following claims.