DE4316816A1 - Circuit arrangement - Google Patents

Abstract

If an individual converter valve cannot be loaded on its own - in particular at high switching frequencies - with the necessary (high) current, two or more converter valves must always be interconnected in parallel. Owing to component-related scatter, structurally identical power semiconductors have characteristics which are different from one another in practice. Consequently, currents of different magnitude flow through them at the same saturation voltage. In order that the currents are divided as uniformly as possible, two converter valves, for example two IGBTs (I, II), are operated in two parallel line branches (1, 2), line sections, which are wound in the opposite winding direction (with the same number of turns) over a common magnet core to form windings (L1, L2), being connected upstream of the converter valves. The difference between the saturation voltages is divided equally between the two windings (L1, L2) in the case of pulse-clocked converter valves. Approximate balancing of the currents can be achieved in this way. <IMAGE>

Description

The invention relates to a circuit arrangement according to the preamble of claim 1.

Such a circuit arrangement is from the book: Heumann, Klemens: Basics of power electronics; Stuttgart: B.G. Teubner, 1985; 3rd edition, page 58. In three connected in parallel Current branches are each an inductor and a thyristor in series switched. There must always be two or more fast-switching Power semiconductors can be connected in parallel if a single Power semiconductors alone with the required (high) current are not is resilient. Because with direct parallel connection of the thyristors would be the voltage at the thyristors in stationary operation (forcibly) the same and there would be - usually easy differing pass characteristics of the thyristors - one uneven power distribution. Especially with (fast) Switching operations leads to a high thermal Power dissipation.

The upstream inductors are said to be dynamic Improve power distribution.

In the previously mentioned book on page 59 there is a wiring of GTO thyristors described; the wiring is parallel to that Thyristor and consists essentially of a capacitance and one Parallel connection of a diode and a resistor. This RCD wiring is intended to prevent impermissible voltage stresses, which means a too high du / dt and overvoltages, prevent.

Mainly by the capacity one can be too quick Prevent voltage rise at the thyristor. This RCD circuit is not associated with parallel operation of Thyristors.

The invention has for its object a circuit arrangement with two converter valves in two parallel current branches to provide in which the division of the currents in the current branches is symmetrized as precisely as possible, both in the static case, thus the case of conductive thyristors with a slight slope increasing current flow is designated, as well as in the dynamic case, d. H. during switching operations.

This object is achieved by a circuit arrangement solved according to claim 1.

It is advantageous that with different sizes of Saturation voltages at the converter valves resulting differential voltage evenly on the two Splits line sections, provided that the total current in one Edge rises or changes rapidly due to a switching operation. Because the line sections form together with the common Kern a transformer with windings wound in opposite directions, and according to the transformer equations - with the same number of turns - the voltages on the windings are always the same.

If the currents in the case of a rising current edge during the Switch-on time of the converter valves are also symmetrical Switching off ensures an almost uniform current distribution. In the branch with the lower saturation voltage on Converter valve is due to the mentioned voltage on the winding caused a magnetizing current. Its size depends on the Duration from, during which the voltage is present, and from the amount of Tension. This magnetizing current creates an "asymmetry" in relation to the currents flowing through the line branches, because without the magnetizing current, the currents would be the same. The magnetizing current can be achieved by dimensioning the circuit in practice easily limited to less than 1% of the total current become. This small difference in currents is in practice negligible.

It is favorable to connect a relief network in parallel the converter valves according to claim 2.  

The relief network prevents thermal destruction of the Converter valves when switching off through the line branches flowing streams. Because when you switch off it forms Line section through which the magnetizing current flows a high voltage, which is caused by the capacitors of the Relief network added and later slowly dismantled is what the power loss at the converter valves is limited. Without the relief networks can turn off Power losses up to a few kilowatts during a period of occur up to a microsecond. Most converter valves cannot withstand this extreme strain.

Another form of a relief network is in claim 3 Are defined. The Zener diodes limit the voltage that Switching off on the line sections through the rapid interruption of the magnetizing current occurs.

Instead of the Zener diodes can according to claim 4 Voltage limitation and thus also to limit the Power loss ohmic resistors can be inserted.

Further preferred embodiments of the invention are in the other subclaims marked.

An exemplary embodiment of the invention is described below four figures described in more detail, from which further details and give advantages.

Show it:

Figure 1a is a circuit diagram of the circuit arrangement according to the invention with a relief network.

Figure 2a is the curve of the total current rising edge during the duty cycle of the power converter valves.

2b shows the course of the voltage at the traversed by the magnetizing winding.

Fig. 2c the course of the magnetizing current, and

Figure 3 is a circuit diagram of the circuit arrangement according to the invention with two exported in another way relief networks.

Fig. 4 shows the design of a magnetic core with the two oppositely wound line sections.

In Fig. 1 is a circuit diagram with two parallel line branches (1, 2) can be seen lying insulated gate bipolar transistors (IGBTs I, II). The IGBTs (I, II) each have their collector and emitter connections in the line branches ( 1 , 2 ). The emitter connections are connected to one another via a switching point (D), which marks the end of the parallel line branches ( 1 , 2 ). With their collector connections, they each lie on a line section which is wound over a magnetic core (common to both line sections); the two line sections are wound in opposite directions to one another, so that the flow in the magnetic core is zero in the case of equal currents. Together with the magnetic core, the two line sections thus form a transformer with windings (L1, L2) which are wound in opposite directions.

At their ends not connected to the IGBTs (I, II), the two line sections in point (A) are directly connected to one another. At point (A), the parallel line branches (I, II) are preceded by an inductance ( 3 ) which, in the case of a DC voltage source, causes a current feed with a slightly rising edge during the switch-on times of the IGBTs (I, II).

There is a DC voltage between the connection of this inductance ( 3 ) and point (D) which is not connected to point (A) and is clocked with the aid of the IGBTs (I, II).

A relief network (E) is connected between the collector connections of the two IGBTs (I, II) and point (D). This relief network (E) has a parallel connection of a diode ( 4 ) and an ohmic resistor ( 5 ). The diode ( 4 ) is connected with its cathode to point (D) and lies with its anode on one connection of two capacitors (C).

These capacitors (C) are connected at their other connections to one of the two line branches ( 1 , 2 ), between the collector side of the two IGBTs (I, II) and the windings (L1, L2).

The base connections ( 6 ) of the two IGBTs (I, II) are each connected to a common control connection ( 8 ) via an ohmic resistor ( 7 ). The two IGBTs (I, II) are either conductive or blocked at the same time, depending on the voltage that is present at the control connection ( 8 ).

Fig. 2a shows schematically the current profile of the current flowing through the inductance ( 3 ) (I₁); the inductance ( 3 ) results in a slightly rising edge for the current up to the point in time at which the IGBTs (I, II) are blocked. The time until the IGBTs (I, II) are blocked is referred to as the switch-on time (t 1).

If the saturation voltages - due to component tolerances - between the collector and emitter of the two IGBTs (I, II) different are large, there is a voltage difference, which is in the present case according to the transformer equations evenly divides the two windings (L1, L2) because both windings (L1, L2) have the same number of turns.

In Fig. 2b the variation of the voltage applied to the winding (L1) voltage (U _L1) can be seen, and for the case that the described voltage difference occurs. It is assumed for simplicity that this voltage difference (as a result of a current edge) appears as a (temporary) DC voltage. The voltage (U _L1 ) is half as large as this voltage difference. During the switch-on time, the winding (L1 or L2), through which a larger current flows, is magnetized; it is now assumed that it is the winding (L1).

With

results for the magnetic induction B at the end of the switch-on time (t₁):

where Phi is the magnetic flux density and A that of the magnetic Induction perpendicular cross-sectional area of the with the Winding (L1) equipped magnetic core.

The following applies to the associated electrical current - the magnetizing current (I _M ) - using the equation for B derived from above and with the quantity L, which denotes the inductance of the winding (L1):

Except for this magnetization current (I _M ), the currents in both IGBTs (I, II) are the same. The magnetizing current (I _M ) flows in the line branch whose IGBT (I, II) has the lower saturation voltage.

Based on the previously described (idealizing) At the end of the switch-on time (t 1), the choke (winding L1) the energy

W (L1) = 1/2 · L · I _M ² (in practice, I _M ≈ 0.01 · I₁).

At the end of the switch-on time (t₁), the voltage (U _L1 ) on the inductance (L₁) changes its sign and is very large in amount (voltage U₂ in Fig. 2b). Since the current (I₁) does not drop abruptly in the real case, but with a (steep) flank, the product of current (I₁) and voltage (U₂) would easily result in a significant power loss in the kilowatt range without a relief network - for the duration of just under a microsecond.

So that the IGBTs (I, II) are protected against thermal destruction, is prevented by the relief network (E) that the Switch off the energy within a very short time in the IGBTs (I, II) is converted into heat.

For this purpose, the capacitors (C) are inserted in the relief network (E), which are charged via the downstream diode ( 4 ) when the current flow through the line branches ( 1 , 2 ) is interrupted. When the IGBTs (I, II) become conductive again in the next cycle, that is to say on the next current edge according to FIG. 2a, the capacitors (C) discharge via the resistor ( 5 ) and (slowly) pass the stored energy on to the resistor ( 5 ).

In Fig. 2c, the course of the magnetizing current (I _M) is shown. In practice, it can easily be dimensioned smaller than 1% of the total current (I 1).

Fig. 3 shows also substantially in two parallel line branches (1, 2) lying IGBTs (I, II), which are driven simultaneously. The two line branches are identical to those described in FIG. 1. In contrast to the circuit according to FIG. 1, two differently designed relief networks (E ') are present in the circuit according to FIG. 3, which are also - in contrast to FIG. 1 - connected in parallel to the two windings (L1, L2). They each consist of a series connection of a diode ( 9 ) and a zener diode ( 10 ). The diodes ( 9 ) are polarized in the reverse direction during the switch-on time (t 1); only when the voltage on the winding (L1 or L2) through which the magnetizing current (I _M ) reverses when switching off, are the diodes ( 9 ) polarized in the forward direction and the Zener diodes ( 10 ) are polarized in the reverse direction, so that the Zener voltage is present.

These relief networks (E ') with the Zener diodes ( 10 ) serve to limit the voltage (U 2), which forms when switching off at the end of the switch-on time (t 1) on each of the two windings (L1, L2) . As already explained above, the rapid drop in the current flowing through the line branches ( 1 , 2 ) by self-induction leads to a high voltage (U₂) at that winding (L1 or L2) through which the magnetizing current (I _M ) flows ( see Fig. 2b). The relief network (E ′) also serves to protect the IGBTs (I, II) in the course of switching operations. Because the Zener diodes ( 10 ) limit the voltage (U₂) which develops after the switch-on time (t₁) to the Zener voltage (ideally conductive diodes ( 9 ) provided), the diodes ( 9 ) being conductive and demagnetizing the windings (L1, L2) allow.

Instead of the Zener diodes ( 10 ), resistors can also be inserted, which likewise cause demagnetization and at the same time limit the current strength.

Fig. 4 shows the configuration of the windings (L1, L2), which are guided in a distorted winding direction over a common magnetic core ( 11 ) ("current-compensated choke"). The number of turns is 1, which keeps the inductance of the windings particularly low.

The windings shown are particularly easy to manufacture in terms of production technology, since one line section is guided in a straight path through the magnetic core ( 11 ) and the other line section is laid in the opposite direction. Inside the tubular magnetic core ( 11 ), the line sections of the two line branches ( 1 , 2 ) in the opposite direction are flowed through by currents (I₁ ', I₁''). If the current distribution is uniform, which cannot be achieved in practice due to the usual component tolerances in the circuit, these two currents are of equal size, and the magnetic core ( 11 ) is then not penetrated by any magnetic field lines. The above-mentioned magnetizing current (I _M ) causes the creation of a magnetic field, which induces the aforementioned voltage (U₂) when the power is interrupted.

Claims (7)

1. Circuit arrangement with several structurally identical converter valves arranged individually in two parallel line branches, the current flow through the line branches being controlled by the converter valves, characterized in that exactly two line branches ( 1 , 2 ) are present, that the line branches ( 1 , 2 ) each have a line section forming a winding (L1, L2) and that both line sections are wound in opposite directions with the same number of turns over a common magnetic core ( 11 ). 2. Circuit arrangement according to claim 1, characterized in that a relief network (E) is connected in parallel to the converter valves, which contains a parallel connection of a diode ( 4 ) and a resistor ( 5 ), that the diode ( 4 ) is poled in the same direction as the direction of the current in the converter valves lying parallel to the relief network (E), that two capacitors (C) are connected in series to the parallel connection, which are connected separately from one another between the line sections and the converter valves and that the parallel connection at their other end is closed directly on the side of the converter valves facing away from the line sections, a current flow occurring in the forward direction of the diode when the converter valves are switched off. 3. Circuit arrangement according to claim 1 or 2, characterized in that in each case a relief network (E ') is connected in parallel to the two line sections, which has a series connection of a diode ( 9 ) and a Zener diode ( 10 ), with respect to the current through the Line sections in conductive converter valves, the Zener diode ( 10 ) in the forward direction and the diode ( 9 ) is polarized in the reverse direction. 4. Circuit arrangement according to claim 3, characterized in that an ohmic resistor is used instead of the Zener diode ( 10 ). 5. Circuit arrangement according to one of the preceding claims, characterized, that the converter valves isulated gate bipolar transistors (IGBTs I, II). 6. Circuit arrangement according to one of claims 1 to 4, characterized, that the converter valves are MOS field effect transistors. 7. Circuit arrangement according to one of claims 1 to 4, characterized, that the converter valves bipolar transistors are.