Current source inverter system

The n-phase current source inverter system simplifies SRM drive systems by using a single switch per phase and open-loop control, achieving efficient torque control and reduced complexity, mimicking a series-excited DC machine for SRMs.

WO2026233913A1PCT designated stage Publication Date: 2026-11-12VIENNA UNIVERSITY OF TECHNOLOGY
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
PCT/EP2026/062777
Authority / Receiving Office
WO · WO
Patent Type
Applications
Current Assignee / Owner
Priority Date
2025-05-05
Filing Date
2026-05-04
Publication Date
2026-11-12

AI Technical Summary

Technical Problem

The complexity and high cost of conventional inverter topologies for switched reluctance motors (SRMs) due to the need for multiple switches and complex control systems hinder their widespread adoption, despite their environmental benefits over permanent magnet synchronous motors.

Method used

A n-phase current source inverter system with a simplified topology using a single switch element per phase, eliminating the need for a DC link inductor and utilizing load inductance, and operating in open-loop mode to control strictly positive currents, reducing complexity and cost.

Benefits of technology

The system simplifies SRM drive systems by allowing self-sustaining torque control through the DC link current, mimicking a series-excited DC machine, enhancing efficiency and reducing torque ripple and acoustic noise while maintaining high-frequency modulation for precise current shaping.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure EP2026062777_12112026_PF_FP_ABST
    Figure EP2026062777_12112026_PF_FP_ABST
Patent Text Reader

Abstract

The invention relates to an n-phase current source inverter system supplying, preferably inductive, loads requiring positive currents, preferably a reluctance motor, comprising a direct current (DC) source, preferably realized by a voltage source operated in current control mode, n parallel current paths (1), each path (1) including a switch element (2) connected in series with a load element (3) between a DC-plus (4) and a DC-minus rail (5), wherein the switch elements (2) are connected to the DC-minus rail (5) and the load elements (3) are connected to the DC-plus rail (4), an output filter comprising n capacitors (6), each connected to a node of one of the n current paths (1) between the load element (3) and the switch element (2), wherein the n capacitors (6) are arranged in a delta or star configuration, and a control unit preferably configured to operate the system in open-loop mode, wherein the current source inverter comprises no DC link inductor The invention further relates to a motor drive system comprising the n-phase current source inverter system.
Need to check novelty before this filing date? Find Prior Art

Description

[0001] CURRENT SOURCE INVERTER SYSTEM

[0002] The present invention relates to a current source inverter system supplying loads requiring positive currents and a motor drive system comprising the current source inverter system.

[0003] BACKGROUND OF THE INVENTION

[0004] Electric motors play a crucial role in industrial applications and account for over 40 % of the electrical energy consumed globally. Improving motor efficiency is essential, but with the sheer number of motors in use, designing motors sustainably is even more critical. Permanent magnet synchronous motors (PMSMs) are widely used in industry due to their high-power density and ease of control. However, PMSMs rely on rare-earth materials like neodymium, which are difficult to recycle and raise environmental concerns. In contrast, switched reluctance motors (SRMs) are made only from copper and iron, making them more sustainable. Unlike PMSMs, where PMs are energy-intensive and difficult to recycle, SRMs eliminate this environmental concern. SRMs can be employed in a wide range of applications, from household appliances and electric vehicles to industrial machinery and renewable energy systems.

[0005] Despite their advantages, the widespread adoption of SRMs has been limited by the high cost of unconventional voltage source inverters (VSIs) for SRMs (12 or 8 switches needed) and the complexity of their control systems. One of the primary reasons SRMs are not yet widely adopted is the complexity of their drive inverters.

[0006] SRMs require positive phase currents to generate torque. Therefore, they utilize open-end windings (OEWs), meaning both ends of each winding must be accessible. SRMs require unidirectional current because their torque generation is based on the reluctance attraction force between the energized winding and an iron tooth on the rotor. As the rotor spins, another coil in the sequence is energized, enabling further attraction of the rotor’s iron piece and allowing continuous rotation. Due to the positive-only currents, designing suitable inverters for SRMs has been the subject of extensive research.

[0007] Thus, SRM currents are strictly positive and can take arbitrary, non-sinusoidal waveforms.These waveforms are typically optimized to minimize torque ripple and acoustic noise while maximizing torque density. For simplicity, one may assume sinusoidal waveforms. Most prior art focuses on three-phase SRMs, with the three-phase current waveforms:

[0008]

[0009] where Î represents the peak value of the AC component, ω is the angular electrical frequency, and idc / 3 is the DC component. It is noteworthy that the sum of these currents is given by i1+ i2+ i3= idc. As a result, the currents do not sum to zero, meaning that SRMs cannot have a star point where the winding ends are connected on one side. For simplicity, one can assume Î = idc / 3, without loosing any generality.

[0010] Consequently, SRMs must utilize what are referred to as ’’open-end windings” (OEW), which means they have six terminals for connection of three-phase SRMs. The positive terminals of the windings are denoted as 1+, 2+, 3+, and negative terminals are 1-, 2-, 3-.

[0011] In prior art, mainly voltage source inverters (VSIs) are used. A conventional two-level, three-phase inverter is unsuitable because it generates alternating phase currents, which sum to zero and switch between positive and negative values. One solution for supplying SRMs with positive-only currents is using a full-bridge inverter. This configuration requires a total of 12 switches (MOSFETs). In this configuration, both ends of each winding are connected to bridge legs to enable the flow of the DC link current component through the SRM. Consequently, a total of 12 MOSFETs are required to drive the SRM, which significantly increases the complexity of the drive inverter. This added complexity makes SRMs less attractive for practical applications.

[0012] Variations of this topology exist, where half of the MOSFETs are replaced by diodes, resulting in a configuration with 6 MOSFETs and 6 diodes to simplify the drive inverter for the SRM, see e.g. W. Kong, D. Jiang, R. Qu, Z. Yu, S. Jia, and L. Jing, “Drive for dc-biased sinusoidal current vernier reluctance motors with reduced power electronics devices f in 2017 IEEE IEMDC, pp. 1-6, IEEE, 2017. While this design reduces the number of gate drivers required to 6—compared to the conventional VSI with 12 MOSFETs, which requires 12 gate drivers—itintroduces larger losses due to the diodes’ forward voltage drop, as diodes generally exhibit higher losses compared to MOSFETs.

[0013] To minimize the number of switching devices, a 4-leg inverter topology can be used for SRMs, as e.g. disclosed in US 10,608,571 B2. This configuration reduces the total number of switches to 8. The approach involves connecting the terminal of one phase winding to the terminal of the next phase. For example, the terminal pair 1- and 2+ are connected, as are 2- and 3+ These connection points are then linked to half-bridge circuits, while the winding terminals 1+ and 3-are connected to separate half-bridges. This topology introduces additional complexity in control because the SRM phase currents are no longer directly controlled by individual bridge legs. For instance, currents such as i1− i2and i2− i3are managed by the middle half-bridges, while the outer half-bridges control the first and last phase currents, i1and i3, respectively.

[0014] The above examples operate in pulse width modulation where the switching frequency is in the range of kilo hertz. Other examples described below are on switching at the fundamental frequency, where it is switched once when the coil in the SRM that conducts is changed. The switching frequency in the pulse width modulation is way higher than the fundamental frequency.

[0015] Other Switch-per-Phase topologies switching at the fundamental frequency from prior art include e.g. a converter configuration that employs one transistor and one diode per phase of the SRM (so-called R-Dump converter), as disclosed in R. Krishnan and P. N. Materu, “Design of a single-switch-per-phase converter for switched reluctance motor drives.'' IEEE transactions on industrial electronics, vol. 37, no. 6, pp. 469-476, 1990. When transistor is turned off, the current freewheels through diode, charging the capacitor. Subsequently, the stored energy flows through the external resistor R, which partially dissipates the energy stored in phase A. This approach has the disadvantage that the current in phase A takes longer to extinguish compared to directly recharging the source. Additionally, the dissipation of energy in the resistor reduces the overall effciency of the motor drive.

[0016] Another approach is the bifilar-winding-based SRM converter, which utilizes a single transistor and diode per phase while employing a bifilar winding to regenerate stored magnetic energy back to the source, see R. Davis, W. Ray, and R. Blake, “Inverter drive for switched reluctance motor: circuits and component ratings f in IEE Proceedings B (Electric Power Applications),vol. 128, pp. 126-136, IET, 1981. When the transistor turns off, the induced EMF forwardbiases the diode, circulating current through the bifilar secondary winding and returning energy to the supply. This topology relies on a voltage-driven approach, where the switch experiences a voltage stress proportional to the winding turns ratio. Thus, this approach needs bifilar windings, a voltage reflection mechanisms, and energy recirculation through secondary windings.

[0017] The split DC supply converter utilizes a dual-capacitor configuration to provide a mid-point voltage, allowing each phase to operate with a single switch while enabling freewheeling and regeneration, see H.-J. Ryoo, et al., ‘M new split source type converter for srm drives,” in PESC 98 record. 29th annual IEEE power electronics specialists conference, vol. 2, pp. 1290-1294, IEEE, 1998. However, this topology inherently derates the DC supply voltage, as only half of Vac is available at any time, leading to reduced utilization of the power supply. Additionally, proper charge balancing between capacitors is necessary, restricting its application to evenphase SRMs.

[0018] The C-dump converter, see e.g. T. J. Miller, “Converter volt-ampere requirements of the switched reluctance motor drive,” IEEE Transactions on Industry Applications, no. 5, pp.

[0019] 1136-1144, 1985, employs an additional inductor and a dedicated recovery switch to transfer stored magnetic energy from the motor windings to the DC link capacitor. This topology allows independent phase current control but introduces additional circuit complexity, requiring precise timing for recovery switch operation and increasing conduction losses due to energy circulation.

[0020] The single-controllable-switch-based SRM drive proposed in J. Kim, K. Ha, and R. Krishnan, “Single-controllable-switch-based switched reluctance motor drive for low cost, variablespeed applications, ” IEEE Transactions on Power Electronics, vol. 27, no. 1, pp. 379-387, 2011, is designed for a single-phase SRM with an asymmetric two-phase winding configuration. It operates using a single controllable switch and a recovery capacitor, enabling energy recirculation within the auxiliary winding. The topology requires a specialized commutation strategy dependent on the auxiliary winding and capacitor voltage, which introduces additional constraints on switching dynamics.Summarizing, the main challenge in the context of SRM drive systems is the complex inverter topologies, with VSIs requiring 12 or 8 switches. This leads to increased system complexity, higher costs and reduced reliability.

[0021] Prior art attempts to simplify inverter designs, such as US 5,870,292 A, have introduced resonant series link current source inverters (CSI) to reduce switching losses through zerocurrent switching. However, these systems still require a mandatory resonant link inductor and rely on complex high-side switching configurations, which maintain a high level of component count and control overhead.

[0022] SHORT DESCRIPTION OF THE INVENTION

[0023] In view of the above description of the background, the objective task of the invention is to provide an inverter topology for supplying loads requiring only positive currents, which features enhanced simplicity and efficiency, thus, reducing complexity while maintaining functionality and preferably pulse width modulation (PWM) operation.

[0024] A solution is provided by a n-phase current source inverter system supplying, preferably inductive, loads requiring positive currents, preferably a reluctance motor, comprising a direct current (DC) source, n parallel current paths, each path including a switch element connected in series with a load element between a DC-plus and a DC-minus rail, wherein the switch elements are connected to the DC-minus rail and the load elements are connected to the DC-plus rail, an output filter comprising n capacitors, each connected to a node of one of the n current paths between the load element and the switch element, wherein the n capacitors are arranged in a delta or star configuration, and a control unit preferably configured to operate the system in open-loop mode, wherein the n-phase current source inverter comprises no DC link inductor.

[0025] The inductance of the load itself facilitates CSI operation. In a conventional CSI a DC link inductor is required. The inventive approach allows to eliminate the need for the DC link inductor by utilizing the inductance of the load.

[0026] In contrast to prior art current source inverter (CSI) systems the inventive CSI topology only uses a single switch element per phase. Thus, each current paths preferably consists of a single switch element connected in series with a load between the DC-minus and DC-plus rail. As forthe inventive system the load requires only positive currents, these currents can be controlled using only the upper switches of a conventional CSI. This eliminates the need for the lower switch set, effectively reducing the number of switch elements by half, resulting in a configuration with just one switch element per phase.

[0027] The capacitors of the output filter allow to filter the high-frequency current components, ensuring smooth voltage supply to the loads. Preferably, the system operates in PWM (pulse width modulation) with high switching frequency commutates between the parallel branches. Due to this high switching frequency, there is a smooth voltage over the filter capacitors and small compact output capacitors. The capacitors may be arranged in a delta or star configuration. From a circuit point of view these configurations are equivalent. A delta configuration can be easier for practical applications to realize in hardware.

[0028] In contrast to the conventional CSI, where the phase currents sum to zero, it is crucial to note that in the inventive topology, the currents remain strictly positive. Consequently, by Kirchhoff’s Current Law, they sum precisely to the DC link current, e.g. for a 3-phase current source inverter system this is given by: i1+ i2+ i3= idc. Thus, a dedicated return path for the load currents is required, which may be provided via a DC-minus connection. This ensures proper current flow and maintains unidirectional conduction.

[0029] Thus, by using the fact that the load requires solely positive currents, it is possible to drastically simplify the topology of the CSI system.

[0030] The preferred open-loop control strategy eliminates the need for complex feedback control loops.

[0031] The load elements may be connected to the CSI via AC terminals.

[0032] Preferably, each load element in each current path is connected to the DC-plus rail, wherein each switch element in each current path is connected to the DC-minus rail. By connecting the switch elements directly to the DC-minus rail the gate driver implementation is simplified, as the switch elements no longer require an isolated gate drive supply. Thus, by connecting the switch elements with their source to the DC-minus rail complexity and cost is reduced by eliminating the need for isolation and allowing for a simplified NON-isolated gate driver circuit for the switch elements.The switch element can be chosen out of the group comprising MOSFETs, diodes, monolithic bidirectional switches, IGBTs, Thyristors, GaN e-FETs, GTOs, or combinations thereof. Preferably, a switch element consists of a MOSFET and a diode connected in series. This prevents reverse conduction to ensure, that blocking of line-to-line voltages as required for CSIs. In another embodiment single monolithic bidirectional switches may be used. Since only positive currents are commuted, the gate driver of the monolithic bidirectional switches can be simplified. For example, a simple low-voltage diode can be employed to control one of the gates, which could eliminate the need for two gate drivers per switch element. Further, it is also possible to use two MOSFETs in series, or a single monolithic bidirectional switch with two gate drivers. Such inverters would allow to change the direction of the DC link current. In this case the load elements can get positive AC currents, or negative AC currents.

[0033] In a preferred embodiment the load is an RL load comprising a resistor and an inductor connected in series. The RL load requires strictly positive currents, thus, allowing for the inventive topology of the CSI with a single switch element per load. The RL loads may correspond to a switched reluctance motor (SRM).

[0034] Further, the CSI can support in certain embodiments an m-phase load, with n ≥ m, wherein the control unit is further configured to map the n inverter phases to m load phases or individual load coils. For example, a SRM has typically a greater number of coils than phases. These coils are often connected in series, as they carry the same current, resulting in a specific number of supplied phases. The inventive inverter can supply individual coils of the SRM, regardless of whether the number of coils exceeds or equals the number of phases. For instance, a 5-phase SRM may contain 10 coils. In such a case, it is possible to supply this SRM using either a 10-phase CSI, where each coil is individually controlled, or a 5-phase CSI, where pairs of coils are connected in series. Similarly, a 5-phase SRM could have 20, 30, or any arbitrary number of coils, which may require different configurations of the CSI phases. Since the proposed CSI topology is highly efficient, requiring only one switch element per phase, it is conceivable to opt for switches with lower voltage ratings, leading to an increased number of CSI phases to directly supply individual coils rather than connecting them in series. This approach could e.g. be advantageous in managing voltage levels across coils in comparison to the blocking voltage of MOSFETs and diodes.Note that the inventive CSI system is applicable to both linear and rotary SRMs without any limitations. The concept is general and can be used for any load, specifically SRM, provided that the phase currents remain strictly positive.

[0035] Preferably, the control unit is configured to control the current source inverter in open-loop mode, wherein the modulation index and the current angle, which is a phase shift between the phase currents and the rotor’s electrical angle, are fixed to a constant value. In the case of a SRM this allows for an effective torque control without the need for complex feedback systems. As the CSI operates in open loop, no user interaction is required. Further, the open-loop commutation of the CSI makes the torque on the shaft of the SRM proportional to the square of the DC link current.

[0036] Another aspect of the invention concerns a motor drive system comprising an inventive n-phase current source inverter system.

[0037] In a preferred embodiment the motor is a reluctance motor, more preferably a switched reluctance motor.

[0038] Further, the current source inverter may be integrated into the housing of the motor. This allows a compact, cost-effective and robust drive system that simplifies installation and maintenance.

[0039] In another embodiment, the control unit is configured to set a fixed modulation index for the inverter and a fixed current angle and achieving torque control by regulating the DC link current. The input voltage of the CSI may be set so that the DC link current is controlled to the desired value. In the case of a SRM this means the following. As the phase currents in a SRM add up to the DC link current, the DC link current determines the torque on the SRM (assuming that the current angle is not changing). The voltage can e.g. be realized with a half-bridge, fullbridge or multilevel converter, for example, flying capacitor with 3, 5, 7, or any number of levels, so any DC-DC converter can be used. Current source rectifier can be used (AC -DC converter) in case the power is taken from power grid. In general, any power converter that will ensure DC link current control can be used.

[0040] In this case the motor drive system is as easy to use as conventional DC machines. In a preferred embodiment the CSI, its modulator, the output capacitors and potential further components willbe integrated into the housing of the motor, preferably the SRM. Consequently, a user will only need to connect the system to the positive and negative terminals on the DC side of the CSI. For the user, this drive system will behave like a DC machine, where the DC current directly controls the torque on the SRM shaft. Summarizing, the inventive CSI system allows the whole motor drive system to behave like a DC machine. This can also foster a wider adoption of e.g. SRMs in various industries.

[0041] For the example of SRMs, the inventive CSI system, thus, allows directly determine the current reference, once the rotor angle information of the SRM along with the desired phase shift for the currents (i.e., the current angle) is obtained. By fixing the modulation index e.g. to m=1 and the initial current angle to θi = π / 2, the duty cycles required for CSI control become fully defined. Note that the current angle may take different values in other implementations, which in practice can depend on the position sensor readings. Preferably, the current angle is chosen such that maximum torque per ampere is ensured.

[0042] Thus, once the CSI receives the rotor angle from the SRM shaft, it has all the necessary information to autonomously steer and control the current through each phase. Looking at the broader perspective, the combination of the SRM and CSI to a combined motor drive system forms a self-sustaining control system, eliminating the need for user intervention to modify or regulate the CSI operation.

[0043] Summarizing, with the help of the inventive motor drive system torque control solely through the DC link current without any interaction with the CSI is possible. This results in a speed-controlled drive system, where the user operates the motor just like a series-excited DC machine. This represents a significant advantage, as it simplifies motor drives, especially SRM drives, for end users, promoting wider acceptance and adoption. Further, the inventive CSI system operates at a high switching frequency — typically in the range of tens of kilohertz and potentially exceeding a hundred kilohertz — allowing e.g. precise shaping of the SRM phase currents to any desired waveform. This flexibility is a key advantage, as it enables optimization for reduced torque ripple, minimized acoustic noise, and enhanced torque density. Unlike conventional methods, where phase current shaping is constrained by the fundamental frequency commutation, the inventive approach leverages high-frequency modulation to achieve superior performance characteristics while maintaining a simple DC-side control interface.In another embodiment where a regulated torque is not needed the DC link current control is not used. For example, the CSI system can be attached to the DC link voltage, which will after the transient set to a constant speed, which is the minimum possible speed if the DC link voltage is kept constant. In case of an SRM the SRM speed can then be adjusted by reducing the modulation index to be less than 1, which would lead to an SRM speed increase.

[0044] DETAILED DESCRIPTION OF THE INVENTION

[0045] The foregoing and other objects, features and advantages of the invention will become more apparent from the following detailed description, which proceeds with reference to the accompanying figures.

[0046] Fig. 1 shows a conventional current source inverter known in prior art utilizing both upper and lower sets of switches.

[0047] Fig. 2 shows an exemplary embodiment of the inventive current source inverter topology with only an upper set of switches.

[0048] Fig. 3a-3f show another embodiment of the inventive current source inverter (CSI) without a DC link inductor for a 3-phase CSI (Fig. 3a), a n-phase CSI (Fig. 3b), an n- phase CSI with two MOSFETs for each phase whose drains are connected (Fig.

[0049] 3c), an n-phase CSI with two MOSFETs for each phase whose sources are connected (Fig. 3d), an n-phase CSI with monolithic bidirectional switches (Fig.

[0050] 3e), and the n-phase CSI with the control of the DC link current (Fig. 3f).

[0051] Fig. 4a – 4c show a control scheme for a SRM CSI drive system, where the CSI utilizes information from the SRM shaft to generate the current reference (Fig. 4a), the simplified notation for the switch used in Fig. 4a (Fig. 4b) and a SRM cross-section depicting the definition of the mechanical angle (Fig. 4c).

[0052] Fig. 5a and 5b show the DC-side speed-torque characteristic of a SRM CSI drive system (Fig. 5a) and the relationship between the DC link current and the torque (Fig. 5b).Fig. 6 shows a speed control diagram of a SRM CSI drive system.

[0053] Fig. 7a – 7e show simulation waveforms validating the operation of a SRM CSI drive system during acceleration from standstill to 3000 rpm for speed (Fig. 7a), torque (Fig.

[0054] 7b), current (Fig. 7c), duty cycle of the input buck converter (Fig. 7d) and CSI duty cycles (Fig. 7e).The inventive n-phase current source inverter system enables a CSI topology specifically designed for loads requiring strictly positive currents. These loads might correspond to reluctance motors, e.g. SRMs.

[0055] In a conventional three-phase CSI, as depicted in Fig. 1, the output currents can be both positive and negative, following typically a sinusoidal pattern:

[0056] w

[0057]

[0058] ere, w m representing the CSI modulation index, which can range from 0 to 1 to regulate the amplitude of the output currents. The electrical angular frequency co is also freely adjustable, providing full control over the system’s operational dynamics. This is well-known and can be considered prior art. The angle 0i is the initial current angle that can be set to any value, and it is typically set to result in the maximum torque per ampere of the load, e.g. the SRM. When used for an SRM this angle depends on how the angle on the SRM is defined. For example, it can be defined as 0i = n / 2, such that it gives the largest torque. In practical applications, this depends on how the encoder is mounted. Further, the initial current angle 0i basically determines the phase shift between the currents and the rotor.

[0059] It is important to note that in this conventional CSI configuration, see e.g. Fig. 1, the load is star-connected, which imposes the constraint that the sum of the phase currents must always be zero, i.e., ii + 12 + is = 0. Summarizing, the conventional CSI as shown in Fig. 1 utilizes both upper and lower sets of switch elements 2. Since CSIs require blocking of line-to-line output voltages, in Fig. 1 eachMOSFET 21, which are indicated in the figures by Ti, is placed in series with a diode 22, which are indicated in the figures by Di, to prevent reverse conduction.

[0060] A key step in the inventive system is recognizing that since the load — specifically an SRM — requires only positive currents, these currents can be controlled using only the upper switch elements 2 of the CSI. Thus, the proposed CSI system is designed to control load currents that remain strictly positive. This eliminates the need for the lower switch set, effectively reducing the number of switch elements 2 by half, resulting in a configuration with just one switch element 2 per phase, as shown exemplary in Fig. 2 for a three-phase CSI system.

[0061] Thus, in contrast to the conventional CSI as in Fig. 1, where the phase currents sum to zero, the currents remain strictly positive for the inventive system. Consequently, by Kirchhoff s CurrentLaw, the phase currents sum precisely to the DC link current, i.e. ii + 12 + is = ide. Thus, a dedicated return path 1 for the load currents is required, which is provided via the DC-minus connection 5, as illustrated in Fig. 2. This ensures proper current flow and maintains unidirectional conduction.

[0062] As shown in Fig. 2 the CSI system comprises an output filter with n capacitors 6, indicated by Cf in the figures. Fig. 2 shows a three-phase CIS, thus, n=3 in Fig. 2. As shown in Fig. 2 the capacitors 6 can be arranged in a star configuration. They can also be arranged in delta configuration to reduce commutation loops. These configurations are schematically equivalent up to recalculations of capacitance values. Each capacitor 6 is connected to a node of one of the n current paths 1, in Fig. 2 three current paths 1, between the load element 3 and the switch element 2. The n parallel current paths 1, in Fig. 2 the three parallel current paths 1, include a switch element 2 connected in series with a load element 3 between the DC-plus 4 and DC-minus rail 5.

[0063] The CSI system further includes a control unit and a DC source for supplying a controlled DC current to the CSI (both not depicted). The DC source is preferably realized by a voltage source operated in current control mode. The voltage source may be a half-bridge converter or a fullbridge converter.

[0064] In Fig. 2 the switch elements 2 are exemplary presented as a MOSFET 21 and a diode 22 connected in series (see Fig. 3a). The load elements 3 are in Fig. 2 and all other figures exemplary depicted as RL (resistor (R) - inductor (L)) loads. The load elements 3 can, however, correspond to any load requiring positive currents, which is preferably also inductive.

[0065] For the general case of an n-phase system, the sinusoidal current waveform for each phase k

[0066] may be given by:

[0067]

[0068] andk G {1, 2,...,n}. The modulation index can be set to m = 1 for optimal efficiency in operation, leading to

[0069] J

[0070]

[0071] y c >

[0072] n. As expected, these currents sum up to ide, i.e.,fc=I

[0073] In the CSI system of Fig. 2 the system also comprises a DC link inductor 7, indicated by Ldc in the figure.

[0074] However, in a preferred embodiment the n-phase current source inverter system does not comprise a DC link inductor 7, see e.g. Fig. 3a - 3f. The conventional CSI requires a DC linkinductor 7. However, for the inventive system, the inductance of the load itself can be utilized for CSI operation, eliminating the need for the DC link inductor 7 in the CSI system.

[0075] In another preferred embodiment the arrangement of the switch elements 2 and the load elements 3 is modified so that the switch elements 2 are connected with their source to the DC-minus rail 5 and the load elements 3 are connected to the DC-plus rail 4, see e.g. Fig. 3a - 3f. This configuration simplifies the gate drive implementation, reducing complexity and cost by eliminating the need for isolation or simplify isolation as e.g. in Fig. 3d, where a single gate driver isolation per phase can be used.

[0076] In Fig. 3a the CSI system without DC link inductor 7 and each load element 3 in each current path 1 directly connected to the DC-plus rail 4 and each switch element 2 in each current path 1 directly connected to the DC-minus rail 5 is shown for a three-phase CSI.

[0077] For generality, Fig. 3b illustrates the generalized case of Fig. 3a for an n-phase CSI.

[0078] In Fig. 3f the same topology as in Fig. 3b is shown, but with the input voltage uacthat is set to that the DC link current ide is controlled to the desired value. For a SRM, where the phase currents add up to the DC link current ide, the DC link current determines the torque on the SRM (assuming the current angle 0i is not changing). Note that this DC link current control is not needed always. For example, if one does not care that the torque shall be regulated, one can attach the system e.g. in Fig. 3b to the DC link voltage Ude, which will after the transient set to a constant speed, which is the minimum possible speed if the DC link voltage is kept constant. For an SRM one can adjust the SRM speed then by reducing the modulation index m to be less than 1, which would lead to an SRM speed increase.

[0079] For the regulated input voltage like shown in Fig. 3c, the voltage can be realized as a halfbridge, full-bridge or multilevel converter (for example, flying capacitor with 3, 5, 7, or any number of levels), so any DC-DC converter or current source rectifier, i.e., AC -DC converter, in case power grid is used as a source, can be used. In Fig. 6, the usage of half-bridge is illustrated.

[0080] In other embodiments the switch element 2 is chosen out of the group comprising MOSFETs 21, diodes 22, monolithic bidirectional switches 23, IGBTs, Thyristors, GaN e-FETs, GTOs, or combinations thereof, wherein a single switch element 2 preferably comprises a MOSFET 21 and a diode 22 connected in series. In Fig. 3a, 3b and 3f each switch element 2 corresponds toa diode 22 and a MOSFET 21 connected in series. In these embodiments the MOSFETs 21 are directly connected to the DC-minus rail 5. This modification simplifies the gate driver implementation, as the MOSFETs 21 no longer require an isolated gate drive supply.

[0081] In Fig. 3c and 3d each switch element 2 corresponds to two MOSFETs 21 connected in series, where in Fig. 3c the drains of the MOSFETs 21 are connected and in Fig. 3d the sources of the MOSFETs 21 are connected, and in Fig. 3e each switch element 2 corresponds to a single monolithic bidirectional switch 23. The embodiments of Fig. 3c - 3e have a higher efficiency, since diodes 22 have a forward voltage drop and therefore higher losses. However, these embodiments are also more expensive than the embodiments of Fig. 3a, 3b and 3f.

[0082] In the embodiment of Fig. 3c with connected drains MOSFETs 21 that are connected to DC-minus 5, indicated as Ti-a,..., T„.ain the figure, can have non-isolated gate drivers, whereas isolated gate drive supplies are needed for MOSFETs 21 connected to the load elements 3 and indicated by Ti-b,..., T„.b in the figure. In the embodiment of Fig. 3d a single isolated gate supply can be shared for the two MOSFETs 21 of the switch element 2, since their sources are connected.

[0083] Thus, it is also possible to use two MOSFETs 21 in series, or a single monolithic bidirectional switch 23 with two gate drivers. Such a CSI system would allow to change the direction of the DC link current ide. In this case the load elements 3 can get positive AC currents that are between 0 and ide, or negative AC currents that are between -ide and 0. The reversion of the DC link current direction allows for example for recuperative breaking of the SRM drive system. This can be e.g. also achieved by keeping ide always positive and reversing the supply voltage. Thus, if needed, the DC link current ide can be positive and negative.

[0084] Note that the inventive n-phase CSI system may supply a load with n-phases. Thus, the number of phases may correspond to the number of load phases. However, the n-phase CSI system can also support a m-phase load with n ≥ m, where the control unit is configured to map the n inverter phases to the m load phases or individual load coils. For example, SRMs typically have a greater number of coils than phases. These coils are often connected in series, as they carry the same current, resulting in a specific number of supplied phases. The inventive n-phase CSI system can supply individual coils of the SRM, regardless of whether the number of coils exceeds or equals the number of phases. For instance, as illustrated in Fig. 4c, a 5-phase SRM may contain 10 coils. In such a case, it is possible to supply this SRM using either a 10-phase CSI, where each coil is individually controlled, or a 5-phase CSI, where pairs of coils are connected in series. Similarly, a 5-phase SRM could have 20, 30, or any arbitrary number ofcoils, which may require different configurations of the CSI phases. Since the proposed CSI system is highly efficient, requiring only one switch element 2 per phase, it is conceivable that designers might opt for switch elements 2 with lower voltage ratings, leading to an increased number of CSI phases to directly supply individual coils rather than connecting them in series. This approach could be advantageous in managing voltage levels across coils in comparison to the blocking voltage of MOSFETs 21 and diodes 22.

[0085] In another embodiment the load elements represent an SRM. The inventive CSI system is applicable to both linear and rotary SRMs without any limitations. The concept is general and can be used for any SRM, provided that the phase currents remain strictly positive.

[0086] Note that the control strategy for the n-phase CSI system is straightforward. The duty cycles, averaged over a switching frequency period for each switch element, are determined as:

[0087] U Jfc=*,*

[0088] e

[0089] This equation remains valid for any phase current waveform, provided that the phase currents remain strictly positive, i.e., ik > 0, where k 6 { 1, 2,..., n}, and n is the number of phases. Since in the inventive CSI system, all phase currents sum to the DC link current, the duty cycles

[0090] also sum to unity:

[0091]

[0092] Examples:

[0093] SRM CSI drive system control:

[0094] An exemplary control concept for the n-phase CSI system supporting an SRM is illustrated in Fig. 4a. In this control concept the CSI utilizes information from the SRM shaft to generate the current reference. The angle information is used to determine the rotor shaft position, which in turn generates the current angle 0i for the phase currents ik and subsequently computes the duty cycles dk. The current angle is a phase shift with respect to the electrical shaft angle.

[0095] The mechanical angle 0 is multiplied by the number of rotor teeth to obtain the current angle 0i.

[0096] Fig. 4a presents a five-phase SRM as an example. The rationale for choosing a five-phase configuration is its significantly lower torque ripple compared to a three-phase SRM. Due to the increased number of phases, a simplified notation is used for the switch element 2, whichin the example of Fig. 4a corresponds to a series connection of a MOSFET 21 and a diode 22, as depicted in Fig. 4b. For completeness and to clarify the rotor angle definition, a cross-section of the five-phase SRM is shown in Fig. 4c. This exemplary cross-sectional view of a SRM shows a stator with 10 teeth and a rotor with 8 teeth. Preferably, the motor consists solely of copper and iron, without permanent magnets (PMs).

[0097] A highlighted feature of the control, enabled by the inventive CSI system, is that once the rotor angle information 9 is obtained, along with the desired phase shift for the current angles 0i, the current reference can be directly determined.

[0098] The current phase shift, or initial current angle 0i, may be set to 0i = 7t / 2, but in other implementations, it may take different values. This variation arises due to differences in encoder mounting and mechanical angle 0 definition. In general, 0i is preferably chosen to ensure maximum torque per ampere.

[0099] By selecting a modulation index of m = 1, the duty cycles required for CSI control become fully defined, resulting in the following expression for an n-phase SRM:

[0100]

[0101] In this equation, cot = NR* 0 where NRis the number of rotor teeth and 0 is the mechanical angle denoted in Fig. 4c. Thus, once the CSI receives the rotor angle from the SRM shaft, it has all the necessary information to autonomously steer and control the current through each phase. Looking at the broader perspective, the combination of the SRM and CSI forms a self-sustaining control system, eliminating the need for user intervention to modify or regulate the CSI operation.

[0102] At this stage, with the CSI fully controlled by the SRM angle information and requiring no further user intervention, one can analyze the entire SRM CSI drive system from its DC-side connection points 4, 5, indicated by ’+’ and in Fig. 4a. Under the exemplary control scheme, the proposed SRM CSI drive exhibits DC machine-like behaviour from the DC side,

[0103]

[0104] specifically resembling a series-excited DC machine, with the torque given by:

[0105] where T is the SRM torque, 1< T is the torque constant, and ide is the DC link current. The torque is proportional to the square of the DC link current, a characteristic also verified by the plots in Fig. 5a and 5b. Fig. 5a illustrates the torque-speed profile of the SRM CSI system, which closely matches the torque-speed characteristics of a series-excited DC machine. In Fig. 5a the solid line represents the theoretical characteristic, while the star values correspond to steady-statetorque and speed values obtained from transient simulations, confirming the validity of the derived speed-torque relationship. This relationship can be expressed as:

[0106]

[0107] mechanical angular speed, Uais the supply voltage, and Racrepresents the equivalent resistance of the SRM per phase as seen from the DC

[0108]

[0109] side. This equivalent resistance is given by: where n is the number of phases, and R is the per-phase resistance of the SRM. Fig. 5b shows the relationship between the DC link current and the torque. Notably, uppercase letters denote steady-state values (Ide), while lowercase letters represent transient values (idc).

[0110] As demonstrated, the inventive CSI system enables SRM torque control solely through the DC link current, without any interaction with the CSI. This results in a speed-controlled drive system, as depicted in Fig. 6, where the user operates the SRM just like a series-excited DC machine. This represents a significant advantage, as it simplifies SRM drives for end users, promoting wider acceptance and adoption.

[0111] Note that all derivations were presented under the assumption of sinusoidal SRM phase currents. However, the same principle applies to any arbitrary phase current waveform, provided the currents remain strictly positive. Regardless of the specific shape of the SRM phase currents, the inventive CSI system supporting an SRM ensures that the user can control the SRM torque purely through the DC link current ide.

[0112] In the control diagram in Fig. 6 the control process begins with the speed reference IT, from which the measured speed > is subtracted, and the result is fed into the speed controller Cu. Typically, Cu may be a PI controller. The controller outputs the torque reference T*, which is then divided by the torque constant and passed through a square root function to generate the reference for the DC link current, i*dc. Position control can be trivially implemented by replacing the speed loop with a position loop, where the position reference and measurement are processed through a position controller that generates the torque reference, from which the DC link current reference is derived. In this case the controller type Cu has to change to PID. As this follows directly from the speed control loop, no separate diagram is provided.

[0113] Next, the DC link current reference i*dc is compared to the measured DC link current ide, and the error is processed by the current controller Ci, which outputs the voltage reference u*afor the input buck converter. This voltage reference is normalized by the input voltage Uintodetermine the duty cycle for the buck converter, which is then fed into a PWM module to generate the gate signals.

[0114] Summarizing, since the CSI operates solely based on the SRM angle information, a user does not interact with the CSI directly. Instead, the SRM can be controlled in the same manner as a series-excited DC machine. Consequently, speed and torque regulation may be achieved through an input buck converter, significantly simplifying the SRM drive system. This approach eliminates the need for a user to have specialized knowledge of SRM drive control, as the proposed CSI system and control strategy make the SRM behave like a conventional DC machine from the user’s perspective.

[0115] Note that the design and implementation of speed and current controllers, as well as their coupling, are well-established and considered prior art. However, with the inventive CSI system an SRM CSI drive system is possible, which enables full control of the SRM drive exclusively through its DC side. This should not be mistaken for the conventional commutation mechanisms found in brushless DC motors or traditional SRM drives, where the DC link current is commutated at the fundamental electrical frequency. In contrast, the CSI system may operate at a high switching frequency — typically in the range of tens of kilohertz and potentially exceeding a hundred kilohertz — allowing precise shaping of the SRM phase currents to any desired waveform. This flexibility is a key advantage, as it enables optimization for reduced torque ripple, minimized acoustic noise, and enhanced torque density. Unlike conventional methods, where phase current shaping is constrained by the fundamental frequency commutation, the CSI system leverages high-frequency modulation to achieve superior performance characteristics while maintaining a simple DC-side control interface.

[0116] Waveforms verifying the operation of the proposed system are presented in Fig. 7a - 7e. The numerical values used in the simulation are provided in the following table:Parameter Symbol Value

[0117] SRM Parameters

[0118] Nominal mecli. power F eeh 5 kW

[0119] Nominal mecli. speed n,nech 3000 rpm

[0120] Nominal torque T 15.9 Nm

[0121] Max. overload torque Tmaj. 31.8 Nm

[0122] Phase number n 5

[0123] Number of stator teeth Ns10

[0124] Number of rotor teeth JVf 8

[0125] Phase resistance R 0.05 fl

[0126] Stator inductance in aligned rot. pos. L&8.8 mH

[0127] Stator inductance in imaligned rot. pos. Lu0.5 mH

[0128] Moment of inertia J 0.001 kgm2

[0129] DC- Side Parameters

[0130] Nominal DC-link voltage v,&. 72 V

[0131] Nominal DC-link current inc 69.24 A

[0132] DC resistance Tide 0.015 fl

[0133] DC inductance £(ic1.4 mH

[0134] Nominal power conversion resistance Rpc1.04 fl

[0135] Tbrque constant fer 0.0033 N mA”2

[0136] Table 1: Simulation Parameters.

[0137] Fig. 7a to 7e show simulation waveforms validating the operation of the proposed SRM CSI drive system during acceleration from standstill to 3000 rpm. In Fig. 7a a speed reference and actual measured speed is simulated, demonstrating the system’s capability to track the desired speed profile are shown. In Fig. 7b it can be seen that the SRM torque reaches its maximum value during acceleration, thus, ensuring optimal torque generation. In Fig. 7c phase currents and the DC link current are simulated, where the quadratic relationship between torque and DC link current is evident when comparing to Fig. 7b, and SRM phase currents remain strictly positive as required. In Fig. 7d the duty cycle of the input buck converter is simulated, illustrating the control of the DC link voltage. In Fig. 7e the CSI duty cycles are simulated, which sum up to 1, as mandated by the invention, ensuring correct current modulation and phase current distribution.

Claims

CLAIMS1. A n-phase current source inverter system supplying, preferably inductive, loads requiring positive currents, preferably a reluctance motor, comprising• a direct current (DC) source, preferably realized by a voltage source operated in current control mode,• n parallel current paths (1), each path (1) including a switch element (2) connected in series with a load element (3) between a DC-plus (4) and a DC- minus rail (5), wherein the switch elements (2) are connected to the DC-minus rail (5) and the load elements (3) are connected to the DC-plus rail (4), • an output filter comprising n capacitors (6), each connected to a node of one of the n current paths (1) between the load element (3) and the switch element (2), wherein the n capacitors (6) are arranged in a delta or star configuration, and • a control unit preferably configured to operate the system in open-loop mode, characterized in that the current source inverter comprises no DC link inductor (7), wherein the current source behavior is established by the inductance of the load.

2. The n-phase current source inverter system according to claim 1, wherein the switch element (2) is chosen out of the group comprising MOSFETs (21), diodes (22), monolithic bidirectional switches (23), IGBTs, Thyristors, GaN e-FETs, GTOs, or combinations thereof, wherein each switch element (2) preferably comprises a MOSFET (21) and a diode (22) connected in series.

3. The n-phase current source inverter system according to any of the claims 1 to 2, wherein the load elements (3) are RL loads comprising a resistor (R) and an inductor (L) connected in series, preferably corresponding to a switched reluctance motor.

4. The n-phase current source inverter system according to any of claims 1 to 3, wherein each load element (3) in each current path (1) is connected to the DC-plus rail (4), wherein each switch element (2) in each current path (1) is connected to the DC-minus rail (5).

5. The n-phase current source inverter system according to any of the claims 1 to 4 supporting a m-phase load, with n ≥ m, wherein the control unit is further configured to map the n inverter phases to m load phases or individual load coils.

6. The n-phase current source inverter system according to any of the claims 1 to 5, wherein the control unit is configured to control the current source inverter system in open-loop mode, wherein a modulation index and a current angle are fixed to constant values.

7. A motor drive system comprising the n-phase current source inverter system according to any of claims 1 to 6, wherein the motor is preferably a reluctance motor.

8. A motor drive system according to claim 7, wherein the current source inverter is integrated into the housing of the motor.

9. A motor drive system according to any of the claims 7 to 8, wherein the control unit is configured to set a fixed modulation index (m) for the inverter and a fixed current angle (θi) and wherein the control unit is configured to regulate the DC link current (idc) by setting the input voltage of the current source inverter.