Improved fault determination in power converters

Abstract

The invention relates to a method (500), a device and a computer program for a fault determination (161) in a dual active bridge converter (101) operating with a switching period TS. The dual active bridge converter (101) comprises: - a tracking device (156) providing a midpoint value set (228), the midpoint value set (228) comprising measured values (220V) of each of the midpoint voltages (221V, 222V, 223V, 224V), and a controller (150) providing a computed estimate set (158) comprising midpoint voltage estimates (158h, 158l), the method (500) comprising the steps of: a) determining (510) the midpoint value set (228) with the tracking device (156), b) computing (520) the computed estimate set (158), c) comparing (530) the midpoint value set (228) with the computed estimate set (158) to provide a comparison result (160), and d) providing (540) the fault determination (161) based on the comparison result (160).

Description

[0001] IMPROVED FAULT DETERMINATION IN POWER CONVERTERS FIELD OF THE INVENTION

[0002] The present invention relates to a method for determining faults in power converters, and more particularly to a method according to preamble of claim 1. The present invention also relates to a power converter provided with fault determination, and more particularly to a power converter according to preamble of claim 11. Further, the present invention also relates to a computer program for fault determination, and more particularly to a computer program according to preamble of claim 13.

[0003] BACKGROUND OF THE INVENTION

[0004] Electrical power conversion is an essential element in modern society. Different devices operate in different voltage and current ratings, and there is no universal DC or AC voltage level. It is advantageous to transfer power over high voltages to minimize power loss in the transfer process, and then considerably safer and more practical to use lower voltages when the electrical energy is consumed. Because of the substantial increase both in DC power generation (e.g. solar cells and fuel cells) and DC power storage (e.g. electric vehicle batteries and home electrical energy storage systems), power also must be converted from one DC level to another DC level in many applications.

[0005] A dual active bridge (DAB) is an isolated bi-directional DC-DC converter which has attracted great attention due to its capability of bi-directional power transferability, galvanic isolation between input and output stages, and zero voltage switching (ZVS) capability. These advantages have made the DAB an attractive choice for many applications, including chargers for electric vehicles (EVs), high-frequency-link power conversion systems, and battery storage systems.

[0006] One of the main challenges in electrical power converters is reliability. Semiconductor power electronic components, including MOSFET and IGBT switches, are the primary cause of failures in converters, in particular in DABs. Methodologies for enhancing the reliability of power converters are typically attempted with "fault determination" methods and devices that may e.g. comprise fault detection (FD), fault isolation (FI), and fault tolerance (FT) stages. FD is the initial step in detecting and timing a fault without identifying its type and location. As a result of this step, the FI sequence is initiated, identifying the location and the type of fault. Consequently, FT isolates the faulty component to prevent power converter failure. Switch faults can be classified into short circuit faults (SCFs) and open circuit faults (OCFs). SCFs can be highly destructive if not addressed promptly within the short circuit withstand time of the components by fault detection and protection schemes. SCFs can occur due to various causes, including faulty control or gate drive circuits, component destruction, miswiring, or dielectric breakdown of the load.

[0007] OCFs occur due to the lifting of the bond wires during thermal cycling, driver failures, or SCF-induced failures. Unlike SCFs, the converter can still operate in a limited fashion during OCFs in the steady state. By operating in this mode, the system efficiency will be severely reduced, and the components will be subjected to high electrical stresses. These often asymmetrical stresses across the power components affect the reliability and safety of the DAB converter during any longer-term operation.

[0008] Left undetected and / or uncared for, OCF and SCF failures may lead to catastrophic failures in the converter, including overheating, component explosions and fire.

[0009] To overcome these issues, various fault determination schemes have been proposed, with different detection speeds, implementation complexity and costs. They are based on multiple fault diagnosis signals, including monitoring the inductor current derivatives, inductor voltages, magnetic near-field signals etc. However, many of these schemes are impractical due to the high implementation cost associated with the need to use of numerous sensors of various kinds, and other limitations, such as heavy computational burdens that cannot be managed in converter real-time operation.

[0010] Hence, to prevent serious failures and to reach higher reliability in a converter, improved fault determination schemes are needed.

[0011] BRIEF DESCRIPTION OF THE INVENTION

[0012] Object of the present invention is to provide fault determination schemes for a power converter so that the problems related to the prior art mentioned above are solved or at least alleviated.

[0013] The objects of the invention are further achieved by a power converter related method, converter and computer program characterized by what is stated in the independent claims.

[0014] 1, 11 and 13.

[0015] The preferred embodiments of the invention are disclosed in the dependent claims. It has been surprisingly found that an improved fault determination in a DAB type of power converter may be achieved by tracking midpoint voltages of the legs of the DAB, and comparing them, with digital control means, to estimates of their expected values. Measuring dynamic voltages in midpoints is straightforward as the converter is often equipped with voltage sensors and their digital control and analysis schemes. As the output and input currents and voltages are determined computationally in a DAB (e.g. by adjusting pulse widths and phasing of the pulse width modulation in ways well known for persons having ordinary skilled in the art) by digital means, the estimated values for the midpoint voltages may be derived with digital computations.

[0016] An advantage of the invention is that faults and their locations in the switching legs of the converter can be determined reliably and quickly, without complicated modifications to the converter hardware.

[0017] As an aspect of the present invention, a method for a fault determination in a dual active bridge converter operating with a switching period Ts is disclosed. The dual active bridge converter comprises a primary side and a secondary side.

[0018] The primary side comprises first leg, the first leg comprising a first midpoint with a first midpoint voltage, and a second leg, the second leg comprising a second midpoint with a second midpoint voltage. The first midpoint voltage and the second midpoint voltage alternate between voltage of a primary side low voltage node, indicative of a low state of the first leg and a low state of the second leg, and voltage of a primary side high voltage node, indicative of a high state of the first leg and a high state of the second leg

[0019] The secondary side comprises a third leg, the third leg comprising a third midpoint with a third midpoint voltage, and a fourth leg, the fourth leg comprising a fourth midpoint with a fourth midpoint voltage. The third midpoint voltage and the fourth midpoint voltage alternate between voltage of a secondary side low voltage node, indicative of a low state of the third leg and a low state of the fourth leg, and voltage of a secondary side high voltage node, indicative of a high state of the third leg and a high state of the fourth leg.

[0020] The dual active bridge converter comprises a tracking device providing a midpoint value set, the midpoint value set comprising measured values of each of the first midpoint voltage, the second midpoint voltage, the third midpoint voltage, and the fourth midpoint voltage. The dual active bridge converter comprises also a controller providing a computed estimate set comprising midpoint voltage estimates for voltages of each of the midpoints.

[0021] The method comprises the steps of:

[0022] a) determining the midpoint value set with the tracking device,

[0023] b) computing the computed estimate set,

[0024] c) comparing the midpoint value set with the computed estimate set to provide a comparison result, and

[0025] d) providing the fault determination based on the comparison result. An advantage of the invention is that faults and their locations in the switching legs of the converter can be determined reliably and quickly, without complicated modifications to the converter hardware.

[0026] In an embodiment, in step b), the computed estimate set is based on computing - a primary side high level range VPHR to be between (VHP - ζ1 * VHP) and (VHP + ζ2 * VHP), - a primary side low level range VPLR to be between (VLP - ζ3 * VLP) and (VLP + ζ4 * VLP), - a secondary side high level range VSHR to be between (VHS - ζ5 * VHS) and (VHS + ζ6 * VHS), and

[0027] - a secondary side low level range VSLR to be between (VLS - ζ7 * VLS) and (VLS + ζ8 * VLS), wherein VHP is the voltage of the primary side high voltage node, VLP is the voltage of the primary side low voltage node, VHS is the voltage of the secondary side high voltage node, and VLS is the voltage of the secondary side low voltage node, and ζ1, ζ2, ζ3, ζ4, ζ5, ζ6, ζ7, and ζ8 each have an independent value which is larger than zero and smaller than 0.2, to adjust the voltage margin.

[0028] Benefit of this embodiment is that some margin of error is provided for the fault determination.

[0029] In an embodiment, in step c), the midpoint values set is compared with the computed estimate set such that

[0030] - the first midpoint voltage of the first leg high state is compared with the primary side high level range VPHR,

[0031] - the first midpoint voltage of the first leg low state is compared with the primary side low level range VPLR,

[0032] - the second midpoint voltage of the second leg high state is compared with the primary side high level range VPHR,

[0033] - the second midpoint voltage of the second leg low state is compared with the primary side low level range VPLR,

[0034] - the third midpoint voltage of the third leg high state is compared with the secondary side high level range VSHR,

[0035] - the third midpoint voltage of the third leg low state is compared with the secondary side low level range VSLR,

[0036] - the fourth midpoint voltage of the fourth leg high state is compared with the secondary side high level range VSHR, and

[0037] - the fourth midpoint voltage of the fourth leg low state is compared with the secondary side low level range VSLR. In an embodiment,

[0038]

[0039] ζ1, ζ2, ζ3, ζ4, ζ5, ζ6, ζ7, and ζ8 each have an independent value which is larger than zero and smaller than 0.1. This is a somewhat loose voltage margin that may lead to false negatives (the system has a fault but the fault is not indicated).

[0040] In an embodiment, ζ1, ζ2, ζ3, ζ4, ζ5, ζ6, ζ7, and ζ8 each have an independent value which is larger than zero and smaller than 0.05. This is advantageous voltage margin which has been noted to work well in practical implementations.

[0041] In an embodiment, ζ1, ζ2, ζ3, ζ4, ζ5, ζ6, ζ7, and ζ8 each have an independent value which is larger than zero and smaller than 0.01. This is a somewhat tight voltage margin that may lead to false positives (the system does not have a fault but the fault is indicated).

[0042] In an embodiment, the midpoint value set comprises:

[0043] - only one measured value for each of the first midpoint voltage, the second midpoint voltage, the third midpoint voltage and the fourth midpoint voltage; or

[0044] - at least two measured values for each of the first midpoint voltage, the second midpoint voltage, the third midpoint voltage and the fourth midpoint voltage; or

[0045] - at least three measured values for each of the first midpoint voltage, the second midpoint voltage, the third midpoint voltage and the fourth midpoint voltage.

[0046] Especially the short circuit faults lead to a very large fluctuation in the voltage values from a normal operation which may be detected with a single or few measurements.

[0047] In an embodiment, the step c) comprises, in comparing, determining if the values of each of the midpoint value set fall between or outside the respective values of the computed estimate set over time periods determined with a pulse width set.

[0048] In an embodiment, the pulse width set is based on pulse widths of the midpoint voltages such that for the high state of each of the legs, each of the high states comprising a high state pulse width, and for the low state of each of the legs, each of the low states comprising a low state pulse width, the fault determination

[0049] - indicating a fault in the first leg is provided, if the pulse width for the first midpoint voltage is not between TSH - TDT1 - α1 and TSH - TDT1 + α2, or

[0050] - indicating a fault in the second leg is provided, if the pulse width for the second midpoint voltage is not between TSH - TDT2 - α3 and TSH - TDT2 + α4, or

[0051] - indicating a fault in the third leg is provided, if the pulse width for the third midpoint voltage is not between TSH - TDT3 - α5 and TSH - TDT3 + α6, or

[0052] - indicating a fault in the fourth leg is provided, if the pulse width for the fourth midpoint voltage is not between TSH - TDT4 - α7 and TSH - TDT4 + α8,

[0053] where TSH is half of the switching period Ts, TSH = 0.5 * Ts. TDT1 is a deadtime of the first leg, TDT2 is a deadtime of the second leg, TDT3 is a deadtime of the third leg and TDT4 is a deadtime of the fourth leg, and α1, α2, α3, α4, α5, α6, α7, and α8 are lower and upper threshold values of each of the legs, respectively.

[0054] It is beneficial to provide some upper and lower threshold (to leave "some slack") so that unideal behavior of the converter does not create too many false positives for the fault detection, that is, a fault is detected even in a non-fault situation.

[0055] In an embodiment,

[0056] - α1, α2, α3, α4, α5, α6, α7, and α8 each have a value which is larger than zero and smaller than 0.25 * Ts; or

[0057] - α1, α2, α3, α4, α5, α6, α7, and α8 each have a value which is larger than zero and smaller than 0.125 * Ts; or

[0058] - α1, α2, α3, α4, α5, α6, α7, and α8 each have a value which is larger than zero and smaller than 0.05 * Ts; or

[0059] - α1, α2, α3, α4, α5, α6, α7, and α8 each have a value which is larger than zero and smaller than 0.02 * Ts; or

[0060] - α1, α2, α3, α4, α5, α6, α7, and α8 each have a value which is larger than zero and smaller than 0.01 * Ts; or

[0061] α1, α2, α3, α4, α5, α6, α7, and α8 each have an equal value which is larger than zero and smaller than 0.25 * Ts;

[0062] α1, α2, α3, α4, α5, α6, α7, and α8 each have an equal value which is larger than zero and smaller than 0.125 * Ts; or

[0063] - α1, α2, α3, α4, α5, α6, α7, and α8 each have an equal value which is larger than zero and smaller than 0.05 * Ts; or

[0064] - α1, α2, α3, α4, α5, α6, α7, and α8 each have an equal value which is larger than zero and smaller than 0.02 * Ts; or

[0065] - α1, α2, α3, α4, α5, α6, α7, and α8 each have an equal value which is larger than zero and smaller than 0.01 * Ts.

[0066] The smaller the threshold is, the more strict the fault determination becomes, and it may provide false positives. If the threshold is large, indicated by a large value of α1, α2, α3, α4, α5, α6, α7, and α8, false negatives can ensue.

[0067] Said 0.125 * Ts is an advantageous upper limit, minimizing false positives and false negatives.

[0068] A converter which has more e.g. noise and coupling related non-ideal aspects, benefits from a larger upper and lower threshold. In all, it is beneficial to provide some upper and lower threshold (to leave "some slack") so that unideal behavior of the converter does not create too many false positives for the fault detection, that is, a fault is detected even in a non- fault situation, and not too many false negatives, that is, a fault is not detected even when a fault occurs, that is, in a fault situation.

[0069] Keeping the upper and lower threshold values equal for each of the legs helps in the implementation as there are considerably less parameters to adjust and store.

[0070] In an embodiment, each of the legs comprises a high side switch and a low side switch. If the fault determination indicating fault in any of the legs is provided, and if, for the midpoint voltage for the leg for which fault is indicated, i) the low state pulse width is shorter than TSH - TDTN - OM, or ii) the high state pulse width is longer than TSH - TDTN + ON,

[0071] - determining that the location of the fault is the low side switch of the respective leg and of an open-circuit-fault type, and

[0072] - otherwise determining that the location of the fault is the high side switch of the respective leg and of an open-circuit-fault type,

[0073] wherein TDTN is the dead time and OM and OCN are the lower and upper threshold values of the respective leg.

[0074] Thus, it is possible, as an advantage in the embodiment, to pinpoint the faulty switch in the converter.

[0075] In an embodiment, each of the legs comprises a high side switch and a low side switch, and

[0076] - if the fault determination indicating fault in the second leg is provided, if, for the second midpoint voltage,

[0077] - the low state pulse width is shorter than TSH - TDT2 - α3, or

[0078] - the high state pulse width is longer than TSH - TDT2 + α4,

[0079] - determining that the location of the fault is the low side switch of the second leg, and - otherwise determining that location of the fault is the high side switch of the second leg,

[0080] wherein TDT2 is the dead time of the second leg, and α3 and α4 are the lower and upper threshold values of the second leg, respectively.

[0081] Thus, it is possible, as an advantage in the embodiment, to pinpoint the faulty switch in the second leg in the converter.

[0082] As another aspect of the invention, a dual active bridge converter configured to perform a fault determination and to operate with a switching period Ts is disclosed. The dual active bridge converter comprises a primary side and a secondary side.

[0083] The primary side comprises first leg, the first leg comprising a first midpoint with a first midpoint voltage, and a second leg, the second leg comprising a second midpoint with a second midpoint voltage. The first midpoint voltage and the second midpoint voltage alternate between voltage of a primary side low voltage node, indicative of a low state of the first leg and a low state of the second leg, and voltage of a primary side high voltage node, indicative of a high state of the first leg and a high state of the second leg.

[0084] The secondary side comprises a third leg, the third leg comprising a third midpoint with a third midpoint voltage, and a fourth leg, the fourth leg comprising a fourth midpoint with a fourth midpoint voltage. The third midpoint voltage and the fourth midpoint voltage alternate between voltage of a secondary side low voltage node, indicative of a low state of the third leg and a low state of the fourth leg, and voltage of a secondary side high voltage node, indicative of a high state of the third leg and a high state of the fourth leg.

[0085] The dual active bridge converter comprises:

[0086] - a tracking device configured to provide a midpoint value set, the midpoint value set comprising measured values of each of the first midpoint voltage, the second midpoint voltage, the third midpoint voltage and the fourth midpoint voltage, and a controller configured to provide a computed estimate set comprising midpoint voltage estimates for voltages of each of the midpoints. The dual active bridge converter is configured to:

[0087] a) determine the midpoint value set with the tracking device,

[0088] b) compute the computed estimate set,

[0089] c) compare the midpoint value set with the computed estimate set to provide a comparison result, and

[0090] d) provide the fault determination based on the comparison result.

[0091] An advantage of the invention is that faults and their locations in the switching legs of the converter can be determined reliably and quickly, without complicated modifications to the converter hardware.

[0092] In an embodiment, the dual active bridge converter is configured to execute the steps of a method disclosed above.

[0093] The disclosed dual active bridge converter is an advantageous device to operate the method disclosed above.

[0094] As yet another aspect of the invention, a computer program is disclosed. The computer program comprises executable instructions which are configured to execute all the steps of a method defined above, in a dual active bridge converter also defined above.

[0095] The disclosed computer program is an advantageous implementation of the method disclosed above, operable in the dual active bridge converter disclosed above.

[0096] Also, as an aspect of the current invention, a method for detecting an open circuit fault in dual active bridge converter is disclosed. The method comprises: analyzing open-circuit and short-circuit faults, identifying midpoint voltage signals for fault detection, and using the midpoint voltage signals as diagnostic signals to detect the open circuit fault. As an embodiment, an apparatus comprising computing unit to implement the method defined above is disclosed.

[0097] BRIEF DESCRIPTION OF THE DRAWINGS

[0098] The invention is described in detail by means of specific embodiments with reference to the enclosed drawings, in which

[0099] • Figure 1 shows a dual active bridge incorporating the invention,

[0100] • Figures 2 and 3 illustrate open circuit and short circuit faults,

[0101] • Figure 4 shows the steps of the inventive method,

[0102] • Figure 5 shows details of the waveforms of midpoint voltages, and related aspects of using the midpoint voltages in the context of the present invention, • Figures 6 and 7 show waveforms and related pulse width aspects, • Figure 8 shows groupings and locations of the faults, and

[0103] • Figure 9 shows schematically the computer program aspect of the present invention.

[0104] DETAILED DESCRIPTION OF THE INVENTION

[0105] In the following description, like numbers (e.g. 210) or like labels (e.g. 200a) denote like elements.

[0106] The following definitions also apply in the present application throughout:

[0107] A "switch" is any electrical component that can be dynamically arranged or switched to a high-conducting state between two of its power nodes, a first power node and a second power node, or to a low-conducting state between two of its power nodes, by a control signal to a control node of the switch, subject to bias voltage conditions between the power nodes and the control node, as is well known to a person skilled in the art.

[0108] The switch may be a transistor.

[0109] The switch may be a transistor arranged with bias voltages to perform switching action, bias voltages arranged in ways known well for a person skilled in the art.

[0110] The switch may be e.g. a Si MOSFET, a SiC MOSFET, a superjunction MOSFET, a cascode transistor, an IGBT, a JFET, a cascode combination of JFET and MOSFET or a GaN HEMT.

[0111] The switch may comprise many switching devices in parallel to increase the current carrying capacity of the switch. An "on state" of a switch, or a switch being "on" means a high-conducting state of the switch. In this state, the two power nodes are electrically connected.

[0112] An "off state" of a switch, or a switch being "off" means a low-conducting or a nonconducting state of the switch. In this state, the two power nodes are electrically disconnected.

[0113] Pulse-width modulation (abbreviated " PWM") means a well-known technique in power electronics to control the amount of power delivered to a load by varying the width of pulses in periodic signals controlling the switches of the power electronics device, e.g. a converter. Similarly, the phases of the periodic signals control the transfer of powers in the power electronics device.

[0114] A "power converter" or a "converter" is a device that is configured to change electrical energy fed to an input of the power converter in a first form and feed the electrical energy in a second form to the output of the power converter, for example to adjust voltage levels, convert between direct current (DC) or alternating current (AC), or alter frequency, to meet the specific needs of an electrical system or load.

[0115] In this document, a "converter" means a " DAB converter".

[0116] A " DC power converter" is a device that is configured to adjust voltage levels. In other words, A DC converter is configured to change a first DC level of voltage of the electrical energy fed to the input of the DC power converter and feed the electrical energy in a second DC level of voltage to the output of the DC power converter.

[0117] A " DAB" is a shorthand for a "dual active bridge" converter, which is a well-known circuit topology for an isolated (typically DC-DC) converter known for bidirectional power flow, galvanic isolation, and high efficiency. The DAB uses a high-frequency transformer for isolation and usually exhibits high efficiency and controllability across a wide range of input / output voltages.

[0118] A "leg" in a DAB (Dual Active Bridge) converter refers to a half-bridge circuit with at least two switches (e.g. MOSFETs) and their associated body diodes, arranged to create an AC voltage waveform, e.g. to one node of a transformer from the midpoint or switching node of the leg. Said AC voltage waveform may have a pulsed time behavior.

[0119] A "midpoint" is a voltage node in the leg, the leg comprising a high side switch and a low side switch, and the midpoint arranged between the high side switch and the low side switch.

[0120] A "midpoint voltage signal" means the voltage of any of the midpoints of any of the legs of the converter.

[0121] A "high state" (also called an "up-state") of the leg means that the midpoint is connected, through switching action of the switches, to a high voltage node. Thus, the voltage of the midpoint is equal to or close to being equal (due to the small non-zero series resistance of the switch in the "on state" switch) to the voltage of the high voltage node.

[0122] A "low state" (also called a "down-state") of the leg means that the midpoint is connected, through switching action of the switches, to the low voltage node. Thus, the voltage of the midpoint is equal or close to being equal (due to the small non-zero series resistance of the switch in the "on state" switch) to the voltage of the low voltage node.

[0123] A "shoot through" or "shoot-through" in a switched-mode power converter or supply (SMPC or SMPS) like a DAB is a dangerous condition where two switching devices, such as MOSFETs, in a leg are "on" at the same time, creating a direct low-impedance path from the high voltage node to the low voltage node or ground. This results in a large, uncontrolled surge of current that can damage or destroy the power supply and other components. Shoot-through can happen due to faults, mis-timed switching signals, or for a brief moment during a switching transition if the dead time is not properly set.

[0124] A "dead time" is an intentional, short time interval during which both switches in the same leg are "off", to avoid the shoot-through. To avoid a shoot-through, a dead time maybe arranged between state transitions of the switches.

[0125] Figure 1 shows a DAB converter 101 topology in a schematic sense, and a controller 150 controlling the operation of DAB converter 101. It also shows different voltage concepts between various voltage nodes of the DAB converter topology.

[0126] The dual active bridge converter 101 operates with a switching period Ts.

[0127] The DAB 101 comprises a primary side 121 and a secondary side 122.

[0128] The primary side 121 and the secondary side 122 are connected with a transformer 205 and a primary inductor 206 (which may be part of the leakage inductance of the transformer 205).

[0129] The winding ratio of the transformer 205 may be 1:1 (primary:secondary).

[0130] The winding ratio of the transformer 205 may be N:1 (primary:secondary).

[0131] The winding ratio of the transformer 205 may be 1: N (primary:secondary).

[0132] The AC side 123 of the DAB 101 comprise the transformer 205 and the primary inductor 206.

[0133] The primary side 121 comprises a first leg 111, which comprises a first midpoint 221 with a first midpoint voltage 221V, and a second leg 112, which comprises a second midpoint 222 with a second midpoint voltage 222V.

[0134] The first midpoint voltage 221V alternates between voltage of a primary side low voltage node 225L and voltage of a primary side high voltage node 225H as determined by the switching action of switches 200a, the first high side switch, and the first low side switch 200b. The second midpoint voltage 222V alternates between voltage of a primary side low voltage node 225L and voltage of a primary side high voltage node 225H as determined by the switching action of switches 200c, the second high side switch, and the second low side switch 200d.

[0135] Similarly, the secondary side 122 comprises a third leg 113, the third leg 113 comprising a third midpoint 223 with a third midpoint voltage 223V, and a fourth leg 114, the fourth leg 114 comprising a fourth midpoint 224 with a fourth midpoint voltage 224V.

[0136] The third midpoint voltage 223V alternates between voltage of a secondary side low voltage node 227L and voltage of a secondary side high voltage node 227H as determined by the switching action of switches 202a, the third high side switch, and the third low side switch 202b.

[0137] The fourth midpoint voltage 224V alternates between voltage of the secondary side low voltage node 227L and voltage of a secondary side high voltage node 227H as determined by the switching action of switches 202c, the fourth high side switch, and the fourth low side switch 202d.

[0138] Primary side high voltage node 225H and secondary side high voltage node 227H are jointly called high voltage nodes and when referring to either one of them, a high voltage node.

[0139] Primary side low voltage node 225L and secondary side low voltage node 227L are jointly called low voltage nodes and when referring to either one of them, a low voltage node.

[0140] Primary side low voltage node 225L and secondary side low voltage node 227L may be in zero potential (0 V).

[0141] Primary side low voltage node 225L and secondary side low voltage node 227L may be in a ground potential.

[0142] Primary side low voltage node 225L may be in a ground potential of the primary side. Secondary side low voltage node 227L may be in a ground potential of the secondary side.

[0143] Referring still to Figure 1, and also to Figures 4 and 5, as an aspect of the current invention, the dual active bridge converter 101 comprises a tracking device 156 providing a midpoint value set 228.

[0144] The midpoint value set 228 comprises measured values 220V of each of the first midpoint voltage 221V, the second midpoint voltage 222V, the third midpoint voltage 223V and the fourth midpoint voltage 224V.

[0145] Thus, the midpoint value set 228 is a sample of the midpoint voltages over time, describing the dynamic behavior of the switching action of the legs 111, 112, 113 and 114. The DAB converter comprises further a controller 150. The controller 150 is arranged to provide a computed estimate set 158. The computed estimate set 158 comprises midpoint voltage estimates 158h and 1581, high state estimates 158h, and low state estimates 1581, for voltages of each of the midpoints 221, 222, 223, 224.

[0146] The controller is also configured to control the switching operation of the DAB 101. The controller 150 may operate digitally, and comprise at least one processor, microcontroller or an application specific integrated circuit (ASIC), at least one bus and / or at least one digital memory. Operation of various parts of the controller 150 may be governed with software or a computer program.

[0147] A pulse-width modulation (" PWM") signal generator 151 generates pulsed signals determining times when the switches 200a - 200d and 202a - 202d of the DAB are to be turned on and off to meet input voltage 211 (Vin), output voltage 212 (Vout) and current requirements of the load 212, which in Figure 1 is represented with a voltage source of value Vout, in turn representing e.g. a battery. Naturally, virtually all types of electrical loads may be fed with the DAB (resistive, inductive, capacitive, reactive, frequency or time dependent, or any combination thereof).

[0148] An isolator 152 protects the PWM signal generator 151 and the rest of the converter control electronics from electrical failures e.g. in switches 200a - 200d or 202 - 202d. A driver 153 generates suitable signal levels or voltage levels for controlling the switches on and off. Finally, a gate drive network 155 connects the driver 153 to the control nodes 201a, 201b, 201c and 201d of the switches 200a, 200b, 200c and 200d, and 203a, 203b, 203c and 203d of the switches 202a, 202b, 202c and 202d respectively, to operate the switches 200a - 200d and 202a - 202d on and off over time.

[0149] Thus, the controller 150 may comprise the PWM signal generator 151, the isolator 152, the driver 153, the gate drive network 155, and the tracking device 156.

[0150] The PWM signal generator 151, the isolator 152, the driver 153 and the gate drive network 155 and the tracking device 156 may be embedded in whole or in part e.g. into one or more microcontroller units or a processors, and controlled by software or a computer program.

[0151] Referring to Figures 1, 4 and 5, the method 500 comprises a step of (step a, 510) determining the midpoint value set 228 with the tracking device 156.

[0152] The tracking device 156 may be e.g. a digital voltage sensor or a group of digital voltage sensors configured to measure voltages relative to one or more ground level voltages.

[0153] Midpoint voltages can reach 1000V in typical DCDC power conversion applications. Thus, the tracking device 156 may comprise a voltage divider comprising low tolerance (close to its specified nominal value) resistors which are selected to meet the design's required accuracy and voltage levels. The voltage divider may be arranged to scale the measured voltage down to suit the voltage ranges of the tracking device.

[0154] After scaling down, the controller 150 may be configured to locally compare and quantize the signal into predefined ranges. Alternatively, the controller 150 maybe configured to forward the signal in its "raw" form to the isolator 152. For example, the signal can be compared using LT1721 by Analog Devices and then transferred to the controller through digital isolators, e.g. ADuM250N by Analog Devices, which provide galvanic isolation and logic level translation. Alternatively, the scaled signal can be digitized using analog-to-digital converters such as the AMC3336, a precision, isolated delta-sigma (AD) modulator by Texas Instruments.

[0155] Further, the method comprises the step of computing (step b), 520) the computed estimate set 158. As with any switched mode converter, it is possible to compute the expected voltage levels of the midpoints in a correct operation of the DAB 101 (that is, when the DAB is operating without faults), if basic operational parameters like the input voltage and output voltage of the DAB 101 are known.

[0156] Thus, the computed estimate set 158 comprises the voltages of the midpoint voltages 221V, 222V, 223 and 224V in the correct operation of the dual active bridge converter 101.

[0157] In other words, the computed estimate set 158 comprises the voltages of the correct midpoint voltages 221V, 222V, 223 and 224V.

[0158] In other words, the computed estimate set 158 comprises values of voltages of the midpoint voltages 221V, 222V, 223 and 224V in the correct operation of the dual active bridge converter 101.

[0159] Still in other words, the computed estimate set 158 comprises values of correct midpoint voltages 221V, 222V, 223 and 224V.

[0160] Said expected voltage levels constitute the computed estimate set 158 comprising the correct ranges for the midpoint voltages when each of the legs is in the high state, represented with a high state estimates 158h, and when each of the legs is in the low state, represented with a low state estimates 1581.

[0161] In the high state of the leg 111, 112, 113, 114, the midpoint 221, 222, 223, 224 is connected the respective high voltage node 225H or 227H through switching action of the switches 200a - 200d, 202a - 202d of the DAB 101.

[0162] In the low state of the leg 111, 112, 113, 114, the midpoint 221, 222, 223, 224 is connected the respective low voltage node 225L or 227L through switching action of the switches 200a - 200d, 202a - 202d of the DAB 101.

[0163] Further, the method 500 comprises comparing (step c), 530) the midpoint value set 228 with the computed estimate set 158 to provide a comparison result 160. Said comparing of step c) is readily achieved in the controller 150, as both the measurements of step a) providing the midpoint value set 228 and the computations of the computed estimate set 158 of step b) are performed digitally, providing digital data which is readily compared with a computer program, and run in processors or microcontrollers and / or related digital technologies.

[0164] In step d), 540 of the method 500, the fault determination 161 based on the comparison result 160 is provided.

[0165] The fault determination may be an open circuit fault, or a short circuit fault, based on the characteristic of the comparison and its result.

[0166] The fault determination may be a determination indicating a failure 161f or that no failures are detected, that is, a pass 161p.

[0167] The fault determination may be a fault detection (FD), comprising an indication that at least one of the switches of the DAB is not operating properly.

[0168] The fault determination may be a fault isolation (FI), comprising an indication on which of the switches of the DAB is not operating properly, or which of the legs of the DAB is not operating properly.

[0169] Looking at Figure 2, an open circuit fault (OCF) is shown. In this case, the switch is permanently rendered to an "off" state, regardless of the control signal, with a very high resistance value between the power nodes of the switch. OCFs occur due to lifting of the bond wires during thermal cycling in the switch component, driver failures or if the channel or base of the device is destroyed to an insulating state e.g. due to a thermal event.

[0170] Looking next at Figure 3, an short circuit fault (SCF) is shown. In this case, the switch is permanently rendered to an "on" state, regardless of the control signal, with a very low resistance value between the power nodes of the switch. An SCF can be highly destructive if not addressed promptly within the short circuit withstand time of the components by fault detection and protection schemes. SCFs can occur due to various reasons, including faulty control or gate drive circuits, component destruction to a permanently conducting state, or mis wiring.

[0171] Referring to Figures 4 and 5, in method 500, in step b) 520, the computed estimate set 158 is based on computing, in step 521,

[0172] - a primary side high level range VTi I to be between Vzi n> - i * VHP and VHP + G * VHP, - a primary side low level range VTI N to be between VLP - G * VHP and VLP + G * VHP, - a secondary side high level range VSHR to be between VHS - G * VHS and VHS + G * VHS, and

[0173] - a secondary side low level range VSLR to be between VLS - G * VHS and VLS + G * VHS, wherein VHP is the voltage of the primary side high voltage node 225H and VHS is the voltage of the secondary side high voltage node 227H, and wherein VLP is the voltage of the primary side low voltage node 225L and VLS is the voltage of the secondary side low voltage node 227L. Multipliers

[0174]

[0175] £3, tk, I, and V each have an independent value which is larger than zero and smaller than 0.2.

[0176] Depending on the voltage levels, sensitivity and response times of the tracking device 156, said value of 0.2 may be excessive, resulting in missed fault events (false negatives). On the other hand, making too strict and narrow ranges may lead in a fault determination which is not really because of a fault (false positives), Thus, in an embodiment of method 500, in step 521, the computed estimate set 158 is based on computing, in step 521,

[0177] - Ci / Cz, Ci / Ci, Ci / CfCz, and C« each have an independent value which is larger than zero and smaller than 0.1; or

[0178] - Ci, Cz, Ci, Ci, Cs, CfCz, and Ci each have an independent value which is larger than zero and smaller than 0.05; or

[0179] - Ci, Cz, Ci, C4, Cs, Ci, Cz, and Ci each have an independent value which is larger than zero and smaller than 0.01.

[0180] In step c, the midpoint value set 228 may be compared with the computed estimate set 158 such that the midpoint voltage value 221V - 224V is compared with the corresponding side (primary side or secondary side) and leg state (up-state or low state) value.

[0181] In other words, specifically, in step c, the midpoint value set 228 may be is compared with the computed estimate set 158 such that

[0182] - the first midpoint voltage 221V of the first leg high state is compared with the primary side high level range VPHR,

[0183] - the first midpoint voltage 221V of the first leg low state is compared with the primary side low level range VPLR,

[0184] - the second midpoint voltage 222V of the second leg high state is compared with the primary side high level range VPHR,

[0185] - the second midpoint voltage 222V of the second leg low state is compared with the primary side low level range VPLR,

[0186] - the third midpoint voltage 223V of the third leg high state is compared with the secondary side high level range VSHR,

[0187] - the third midpoint voltage 223V of the third leg low state is compared with the secondary side low level range VSLR,

[0188] - the fourth midpoint voltage 224V of the fourth leg high state is compared with the secondary side high level range VSHR, and

[0189] - the fourth midpoint voltage 224V of the fourth leg low state is compared with the secondary side low level range VSLR. The midpoint value set 228 may comprise only a single measured value 220V for each of the first midpoint voltage 221V, the second midpoint voltage 222V, the third midpoint voltage 223V and the fourth midpoint voltage 224V.

[0190] If a fault may be detected with only a single value, the detection likely indicates a short circuit fault (SCF). This is because a short circuit fault creates a substantial deviation from the normal operation of the DAB converter 101. Similarly, the short circuit fault must be detected and managed quickly owing to the likely very high short circuit currents flowing in the faulty legs of the DAB converter 101.

[0191] Next turning to Figure 6 in addition to Figure 4, in an embodiment, the method 500 comprises (step c), 531) comparing 531 the values of the midpoint value set 228 to the respective values of the computed estimate set 158 over time periods determined with a pulse width set 229.

[0192] The pulse width set 229 comprises the information of the expected pulse widths of each of the midpoint voltages 221V - 224V when the converter 101 is operating in a non-fault state.

[0193] Pulse widths of the midpoint voltages are readily computed when the switching period Ts of the converter is determined. The switching period, in turn, is a fundamental parameter in the operation of DAB, well established in the literature of switched mode power converters. In general, power transferred from input to output is inversely proportional to the switching period. In practice, the switching period (which is the inverse of switching frequency / s, Ts = 1 / / s) is limited by time constants of the switching elements, by the maximum rate the magnetic flux can change in the inductors and transformers etc.

[0194] In a DAB with a duty cycle of D = 50%, in an ideal case, pulse widths are half of the switching period, and in practice, a small provision for the dead times must be made. Thus, in an embodiment, the duty cycle of each of the legs 111, 112, 113 and 114, is 50%.

[0195] Thus, in an embodiment, the duty cycle of each of the legs 111, 112, 113 and 114, is 50%, with provisions for a dead time.

[0196] In an embodiment, the duty cycle of each of the legs 111, 112, 113 and 114, is 50%, with provisions for dead times of transitions of each of the legs 111, 112, 113 and 114.

[0197] The pulse width set 229 may comprise pulse width data for each of the midpoints 221 - 224 and related midpoint voltages 221V, 222V, 223V, and 224V, and pulse width data for the high state and low state of the legs 111, 112, 113 and 114.

[0198] The comparing, in step 531, may be based on if the values of each of the midpoint value set 228 fall between or outside the respective values of the computed estimate set 158 over time periods determined with a pulse width set 229. Again, this comparing is readily achieved with digital devices of the controller 150.

[0199] Turning next to Figure 7 and Figure 8, for the locations of the faults relative to legs, in addition to Figure 6, the pulse width set 229 may be based on pulse widths of the midpoint voltages such that for a high state of each of the legs 111, 112, 113, 114, each of the high states comprising a high state pulse width 229H, and for a low state of each of the legs 111, 112, 113, 114, each of the low states comprising a low state pulse width 229 L, the fault determination 161 indicating a fault 161f 1 in the first leg is provided, if the pulse width for the first midpoint voltage is not between TSH - TDTI - ai and TSH - TDTI+ 02.

[0200] Further, the fault determination 161 indicating a fault 161f2 in the second leg is provided, if the pulse width for the second midpoint voltage is not between TSH - TDT2 - ots and TSH - TDT2 + c.

[0201] Further, the fault determination 161 indicating a fault 161 f3 in the third leg is provided, if the pulse width for the third midpoint voltage is not between TSH - TDTS - as and TSH - TDTS + Ot6.

[0202] Finally, the fault determination 161 indicating a fault 161f4 in the fourth leg is provided, if the pulse width for the fourth midpoint voltage is not between TSH - TDT4- as and TSH - TDT4+ as.

[0203] Above, TSH is half of the switching period, TSH = 0.5 * Ts, TDTI is a deadtime of the first leg 111, TDT2 is a deadtime of the second leg 112, TDTS is a deadtime of the third leg 113 and TDT4 is a deadtime of the fourth leg 114, and ai, as, as, at, as, as, as, and as are lower and upper threshold values of each of the legs 111, 112, 113, 114, respectively.

[0204] Each of the legs 111, 112, 113 and 114 may comprise an individual high state pulse width 229H and a low state pulse width 229L.

[0205] It is advantageous to tie the threshold values to the switching period of the DAB converter 101. Threshold values have an unit of time (seconds, e.g. microseconds or nanoseconds).

[0206] For example, cu, 02, as, at, as, as, as, and as each may have a value which is larger than zero and smaller than 0.25 * Ts.

[0207] For example, cu, as, as, c, as, as, as, and as each may have a value which is larger than zero and smaller than 0.125 * Ts.

[0208] For example, cu, as, as, at, as, as, as, and as each may have a value which is larger than zero and smaller than 0.05 * Ts.

[0209] For example, cu, as, as, at, as, as, as, and as each may have a value which is larger than zero and smaller than 0.02 * Ts. For example, oti, 02, ots, cu, as, ae, a?, and as each may have a value which is larger than zero and smaller than 0.01 * Ts..

[0210] For example, cu, as, as, a-t, as, as, a?, and as each may have an equal value which is larger than zero and smaller than 0.25 * Ts.

[0211] For example, cu, as, as, at, as, as, a?, and as each may have a value which is larger than zero and smaller than 0.125 * Ts.

[0212] For example, at, as, as, at, as, as, a?, and as each may have an equal value which is larger than zero and smaller than 0.05 * Ts.

[0213] For example, at, as, as, at, as, as, a?, and as each may have an equal value which is larger than zero and smaller than 0.02 * Ts.

[0214] For example, at, as, as, at, as, as, a?, and as each may have an equal value which is larger than zero and smaller than 0.01 * Ts.

[0215] In an embodiment, referring to Figure 8, each of the legs 111, 112, 113, 114 comprises a high side switch 200H, 200a, 200c, 202a, 202c and a low side switch 200L, 200b, 200d, 202b, 202d. If the fault determination indicating fault 161fl - 161f4 in any of the legs 111, 112, 113 and 114 is provided, the method 500 may further comprise steps to determine if the fault is in the low side switch 200L, or in the high side switch 200H.

[0216] The method comprises further a step of determining if, for the midpoint voltage 221V, 222V, 223V, 224V and for the respective leg 111, 112, 113, 114 for which fault is indicated, the low state pulse width is shorter than TSH - TDTN - OM, or the high state pulse width is longer than TSH - TDTN + ON, the method determines that the location of the fault 161fnl is the low side switch 200b, 200d, 202b, 202d of the respective leg 111, 112, 113, 114. Otherwise, the method determines that the location of the fault 161fnh is the high side switch 200a, 200c, 202a, 202c of the respective leg 111, 112, 113, 114.

[0217] In both cases (high side switch fault, or low side switch fault), it may further be determined that the fault is of an open-circuit-fault type.

[0218] If the low state pulse width is shorter than what occurs in the normal operation, or the high state pulse width is longer than what occurs in the normal operation, it is a sign that the leg has a problem of "going down", that is, enter the "down-state" or the "low-state". This may indicate a problem of the low side switch 200L to be turned on, indicative of an open circuit fault (OCF) on the low switch side.

[0219] Similarly, if the high state pulse width is shorter than what occurs in the normal operation, or the low state pulse width is longer than what occurs in the normal operation, it is a sign that the leg has a problem of "going up", that is, enter the "up-state" or the "high-state". This may indicate a problem of the high side switch 200H to be turned on, indicative of an open circuit fault (OCF) on the high switch side. Above, wherein TDTN is the dead time and XM and ON are the lower and upper threshold values of the respective leg 111, 112, 113, 114 such that cu (M = 1) is the lower threshold value of the first leg 111, as (N = 2) is the upper threshold value of the first leg 111, as (M = 3) is the lower threshold value of the second leg 112, c (N = 4) is the upper threshold value of the second leg 112, as (M = 5) is the lower threshold value of the third leg 113, c (N = 6) is the upper threshold value of the third leg 113, a? (M = 7) is the lower threshold value of the fourth leg 114, and finally as (N = 8) is the upper threshold value of the fourth leg 114.

[0220] In particular, for more clarity and example, in an embodiment, for leg 112, if the fault determination indicating fault in the second leg 161f2 is provided, the method may determine that if, for the second midpoint voltage 112,

[0221] - the low state pulse width 229L is shorter than TSH - TDT2 - as, or

[0222] - the high state pulse width 229H is longer than TSH - TDT2 + c,

[0223] the location of the fault 161f2 is the low side switch 200d of the second leg 112, shown as 161f21. Otherwise, as determined by the method, the location of the fault 161f2h is the high side switch 200c of the second leg 112, shown as 161f2h. Here, TDT2 is the dead time of the second leg, and as and c are the lower and upper threshold values of the second leg 112, respectively, and TSH is half of the switching period.

[0224] Looking back to Figures 1, 4 and 5, a dual active bridge converter 101 configured to perform a fault determination 161 and to operate with a switching period Ts is disclosed. The dual active bridge converter 101 comprises a primary side 121 and a secondary side 122. The primary side 121 comprises a first leg 111, the first leg 111 comprising a first midpoint 221 with a first midpoint voltage 221V, and a second leg 112, the second leg comprising a second midpoint 222 with a second midpoint voltage 222V.

[0225] The first midpoint voltage 221V and the second midpoint voltage 222V alternate between voltage of a primary side low voltage node 225L, indicative of a low state of the first leg 111 and a low state of the second leg 112, and voltage of a primary side high voltage node 225H, indicative of a high state of the first leg 111 and a high state of the second leg 112.

[0226] The secondary side 122 of the converter 101 comprises a third leg 113, the third leg 113 comprising a third midpoint 223 with a third midpoint voltage 223V, and a fourth leg 114, the fourth leg 114 comprising a fourth midpoint 224 with a fourth midpoint voltage 224V.

[0227] The third midpoint voltage 223V and the fourth midpoint voltage 224V alternate between voltage of a secondary side low voltage node 227L, indicative of a low state of the third leg 113 and a low state of the fourth leg 114, and voltage of a secondary side high voltage node 227H, indicative of a high state of the third leg 113 and a high state of the fourth leg 114. The dual active bridge converter 101 further comprises a tracking device 156 configured to provide a midpoint value set 228, the midpoint value set comprising measured values 220V of each of the first midpoint voltage 221V, the second midpoint voltage 222V, the third midpoint voltage 223V and the fourth midpoint voltage 224V.

[0228] The DAB converter 101 comprises also a controller 150 configured to provide a computed estimate set 158 comprising midpoint voltage estimates 158h, 1581 for voltages of each of the midpoints 221, 222, 223, 224.

[0229] The dual active bridge converter 101 is configured to:

[0230] a) determine the midpoint value set 228 with the tracking device 156,

[0231] b) compute the computed estimate set 158,

[0232] c) compare the midpoint value set 228 with the computed estimate set 158 to provide a comparison result 160, and

[0233] d) provide the fault determination 161 based on the comparison result 160.

[0234] A dual active bridge converter 101 may also be configured to execute the steps of the method as defined above.

[0235] Referring to Figure 9, a computer program 600 comprises executable instructions 610 which are configured to execute all the steps of the method 500 defined above method in a dual active bridge converter 101, also defined above.

[0236] The computer program may be run e.g. in the controller 150 of the DAB converter 101. The invention has been described above with reference to the examples shown in the figures. However, the invention is in no way restricted to the above examples but may vary within the scope of the claims.

Claims

CLAIMS1. A method (500) for a fault determination (161) in a dual active bridge converter (101) operating with a switching period Ts, the dual active bridge converter (101) comprising a primary side (121) and a secondary side (122),- the primary side (121) comprising first leg (111), the first leg (111) comprising a first midpoint (221) with a first midpoint voltage (221V), and a second leg (112), the second leg comprising a second midpoint (222) with a second midpoint voltage (222V), the first midpoint voltage (221V) and the second midpoint voltage (222V) alternating between voltage of a primary side low voltage node (225L), indicative of a low state of the first leg (111) and a low state of the second leg (112), and voltage of a primary side high voltage node, (225H) indicative of a high state of the first leg (111) and a high state of the second leg (112), - the secondary side (122) comprising a third leg (113), the third leg (113) comprising a third midpoint (223) with a third midpoint voltage (223V), and a fourth leg (114), the fourth leg (114) comprising a fourth midpoint (224) with a fourth midpoint voltage (224V), the third midpoint voltage (223V) and the fourth midpoint voltage (224V) alternating between voltage of a secondary side low voltage node (227L), indicative of a low state of the third leg (113) and a low state of the fourth leg (114), and voltage of a secondary side high voltage node (227H), indicative of a high state of the third leg (113) and a high state of the fourth leg (114) characterized in that the dual active bridge converter (101) comprises:- a tracking device (156) providing a midpoint value set (228), the midpoint value set (228) comprising measured values (220V) of each of the first midpoint voltage (221V), the second midpoint voltage (222V), the third midpoint voltage (223V) and the fourth midpoint voltage (224V), and a controller (150) providing a computed estimate set (158) comprising midpoint voltage estimates (158h, 1581) for voltages of each of the midpoints (221, 222, 223, 224), the method (500) comprising the steps of:a) determining (510) the midpoint value set (228) with the tracking device (156), b) computing (520) the computed estimate set (158),c) comparing (530) the midpoint value set (228) with the computed estimate set (158) to provide a comparison result (160), andd) providing (540) the fault determination (161) based on the comparison result (160).

2. A method (500) according to claim 1, characterized in that in step b), the computed estimate set (158) is based on computing (521)- a primary side high level range VPHR to be between (VHP - CI * HP) and (VHP + 2 * VHP), - a primary side low level range VPLR to be between (VLP - s * VLP) and (VLP + I * VLP),- a secondary side high level range VSHR to be between ( V is - Cs * VHS) and (VHS + Cf> * HS), and- a secondary side low level range VSLR to be between (VLS - z * V / i s) and (VLS + * VLS), wherein Vnpis the voltage of the primary side high voltage node (225H), VLpis the voltage of the primary side low voltage node (225L), VHS is the voltage of the secondary side high voltage node (227H), and VLS is the voltage of the secondary side low voltage node (227L), and Ci / Cz, Ci / Ci, Ci / CfCz,and each have an independent value which is larger than zero and smaller than 0.2.

3. A method (500) according to claim 2, characterized in that in step c), the midpoint values set (228) is compared with the computed estimate set (158) such that- the first midpoint voltage (221V) of the first leg high state is compared with the primary side high level range VPHR,- the first midpoint voltage (221V) of the first leg low state is compared with the primary side low level range VPLR,- the second midpoint voltage (222V) of the second leg high state is compared with the primary side high level range VPHR,- the second midpoint voltage (222V) of the second leg low state is compared with the primary side low level range VPLR,- the third midpoint voltage (223V) of the third leg high state is compared with the secondary side high level range VSHR,- the third midpoint voltage (223V) of the third leg low state is compared with the secondary side low level range VSLR,- the fourth midpoint voltage (224V) of the fourth leg high state is compared with the secondary side high level range VSHR, and- the fourth midpoint voltage (224V) of the fourth leg low state is compared with the secondary side low level range VSLR.

4. A method (500) according to claims 2 or 3, characterized in that- Ci / Cz, Cz, Ci / Cs / CfCz, and Cs each have an independent value which is larger than zero and smaller than 0.1; or- Ci, Cz, Cz, Ci, Cs, CfCz, and Cs each have an independent value which is larger than zero and smaller than 0.05; or- Ci, Cz, Cz, Ci, Cs, CfCz, and Cs each have an independent value which is larger than zero and smaller than 0.01.

5. A method (500) according to any one of claims 1 - 4, characterized in that the midpoint value set (228) comprises:- only one measured value (220V) for each of the first midpoint voltage (221V), the second midpoint voltage (222V), the third midpoint voltage (223V) and the fourth midpoint voltage (224V); orat least two measured values (220V) for each of the first midpoint voltage (221V), the second midpoint voltage (222V), the third midpoint voltage (223V) and the fourth midpoint voltage (224V); orat least three measured values (220V) for each of the first midpoint voltage (221V), the second midpoint voltage (222V), the third midpoint voltage (223V) and the fourth midpoint voltage (224V).

6. A method (500) according to any one of claims 1 - 5, characterized in that- the step c) comprises, in comparing (531), determining if the values of each of the midpoint value set (228) fall between or outside the respective values of the computed estimate set (158) over time periods determined with a pulse width set (229).

7. A method (500) according to claim 6, characterized in that the pulse width set (229) is based on pulse widths of the midpoint voltages such that for the high state of each of the legs (111, 112, 113, 114), each of the high states comprising a high state pulse width (229H), and for the low state of each of the legs (111, 112, 113, 114), each of the low states comprising a low state pulse width (229L), the fault determination (161)- indicating a fault (161fl) in the first leg is provided, if the pulse width for the first midpoint voltage is not between TSH - TDTI - cu and TSH - TDTI+ CU, or- indicating a fault (161f2) in the second leg is provided, if the pulse width for the second midpoint voltage is not between TSH - TDT2 - as and TSH - TDT2 + c, or- indicating a fault (161f3) in the third leg is provided, if the pulse width for the third midpoint voltage is not between TSH - TDTS - as and TSH - TDTS + c, or- indicating a fault (161f4) in the fourth leg is provided, if the pulse width for the fourth midpoint voltage is not between TSH - TDT4- a? and TSH - TDT4+ as,where TSH is half of the switching period Ts, TDTI is a deadtime of the first leg (111), TDT2 is a deadtime of the second leg (112), TDTS is a deadtime of the third leg (113) and TDT4 is a deadtime of the fourth leg (114), and cu, 02, as, a-t, as, as, ay, and as are lower and upper threshold values of each of the legs (111, 112, 113, 114), respectively.

8. A method (500) according to claim 7, characterized in that- oti, ot2, 03, ai, as, 06, 07, and os each have a value which is larger than zero and smaller than 0.25 * Ts; or- cu, ot2, as, c, as, ot6, 07, and os each have a value which is larger than zero and smaller than 0.125 * Ts; or- cu, ot2, os, ot4, as, ot6, 07, and os each have a value which is larger than zero and smaller than 0.05 * Ts; or- ai, ot2, 03, ot4, as, as, 07, and os each have a value which is larger than zero and smaller than 0.02 * Ts; or- ai, ot2, 03, ot4, as, as, 07, and os each have a value which is larger than zero and smaller than 0.01 * Ts; orai, 02, 03, ot4, os, os, 07, and os each have an equal value which is larger than zero and smaller than 0.25 * Ts; oroi, 02, os, 04, os, os, 07, and os each have an equal value which is larger than zero and smaller than 0.125 * Ts; or- oi, 02, os, 04, os, 06, 07, and os each have an equal value which is larger than zero and smaller than 0.05 * Ts; or- oi, 02, os, 04, os, 06, 07, and os each have an equal value which is larger than zero and smaller than 0.02 * Ts; or- oi, 02, os, 04, os, 06, 07, and os each have an equal value which is larger than zero and smaller than 0.01 * Ts.

9. A method (500) according to claim 7 or claim 8, characterized in that each of the legs (111, 112, 113, 114) comprises a high side switch (200H, 200a, 200c, 202a, 202c) and alow side switch (200L, 200b, 200d, 202b, 202d), and if- the fault determination indicating fault in any of the legs (111, 112, 113, 114) is provided, and if, for the midpoint voltage (221V, 222V, 223V, 224V) for the leg (111, 112, 113, 114) for which fault is indicated,- the low state pulse width is shorter than TSH - TDTN - OM, or- the high state pulse width is longer than TSH - TDTN + XN,- determining that the location of the fault (161fnl) is the low side switch (200L, 200b, 200d, 202b, 202d) of the respective leg (111, 112, 113, 114) and of an open-circuit-fault type, and - otherwise determining that the location of the fault (161fnh) is the high side switch (200H, 200a, 200c, 202a, 202c) of the respective leg (111, 112, 113, 114) and of an open-circuit-fault type, wherein TDTN is the dead time and XM and ON are the lower and upper threshold values of the respective leg (111, 112, 113, 114).

10. A method (500) according to claim 7 or claim 8, characterized in that each of the legs (111, 112, 113, 114) comprises a high side switch (200H, 200a, 200c, 202a, 202c) and a low side switch (200L, 200b, 200d, 202b, 202d), and- if the fault determination indicating fault in the second leg (161f2) is provided, if, for the second midpoint voltage (112),- the low state pulse width is shorter than TSH - TDT2 - α3, or- the high state pulse width is longer than TSH - TDT2 + α4,- determining that the location of the fault (161f21) is the low side switch (200d) of the second leg (112), and- otherwise determining that location of the fault (161f2h) is the high side switch (200c) of the second leg (112), wherein TDT2 is the dead time of the second leg, and as and c are the lower and upper threshold values of the second leg (112), respectively.

11. A dual active bridge converter (101) configured to perform a fault determination (161) and to operate with a switching period Ts, the dual active bridge converter (101) comprising a primary side (121) and a secondary side (122),- the primary side (121) comprising first leg (111), the first leg (111) comprising a first midpoint (221) with a first midpoint voltage (221V), and a second leg (112), the second leg comprising a second midpoint (222) with a second midpoint voltage (222V), the first midpoint voltage (221V) and the second midpoint voltage (222V) alternating between voltage of a primary side low voltage node (225L), indicative of a low state of the first leg (111) and a low state of the second leg (112), and voltage of a primary side high voltage node (225H) indicative of a high state of the first leg (111) and a high state of the second leg (112), - the secondary side (122) comprising a third leg (113), the third leg (113) comprising a third midpoint (223) with a third midpoint voltage (223V), and a fourth leg (114), the fourth leg (114) comprising a fourth midpoint (224) with a fourth midpoint voltage (224V), the third midpoint voltage (223V) and the fourth midpoint voltage (224V) alternating between voltage of a secondary side low voltage node (227L), indicative of a low state of the third leg (113) and a low state of the fourth leg (114), and voltage of a secondary side high voltage node (227H), indicative of a high state of the third leg (113) and a high state of the fourth leg (114) characterized in that the dual active bridge converter (101) comprises:- a tracking device (156) configured to provide a midpoint value set (228), the midpoint value set (228) comprising measured values (220V) of each of the first midpoint voltage (221V), the second midpoint voltage (222V), the third midpoint voltage (223V) and the fourth midpoint voltage (224V), and a controller (150) configured to provide a computed estimate set(158) comprising midpoint voltage estimates (158h, 1581) for voltages of each of the midpoints (221, 222, 223, 224), the dual active bridge converter (101) configured to:a) determine the midpoint value set (228) with the tracking device (156),b) compute the computed estimate set (158),c) compare the midpoint value set (228) with the computed estimate set (158) to provide a comparison result (160), andd) provide the fault determination (161) based on the comparison result (160).

12. A dual active bridge converter (101) according to claim 11, characterized in that the dual active bridge converter (101) is configured to execute the steps of a method (500) according to any one of claims 1 - 10.

13. A computer program (600), characterized in that the computer program (600) comprises executable instructions (610) which are configured to execute all the steps of a method (500) according to any one of claims 1 - 10 in a dual active bridge converter (101) according to any one of claims 11 - 12.

Images