Noise analysis device, noise analysis method, and program

Abstract

The time range setting section (14) acquires a plurality of occurrence times (ton(i), toff(i)) at which a plurality of times of opening and closing of the semiconductor element respectively occur. The phase conversion sections (14a, 14b) generate a plurality of phase difference information for performing phase conversion of noise spectra at the times of opening and closing of the semiconductor element, which reflects time differences of the plurality of occurrence times. The adding section (18) calculates a calculation result (RTNS) of a total spectrum generated by the opening and closing of the semiconductor element by adding a plurality of noise spectra (NSon(i), NSoff(i)) obtained by phase-converting the noise spectra (SPNon, SPNoff) at the times of opening and closing of the semiconductor element using the plurality of phase difference information respectively.

Description

TECHNICAL FIELD

[0001] The present disclosure relates to a noise analysis apparatus, a noise analysis method, and a program. BACKGROUND

[0002] In order to predict electromagnetic noise generated due to switching of a semiconductor element of an electronic device or a power converter, a technique is known in which a noise observation point is simulated and noise is derived using an information processing apparatus (hereinafter referred to as "noise analysis technique").

[0003] For example, a noise analysis technique is known in which a propagation path of noise is modeled by a combination of circuit elements of a resistor, an inductor, and a capacitor in circuit simulation. On the other hand, in a case where modeling by a combination of circuit elements is difficult, a noise analysis using a noise transfer function obtained by electromagnetic field analysis is known to be effective.

[0004] For example, in Japanese Patent Application Publication No. 2013-242649 (Patent Literature 1), a noise analysis technique is disclosed in which a semiconductor element that is subjected to switching control is taken as a noise source, and a transient waveform (for example, a voltage waveform) including both a turn-on time and a turn-off time of the semiconductor element is Fourier-transformed to calculate a frequency spectrum of the noise source (hereinafter referred to as "noise source spectrum" for short).

[0005] Further, according to the noise analysis technique of Patent Literature 1, by multiplying a noise transfer function from the noise source to a noise observation point by the noise source spectrum after the noise source spectrum is calculated, noise at the noise observation point can be derived with high accuracy while taking into account complex propagation characteristics of the noise.

[0006] PRIOR ART DOCUMENTS

[0007] PATENT LITERATURE

[0008] Patent Literature 1: Japanese Patent Application Publication No. 2013-242649 SUMMARY

[0009] PROBLEMS TO BE SOLVED BY THE INVENTION

[0010] In a DCDC converter in which relatively simple switching control is performed, in a stable operation state, an on period length and an off period length of a semiconductor element are mostly regarded as constant. In this case, by processing a transient waveform of one switching cycle, which includes one transient waveform at the turn-on time and one transient waveform at the turn-off time corresponding to the on period length and the off period length of the semiconductor element in the stable operation state, respectively, as a noise source, noise analysis can be achieved.

[0011] The higher the noise analysis accuracy in the high frequency domain, the shorter the time step of the transient waveform needs to be made, for example, typically set to several [ns] or so. In contrast, the on-period length and the off-period length of the semiconductor element, for example, need several tens of [μs], which is longer compared to the above-mentioned time step.

[0012] Therefore, when the transient waveform of 1 on-off cycle, i.e., 1 turn-on and 1 turn-off, is taken as the noise source data, the number of time steps also increases. Furthermore, when the transient waveform of multiple on-off cycles is taken as the noise source data, there is a concern that the Fourier transform takes a long time due to the increase in the number of time steps as the object of the Fourier transform.

[0013] As described above, if it is a relatively simple DCDC converter, the noise analysis can be performed using the noise source spectrum obtained by Fourier transforming the transient waveform of 1 on-off cycle, with the time width as the object of the Fourier transform being suppressed.

[0014] However, in recent years, the on-off control of the power conversion circuit has been advancing, and even in a stable operating state, the on-period length and the off-period length sometimes change. In particular, it is known that in a PWM (Pulse Width Modulation) inverter, in principle, the semiconductor element is subjected to on-off control in conjunction with changes in the on-period length and the off-period length. When such a power converter is taken as the object of noise analysis, when the transient waveform containing 1 turn-on and 1 turn-off, respectively, as described above, is taken as the noise source and analyzed, the behavior in which the phase of the noise changes due to changes in the on-period length and the off-period length can no longer be reflected in the noise source. As a result, there is a concern that the accuracy of the noise analysis will decrease.

[0015] In order to reflect the behavior in which the on-period length and the off-period length change in the noise source, there is an option to perform noise analysis using noise source data obtained by Fourier transforming the transient waveform of multiple on-off cycles each containing multiple turn-ons and turn-offs. On the other hand, since the time width as the object of the Fourier transform increases in order to obtain such noise source data, the time required for the Fourier transform becomes longer. As a result, the time for the noise analysis becomes longer, which becomes a technical problem.

[0016] The present disclosure is made in view of the above-mentioned problems, and aims to provide a noise analysis technology that can accurately derive the calculation result of the observed noise at high speed even in the case where the on-period length and the off-period length of the semiconductor element change.

[0017] Technical solution for solving the technical problem

[0018] According to one aspect of the present disclosure, a noise analysis device is provided. The noise analysis device calculates a total spectrum of noise generated by switching, which is at least one of turning on and turning off, of a semiconductor element. The noise analysis device includes a first acquisition unit, a phase conversion unit, and a first addition unit. The first acquisition unit acquires information indicating a plurality of occurrence times at which a plurality of switching operations occur respectively in a noise analysis target period including a plurality of switching operations of the semiconductor element. The phase conversion unit generates a plurality of phase difference information corresponding to the plurality of occurrence times respectively, the plurality of phase difference information being used to perform phase conversion of a noise spectrum at the time of switching of the semiconductor element, which reflects a time difference of the plurality of switching operations. The first addition unit calculates a total spectrum obtained by adding a plurality of noise spectra, the plurality of noise spectra being obtained by performing phase conversion of the noise spectrum at the time of switching of the semiconductor element respectively using the plurality of phase difference information.

[0019] According to another aspect of the present disclosure, a noise analysis method is provided. The noise analysis method calculates a total spectrum of noise generated by switching, which is at least one of turning on and turning off, of a semiconductor element. In the noise analysis method, (1) information indicating a plurality of occurrence times at which a plurality of switching operations occur respectively in a noise analysis target period including a plurality of switching operations of the semiconductor element is acquired, (2) a plurality of phase difference information corresponding to the plurality of occurrence times respectively is generated, the plurality of phase difference information being used to perform phase conversion of a noise spectrum at the time of switching of the semiconductor element, which reflects a time difference of the plurality of switching operations, and (3) a total spectrum obtained by adding a plurality of noise spectra is calculated, the plurality of noise spectra being obtained by performing phase conversion of the noise spectrum at the time of switching of the semiconductor element respectively using the plurality of phase difference information.

[0020] Effects of Invention

[0021] According to the present disclosure, it is possible to provide a noise analysis technique capable of deriving a calculation result of observed noise at high speed and accurately even in a case where a length of a conduction period and a length of an off period of a semiconductor element change. BRIEF DESCRIPTION OF DRAWINGS

[0022] Figure 1 A conceptual diagram for explaining a hardware structure example of the noise analysis device of the present embodiment.

[0023] Figure 2 A block diagram for explaining a structure example of the noise analysis device of Embodiment 1.

[0024] Figure 3 A circuit diagram for explaining an example of a noise analysis target.

[0025] Figure 4 A waveform diagram showing an example of a turn-on transient waveform (voltage waveform).

[0026] Figure 5This is a waveform diagram showing an example of a turn-off transient waveform (voltage waveform).

[0027] Figure 6 This is a waveform diagram showing another example of the turn-off transient waveform (current waveform).

[0028] Figure 7 To show Figure 2 The waveform diagram of an example of the opening and closing control signal shown.

[0029] Figure 8 3 is a waveform diagram showing a turn-on transient waveform (voltage waveform) after polygonal line approximation.

[0030] Figure 9 To show the Figure 8 Characteristic diagram of an example of a noise source spectrum obtained by Fourier transforming the conduction waveform of .

[0031] Figure 10 For illustration Figure 2 1 is a block diagram showing an example of the configuration of an observation noise calculation unit.

[0032] Figure 11 For illustration Figure 10 A block diagram showing an example of the configuration of a total calculation unit is shown.

[0033] Figure 12 This is a block diagram illustrating a configuration example of an observation noise calculation unit for calculating observation noise from multiple noise sources.

[0034] Figure 13 For illustration Figure 2 1 is a block diagram of a modified example of the configuration of the observation noise calculation unit shown in FIG.

[0035] Figure 14 This is a block diagram illustrating a modified example of the noise analysis function unit.

[0036] Figure 15 This is a block diagram for explaining a first configuration example of a transient waveform timing deviation correction process.

[0037] Figure 16 This is a block diagram for explaining a second configuration example of the transient waveform timing deviation correction process.

[0038] Figure 17 This is the first example of a wiring diagram for a noise transfer function.

[0039] Figure 18 This is the second example of the wiring diagram of the noise transfer function.

[0040] Figure 19 This is the third example of the wiring diagram for the noise transfer function.

[0041] Figure 20This is the fourth example of the wiring diagram for the noise transfer function.

[0042] Figure 21 This is a block diagram illustrating an example of a structure for integrating the noise transfer function.

[0043] Figure 22 This is a circuit diagram illustrating another example of a noise analysis target.

[0044] Figure 23 This is a circuit diagram illustrating another example of a noise analysis target.

[0045] Figure 24 To show Figure 23 The spectrum diagram shows an example of the observed noise calculation results of the noise analysis target circuit shown.

[0046] Figure 25 for Figure 24 A magnified view of part of the frequency domain.

[0047] Figure 26 This is a block diagram illustrating a configuration example of a noise analysis device according to a second embodiment.

[0048] Figure 27 For illustration Figure 26 A block diagram of a structural example of the on / off separation unit in FIG.

[0049] Figure 28 This is a block diagram illustrating a configuration example of a noise analysis device according to a third embodiment.

[0050] Figure 29 This is a block diagram for explaining functions added to the observation noise calculation unit 10 in the noise analysis of the third embodiment.

[0051] Figure 30 This is a spectrum diagram showing an example of the observation noise calculation result obtained by the noise analysis according to the third embodiment.

[0052] Figure 31 This is a block diagram illustrating a configuration example of a noise analysis device according to a fourth embodiment.

[0053] Figure 32 This is a block diagram for explaining the function added to the observation noise calculation unit in the noise analysis of the fourth embodiment.

[0054] Figure 33 For illustration Figure 1 A block diagram of an example of the hardware configuration of an information processing device shown.

[0055] Figure 34 A block diagram illustrating a variation of performing phase transformation in the time domain.

[0056] Figure 35FIG. 1 is a block diagram showing a first configuration example of a noise analysis device that calculates observation noise using a noise source as a noise observation point.

[0057] Figure 36 FIG. 2 is a block diagram showing a second configuration example of a noise analysis device that calculates observation noise using a noise source as a noise observation point. Figure 35

[0058] Figure 37 Figure 36 FIG. 3 is a block diagram showing a structure of a total calculation section.

[0059] Figure 38 FIG. 4 is a block diagram showing a third configuration example of a noise analysis device that calculates observation noise using a noise source as a noise observation point. Figure 35

[0060] Figure 39 FIG. 5 is a block diagram showing a fourth configuration example of a noise analysis device that calculates observation noise using a noise source as a noise observation point.

[0061] Figure 40 FIG. 6 is a block diagram showing a fifth configuration example of a noise analysis device that calculates observation noise using a noise source as a noise observation point.

[0062] Figure 41 FIG. 7 is a block diagram showing a sixth configuration example of a noise analysis device that calculates observation noise using a noise source as a noise observation point.

[0063] Figure 42 FIG. 8 is a block diagram showing a seventh configuration example of a noise analysis device that calculates observation noise using a noise source as a noise observation point.

[0064] Reference Signs

[0065] ​​​1,1Y: noise analysis function section; 3: on-off control signal acquisition section; 4: noise transfer function acquisition section; 4X: transfer function integration section; 5: measurer parameter acquisition section; 6: load current waveform acquisition section; 10, 11: observed noise calculation section; 10X, 10Y, 10Y', 10Z: total calculation section; 12, 12Y, 13, 15, 16, 43, 44: multiplication section; 14: time range setting section; 14a-14d: phase conversion section; 17, 18, 19: addition section; 20: transient waveform acquisition section; 21: on transient waveform acquisition section; 22: off transient waveform acquisition section; 23, 24: Fourier transform section; 25: on noise source spectrum acquisition section; 26: off noise source spectrum acquisition section; 27a, 27b: time deviation detection function section; 28a, 28b: time deviation correction function section; 29: on-off separation section; 31: on time detection section; 32: off time detection section; 33: on transient waveform output section; 34: off transient waveform output section; 40: weighting operation function section; 41: window function calculation section; 42: weighting coefficient setting section; 51, 53: information processing device; 52: information display device; 54: cloud; 60: noise source spectrum correction section; 61, 62: load current value acquisition section; 65, 66: interpolation calculation function section; 71: 1st noise source; 72: 2nd noise source; 73: 1st noise observation point; 74: 2nd noise observation point; Gcmn, Goff, Gon: noise transfer function; 81: noise transfer function; 81a: 1st noise transfer function; 81b: 2nd noise transfer function; 82-84: relay point; 101, 102: semiconductor element (noise source); 200: analysis target circuit; 201: input positive terminal; 202: input negative terminal; 203, 303: load; 204: capacitor; 205, 411, 412, 413: intermediate terminal; 220: CPU; 230: memory; 240: I / O circuit; 250: bus; 301: output positive terminal; 302: output negative terminal; 304: reactor; 401: 1st arm; 402: 2nd arm; 403: 3rd arm; 405: AC load; NS(i): observed noise; NSoff(i): off observed noise; NSon(i): on observed noise; Prt1-Prt4: port; RTNS: observed noise calculation result; SPNoff: off noise source spectrum; SPNon: on noise source spectrum; Tstr: start time; Tend: end time; Vdi: diode voltage; Vtr: transistor voltage; WGoff(i), WGon(i): weighting coefficient; X1-XJ, X(i): load current value; toff(i): off time; ton(i): on time; w(t): window function. DETAILED DESCRIPTION

[0066] Embodiments of the present disclosure will be described below in detail with reference to the accompanying drawings. Also, the same reference numbers are added to the same or equivalent parts in the drawings, and their descriptions are not repeated.

[0067] Embodiment 1

[0068] First, the noise analysis device of the present embodiment is described. Figure 1 and Figure 2 A configuration example of the noise analysis device of the present embodiment is described. Figure 1 A hardware configuration example of the noise analysis device is shown, Figure 2 A block diagram showing a configuration example of the noise analysis device of Embodiment 1 is shown.

[0069] With reference to Figure 1 , the functions of the noise analysis device of the present embodiment are realized, for example, by executing a predetermined noise analysis program by the information processing device 51. That is, when the noise analysis program is executed, the functions of each block shown in the noise analysis function section 1 are realized by the arithmetic processing by the information processing device 51, thereby constituting the noise analysis device of the present embodiment, or executing the noise analysis method. By this, the simulation of the observation noise to which the noise analysis technology of the present embodiment is applied can be executed. Figure 2 The functions of each block included in the noise analysis function section 1 shown in the block diagram are realized, thereby constituting the noise analysis device of the present embodiment, or executing the noise analysis method. By this, the simulation of the observation noise to which the noise analysis technology of the present embodiment is applied can be executed.

[0070] That is, the functions of each block described in the block diagram shown below are basically realized by software processing based on the execution of a program. However, for at least a part of the functions of any block, a digital circuit or an analog circuit such as an FPGA (Field Programmable Gate Array) or an ASIC (Application Specific Integrated Circuit) can be constituted.

[0071] Figure 33 A hardware configuration example of the information processing device 51 is shown.

[0072] As shown in, for example, Figure 33 The information processing device 51 is constituted based on a computer to include a CPU (Central Processing Unit) 220, a memory 230, and an input / output (I / O) circuit 240. The CPU 220, the memory 230, and the I / O circuit 240 can transmit and receive data to and from each other via a bus 250.

[0073] A program including a noise analysis program is preliminarily stored in a part of the memory 230, and the noise analysis described later can be executed by the CPU 220 executing the program. The I / O circuit 240 transmits and receives data to and from other devices, for example, Figure 1The information display device 52, the information processing device 53, and the cloud 54 are connected to each other to input and output signals and data.

[0074] The observation noise calculation result obtained as the simulation result in the information processing device 51 can be displayed on the information display device 52. Further, although the information processing device 51 and the information display device 52 are separate in Figure 1 , they can be integrated.

[0075] Further, the observation noise calculation result can be stored in the information processing device 51 in addition to being displayed on the information display device 52. In addition, the storage location of the observation noise calculation result is not limited to the information processing device 51 in which the simulation is performed, but can be another information processing device 53 connected to the information processing device 51 via a wired or wireless network, or can be the cloud 54.

[0076] Next, the noise analysis function section 1 will be described with reference to Figure 2 , the noise analysis function section 1 includes an on-off control signal acquisition section 3, a noise transfer function acquisition section 4, an on transient waveform acquisition section 21, and an off transient waveform acquisition section 22.

[0077] The on transient waveform acquisition section 21 acquires an on transient waveform (voltage waveform) on the time axis of a semiconductor element (noise source) that is controlled on and off. Similarly, the off transient waveform acquisition section 22 acquires an off transient waveform (voltage waveform) on the time axis of the semiconductor element.

[0078] Here, Figure 3 An example of a circuit diagram of a noise analysis target is shown. In Figure 3 , the cables, power supplies, and peripheral circuits are omitted for simplicity of illustration.

[0079] With reference to Figure 3 , the analysis target circuit 200 has an upper branch semiconductor element 101 and a lower branch semiconductor element 102 connected in series between an input positive terminal 201 and an input negative terminal 202. A smoothing capacitor 204 is connected between the input positive terminal 201 and the input negative terminal 202.

[0080] The high potential side terminal of the upper branch semiconductor element 101 is connected to the input positive terminal 201, and the low potential side terminal of the lower branch semiconductor element 102 is connected to the input negative terminal 202. The low potential side terminal of the upper branch semiconductor element 101 and the high potential side terminal of the lower branch semiconductor element 102 are connected to an intermediate terminal 205. A load 203 is connected between the input positive terminal 201 and the intermediate terminal 205. In Figure 3 , the upper branch semiconductor element 101 mainly operates as a diode, and the lower branch semiconductor element 102 mainly operates as a transistor.

[0081] Alternatively, in the configuration of Figure 3 In this case, in the analysis target circuit 200, the upper branch semiconductor element 101 mainly operates as a transistor, and the lower branch semiconductor element 102 mainly operates as a diode.

[0082] Further, the analysis target circuit 200 can be configured to connect a plurality of sets (branches) of series-connected upper branch semiconductor elements and lower branch semiconductor elements in parallel between the input positive terminal 201 and the input negative terminal 202. In each branch, the number of series-connected semiconductor elements is not limited to two, and can be any number.

[0083] In the analysis target circuit 200, each of the upper branch semiconductor element 101 and the lower branch semiconductor element 102 (hereinafter, also simply referred to as "semiconductor element") is subjected to on-off control to perform a desired power conversion between the direct current power between the input positive terminal 201 and the input negative terminal 202 and the power (direct current power or alternating current power) input and output to and from the load 203. With the on-off control, the turn-on from the off state to the on state and the turn-off from the on state to the off state are repeatedly performed in the semiconductor element.

[0084] Figures 4-6 An example of the turn-on transient waveform and the turn-off transient waveform is shown. Figure 4 and Figure 5 A voltage waveform is shown as a transient waveform.

[0085] Figure 4 A waveform of the terminal-to-terminal voltage Vtr (hereinafter, also referred to as transistor voltage Vtr) of the semiconductor element (operating as a transistor) and the terminal-to-terminal voltage Vdi (hereinafter, also referred to as diode voltage Vdi) of the semiconductor element (operating as a diode) at the time when the semiconductor element (operating as a transistor) is turned on at tl is shown. During the off period of the semiconductor element (operating as a transistor), Vdi is around zero due to diode conduction, whereas Vtr > 0 and the turn-on of the transistor is prevented. In response to the turn-on of the transistor, Vtr decreases to zero, whereas Vdi rises due to diode off, and at this time, the voltage jump that occurs in Vtr and Vdi can become a noise source. That is, the voltage waveform after tl corresponds to the turn-on transient waveform.

[0086] Further, tl is a reference time in the turn-on transient waveform, and is defined as tl = 0. In addition, although in the above description, the turn-off transient waveform is described as a waveform in which the voltage of the semiconductor element (operating as a diode) is reduced to zero, the turn-off transient waveform is not limited to this. For example, the turn-off transient waveform can be a waveform in which the voltage of the semiconductor element (operating as a diode) is reduced to a voltage lower than zero. Figure 4In the present embodiment, t1 is determined to correspond to the timing at which the voltage of Vtr and Vdi starts to change, but t1 as a reference can be determined to correspond to the timing at which the voltage of the gate voltage of the semiconductor element starts to change. Alternatively, t1 can be determined to correspond to the timing at which the change of the on-off control signal described later starts.

[0087] Likewise, Figure 5 The terminal-to-terminal voltage Vtr of the semiconductor element (operating as a transistor) and the terminal-to-terminal voltage Vdi of the semiconductor element (operating as a diode) at the time when the semiconductor element (operating as a transistor) is turned off at t2 are shown. During the on period of the transistor, Vtr is around zero, whereas Vdi > 0 and the conduction of the diode is prevented. In response to the turn-off of the transistor, Vtr starts to rise from 0, whereas Vdi decreases to 0 due to the conduction of the diode, and the voltage jump that occurs at this time in Vtr and Vdi can become a noise source. That is, the voltage waveform after t2 corresponds to the turn-off transient waveform.

[0088] Further, t2 is a reference timing in the turn-off transient waveform, and is defined as t2 = 0. In addition, although in the present embodiment, t2 is determined to correspond to the timing at which the voltage of Vtr and Vdi starts to change, t2 as a reference can be determined to correspond to the timing at which the voltage of the gate voltage of the semiconductor element starts to change. Alternatively, t2 can be determined to correspond to the timing at which the change of the on-off control signal described later starts. Figure 5

[0089] Each time the semiconductor element is turned on and turned off, the voltage change shown in Figure 4 and Figure 5 is generated, so the turn-on transient waveform and the turn-off transient waveform of the semiconductor element can be obtained in advance based on the simulation results or the measured data.

[0090] For example, the measured data can be acquired using an oscilloscope (not shown) having a trigger function and a memory function. Specifically, the measured waveform shown in Figure 4 or Figure 5 is stored by triggering the turn-on or turn-off of the semiconductor element, and the turn-on transient waveform and the turn-off transient waveform can be obtained.

[0091] Alternatively, as shown in Figure 6 , the current waveform (e.g., the transistor current Itr) can be used as the transient waveform. As an example, Figure 6 the turn-on transient waveform and the turn-off transient waveform are shown in Figure 3 ​Fig. 2 is an example of a current measured waveform of the lower branch semiconductor element 102 (operating mainly as a transistor) in the analysis target circuit 200. As for the current waveform, the transient waveform can also be determined in advance based on simulation results or measured data (oscilloscope). As such, the turn-on transient waveform and the turn-off transient waveform acquired by the turn-on transient waveform acquisition section 21 and the turn-off transient waveform acquisition section 22 do not necessarily have to be voltage waveforms, and a current waveform can be used as either one or both.

[0092] Further, as for the turn-on transient waveform and the turn-off transient waveform of the semiconductor element obtained from measurement or simulation, a database of the transient waveforms can be pre-stored in any of the information processing devices 51, 53, and the cloud 54. Figure 2 The turn-on transient waveform acquisition section 21 and the turn-off transient waveform acquisition section 22 illustrated can acquire the turn-on transient waveform and the turn-off transient waveform by accessing the database.

[0093] Thus, the turn-on transient waveform acquisition section 21 acquires information indicating a voltage change or a current change on a time axis at the time of one-time turn-on. Likewise, the turn-off transient waveform acquisition section 22 acquires information indicating a voltage change or a current change on a time axis at the time of one-time turn-off.

[0094] Further, in the circuit in which a transistor and a diode are combined as illustrated in Figure 3 As illustrated in Figure 4 and Figure 5 , the diode is turned off when the transistor is turned on, and the diode is turned on when the transistor is turned off. Thus, strictly speaking, the turn-on transient waveform of the semiconductor element ( Figure 4 ) includes the turn-on transient waveform of Vtr and the turn-off transient waveform of Vdi. Likewise, strictly speaking, the turn-off transient waveform of the semiconductor element ( Figure 5 ) includes the turn-off transient waveform of Vtr and the turn-on transient waveform of Vdi.

[0095] In addition, in a case where noise of only one of the transistor and the diode is dominant, only one of the transient voltage waveform or the transient current waveform of the transistor and the transient voltage waveform or the transient current waveform of the diode can be used as the turn-on transient waveform and the turn-off transient waveform.

[0096] Further, in the analysis target circuit 200 illustrated in Figure 3 , there are two semiconductor elements as noise sources, but in Figure 2 , a structure for analyzing noise output from one noise source (semiconductor element) is illustrated for simplicity.

[0097] Referring again to Figure 2The on-off control signal acquisition unit 3 acquires an on-off control signal including information of the on time and the off time of the semiconductor element. The on-off control signal includes time series information of the on and off of the semiconductor element as a noise source.

[0098] For example, a gate signal for specifying the on period and the off period of the semiconductor element as shown in FIG. 6 can be used as the on-off control signal. Figure 7

[0099] Referring to FIG. 6, Figure 7 The gate signal is time-series data that is set to "1" during a period in which the semiconductor element should be controlled to the on state and is set to "0" during a period in which the semiconductor element should be controlled to the off state. In this case, the timing at which the value of the gate signal as the on-off control signal changes from "0" to "1" corresponds to the on occurrence timing, and the timing at which the value changes from "1" to "0" corresponds to the off occurrence timing.

[0100] Further, the value of the on-off control signal is not limited to the digital values "0" and "1", but can be an analog value that is set to be different between the on state period and the off state period. In this case, the timing at which the magnitude relationship of the analog value and a predetermined threshold value is reversed corresponds to the on timing or the off timing.

[0101] As such, one on-off control signal is prepared for one noise source (semiconductor element), and each on-off control signal contains information on the time axis indicating the on occurrence timing and the off occurrence timing of the corresponding semiconductor element (noise source). Further, it can be defined that the on timing of the semiconductor element (operating as a transistor) is the off timing of the semiconductor element (operating as a diode), and the off timing of the semiconductor element (operating as a transistor) is the on timing of the semiconductor element (operating as a diode), and only the on-off control signal of the semiconductor element (operating as a transistor) is prepared.

[0102] Specifically, the on-off control signal contains information on the time axis indicating a plurality of on occurrence timings and a plurality of off occurrence timings for one semiconductor element (noise source). In addition, the on-off control signal can be constituted by information indicating these on occurrence timings and off occurrence timings themselves.

[0103] As such, by including a plurality of on occurrence timings and off timings, it is possible to reflect changes in the on period length and the off period length of the semiconductor element in the on-off control signal data.

[0104] Referring again to FIG. 6, Figure 2 ​The noise analysis function unit 1 further includes Fourier transform units 23 and 24 and an observed noise calculation unit 10. The Fourier transform unit 23 outputs a frequency-domain on-state noise source spectrum SPNon, obtained by Fourier transforming the on-state transient waveform acquired by the on-state transient waveform acquisition unit 21. Similarly, the Fourier transform unit 24 outputs a frequency-domain off-state noise source spectrum SPNoff, obtained by Fourier transforming the off-state transient waveform acquired by the off-state transient waveform acquisition unit 22. The on-state noise source spectrum SPNon and the off-state noise source spectrum SPNoff, obtained by Fourier transforming the transient waveform data on the time axis by the Fourier transform units 23 and 24, are input to the observed noise calculation unit 10, along with the on / off control signal from the on / off control signal acquisition unit 3 and the noise transfer function from the noise transfer function acquisition unit 4.

[0105] Here, use Figure 8 and Figure 9 An example of Fourier transform processing of a transient waveform will be described.

[0106] Figure 8 Show Figure 4 The following is an example of a broken line approximation of the semiconductor device's turn-on transient waveform (transistor voltage Vtr). When using the actual measured waveform, it is preferable to use Figure 8 The measured waveform indicated by the solid line is approximated to the broken line as indicated by the dotted line, and the waveform approximated by the broken line is the target of Fourier transform.

[0107] The waveform after the broken line approximation can be generated by automatically executing the broken line approximation within the noise analysis function unit 1, or waveforms obtained by applying the broken line approximation to the on-transient waveform and the off-transient waveform acquired by the on-transient waveform acquisition unit 21 and the off-transient waveform acquisition unit 22 can be prepared in advance. Furthermore, the approximation is not limited to the broken line approximation, and any regression curve can be used for approximation.

[0108] Figure 9 Show the Figure 8 The waveform approximated by the broken line shown is subjected to Fourier transform. Typically, in Fourier transform, calculations are performed to obtain a complex spectrum. Figure 9 The absolute value of the spectrum at each frequency is plotted. The Fourier transform is obtained by, for example, the following equation (1).

[0109] [Mathematical formula 1]

[0110] … (1)

[0111] In formula (1), N(f) is a noise source spectrum at frequency f. n(t) is a voltage value or a current value indicating the turn-on transient waveform or the turn-off transient waveform on a time axis defined by t = 0 for t1 and t2 described above. w(t) is a window function, and T is a time width of the window function.

[0112] Alternatively, in the Fourier transform sections 23 and 24, a modified formula or an approximate formula capable of obtaining a Fourier transform result equivalent to formula (1) can be used. Further, the time width T of the window function is arbitrarily set, and in a case where the time width T of the window function is longer than the time length of the transient waveform data, the transient waveform data can be interpolated by extrapolation. In addition, although the noise source spectrum at f = 0, that is, direct current, does not need to be calculated, in the case of f = 0, the value obtained by formula (1) needs to be divided by 2.

[0113] The Fourier transform sections 23 and 24 take the transient waveform of one turn-on and one turn-off Figures 4-6 as an object of Fourier transform. That is, it is generally understood that the number of transient waveforms (turn-on and turn-off) taken as an object of Fourier transform in the Fourier transform sections 23 and 24 is less than the number of times of turn-on occurrence times and the number of times of turn-off occurrence times acquired from the on-off control signal.

[0114] Referring again to Figure 2 , although in the example of Figure 2 , the Fourier transform sections 23 and 24 perform Fourier transform on the turn-on transient waveform and the turn-off transient waveform in parallel, the Fourier transform sections 23 and 24 can sequentially perform Fourier transform on the turn-on transient waveform and the turn-off transient waveform using a common Fourier transform section.

[0115] The observation noise calculation section 10 calculates an observation noise calculation result RTNS using the turn-on noise source spectrum SPNon and the turn-off noise source spectrum SPNoff from the Fourier transform sections 23 and 24, the on-off control signal data from the on-off control signal acquisition section 3, and the noise transfer function from the noise transfer function acquisition section 4.

[0116] Figure 10 A block diagram showing a configuration example of the observation noise calculation section 10 shown in Figure 2 is shown.

[0117] Referring to Figure 10 , the observation noise calculation section 10 includes multiplication sections 12 and 13 and a total calculation section 10X. The noise transfer function acquisition section 4 acquires a noise transfer function Gon at the time of turn-on and a noise transfer function Goff at the time of turn-off. That is, in the example of Figure 10 , the noise transfer functions at the time of turn-on and the time of turn-off are separately set.

[0118] The multiplication unit 12 multiplies the on-state noise source spectrum SPNon by the on-state noise transfer function Gon. The multiplication unit 13 multiplies the off-state noise source spectrum SPNoff by the off-state noise transfer function Goff.

[0119] Here, the noise transfer function refers to data that includes the frequency-domain transfer function from the noise source to the noise observation point. This function can be determined in advance through, for example, electromagnetic field analysis or circuit analysis. For example, this noise transfer function is represented by the voltage or current at the noise observation point when a voltage of 1 [V] or a current of 1 [A] is applied to the noise source. To account for the phase difference between the noise source and the noise observation point, the voltage or current at the noise observation point must be represented as a complex number.

[0120] As described later, when there are multiple noise sources and noise observation points, the noise transfer function data includes noise transfer functions for each combination of noise source and noise observation point. The objects observed at the noise observation points are not limited to voltage and current; electric fields and magnetic fields may also be observed. However, in this specification, the use of electric and magnetic fields is omitted, and voltage and current are used as noise.

[0121] The noise transfer function data may be S-parameter data of a port having a noise source and a noise observation point. The noise transfer function obtained by the noise transfer function acquisition unit 4 may be derived by performing S-parameter transformation calculations to derive an equation for the voltage or current at the noise observation point when a voltage of 1 [V] or a current of 1 [A] is applied to the noise source.

[0122] Furthermore, when using S-parameters for ports with multiple noise sources and multiple noise observation points, the noise transfer function can be derived for each combination of noise source and noise observation point based on a single S-parameter. When performing S-parameter conversion, any impedance can be applied in series or parallel to the noise source. For example, a voltage of 1 [V] can be applied in series with the resistance of a semiconductor device when it is on, or a current of 1 [A] can be applied in parallel with the capacitance of a semiconductor device when it is off.

[0123] Alternatively, the noise transfer function may be set using Y parameters, Z parameters, or F parameters that can be converted to or from S parameters instead of S parameters.

[0124] like Figure 10 As shown, the on-state observed noise and off-state observed noise are calculated by multiplying the on-state noise source spectrum SPNon and the off-state noise source spectrum SPNoff by the on-state noise transfer function Gon and the off-state noise transfer function Goff, respectively. Furthermore, the observed noise is calculated by summing these on-state observed noise and off-state observed noise based on the data of the switching control signal.

[0125] Figure 11 Show instructions Figure 10A block diagram of a structure example of the total calculation section 10X shown.

[0126] With reference to Figure 11 The total calculation section 10X has a time range setting section 14, phase conversion sections 14a, 14b, multiplication sections 15, 16, and addition sections 17, 18.

[0127] The time range setting section 14 extracts the on occurrence time and the off occurrence time of the 1st to Nth (N: a natural number) from the designated noise analysis target period (start time Tstr to end time Tend) from the on-off control signal from the on-off control signal acquisition section 3. Thus, the on time ton(i) of the i-th (i: 1 to N) and the off time toff(i) of the i-th (i: 1 to N) are obtained from the on-off control signal. In addition, the noise analysis target period can be arbitrarily set, and can be a part or all of the time domain corresponding to the on-off control signal acquired by the on-off control signal acquisition section 3.

[0128] Hereinafter, for the sake of simplicity of explanation, an example in which the same number (N each) of on times and off times are extracted will be described, but in fact, there can be a case where only one of the on time and the off time is extracted from the pulse at the most edge of the noise analysis target period, and thus the number of on times and off times differs.

[0129] The phase conversion section 14a sets the i-th on time ton(i) = t, and calculates exp(-j-2πft) in order to apply a phase difference caused by a difference in on time to the noise source spectrum. Similarly, the phase conversion section 14b sets the i-th off time toff(i) = t, and calculates exp(-j-2πft) in order to express a time difference of off with a phase difference. The exp(-j-2πft) calculated in correspondence with the on time ton(i) and the off time toff(i), respectively, corresponds to one embodiment of "a plurality of phase difference information". By multiplying these phase difference information with the noise source spectrum, it is possible to include the phase change of each of the plurality of on times and the plurality of off times in the noise source spectrum.

[0130] The multiplication section 15 outputs the multiplication result of the output value (on observation noise) of the multiplication section 12 and the phase difference information of each of the plurality of on times from the phase conversion section 14a. Thus, the on observation noise NSon(i) corresponding to each of the i-th (i: 1 to N) on times ton(i) is calculated.

[0131] Likewise, the multiplication section 16 outputs the multiplication result of the output value of the multiplication section 13 (off observation noise) and the phase difference information of each of the plurality of off timing. Thus, the off observation noise NSoff(i) corresponding to the i-th (i: 1 to N) off timing toff(i) is calculated.

[0132] The on observation noise NSon(i) and the off observation noise NSoff(i) correspond to one embodiment of the "plurality of multiplication values" which are the multiplication values of the plurality of noise source spectra obtained by performing the phase transformation on the noise source spectrum (on noise source spectrum SPNon or off noise source spectrum SPNoff) according to the phase difference information corresponding to the on timing ton(i) and the off timing toff(i), respectively, and the noise transfer function (Gon or Goff). Further, in Figure 10 、 Figure 11 In the above, even if the positions (the order of multiplication operation) of the multiplication sections 12, 13 and the multiplication sections 15, 16 are exchanged, the same "plurality of multiplication values" can be calculated, and the function of the "first multiplication section" is realized by the multiplication sections 12, 13, 15, 16.

[0133] The addition section 17 adds the on observation noise NSon(i) from the multiplication section 15 and the off observation noise NSoff(i) from the multiplication section 16 to calculate the observation noise NS(i) of the i-th pulse (i.e., 1 on and off). Thus, the observation noise NS(1) to NS(N) of the 1st to N-th pulses, respectively, are calculated.

[0134] The addition section 18 outputs the observation noise calculation result RTNS by adding the observation noises NS(1) to NS(N) calculated by the addition section 17. The observation noise calculation result RTNS is expressed as a set of data of the noise intensity (e.g., noise voltage [dBV]) at each frequency which is the same as the noise source spectrum exemplified in Figure 9

[0135] Further, at this time, the addition can be performed by changing the weight between the plurality of pulses. For example, a weighting coefficient kw(i) can be introduced, and the multiplication value of the observation noise NS(i) from the addition section 17 and the weighting coefficient kw(i) is totaled by the addition section 18. As an example, the weighting coefficient kw(i) can be set to be large in the central part and small in the edge part during the period of the noise analysis object.

[0136] In the above, even if the positions (the order of multiplication operation) of the multiplication sections 12, 13 and the multiplication sections 15, 16 are exchanged, the same "plurality of multiplication values" can be calculated, and the function of the "first multiplication section" is realized by the multiplication sections 12, 13, 15, 16. Figure 11 ​In the above embodiment, the on-off control signal acquisition section 3 corresponds to an embodiment of the "first acquisition section", and the noise transfer function acquisition section 4 corresponds to an embodiment of the "second acquisition section". In addition, the functions of the "first addition section" are implemented by the addition sections 17 and 18. In addition, the on time ton(i) and the off time toff(i) correspond to the "a plurality of occurrence times" in which the semiconductor element is turned on and off a plurality of times.

[0137] The observation noise calculation result RTNS is calculated for each of the noise analysis target periods set by the time range setting section 14. Figure 11 Thus, when the time range setting section 14 sets a plurality of noise analysis target periods, the observation noise calculation result RTNS can be calculated for each of the plurality of noise analysis target periods. In this case, a function of performing statistical calculation (calculating the average, maximum value, minimum value, etc.) of the calculated plurality of observation noise calculation results RTNS can also be provided, so that the result of the statistical calculation is output from the noise analysis function section 1 as the final observation noise calculation result RTNS. Figure 2

[0138] In addition, in the configuration example of Figure 11 , it is confirmed that the same observation noise calculation result RTNS is obtained even if the order of the addition operations performed by the addition sections 17 and 18 is exchanged. That is, a configuration in which the addition section 17 performs the operation of adding the on noise and the off noise to the result of the N-time addition operation by the addition section 18 can also be adopted.

[0139] Figure 12 A configuration example of the observation noise calculation section 11 for calculating the observation noise from a plurality of noise sources (semiconductor elements) is shown. For example, in the analysis target circuit 200 exemplified in Figure 3 , when the observation noise is calculated with the semiconductor elements 101 and 102 as independent noise sources, the observation noise calculation section shown in Figure 12 can be applied.

[0140] As shown in Figure 12 , the observation noise calculation section 11 has a total calculation section 10X (the same configuration as Figure 11 ) and an addition section 19 provided corresponding to each noise source. From each total calculation section 10X, the observation noise calculation result is calculated for each noise source as the output of the addition section 18.

[0141] ​Adder 19 adds the observed noise calculation results from summing calculation unit 10X, provided for each noise source, and outputs the sum of the observed noise from multiple noise sources as observed noise calculation result RTNS. This allows for the calculation of observed noise results RTNS for any number of noise sources. In other words, adder 19 corresponds to one embodiment of the "second adder."

[0142] As described above, the noise analysis technique of Embodiment 1 enables noise analysis by minimizing the time width of the Fourier transform target to a minimum number of pulses (typically, one pulse) and using the on- and off-state behavior of the semiconductor device during multiple pulses, including information on changes in the on- and off-period lengths, as noise sources. This eliminates the need to directly Fourier transform multiple pulses, which would increase the time required for the Fourier transform, while reflecting the behavior of changes in the on- and off-period lengths of the semiconductor device. This allows for high-speed and accurate noise analysis.

[0143] The basic configuration of noise analysis for one or more noise sources has been described above. Modifications and detailed specific examples of the noise analysis technique of the first embodiment will be described as appropriate below.

[0144] Figure 13 Shown with Figure 10 The observation noise calculation unit 10 ( Figure 2 ) is a modification of the structure.

[0145] Reference Figure 13 The noise transfer function acquisition unit 4 may also acquire a noise transfer function Gcmn that is set commonly for both the ON and OFF states. In this case, the observed noise calculation unit 10 may be configured to include a total calculation unit 10Y and a multiplication unit 12Y.

[0146] The total calculation unit 10Y has Figure 11 The structure of the total calculation unit 10X shown in FIG. 1 is obtained by deleting the multiplication units 12 and 13. In addition, the multiplication unit 12Y is configured to Figure 11 The output value of the adding unit 18 is multiplied by the noise transfer function Gcmn. Figure 11 In the calculation, the observation noise result RTNS of Gon=Goff=Gcmn can be obtained.

[0147] In addition, although Figure 11 In the structure of , Gon = Goff = Gcmn is input to the multiplication units 12 and 13, and the same observation noise calculation result RTNS as above can be obtained. However, by adopting Figure 13 The structure can reduce one multiplication operation, thus shortening the noise analysis time.

[0148] Figure 14 Show Figure 2 A modified example of the noise analysis function unit 1 is shown.

[0149] exist Figure 2 In the structure, the Fourier transform units 23 and 24 are provided in the noise analysis function unit 1 to obtain the on-state noise source spectrum SPNon and the off-state noise source spectrum SPNoff of the on-state transient waveform and the off-state transient waveform.

[0150] In contrast, in Figure 14 In the modified example, the noise analysis function unit 1 receives input of the on-noise source spectrum SPNon and the off-noise source spectrum SPNoff obtained by Fourier transforming the on-transient waveform and the off-transient waveform in advance.

[0151] In this case, the on-noise source spectrum acquisition unit 25 and the off-noise source spectrum acquisition unit 26 acquire the on-noise source spectrum SPNon and the off-noise source spectrum SPNoff input to the noise analysis function unit 1. Figure 14 , the on-noise source spectrum acquisition unit 25 and the off-noise source spectrum acquisition unit 26 correspond to an embodiment of the “third acquisition unit”.

[0152] By inputting the acquired on-noise source spectrum SPNon and off-noise source spectrum SPNoff into the observation noise calculation unit 10, it is possible to use Figure 10 or Figure 13 The structure calculates the observation noise calculation result RTNS.

[0153] Alternatively, you can also Figure 2 and Figure 14 In this case, the following structure can be used to select the noise source data that has been automatically determined: Figure 2 The on-noise source spectrum SPNon and the off-noise source spectrum SPNoff are input to the observed noise calculation unit 10 via a path passing through the Fourier transform units 23 and 24 and a path bypassing the Fourier transform units 23 and 24. This allows both the transient waveform (time domain) and the spectrum (frequency domain) to be used as on-noise source data and off-noise source data for noise analysis.

[0154] In addition, Figure 4 and Figure 5 In the example, t1 and t2 are used as the reference for the transient waveform. However, sometimes the time when the transient waveform data is input may deviate from the reference time. In this case, you can also apply Figure 15The structure shown automatically detects the occurrence time of the switch-on or switch-off from the transient waveform and automatically performs the correction process of the time deviation.

[0155] Reference Figure 15 The connection transient waveform (eg Figure 4 ) is input to the time deviation detection function unit 27a. The time deviation detection function unit 27a takes the initial time of the transient waveform data ( Figure 4 The voltage value (or current value) at the origin time in the circuit is saved as the initial value. In addition, at each subsequent time, the time deviation detection function unit 27a compares the difference between the voltage value (or current value) at that time and the initial value with a predetermined threshold value, and detects that the connection has occurred when the absolute value of the difference is greater than the threshold value. The threshold value can be, for example, Figure 4 It is determined by detecting the connection near t1.

[0156] The time deviation detection function unit 27a detects the time difference between zero time (the time corresponding to the origin) and the time when the switch-on is detected as the time deviation τ. The time deviation τ is assumed to be a positive value (τ > 0) when the switch-on occurs later than the reference zero time, and a negative value (τ < 0) when the switch-on occurs earlier than the zero time, thereby providing a suitable response.

[0157] The timing deviation correction function unit 28 a inputs the switching-on transient waveform obtained by correcting the time axis data of the switching-on transient waveform acquired by the switching-on transient waveform acquisition unit 21 based on the timing deviation amount τ detected by the timing deviation detection function unit 27 a to the Fourier transform unit 23 .

[0158] Similarly, for the disconnection transient waveform (eg Figure 5 ), a time deviation detection function unit 27b and a time deviation correction function unit 28b are also provided, similar to the time deviation detection function unit 27a and the time deviation correction function unit 28a. Specifically, the time deviation detection function unit 27b detects the time difference between zero time (the time corresponding to the origin) and the time when disconnection is detected as the time deviation τ. Furthermore, although the time deviation τ for connection (time deviation detection function unit 27a) and the time deviation τ for disconnection (time deviation detection function unit 27b) are denoted by the same reference numerals for simplicity of explanation, they can actually be different values. The time deviation τ for disconnection can also be set to a positive value (τ>0) if disconnection occurs later than the reference zero time, and to a negative value (τ<0) if disconnection occurs earlier than zero time.

[0159] By adopting this structure, even if a voltage waveform or a current waveform that makes the on-time and the off-time zero time is not input, the amount of time deviation between the on-time or the off-time can be automatically corrected to obtain the transient waveform data for deriving the on-noise source spectrum SPNon and the off-noise source spectrum SPNoff.

[0160] Alternatively, as shown in Figure 16 , the time deviation correction function sections 28a, 28b can be arranged at the stage subsequent to the Fourier transform sections 23, 24. In this case, the time deviation correction function sections 28a, 28b are configured to multiply the output of the Fourier transform sections 23, 24 by exp(-j-2pft(-t)) = exp(j-2pft) in accordance with the amount of time deviation t detected by the time deviation detection function sections 27a, 27b. The structure example using Figure 16 can also obtain the on-noise source spectrum SPNon and the off-noise source spectrum SPNoff as in Figure 15 .

[0161] Although, for example, in Figure 11 , the phase transformation is performed by multiplying by exp(-j-2pft) calculated in correspondence with the on-time ton(i) and the off-time toff(i) respectively in the frequency domain, a modification example in which the phase transformation is performed in the time domain is also possible according to the structure shown in Figure 34 .

[0162] Figure 34 A block diagram for explaining the modification example in which the phase transformation is performed in the time domain.

[0163] Referring to Figure 34 , the on-transient waveform acquired by the on-transient waveform acquisition section 21 is input to the phase transformation section 14c. The phase transformation section 14c accepts the on-time ton(i) extracted by the time range setting section 14, and outputs the on-transient waveform shifted on the time axis in accordance with the time difference corresponding to the on-time ton(i) respectively. The on-transient waveform (waveform data) in which the phase difference is reflected in the time domain like this is input to the Fourier transform section 23. The Fourier transform section 23 performs the Fourier transform on the on-transient waveform (waveform data) of each on-time ton(i) output from the phase transformation section 14c.

[0164] Likewise, the opening transient waveform acquired by the opening transient waveform acquisition section 22 is input to the phase conversion section 14d. The phase conversion section 14d receives the opening time toff(i) extracted by the time range setting section 14, and outputs the opening transient waveform shifted on the time axis according to the time difference corresponding to the opening time toff(i). The Fourier transform section 24 performs Fourier transform on the opening transient waveform (waveform data) of each opening time toff(i) from the phase conversion section 14d, which reflects the phase difference in the time domain.

[0165] Figure 11 The illustrated total calculation section 10X can be deformed as follows: instead of the phase conversion sections 14a, 14b and the multiplication sections 15, 16, a structure of multiplying the outputs of the Fourier transform sections 23, 24 by the noise transfer functions Gon, Goff using the multiplication sections 12, 13 is applied to calculate the opening observation noise NSoff(i) and the closing observation noise NSon(i). Figure 34

[0166] Even if the deformed example of Figure 34 performs phase conversion based on the phase difference information in the time domain, the entire pulse including the opening and the closing is not taken as the object of Fourier transform, so a certain effect can be obtained for reducing the number of time steps taken as the object of Fourier transform. However, Fourier transform is performed after phase conversion, and the amount of Fourier transform corresponding to the number of occurrences of the opening and the closing is required, so if phase conversion is performed using multiplication in the frequency domain, the effect of reducing the operation burden of Fourier transform is better.

[0167] Next, a case where the noise transfer function is an S parameter is illustrated, and a wiring example of the noise transfer function is explained using Figures 17-21 As described above, the noise transfer function means the transfer function of the noise in the frequency domain on the propagation path from the noise source to the noise observation point.

[0168] Figure 17 A wiring diagram is shown as a first example when the number of noise sources and the number of noise observation points are each two.

[0169] As illustrated in Figure 17 , the noise transfer function 81 has four ports Prtl to Prt4, and the ports Prtl to Prt4 are connected to the first noise source 71, the second noise source 72, the first noise observation point 73, and the second noise observation point 74, respectively. The first noise source 71 and the second noise source 72 correspond to the upper branch semiconductor element 101 and the lower branch semiconductor element 102 in the analysis target circuit 200 illustrated in Figure 3 , for example. Furthermore, the number of ports is not limited to four, but can be any number.

[0170] Figure 18 ​As a second example, a wiring diagram in which the noise source and the noise observation point are each a single point is shown.

[0171] In Figure 18 , the noise transfer function 81 has two ports Prtl, Prt2, the port Prtl is connected to the first noise source 71, and the port Prt2 is connected to the first noise observation point 73.

[0172] The noise transfer function (Gon, Goff, Gcmn) used for the noise analysis can be acquired by previously deriving the transfer function between the noise source and the noise observation point that should be the object of the noise analysis, and reading the data of the transfer function by the noise transfer function acquisition section 4.

[0173] Figure 19 As a third example, a wiring diagram in which the transfer function is divided into two via a relay point is shown.

[0174] As shown in Figure 19 , the transfer function between the first noise source 71 and the first noise observation point 73 is divided into a first noise transfer function 81a between the first noise source 71 and the relay point 82, and a second noise transfer function 81b between the first noise observation point 73 and the relay point 82. That is, in Figure 19 , the first noise source 71 and the first noise observation point 73 are divided into two propagation paths via the relay point 82, and the first noise transfer function 81a and the second noise transfer function 81b each correspond to the transfer function of the two propagation paths.

[0175] The first noise transfer function 81a and the second noise transfer function 81b each have two ports Prtl, Prt2. The ports Prtl and Prt2 of the first noise transfer function 81a are connected to the first noise source 71 and the relay point 82, respectively. The ports Prtl and Prt2 of the second noise transfer function 81b are connected to the relay point 82 and the first noise observation point 73, respectively.

[0176] Figure 20 As a fourth example, a wiring diagram in which the transfer function is divided into two via a plurality of relay points is shown.

[0177] As shown in Figure 20 , the transfer function between the first noise source 71 and the first noise observation point 73 is divided into a first noise transfer function 81a between the first noise source 71 and the relay points 83, 84, and a second noise transfer function 81b between the first noise observation point 73 and the relay points 82, 83. In Figure 20 , also, between the first noise source 71 and the first noise observation point 73, via the relay points 83, 84, the two propagation paths are divided, and the first noise transfer function 81a and the second noise transfer function 81b each correspond to the transfer function of the two propagation paths.

[0178] The first noise transfer function 81a and the second noise transfer function 81b each have three ports Prtl to Prt3. In the first noise transfer function 81a, the port Prtl is connected to the first noise source 71, the port Prt2 is connected to the relay point 83, and the port Prt3 is connected to the relay point 84. In addition, in the second noise transfer function 81b, the port Prtl is connected to the relay point 83, the port Prt2 is connected to the first noise observation point 73, and the port Prt3 is connected to the relay point 84.

[0179] As shown in FIG. 8, in a case where the noise transfer function between the noise source to be analyzed and the noise observation point is divided into a plurality of cases, the noise transfer function (Gon, Goff, Gcmn) used in the noise analysis function section 1 can be acquired by applying the configuration of FIG. 7. Figure 19 and Figure 20 As shown in FIG. 8, in a case where the noise transfer function between the noise source to be analyzed and the noise observation point is divided into a plurality of cases, the noise transfer function (Gon, Goff, Gcmn) used in the noise analysis function section 1 can be acquired by applying the configuration of FIG. 7. Figure 21

[0180] Referring to FIG. 8, the transfer function integration section 4X is input with data of the first noise transfer function 81a and the second noise transfer function 82a in Figure 21 and Figure 19 and outputs the noise transfer functions Gon, Goff using the transfer function data obtained by integrating the two. The transfer function integration section 4X is arranged at a later stage of the noise transfer function acquisition section 4, whereby the noise transfer function used in the noise analysis function section 1 can be acquired by integrating the divided transfer functions. Figure 20

[0181] Figure 21 An example of applying the transfer function integration section 4X in a configuration in which the noise transfer functions for the on noise source and the off noise source are independent Figure 10 is shown. Similarly, in a configuration in which the noise transfer functions for the on noise source and the off noise source are common Figure 13 , the transfer function integration section 4X can also be applied to acquire the noise transfer function Gcmn.

[0182] By dividing the noise transfer function into a plurality of cases, the noise transfer function can be derived independently for each interval. Thus, even if the design of a partial interval is changed, only the noise transfer function corresponding to the changed interval needs to be derived again, and the entire noise transfer function does not need to be derived again. As a result, the burden of preparing data of the noise transfer function can be reduced.

[0183] Further, the number of divisions of the noise transfer function is not limited to two, and can be divided into any number of three or more. In this case, the noise transfer function can also be divided into a plurality of cases as in Figure 20 ​​Similarly, by appropriately integrating the transfer functions of the plurality of propagation paths defined between the noise source to be analyzed and the noise observation point, the noise transfer functions (Gon, Goff, Gcmn) used in the noise analysis function unit 1 can be obtained.

[0184] Furthermore, the noise analysis program's wiring configuration interface allows for the definition of the connection relationships between each port of the noise transfer function and noise sources, noise observation points, or relay points using a wiring diagram or correspondence table. The noise transfer function acquisition unit 4 inputs the propagation path between the noise source to be analyzed and the noise observation point into this interface, and then configures the transfer function integration unit 4X as needed to acquire the noise transfer function (Gon, Goff, Gcmn) used for noise analysis.

[0185] Next, other examples of noise analysis target circuits and examples of observed noise calculation results obtained by the noise analysis device of the first embodiment will be further described.

[0186] Figure 22 To illustrate and Figure 3 A circuit diagram showing another example of a different noise analysis target.

[0187] exist Figure 22 In the example, the analysis target circuit 200 has a structure similar to Figure 3 The upper arm semiconductor device 101 and the lower arm semiconductor device 102 of the same arm are connected in series between the output positive terminal 301 and the output negative terminal 302 .

[0188] The intermediate terminal 205 corresponding to the connection point between the upper arm semiconductor element 101 and the lower arm semiconductor element 102 is connected to the input positive terminal 201 via the reactor 304. Figure 3 Similarly, the input negative terminal 202 is connected to the low potential side terminal of the lower arm semiconductor element 102, and the capacitor 204 is connected between the input positive terminal 201 and the input negative terminal 202. The output negative terminal 302 and the input negative terminal 202 are both connected to the low potential side terminal of the lower arm semiconductor element 102.

[0189] The load 303 is connected between the output positive terminal 301 and the output negative terminal 302. Figure 22 In the structure of , the lower branch semiconductor element 102 mainly works as a transistor, and the upper branch semiconductor element 101 mainly works as a diode. Figure 3As also described in

[15] , the analysis target circuit 200 can include a plurality of branches of a set of upper-arm semiconductor elements and lower-arm semiconductor elements connected in series. The number of series-connected semiconductor elements is not limited to two but can be any number. Alternatively, the analysis target circuit 200 can be a DC-DC converter or an ACDC converter.

[0190] Figure 23 Another example of a noise analysis target is shown.

[0191] exist Figure 23 In the example of , the analysis target circuit 200 includes a first branch 401, a second branch 402, and a third branch 403 connected in parallel between the input positive terminal 201 and the input negative terminal 202. The first branch 401, the second branch 402, and the third branch 403 each include an upper branch semiconductor element and a lower branch semiconductor element connected in series between the input positive terminal 201 and the input negative terminal 202. In each of the first branch 401, the second branch 402, and the third branch 403, intermediate terminals 411 to 413 corresponding to the connection points of the upper branch semiconductor element and the lower branch semiconductor element are connected to the AC load 405. That is, Figure 23 The analysis target circuit 200 operates as a three-phase inverter.

[0192] As described above, the analysis target circuit 200 can be an inverter that performs DC-AC conversion. The number of phases of the inverter is not limited to three, but can be any number. In addition, the number of semiconductor elements connected in series in each arm is not limited to two, but can be any number.

[0193] Figure 24 Shows the Figure 23 The example shown is an example of the observed noise calculation results for the switching control of the analysis target circuit (three-phase inverter). Here, the noise analysis is performed by switching the semiconductor devices (noise sources) of each phase branch of the three-phase inverter using PWM control with an AC waveform as the modulation signal. In addition, the switching frequency of each semiconductor device under PWM control is set to 10 [kHz]. In the PWM control of the inverter, it is known that Figure 7 Similarly to the switching control signal shown, the length of the on-period and the length of the off-period of the switching control signal of each semiconductor element vary.

[0194] Figure 24 The noise terminal voltage [dBV], which indicates the noise intensity at each frequency, obtained as the observed noise calculation result RTNS, is plotted. Figure 25 Show Figure 24 A magnified image of the 0.1[MHz]~0.5[MHz] region.

[0195] Under switching control with a fixed switching frequency of 10 [kHz], when the on-period length is constant, peaks of the noise spectrum occur at integer multiples of 10 [kHz], so that concavities and convexities occur at 10 [kHz] intervals in the spectrum.

[0196] However, according to the three-phase inverter ( Figure 23 ) is the noise analysis of the embodiment 1 of the object, such as Figure 25 As shown, the above-mentioned unevenness at intervals of 10 [kHz] is not generated.

[0197] That is, it is generally understood that according to the above-mentioned embodiment 1, it is possible to Figure 7 The plurality of pulses included in the gate signal shown are directly subjected to Fourier transform, thereby realizing noise analysis of switching control associated with changes in the on-period length and the off-period length.

[0198] Implementation method 2.

[0199] In the first embodiment, the on-transient waveform and the off-transient waveform are input separately to the noise analysis function unit 1 . In the second embodiment, a configuration in which the on-transient waveform and the off-transient waveform can be separated within the noise analysis function unit 1 will be described.

[0200] Figure 26 This is a block diagram illustrating a configuration example of a noise analysis device according to a second embodiment.

[0201] Reference Figure 26 The noise analysis function unit 1 of the noise analysis device of embodiment 2 and Figure 2 Compared with the configuration of the first embodiment shown, the configuration is different in that a transient waveform acquisition unit 20 is provided in place of the ON transient waveform acquisition unit 21 and the OFF transient waveform acquisition unit 22 , and an ON / OFF separation unit 29 is further provided.

[0202] The transient waveform acquisition unit 20 acquires a transient waveform including both an ON transient waveform and an OFF transient waveform. That is, in the second embodiment, there is no need to extract the ON transient waveform and the OFF transient waveform from the transient waveform input to the noise analysis function unit 1 (noise analysis device).

[0203] Figure 27 Show Figure 26 The configuration example of the on / off separation unit 29 is shown.

[0204] Reference Figure 27 The on / off separation unit 29 includes an on-time detection unit 31 , an off-time detection unit 32 , an on-transient waveform output unit 33 , and an off-transient waveform output unit 34 .

[0205] The on-time detecting section 31 detects the on-time of the semiconductor element in correspondence with the time at which the voltage value included in the transient waveform obtained by the transient waveform obtaining section 20 (for example, Vtr in Figure 4 and Figure 5 decreases across a predetermined threshold value. The on-transient waveform output section 33 extracts the transient waveform of a certain period including the on-time detected by the on-time detecting section 31 from the transient waveform obtained by the transient waveform obtaining section 20 and outputs the on-transient waveform.

[0206] The off-time detecting section 32 detects the off-time of the semiconductor element in correspondence with the time at which the voltage value included in the transient waveform obtained by the transient waveform obtaining section 20 (for example, Vtr in Figure 4 and Figure 5 increases across a predetermined threshold value. The off-transient waveform output section 34 extracts the transient waveform of a certain period including the off-time detected by the off-time detecting section 32 from the transient waveform obtained by the transient waveform obtaining section 20 and outputs the off-transient waveform.

[0207] The on-transient waveform and the off-transient waveform output from the on-transient waveform output section 33 and the off-transient waveform output section 34 are the same as the on-transient waveform and the off-transient waveform obtained by the on-transient waveform obtaining section 21 and the off-transient waveform obtaining section 22 in Figure 2 . The output on-transient waveform and off-transient waveform are input to the Fourier transform sections 23 and 24 of Figure 26 , respectively. Alternatively, the on-transient waveform and the off-transient waveform output from the on-transient waveform output section 33 and the off-transient waveform output section 34 can be input to the phase transform sections 14c and 14d of Figure 34 .

[0208] Further, in the case where the transient waveform includes a current value that changes with the passage of time (for example, Itr in Figure 6 ), the on-time detecting section 31 can be configured to detect the on-time in response to the rise of the current value across a threshold value, and the off-time detecting section 32 can be configured to detect the off-time in response to the decrease of the current value across a threshold value.

[0209] Referring again to Figure 26 , the same as Embodiment 1 except for the structure for obtaining the on-transient waveform and the off-transient waveform, so a detailed description is not repeated.

[0210] As such, according to the noise analysis technique of Embodiment 2, for the transient waveform input to the noise analysis function section 1 (noise analysis device), it is possible to not need the process of separately preparing the on-transient waveform and the off-transient waveform.

[0211] Embodiment 3.

[0212] Generally, it is known that the higher the resolution bandwidth of a noise meter that actually measures noise is compared to the on-off frequency of a semiconductor element that is a noise source, the larger the measured noise appears. In Embodiment 3, a noise analysis technique that takes into account the resolution bandwidth assumed at a noise observation point is described.

[0213] Figure 28 A block diagram for explaining the configuration example of the noise analysis apparatus of Embodiment 3.

[0214] Figure 28 Compared with Embodiment 1, Figure 2 the noise analysis function section 1 of the noise analysis apparatus of Embodiment 3 further has a meter parameter acquisition section 5.

[0215] Further, in the observation noise calculation section 10, in order to perform noise analysis that reflects information on the resolution bandwidth assumed at the noise observation point acquired by the meter parameter acquisition section 5, it has a Figure 29 structure as shown in FIG. 10. Typically, the information includes the resolution bandwidth of the noise meter used at the noise observation point. As the noise meter, a spectrum analyzer or an EMI (Electro Magnetic Interference) receiver or the like can be applied. The structure and operation of the other parts of Figure 28 Embodiment 3 are the same as those of Embodiment 1, and thus detailed description thereof is not repeated. Figure 2

[0216] Figure 29 A block diagram for explaining the function added to the observation noise calculation section 10 in the noise analysis of Embodiment 3.

[0217] Referring to Figure 29 Embodiment 3, the observation noise calculation section 10 further includes a weighting operation function section 40. The weighting operation function section 40 has a window function calculation section 41, a weighting coefficient setting section 42, and multiplication sections 43, 44.

[0218] The window function calculation section 41 sets a window function w(t) based on information on the start time Tstr~the end time Tend of the noise analysis target period from the time range setting section 14 Figure 11 The window function w(t) is set to w(t) = 0 outside the noise analysis target period. The window function w(t) is set to a shape in which the value at the edge portion of the noise analysis target period is small and the value at the center portion is large.

[0219] ​Furthermore, in Embodiment 3, the shape of the window function w(t) is set to match the frequency resolution during noise measurement by a noise measurement device or the like, based on the resolution bandwidth acquired by the measurement device parameter acquisition unit 5. For example, when Fourier transforming the window function w(t) based on a predetermined integrated selectivity characteristic, the shape of the window function w(t) can be determined so that the DC component is 0 dB and the frequency component at half the resolution bandwidth is -6 dB or -3 dB.

[0220] The window function w(t) set by the window function calculation unit 41 and the time range setting unit 14 ( Figure 11 ) is input to the weighting coefficient setting unit 42 .

[0221] The weighting coefficient setting unit 42 outputs the value of the window function w(t) at each on-time ton(i) as the weighting coefficient WGon(i) for each on-time. Similarly, the weighting coefficient setting unit 42 outputs the value of the window function w(t) at each off-time toff(i) as the weighting coefficient WGoff(i) for each off-time.

[0222] The multiplication unit 43 multiplies the ON noise source data at the ON time ton(i) by the corresponding weighting coefficient WGon(i). Similarly, the multiplication unit 44 multiplies the OFF noise source data at the OFF time toff(i) by the corresponding weighting coefficient WGoff(i).

[0223] exist Figure 29 In the data, the "ON" noise source data includes data after the information for distinguishing the ON time ton(i) (the output value of the phase conversion unit 14a) is reflected in the ON noise source spectrum SPNon within the observed noise calculation unit 10. Similarly, the "OFF" noise source data includes data after the information for distinguishing the OFF time toff(i) (the output value of the phase conversion unit 14b) is reflected in the OFF noise source spectrum SPNoff within the observed noise calculation unit 10.

[0224] exist Figure 29 In FIG. 5 , the measuring device parameter acquiring unit 5 corresponds to an embodiment of the “fourth acquiring unit.” In addition, the multiplying units 43 and 44 implement the function of the “second multiplying unit.”

[0225] Furthermore, by multiplying the noise source data in the time domain by the weighting coefficients WGon(i) and WGoff(i), the weighting coefficients WGon(i) and WGoff(i) can be reflected in the on-state observation noise NSon(i) and the off-state observation noise NSoff(i). Figure 34In the configuration, the phase conversion sections 14c, 14d can input the weighting coefficients WGon(i), Goff(i) to the on-time ton(i) and the off-time toff(i), respectively. Then, the phase conversion sections 14c, 14d can output, to the Fourier transform sections 23 and 24, the multiplication results of the on-transient waveform and the off-transient waveform to which the phase difference is imparted in the time domain and the weighting coefficients WGon(i) and Goff(i), respectively.

[0226] In addition, with respect to the weighting coefficients WGon(i), Goff(i) reflected in the frequency domain, for Figure 29 the example, even if the multiplication sections 43, 44 (second multiplication sections) and the multiplication sections 12, 13, 15, 16 (first multiplication sections) shown in FIG. 1 are exchanged in order, the weighting coefficients WGon(i), WGoff(i) can be reflected in the multiplication values calculated by the multiplication sections 12, 13, 15, 16. That is, as is obvious from the description in Figure 11 Figure 11

[0227] As a result, the addition section 17 can calculate, for the observation noise NS(i) of the i-th pulse, the values obtained by multiplying the weighting coefficients WGon(i), WGoff(i) corresponding to the shape of the window function w(t). Thus, the finally calculated observation noise calculation result RTNS can also be calculated with the weight considering the positions of the respective on-times and the respective off-times within the noise analysis target period. In particular, according to Embodiment 3, by reflecting the resolution bandwidth of the noise meter in the setting of the weighting coefficients, the analysis noise calculation result considering the resolution bandwidth can be obtained.

[0228] In addition, it is confirmed that in the configuration example of Figure 29 even if the order of the addition performed by each of the addition sections 17 and 18 is exchanged, the same observation noise calculation result RTNS is obtained. That is, a configuration in which the addition section 17 performs the operation of adding the N-time addition operation result of the addition section 18 to the on-noise and the off-noise can also be adopted.

[0229] Figure 30 A spectrum diagram showing an example of the observation noise calculation result obtained by the noise analysis according to Embodiment 3. Figure 30 The observation noise calculation result RTNS obtained by setting the resolution bandwidth to 9 [kHz] or less than 30 [MHz] and setting the resolution bandwidth to 120 [kHz] or more than 30 [MHz] is plotted. In addition, the on-off frequency of the semiconductor element as the noise source is 10 [kHz] as well. Figure 24

[0230] As shown in FIG. 6, the observation noise calculation result RTNS obtained by the noise analysis according to Embodiment 3 is plotted. Figure 30 ​​​As shown, in the frequency domain where the resolution bandwidth is higher than 30 [MHz] or more than the on-off frequency, a calculation result where noise appears large is obtained. According to this result, it is generally understood that high-precision noise analysis that takes into account the influence of the resolution bandwidth of the noise measuring device can be performed.

[0231] Further, in Figure 29 , a configuration where the parameter (resolution bandwidth) that does not reflect the noise measurement is set in combination with the structure and the embodiment 1 or 2 of the window function w(t) can also be possible. In this case, the noise analysis can also be made high-precision by performing weighting that takes into account the position of each on-time and each off-time within the noise analysis target period.

[0232] Embodiment 4.

[0233] In the embodiments 1 to 3, the on-noise source spectrum SPNon of each on-time ton(i) is set to be common and the off-noise source spectrum SPNoff of each off-time toff(i) is also set to be common to perform the operation of the observation noise calculation result RTNS.

[0234] On the other hand, the noise source spectrum that occurs with the on-off of the semiconductor element varies depending on the current (load current) that is turned on and off. For example, when the load current becomes small, the current change of the semiconductor element at the time of on-off becomes small, and the timing at which the voltage change of the semiconductor element starts and the gradient of the voltage change change accordingly. Thus, in the embodiment 4, a noise analysis technique that further reflects the load current at the on-time and the off-time of the semiconductor element is described.

[0235] Figure 31 A block diagram for explaining the structure example of the noise analysis device of the embodiment 4.

[0236] Referring to Figure 31 , the noise analysis function section 1 of the noise analysis device of the embodiment 4 is different from the structure of the embodiment 1 shown in Figure 2 in that it further has a load current waveform acquisition section 6. The load current waveform acquisition section 6 acquires the load current waveform (time axis) in the time range including the noise analysis target period.

[0237] Further, in the embodiment 4, the on-transient waveform acquisition section 21 and the off-transient waveform acquisition section 22 acquire a plurality of on-transient waveforms and a plurality of off-transient waveforms in which the load current is different. In Figure 31 , J (J: a natural number of 2 or more) on-transient waveforms and J off-transient waveforms are acquired and input to the Fourier transform sections 23, 24, respectively. As a result, J on-noise source spectra and J off-noise source spectra in which the load current levels are different are input to the observation noise calculation section 10.

[0238] In Embodiment 4, a noise source spectrum correction section 60 as shown in FIG. 6 is added to the observation noise calculation section 10. Figure 32 The noise source spectrum correction section 60 has the functions of the noise source spectrum correction section 60 shown in FIG. 6.

[0239] Referring to FIG. 6, Figure 32 the noise source spectrum correction section 60 includes load current value acquisition sections 61, 62 and interpolation calculation function sections 65, 66.

[0240] The load current value acquisition section 61 is inputted with the load current waveform (time axis) acquired by the load current waveform acquisition section 6 and the ON time ton(i) from the time range setting section 14. Figure 11 The load current value acquisition section 61 outputs the current value of the load current waveform at each of the ON times ton(i) as the load current value X(i) at each ON time. The load current value X(i) is inputted to the interpolation calculation function section 65.

[0241] Further, the interpolation calculation function section 65 is inputted with the ON noise source spectrum of each of the different J load current values X1~XJ from the Fourier transform section 23.

[0242] The interpolation calculation function section 65 outputs the ON noise source spectrum SPNon corresponding to the load current value X(i) from the ON noise source spectrum of each of the load current values X1~XJ in accordance with the relationship between the load current values X1~XJ and the inputted load current value X(i).

[0243] For example, the ON noise source spectrum SPNon corresponding to the load current value X(i) can be obtained by linear interpolation based on interpolation or extrapolation using the ON noise source spectrum corresponding to the two load current values closest to the load current value X(i) among the load current values X1~XJ.

[0244] As a result, the interpolation calculation function section 65 can calculate the ON noise spectrum SPNon depending on the load current value X(i) at each ON time ton(i).

[0245] Similarly, the load current waveform (time axis) from the load current waveform acquisition section 6 and the OFF time toff(i) from the time range setting section 14 are inputted to the load current value acquisition section 62. The load current value acquisition section 62 outputs the current value of the load current waveform at each of the OFF times toff(i) as the load current value X(i) at each OFF time. The load current value X(i) is inputted to the interpolation calculation function section 66. Figure 11 Further, the interpolation calculation function section 66 is inputted with the OFF noise source spectrum of each of the different J load current values X1~XJ from the Fourier transform section 24.

[0246]

[0247] ​The interpolation calculation function section 66 outputs the turn-off noise source spectrum SPNoff corresponding to the load current value X(i) in accordance with the turn-off noise source spectrum of each of the load current values X1 to XJ based on the relationship between the load current values X1 to XJ and the input load current value X(i).

[0248] For example, the turn-off noise source spectrum SPNoff corresponding to the load current value X(i) can be obtained by linear interpolation based on interpolation or extrapolation using the turn-off noise source spectra corresponding to the 2 load current values closest to the load current value X(i) among the load current values X1 to XJ.

[0249] As a result, the interpolation calculation function section 66 can calculate the turn-off noise source spectrum SPNoff depending on the load current value X(i) at each turn-off timing toff(i). Furthermore, the interpolation calculation by the interpolation calculation function sections 65, 66 is not limited to the above-described linear interpolation. For example, the turn-on noise source spectrum SPNon and the turn-off noise source spectrum SPNoff for each load current value X(i) can be obtained by n-order spline interpolation (n: natural number of n > 2).

[0250] As such, in Embodiment 4, the turn-on noise source spectrum SPNon and the turn-off noise source spectrum SPNoff can be calculated in accordance with the load current at each timing for each turn-on timing ton(i) and each turn-off timing toff(i) by the noise source spectrum correction section 60.

[0251] According to the noise analysis technology of Embodiment 4, the following operation is performed: the observation noise calculation result RTNS explained in Embodiments 1 to 3 is derived using the turn-on noise source spectrum SPNon and the turn-off noise source spectrum SPNoff obtained by the noise source spectrum correction section 60. At this time, Figure 11 The turn-on noise source spectrum SPNon and the turn-off noise source spectrum SPNoff used in the calculation of the observation noise of the 1st to Nth (i: 1 to N) of the above-described 1st to Nth (i: 1 to N) can be different depending on the load current X(i).

[0252] As such, according to the noise analysis technology of Embodiment 4, the load current dependency of the noise intensity from the noise source (semiconductor element) can be taken into account, and thus the noise analysis can be made more highly accurate.

[0253] Furthermore, Embodiments 2 to 4 can be combined with Figure 13 , Figure 14 , Figure 21 and the like explained in Embodiment 1 as a modification example.

[0254] (Noise analysis device and noise analysis method for calculating observation noise using noise source as noise observation point)

[0255] In Embodiments 1 to 4, an example noise analysis device was described in which an observed noise calculation result RTNS, including multiplication by noise transfer functions (Gon, Goff, Gcmn), is calculated as the total spectrum of noise generated by the switching of semiconductor switching elements. However, it is generally understood that in order to reflect the behavior of changes in the on-period and off-period lengths of semiconductor elements in PWM waveforms, etc., Fourier transforming a transient waveform, each consisting of multiple switching cycles, increases the time span of the Fourier transform. Considering this technical issue, Embodiments 1 to 4 employ a configuration that calculates the observed noise calculation result RTNS as the total spectrum without multiplication by the noise transfer function, thus resolving the technical issue. Observed noise without multiplication by the noise transfer function corresponds to observed noise with the noise source as the noise observation point. In this case, the observed noise calculation result RTNS, similar to that in Embodiments 1 to 4, can be obtained by multiplying the calculated observed noise calculation result RTNS by the noise transfer functions (Gon, Goff, Gcmn).

[0256] Therefore, the noise analysis devices of the above-mentioned embodiments 1 to 4 can achieve the above-mentioned effects even when the noise source is used as the noise observation point to calculate the observed noise while excluding the multiplication element with the noise transfer function. Figure 35 、 Figures 39-42 The functions of the blocks included in the noise analysis function unit 1Y are realized by calculation processing performed by the information processing device 51 , thereby executing a noise analysis device or a noise analysis method for calculating observed noise using a noise source as a noise observation point.

[0257] Figure 35 This is a block diagram illustrating a first configuration example of a noise analysis device according to this embodiment that calculates observation noise using a noise source as a noise observation point.

[0258] Reference Figure 35 In the first structural example, the noise analysis function unit 1Y of this embodiment has Figure 2 The noise analysis function unit 1 shown in FIG. 1 is a structure in which the noise transfer function acquisition unit 4 is removed. Figure 36 As shown, Figure 35 The observation noise calculation unit 10 is configured to include a total calculation unit 10Z instead of Figure 10 and Figure 11 Total calculation unit 10X.

[0259] Figure 37 Show Figure 36 The structure of the total calculation unit 10Z is shown.

[0260] likeFigure 37 The total calculation section 10Z differs from the total calculation section 10X shown in FIG. 6 in that the configuration of the multiplication section 12, 13 and the addition section 17 for multiplying the noise transfer functions Gon, Goff from the noise transfer function acquisition section 4 is omitted. Figure 11

[0261] As a result, in the total calculation section 10Z, the ΣNSon(i) of the on noise source spectrum and the ΣNSoff(i) of the off noise source spectrum are calculated as the observation noise calculation result RTNS with the noise source as the noise observation point. In this case, the ΣNSon(i) and the ΣNSoff(i) correspond to the total spectrum obtained by adding the plurality of noise spectra obtained by the phase transform. In contrast, it is generally understood that, in Embodiments 1 to 4, the total spectrum obtained by adding the plurality of noise spectra after multiplying the noise transfer function and performing the phase transform is calculated as the observation noise calculation result RTNS. As shown by the arrow with a diagonal line in FIG. 7, at the input of the addition section 18, the on NSon(1) to NSon(N) and the off NSoff(1) to NSoff(N) are processed separately, and as a result, the ΣNSon(i) and the ΣNSoff(i) are separately output from the addition section 18. Figure 37

[0262] In the total calculation section 10Z, the ΣNSon(i) of the on noise source spectrum and the ΣNSoff(i) of the off noise source spectrum are calculated as the observation noise calculation result RTNS with the noise source as the noise observation point. In this case, the ΣNSon(i) and the ΣNSoff(i) correspond to the total spectrum obtained by adding the plurality of noise spectra obtained by the phase transform. In contrast, it is generally understood that, in Embodiments 1 to 4, the total spectrum obtained by adding the plurality of noise spectra after multiplying the noise transfer function and performing the phase transform is calculated as the observation noise calculation result RTNS. As shown by the arrow with a diagonal line in FIG. 7, at the input of the addition section 18, the on NSon(1) to NSon(N) and the off NSoff(1) to NSoff(N) are processed separately, and as a result, the ΣNSon(i) and the ΣNSoff(i) are separately output from the addition section 18. Figure 37 Figure 37 Figure 37

[0263] In addition, for the total calculation section 10Z, a modification example in which the phase transform is performed in the time domain to which the structure of Embodiment 1 is applied can also be adopted. Specifically, in the structure of Embodiment 1, the phase transform sections 14a, 14b and the multiplication sections 15, 16 are deleted, and the total calculation section 10Z is configured by the addition section 18. Figure 37 Figure 34 Figure 37 Figure 34 ​​​​​​​​The Nth on-state observation noise NSon(i) after Fourier transform by the Fourier transform section 23 and the Nth off-state observation noise NSoff(i) after Fourier transform by the Fourier transform section 24 are input to the addition section 18, so that the same ΣNSon(i) and ΣNSoff(i) can be calculated as the observation noise calculation result RTNS as the noise observation point at the noise source.

[0264] Likewise, regarding the use of the noise transfer function Gcmn which is commonly set at the on-state and the off-state Figure 13 of the structure, it is also possible to operate as a noise analysis device which calculates the observation noise as the noise observation point at the noise source by removing the multiplication with the noise transfer function as shown in Figure 38 .

[0265] Referring to Figure 38 , Figure 35 , the observation noise calculation section 10 can be configured to delete the multiplication section 12Y from Figure 13 the structure, and further have a total calculation section 10Y' instead of the total calculation section 10Y.

[0266] The total calculation section 10Y' can be configured in such a manner that, in the total calculation section 10Z of Figure 37 , the addition section 18 performs the addition operation of NSon(1) to NSon(N) from the 1st to the Nth and the addition operation of NSoff(1) to NSoff(N) from the 1st to the Nth, and further the addition section 17 equivalent to Figure 11 performs the addition operation of the on-state noise and the off-state noise. Thereby, it is possible to output ΣNS(i) equivalent to the sum of ΣNSon(i) and ΣNSoff(i) in Figure 37 as the observation noise calculation result RTNS as the noise observation point at the noise source. In this case, ΣNS(i) corresponds to the total spectrum obtained by adding the plurality of noise spectra after phase transformation.

[0267] At this time, it is also possible to calculate the same observation noise calculation result RTNS even if the order of addition by the addition sections 17 and 18 is exchanged. That is, even if the following structure is adopted, that is, after calculating "NSon(i) + NSoff(i)" for N times by the addition operation of the on-state noise and the off-state noise by the addition section 17, the sum of "NSon(1) + NSoff(1)" to "NSon(N) + NSoff(N)" is found by the addition section 18, it is possible to calculate the same ΣNS(i) as the observation noise calculation result RTNS.

[0268] In Figure 38 , it is also possible to adopt the application Figure 34a modification of the configuration of the structure that performs the phase transformation in the time domain. Specifically, the phase transformation sections 14a, 14b and the multiplication sections 15, 16 are deleted, and Figure 38 in the configuration of the structure of the noise analysis device, the input to the summation calculation section 10Y' is set to be the Nth on observation noise NSon(i) after the Fourier transformation by the Fourier transformation section 23 and the Nth off observation noise NSoff(i) after the Fourier transformation by the Fourier transformation section 24, and thus the same ΣNS(i) can be calculated as the observation noise calculation result RTNS with the noise source as the noise observation point. Figure 34

[0269] As described above, the observation noise calculation result RTNS (i.e., the summation spectrum not including the multiplication by the noise transfer function) obtained by the noise analysis function section 1Y of the present embodiment can obtain the same observation noise calculation result RTNS as in Embodiments 1 to 4, for example, by multiplying the noise transfer function (Gon, Goff, Gcmn) outside the noise analysis device. That is, the same noise analysis device as in Embodiments 1 to 4 can be configured by using Figures 35-42 The noise analysis device that calculates the observation noise with the noise source as the noise observation point and the noise transfer function acquisition section 4 and the multiplication element (multiplication section 12, 13, 12Y) in Embodiments 1 to 4 constitute the same noise analysis device as in Embodiments 1 to 4.

[0270] As such, the noise analysis device of the present embodiment can output the summation spectrum that reflects the behavior that changes due to the change in the length of the on period and the length of the off period as the observation noise calculation result RTNS even if it is configured in a manner that calculates the observation noise with the noise source as the noise observation point. At this time, the increase in the time required for the Fourier transformation due to the increase in the time width that is the object of the Fourier transformation can be suppressed, and thus the speed of the noise analysis can be increased.

[0271] In the second configuration example of the noise analysis device that calculates the observation noise with the noise source as the noise observation point shown in Figure 39 The noise analysis function section 1Y has a configuration in which the noise transfer function acquisition section 4 is removed from the noise analysis function section 1 shown in Figure 14 Thus, the summation value (summation spectrum) of the plurality of noise spectra after the phase transformation that does not include the multiplication by the noise transfer function (Gon, Goff, Gcmn) can be calculated from the observation noise calculation section 10 as the observation noise calculation result RTNS.

[0272] Similarly, as shown in Figures 40-42 The noise analysis device that calculates the observation noise with the noise source as the noise observation point can also be configured based on the configuration of the noise analysis device of Embodiments 2 to 4.

[0273] ​In the third configuration example of the noise analysis device (1) that calculates the observed noise using the noise source as the noise observation point Figure 40 , the noise analysis function section 1Y has a configuration in which the noise transfer function acquisition section 4 is removed from the noise analysis function section 1 shown in Figure 26 (Embodiment 2). Thus, the observed noise calculation result RTNS corresponding to the total spectrum obtained by adding the plurality of noise spectra after the phase transformation without the multiplication of the noise transfer functions (Gon, Goff, Gcmn) can be calculated.

[0274] In addition, in the fourth configuration example of the noise analysis device (1) that calculates the observed noise using the noise source as the noise observation point Figure 41 , the noise analysis function section 1Y has a configuration in which the noise transfer function acquisition section 4 is removed from the noise analysis function section 1 shown in Figure 28 (Embodiment 3). Thus, as in Figure 35 , Figure 39 and Figure 40 , the observed noise calculation result RTNS corresponding to the total spectrum obtained by adding the plurality of noise spectra after the phase transformation without the multiplication of the noise transfer functions (Gon, Goff, Gcmn) can be calculated.

[0275] Similarly, in the fifth configuration example of the noise analysis device (1) that calculates the observed noise using the noise source as the noise observation point Figure 42 , the noise analysis function section 1Y has a configuration in which the noise transfer function acquisition section 4 is removed from the noise analysis function section 1 shown in Figure 31 (Embodiment 4). Thus, as in Figure 35 and Figures 39-41 , the observed noise calculation result RTNS corresponding to the total spectrum obtained by adding the plurality of noise spectra after the phase transformation without the multiplication of the noise transfer functions (Gon, Goff, Gcmn) can be calculated. In each of the drawings of Figures 39-42 , the observed noise calculation section 10 can be configured as in the configuration described in Figures 36-38 .

[0276] As such, each of the embodiments 1 to 4 that have been described also discloses the use of Figures 35-42A noise analysis device that calculates observation noise with a noise source as a noise observation point. In the case of a noise source as a noise observation point, the noise analysis device can also suppress an increase in the time required for Fourier transform caused by an increase in the time width as an object of Fourier transform, and can output a total noise spectrum that reflects behavior that changes depending on changes in the length of the on period and the length of the off period, as a calculation result RTNS of observation noise. Thus, a noise analysis technique can be provided that can prevent the time for noise analysis from becoming long and can accurately derive a final calculation result of observation noise at high speed.

[0277] Furthermore, it is confirmed that, for the above-described embodiments, including combinations not mentioned in the specification, it is intended from the outset of the application that the structures described in each embodiment are combined as appropriate, within a range that does not produce mismatches and contradictions.

[0278] In addition, in the above embodiments, a structure example in which noise from both the on noise source and the off noise source is analyzed, that is, a noise analysis technique that calculates observation noise related to noise generated by "opening and closing" that includes both the on and off of the semiconductor element, is described. That is, in the present embodiment, an example in which the noise analysis of the present disclosure is performed with both the on time ton(i) and the off time toff(i) as "a plurality of occurrence times" is described.

[0279] On the other hand, a case in which only either the on-induced noise or the off-induced noise of the semiconductor element is mainly observed is also assumed. In this case, noise analysis can be performed without considering the noise source of the non-main side.

[0280] For example, in a case in which noise at the time of on is dominant, the observation noise calculation result RTNS can be calculated without considering the off noise source, that is, by deleting the NSoff term. In this case, a structure for acquiring the off transient waveform or the off noise source spectrum can be deleted from the noise analysis function section 1. As such, the noise analysis technique (noise analysis technique and noise analysis method) of the present embodiment can be applied to the calculation of observation noise related to noise generated by "opening and closing" that is at least one of the on and off of the semiconductor element. In this case, the noise analysis of the present disclosure is performed with only one of the on time ton(i) and the off time toff(i) as "a plurality of occurrence times".

[0281] In addition, although a structure example in which the noise analysis device is realized by the information processing device 51 executing a program (noise analysis program) for executing the noise analysis technique of the present embodiment is described in Figure 1 , the noise analysis program can also be executed by the information processing device 51 in Figure 1The execution of the noise analysis program operating as the noise analysis device can be shared among a plurality of devices. That is, the noise analysis program can be executed by some or all of the information processing devices 51 and 53 and the cloud 54, thereby constituting the noise analysis device.

[0282] Further, the storage location of the data of the on-transient waveform and the off-transient waveform (time domain or frequency domain), the data of the on-off control signal, and the data of the noise transfer function can be any of the information processing devices 51, 53, and the cloud 54. Similarly, the storage location of the observation noise calculation result RTNS is not limited to the information processing device 51, but can be the information processing device 53 and / or the cloud 54.

[0283] Similarly, the display location of the observation noise calculation result RTNS is not limited to the information display device 52, but can be any of the information display device of the information processing device 53 and the virtual information display environment of the cloud 54. For example, the content of the virtual information display environment can be displayed on the information display device 52 by accessing the cloud 54 from the information processing device 51.

[0284] As such, by sharing the processing of the noise analysis technology of the present embodiment among a plurality of execution subjects, even if the data storage performance or the processing performance of the information processing device 51 is not high, the noise analysis can be easily performed. In particular, by utilizing the cloud 54, the noise analysis technology (noise analysis program) that can accurately derive the noise even when the on-period length and the off-period length of the semiconductor element change can be easily provided as a noise analysis service to a large number of users via a network.

[0285] It should be understood that the embodiments disclosed herein are illustrative only and not restrictive of the scope of the application. The scope of the application is not represented by the above description but by the claims, and is intended to include all modifications within the meaning and range of equivalency of the claims.

Claims

1. A noise analysis apparatus that calculates a total spectrum of noise generated by switching of a semiconductor element at least one of turn-on and turn-off, the noise analysis apparatus comprising: a first acquisition section that acquires information indicating a plurality of occurrence times at which a plurality of switching of the semiconductor element occurs respectively during a noise analysis target period including the plurality of switching of the semiconductor element; a phase conversion section that generates a plurality of phase difference information corresponding to the plurality of occurrence times acquired by the first acquisition section respectively, the plurality of phase difference information being used to perform phase conversion of a noise source spectrum of the semiconductor element at the switching reflecting a time difference of the plurality of switching; and a first addition section that calculates the total spectrum obtained by adding a plurality of noise spectra obtained by performing phase conversion of the noise source spectrum of the semiconductor element at the switching using the plurality of phase difference information respectively.

2. The noise analysis apparatus according to claim 1, wherein the noise source spectrum of the semiconductor element at the switching is a Fourier transform of transient waveform data of the semiconductor element at the switching.

3. The noise analysis apparatus according to claim 1, further comprising: a second acquisition section that acquires a noise transfer function of a propagation path between the semiconductor element and a noise observation point; and a first multiplication section that calculates a multiplication value of the noise source spectrum of the semiconductor element at the switching obtained by Fourier transforming the transient waveform data of the semiconductor element at the switching and the noise transfer function, the noise source spectrum of the semiconductor element at the switching being the multiplication value.

4. The noise analysis apparatus according to claim 3, wherein the second acquisition section acquires a plurality of transfer functions for each of a plurality of propagation paths divided between the semiconductor element and the noise observation point, and the noise analysis apparatus further comprises a transfer function integration section that integrates the plurality of transfer functions acquired by the second acquisition section to output the noise transfer function.

5. The noise analysis apparatus according to any one of claims 2 to 4, further comprising: a third acquisition section that acquires the noise source spectrum that has been Fourier transformed.

6. The noise analysis apparatus according to any one of claims 2 to 4, further comprising: a third acquisition section that acquires the transient waveform data; and a Fourier transform section that generates the noise source spectrum by Fourier transforming the transient waveform data acquired by the third acquisition section.

7. The noise analysis apparatus according to claim 6, wherein the plurality of occurrence times are individually set corresponding to the turn-on and the turn-off respectively, the third acquisition section acquires the transient waveform data including transient waveforms of 1 or more of the turn-on and 1 or more of the turn-off of the semiconductor element, and the noise analysis apparatus further comprises a data separation section that individually extracts transient waveform data at the turn-on and transient waveform data at the turn-off from the transient waveform data acquired by the third acquisition section. ​ ​ ​ ​ ​ ​ ​ ​ ​ ​ ​ ​ ​ ​ ​ The Fourier transform section individually outputs the noise source spectrum obtained by Fourier transforming the transient waveform data at the on time and the noise source spectrum obtained by Fourier transforming the transient waveform data at the off time.

8. The noise analysis apparatus according to any one of claims 2 to 6, wherein The noise source spectrum and the plurality of occurrence times are individually set in correspondence with the on time and the off time, respectively.

9. The noise analysis apparatus according to any one of claims 2 to 8, further comprising: a 5th acquisition section that acquires information on a load current flowing through the semiconductor element during the noise analysis target period; and a noise source spectrum correction section that corrects the magnitude of the noise source spectrum in correspondence with the load current at each of the plurality of occurrence times.

10. The noise analysis apparatus according to any one of claims 1 to 9, further comprising: a 4th acquisition section that acquires information on a resolution bandwidth; a window function setting section that sets a window function in a shape that reflects the resolution bandwidth; a weighting coefficient setting section that outputs, as a plurality of weighting coefficients corresponding to the plurality of occurrence times, respectively, values of the window function at the plurality of occurrence times acquired by the 1st acquisition section; and a 2nd multiplication section that calculates a plurality of multiplication values obtained by multiplying the plurality of noise spectra by the plurality of weighting coefficients, respectively, the 1st addition section calculates the calculation result of the aggregate spectrum by adding the plurality of multiplication values obtained by the 2nd multiplication section.

11. The noise analysis apparatus according to any one of claims 1 to 10, wherein the aggregate spectrum relating to the noise generated by the opening and closing of the plurality of semiconductor elements is calculated, further comprising a 2nd addition section that calculates the calculation result of the aggregate spectrum of the plurality of semiconductor elements by adding the calculation results of the aggregate spectrum with respect to each of the semiconductor elements.

12. The noise analysis apparatus according to any one of claims 1 to 11, wherein a plurality of the noise analysis target periods are set, and a statistical calculation result of a plurality of the aggregate spectra corresponding to the plurality of set noise analysis target periods, respectively, is output.

13. A noise analysis method of calculating an aggregate spectrum of noise generated by opening and closing of at least one of an on time and an off time of a semiconductor element, wherein information indicating a plurality of occurrence times at which a plurality of opening and closing of the semiconductor element occur, respectively, during a noise analysis target period including the plurality of opening and closing of the semiconductor element is acquired, a plurality of phase difference information corresponding to the plurality of occurrence times, respectively, is generated, the plurality of phase difference information being used to perform phase transformation of a noise source spectrum of the semiconductor element at the opening and closing, which reflects a time difference of the plurality of opening and closing, the aggregate spectrum obtained by adding a plurality of noise spectra, which are obtained by phase transforming the noise source spectrum of the semiconductor element at the opening and closing, respectively, using the plurality of phase difference information, is calculated.

14. The noise analysis method according to claim 13, wherein The noise source spectrum of the semiconductor element at the opening and closing is obtained by Fourier transform of transient waveform data of the semiconductor element at the opening and closing.

15. The noise analysis method according to claim 13, wherein In the noise analysis method, a noise transfer function of a propagation path between the semiconductor element and a noise observation point is acquired, a multiplication value of a noise source spectrum obtained by Fourier transform of transient waveform data of the semiconductor element at the opening and closing and the noise transfer function is calculated, the noise source spectrum of the semiconductor element at the opening and closing is the multiplication value.

16. The noise analysis method according to claim 15, wherein In the noise analysis method, a plurality of transfer functions of respective propagation paths for a plurality of propagation paths divided between the semiconductor element and the noise observation point are acquired, the multiplication value is calculated using the noise transfer function obtained by integrating the plurality of acquired transfer functions.

17. The noise analysis method according to claim 15 or 16, wherein In the noise analysis method, the noise source spectrum subjected to Fourier transform is acquired, the multiplication value is calculated using the acquired noise source spectrum.

18. The noise analysis method according to claim 15 or 16, wherein In the noise analysis method, the transient waveform data is acquired and the noise source spectrum is generated by Fourier transform of the acquired transient waveform data, the multiplication value is calculated using the generated noise source spectrum.

19. The noise analysis method according to any one of claims 14 to 18, wherein the noise source spectrum and the plurality of occurrence times are individually set in correspondence with the turn-on and the turn-off, respectively.

20. The noise analysis method according to any one of claims 14 to 19, wherein In the noise analysis method, information of a load current flowing through the semiconductor element during the noise analysis target period is acquired, the noise source spectrum is corrected in correspondence with the load current of each of the plurality of occurrence times.

21. The noise analysis method according to any one of claims 13 to 20, wherein In the noise analysis method, information about a resolution bandwidth is acquired, a window function is set in a shape reflecting the resolution bandwidth, a value of the window function of each of the plurality of occurrence times acquired is set as a plurality of weighting coefficients corresponding to the plurality of occurrence times, respectively, a plurality of multiplication values obtained by multiplying the plurality of noise spectra by the plurality of weighting coefficients, respectively, are calculated, the calculation result of the aggregate spectrum is calculated by adding the plurality of multiplication values multiplied by the plurality of weighting coefficients.

22. The noise analysis method according to any one of claims 13 to 21, wherein In calculating the aggregate spectrum related to noise generated by the opening and closing of a plurality of semiconductor elements, the calculation results of the aggregate spectrum of each of the semiconductor elements are aggregated to calculate the calculation result of the aggregate spectrum of the plurality of semiconductor elements.

23. The noise analysis method according to any one of claims 13 to 22, wherein During setting of a plurality of the noise analysis target periods, a statistical calculation result of a plurality of the integrated spectra corresponding to the plurality of set noise analysis target periods, respectively, is calculated.

24. A computer program product for causing a computer to execute the method of any one of claims 13 to 23.

Images