A compact Marx generator based on shock ionization dynistors Authors: Sergey Korotkov, Yury Aristov, Alexander Zhmodikov, Dmitry Korotkov Published: 2026 ABSTRACT SummaryIn this article, we describe a four-stage Marx generator that utilizes shock ionization dynistors as switches. The generator can generate megawatt pulses of nanosecond duration, which is important for many areas of electrophysics. The dynistor in the first stage is turned on by an overvoltage pulse rising at approximately 1 kV/ns. The remaining dynistors are turned on by an overvoltage wave created using triggering capacitors. The main unit of the generator has a volume of 500 cm2 and can switch voltage pulses reaching 12 kV in 8 ns, as well as current pulses with an amplitude of 3 kA and a rise rate of 200 A/ns, into the load. The features of the generator offer the possibility to significantly increase both the output current and the output voltage.Keywords: high-voltage techniques, power semiconductor switches, pulse generation, silicon devices Introduction Generators of high-power nanosecond pulses are used in various plasma-chemical devices,1,2,3,4 in laser5 and radar6 systems, as well as in medicine7 and biology.8 Key generator components are switches. They must have high static blocking voltage, high di/dt tolerance, low switching loss, short turn-on time, small jitter, and a long service life. Semiconductor switches are characterized by small jitter and a long service life. However, only those switches that have short turn-on time and low switching loss are suitable for use in high-power nanosecond pulse generators. Grekhov and Kardo-Sysoev9 demonstrated the possibility of radically reducing the turn-on time of silicon semiconductor devices when they are triggered by an overvoltage pulse rising at a rate of hundreds of kilovolts per microsecond. With such a rise rate of the triggering voltage, a local avalanche breakdown, which is dangerous for the semiconductor device, does not have enough time to develop, and the electric field strength can increase to a value sufficient for impact ionization of silicon. The impact ionization process is initiated by “seed” carriers in regions with a maximum field strength near the collector junction and ensures the rapid filling of the base regions of the semiconductor structure with electron-hole plasma. As a result, the semiconductor device switches to a highly conductive state in approximately 1 ns, creating unique capabilities for switching high-power current pulses of nanosecond duration. High-voltage switches with nanosecond turn-on times, consisting of series-connected diode structures with impact ionization,10,11,12 are well known. They are able to switch pulsed power of tens of kW into a load in fractions of a nanosecond. The disadvantage of these switches is that the maximum charge that can be switched is limited by the amount of the electron-hole plasma stored during the impact ionization process. Significantly higher charge can be switched using four-layer impact ionization devices, in which high conductivity after switching is maintained by injection of charge carriers from the emitters. Generators of nanosecond pulses of megawatt range based on semiconductor four-layer impact ionization devices are known.13,14,15,16,17,18,19 They contain a high-voltage capacitor and a closing switch in the form of a block of series-connected four-layer devices. The block is turned on by applying to it an overvoltage pulse with a nanosecond front edge. A disadvantage of such generators is the high charging voltage of the capacitor. This drawback is eliminated in semiconductor generators developed using the Marx scheme.20,21,22 This scheme utilizes parallel charging of several capacitors to a low voltage. The charging voltage is applied to the closing switches. The output voltage is generated by connecting capacitors in series with the load by turning on the closing keys. The value of the output voltage is determined as the sum of the charging voltages of the capacitors. The closing switches are turned on either simultaneously using separate control units or in relay-race mode by a voltage wave created after switching one of the switches. The disadvantage of the first method is the need to ensure synchronous turn-on of a large number of closing switches. The disadvantage of the second method is the need to generate a triggering voltage wave. In this article, we describe a Marx generator that utilizes four closing switches in the form of impact ionization dynistors (SIDs) that are turned on in relay-race mode.23 A distinctive feature of the SID is that the anode emitter p+ and the cathode emitter n+ contain small sections n+ and p+, respectively (shunts). They are highlighted in yellow in the inset. The shunts are arranged coaxially and are uniformly distributed across the SID structure, spaced less than 1 mm apart. As the triggering voltage increases, the shunts limit injection of electrons and holes from adjacent emitter regions and reduce the intensity of generation of “seed” carriers flowing from the emitters into high-field regions near the collector p-n junction. This increases the time during which the required density of the “seed” carriers is created in the near-collector regions of the SID, and the impact ionization process is initiated when the triggering voltage reaches a higher value. This leads to an increase in the intensity of impact ionization and to an increase in the density of the resulting electron-hole plasma. The delay in the initiation of the impact ionization process relative to the application of the trigger voltage determines the possibility of increasing the intensity of this process by increasing the rise rate of the triggering voltage. The rise rate of the triggering voltage is determined by the amplitude and rise time of the control current, which is passed through the SID structure and ensures the charging of its own capacitance. If this current is sufficiently strong, the SID’s switching capabilities are high. For example, in a generator that we described earlier,24 the SIDs with a structure diameter of 20 mm were capable of switching current pulses with an amplitude of 8.5 kA and at a duration of 200 ns, rising at a rate of 150 A/ns, which was largely limited by the mounting inductance of the generator’s power circuit. The objective of this study is to develop a compact high-power nanosecond pulse generator based on the Marx circuit and containing four low-voltage stages based on SIDs with an operating voltage of 2.3 kV. The research objectives include developing SID control circuits capable of generating control current pulses with amplitude greater than 200 A and with a rise time of less than 3 ns, which are necessary to achieve the required triggering voltage rise rate. The study also includes developing a low-inductance generator design that ensures a high output current rise rate and the ability to promptly increase the number of stages and the amount of energy stored within them. Results Description of the generator The electrical circuit diagram of the developed Marx generator is shown in Figure 1. High-voltage pulses are switched to the load Rload by the four-stage main unit (MU), which contains storage capacitors C01–C04 and shock ionization dynistors SID1–SID4. The dynistor SID1 is triggered by the control unit (CU), which contains storage capacitor C1. Triggering capacitors CC1–CC3 ensure switching of the dynistors SID2, SID3, and SID4 in relay-race mode. In the initial state, capacitors C01–C04 and CC1–CC3 are charged from power source (PS) to voltage U0 through resistors R0 (500 Ω each), and capacitor C1 is charged to voltage U. Diode D1 (K25F) and resistor R1 (5 kΩ) prevent the possibility of applying voltage U0 to the elements of CU. Generator diagram First, SID1 is triggered using the CU unit. The output current of the CU unit (ICU) is branched into the own capacitance of dynistor SID1 (CSID1) and into the circuit of dynistor SID2, which contains capacitor C01, the own capacitance of dynistor SID2 (CSID2), and the control capacitor CC1. The current IC1 flowing through CSID1 is the control current of dynistor SID1. It ensures rapid charging of CSID1 to a voltage that initiates the impact ionization process. Since the circuit of dynistor SID2 contains a capacitor CC1 with a small capacitance, the electrical resistance of this circuit is relatively high, and current I = ICU − IC1 that flows through CSID2 is significantly less than the IC1 current. After SID1 is turned on, the discharge current of capacitor CC1 (IC2) quickly charges CSID2 to a voltage that initiates the impact ionization process in the SID2 structure. After SID2 is turned on, the control current of IC3, whose magnitude is determined by the discharge current of capacitor CC2, flows through the own capacitance of dynistor SID3 (CSID3). As a result, the voltage across SID3 quickly rises to the turn-on level. Turning on SID3 ensures that control current IC4, which is generated during the discharge of capacitor CC3, flows through the own capacitance of dynistor SID4 (CSID4). After SID4 is turned on, the discharge current of series-connected capacitors C01–C04 is switched to the load Rload. The discussed relay-race switching mode of the dynistors SID1, SID2, SID3, and SID4 is illustrated in Figures 2A–2D, respectively. Figure 2E illustrates the process of switching the output current Iload to the load Rload. Diagrams illustrating the processes of currents flow through main unit (A) The process of control current IC1 flow through dynistor SID1. (B) The process of control current IC2 flow through dynistor SID2. (C) The process of control current IC3 flow through dynistor SID3. (D) The process of control current IC4 flow through dynistor SID4. (E) The process of main current switching to the load Rload. The unit CU contains an opening switch in the form of a block of series-connected drift step recovery diodes (DSRDs),25 which have a nanosecond turn-off time. The conditions for the efficient operation of the DSRD block are ensured by the control circuit CC. It is developed based on the circuit design proposed in 201226 and differs significantly from the well-known control circuit based on a saturable core transformer described in various articles.18,19 The absence of a transformer improves the operational stability of the CC circuit and ensures its small dimensions (50 × 30 × 30 mm). The operation of unit CU is illustrated by diagrams “a” and “b” in Figure 3. Initially, capacitor C1 is charged to voltage U. After thyristor T (CCSAB53N30A10) is turned on, the capacitor begins to discharge. As a result, short current pulses IL1 and I(+), shown in diagram “a,” flow through both inductor L1 and the DSRD block. These currents cause capacitor C1 to recharge to a significant reverse voltage, capacitor C2 to charge, and electrons and holes to accumulate in the base regions of the DSRD block’s diode structures. Capacitors C2 and C1 then discharge through inductor L2, as shown in diagram “b.” As a result, a rapidly rising reverse current I(−) flows through the diode structures of the DSRD block, removing the accumulated electron-hole plasma. After all accumulated charge carriers are removed, the DSRD block turns off, and the current flowing through inductor L2 quickly charges the own capacitance of this block. When the voltage across the DSRD block becomes greater than the voltage U0 initially applied to the dynistor SID1, diode D1 turns on to a conducting state, and the output current of the CU unit, determined by the current flowing through inductor L2, flows through it. Diagrams illustrating the process of generating the output current of the CU unit (A) Formation of forward current through the DSRD block. (B) Formation of reverse current through the DSRD block. In the developed Marx generator, semiconductor structures of SID and DSRD with a diameter of 16 mm were used. The key parameters of the SID structures: the thickness is 400 μm, the capacitance at a voltage of 2 kV is 200 pF, the permissible constant voltage in the off state is 2.3 kV, the turn-on voltage is 4 kV, and the turn-on time is 1 ns. The key parameters of the DSRD structures: the thickness is 300 μm, the capacitance at a voltage of 1 kV is 300 pF, the permissible pulse voltage is 1.2 kV, the permissible pulse reverse current density is 250 A/cm2, and the current interruption time is 2 ns. The SID structures were placed in fluoroplastic cases with a height of 7 mm and an outer diameter of 20 mm. The DSRD block consisted of 12 series-connected diode structures, soldered together and placed in a fluoroplastic case with a height of 10 mm and an outer diameter of 20 mm. Effective operation of the DSRD and SID that were used in this study was achieved under the following conditions: •Forward current pulse duration through DSRD: no more than 350 ns; Forward current pulse duration through DSRD: no more than 350 ns; •Reverse current pulse duration through DSRD: no more than 250 ns; Reverse current pulse duration through DSRD: no more than 250 ns; •Rise rate of the triggering voltage on SID: no less than 800 V/ns; Rise rate of the triggering voltage on SID: no less than 800 V/ns; •Control current amplitude at the moment of SID switching: no less than 200 A. Control current amplitude at the moment of SID switching: no less than 200 A. Results of computer simulation Computer simulation was carried out using a model of SID24 and a model of DSRD27 developed previously. As a result, the values of the generator elements were determined, at which the conditions for efficient operation of the DSRD and SID are met at low energy costs and a high amplitude of the output current is ensured: C1 = C2 = 44 nF; L1 = 600 nH; L2 = 350 nH; C01 = C02 = C03 = C04 = 120 nF; CC1 = CC2 = CC3 = 94 pF. The values of voltages U0 and U were set equal to 2.3 kV. Figure 4 shows the calculated curves of current IL2 through inductor L2 and voltage Uload on load Rload = 50 Ω. Figure 5 shows the curve of voltage UDSRD on the DSRD block and the curves of voltages USID1–USID4 and control currents IС1–IС4, obtained by computer simulation of the switching processes of dynistors SID1–SID4. Calculated curves of current through inductor L2 and voltage on the load 50 Ω IL2 is the current through inductor L2, and Uload is voltage on the load. Calculated curves of the voltage on the DSRD block, the voltages on dynistors SID1, SID2, SID3, and SID4, and the control currents of the dynistors UDSRD is the voltage on the DSRD block, USID1, USID2, USID3, and USID4 are the voltages on the dynistors SID1, SID2, SID3, and SID4, respectively, and IC1, IC2, IC3, and IC4 are the control currents of the dynistors SID1, SID2, SID3, and SID4, respectively. Results of experimental studies Using the results of computer modeling carried out according to the diagram in Figure 1, an experimental SID generator was manufactured, and its experimental studies were conducted. The resulting oscillograms closely matched the calculated curves. Figure 6 shows oscillograms of current IL2 flowing through inductor L2 in control circuit CC and of voltage UDSRD on the DSRD block, obtained during switching of dynistor SID1. The inset in Figure 6 shows the oscillogram of UDSRD at a different scale. Figure 7 shows oscillograms of voltages USID1, USID2, USID3, and USID4 on dynistors SID1, SID2, SID3, and SID4, respectively, illustrating their switching process. With the generator operating at a frequency of 1 Hz, the jitter of the output voltage pulse generation moments relative to the turn-on moment of the thyristor T in the CU was determined. The jitter value did not exceed 2 ns. Oscillogram of current through inductor L2 and of voltage on the DSRD block The inset shows the оscillogram of the voltage on the DSRD block on a smaller scale. IL2 is the current through inductor L2, and UDSRD is the voltage on the DSRD block. Oscillograms of voltages on the dynistors The oscillograms in Figure 6 show that the durations of the forward and reverse current pulses through the DSRD unit are 300 ns and 200 ns, respectively, which meet the requirements for efficient DSRD operation. The output current of the CC control circuit, determined by the current flowing through inductor L2 when the DSRD unit is turned off, is 270 A and significantly exceeds the minimum control current required for efficient SID operation. The amplitude of voltage UDSRD is approximately 10 kV and is determined by the sum of the voltages on diode D1, on SID1, and on the inductance of the mounting wires. Figure 7 shows oscillograms of voltages on dynistors SID1, SID2, SID3, and SID4 (USID1, USID2, USID3, and USID4, respectively). The oscillograms show that during the switching process of the dynistors, the rise rate of the triggering voltage is at least 900 V/ns, which exceeds the minimum permissible value for effective operation of SIDs. The dynistors turn on when the triggering voltage rises to approximately 4 kV. Figures 8 and 9 show oscillograms of the generator output voltage Uload, obtained at voltages U0 and U equal to 2.3 kV. Low-inductance resistors were used as the load Rload. Oscillogram of voltage on a load with a resistance of 50 Ω Oscillogram of voltage on a load with a resistance of 1 Ω The oscillogram in Figure 8 corresponds to Rload = 50 Ω. The oscillogram shows that the amplitude and rise rate of the output voltage pulse are 12 kV and 1.5 kV/ns, respectively. The amplitude of Uload exceeds the total charging voltage of the MU unit’s storage capacitors and is determined by the total voltage of the triggering capacitors CC1–CC3, which recharge to a significant voltage during the switching of the dynistors. When the generator was operating at a frequency of 10 Hz, the jitter of the output voltage pulses did not exceed 2 ns. Figure 9 shows oscillogram of generator’s output voltage obtained at a load Rload = 1 Ω. The generator output current amplitude and the energy transferred to the load, as determined from the oscillogram, are 3 kA and 0.67 J, respectively. The maximum rise rate of the output current is approximately 200 A/ns. The maximum power transferred to the load is 9 MW. The amount of energy initially stored in the capacitors in the units MU and CU, which were charged up to a voltage of 2.3 kV, is approximately 1.4 J. It is approximately twice the amount of energy transferred to the load. The experimental studies were conducted using a digital oscilloscope with a bandwidth of 500 MHz, powered by a standard battery. The current waveform was measured using a Rogowski coil. А voltage probe with a well-known design was used in the experiments.23 It allowed recording high-voltage signals with a minimum rise time of 0.4 ns. The high-rise rate of the output current was achieved through the development of a low-inductance design of the MU, which was achieved through the small block dimensions (80 × 95 × 65 mm) and the bifilar arrangement of the mounting buses. The MU unit design is illustrated in Figures 10 and 11. Design of unit MU Photo of unit MU The components of unit MU are clamped in a frame consisting of textolite side plates 1 and 2 and aluminum flanges 3 and 4. The clamping force is created by a screw fixed in flange 3. Output buses 5 are located on the side plates and connected to grounded flange 4. Triggering capacitors 6 (47 pF each) are connected to buses 5 and to dynistors 7 located inside the frame (the dynistors are not shown in Figure 10). Storage capacitors 8 (15 nF each) are connected to bifilar buses 9, insulated from each other by thin textolite plates 10. The load is connected to terminals “a” and “b.” The developed design makes it possible to expand the number of stages and the number of storage capacitors 8 in each stage. To connect additional capacitors, it is necessary to increase the length of bifilar buses 9. Additional stages can be connected by increasing the length of side plates 1 and 2. Discussion The developed experimental generator based on SID is able to switch pulse power of more than 5 MW into a load in a few nanoseconds. The main advantages of the generator are its small dimensions (500 cm2), the low jitter of the output pulses (less than 2 ns) and a unique rise rate of the output current for semiconductor generators (200 kA/μs). The operating principles and design of the generator make it possible to significantly increase its output parameters. The amplitude of the output current and the amount of the output energy can be increased by connecting additional storage capacitors to bifilar buses in the MU of the generator. Increased switching capabilities of SIDs can be achieved by increasing the area of their semiconductor structures. The output voltage of the generator can be increased by increasing the number of stages. A radical increase in output voltage can be achieved using high-voltage closing switches in the form of blocks of series-connected SIDs. Low jitter of the output pulses allows several similar generators to operate synchronously. A good agreement between the results of computer simulation and experimental data makes it possible to develop SID generators of nanosecond pulses with an output voltage of tens of kV and an output power of hundreds of megawatts. Applications for such generators include plasma chemistry, electric discharge purification of wastewater and atmospheric air, powering pulsed lasers, biological research, electrohydraulic rock fracture, pulsed mass acceleration, material surface modification, and the generation of strong magnetic fields and powerful shock waves in large-scale experiments. Limitations of the study The operating principles and design of the described experimental generator make it possible to develop multistage SID nanosecond pulse generators with output voltages of tens of kilovolts and output powers of hundreds of megawatts. The main limitation in developing such generators is the need for high-voltage insulation of their components as well as their relatively high cost. Resource availability Lead contact Requests for further information and resources should be directed to and will be fulfilled by the lead contact, Sergey Korotkov (korotkov@mail.ioffe.ru). Materials availability No new materials were produced in this study; the materials used are commercially available. Data and code availability •All data reported in this paper will be shared by the lead contact upon request. All data reported in this paper will be shared by the lead contact upon request. •No new code was generated in this study. No new code was generated in this study. •Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request. Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request. Acknowledgments The authors would like to thank Dr. Igor Grekhov for a helpful discussion of the objectives of this study. Author contributions S.K., generation of the main ideas of the research and writing the manuscript; Y.A., computer simulation; A.Z., experimental research; D.K., design solutions. Declaration of interests The authors declare no competing interests. Declaration of generative AI and AI-assisted technologies in the writing process No AI was used in this study. STAR★Methods Key resources table REAGENT or RESOURCESOURCEIDENTIFIERSoftware and algorithmsOrCADCadence Design Systemswww.cadence.comMicrosoft ExcelMicrosoft Corporationwww.microsoft.comOtherSIDPJSC Electrovypryamitel, Saransk, RussiaN/ADSRDPJSC Electrovypryamitel, Saransk, RussiaN/ATektronix TDS3052C oscilloscopeTektronix, Inc.www.tek.comTektronix TDS3BATB batteryTektronix, Inc.www.tek.comCurrent monitor (model 410)Pearson Electronics, Inc.www.pearsonelectronics.comLeCroy 9210 pulse generatorLeasametricwww.leasametric.com Method details 1.Nanosecond voltage pulses were measured using a Tektronix TDS3052C oscilloscope, a current monitor (model 410, Pearson Electronics, Inc.), and a laboratory-built wideband divider that could reproduce the front edge of high-voltage pulses with a minimum duration of 0.4 ns. The lower arm of the divider was a coaxial cable with a characteristic impedance of 50 Ω. The upper arm was a TVO-0.5 resistor (470 Ω, 0.5 W) compensated with a copper screen. When measuring voltage oscillograms, the divider was connected to a dynistor via a decoupling capacitor (500 pF). The divider was calibrated using a LeCroy 9210 pulse generator by measuring voltage pulses with a front edge duration of 300 ps. Nanosecond voltage pulses were measured using a Tektronix TDS3052C oscilloscope, a current monitor (model 410, Pearson Electronics, Inc.), and a laboratory-built wideband divider that could reproduce the front edge of high-voltage pulses with a minimum duration of 0.4 ns. The lower arm of the divider was a coaxial cable with a characteristic impedance of 50 Ω. The upper arm was a TVO-0.5 resistor (470 Ω, 0.5 W) compensated with a copper screen. When measuring voltage oscillograms, the divider was connected to a dynistor via a decoupling capacitor (500 pF). The divider was calibrated using a LeCroy 9210 pulse generator by measuring voltage pulses with a front edge duration of 300 ps. 2.Simulation of the DSRD block turn-off processes and the turn-on processes of the SID structures was performed using OrCAD software using previously developed simple DSRD and SID models, which provided sufficient agreement with the experimental results. The DSRD block model contained the block’s inductance and resistor R1, simulating energy losses. The block utilized OrCAD diode models, the operating area of which was equal to the operating area of the DSRD structures. The coefficients in the formulae describing the capacitance-voltage characteristics of these diodes were selected to ensure agreement with the experimentally determined capacitance-voltage characteristics of the DSRD structures. The equality of the turn-off times of the OrCAD diode models to the turn-off times of the DSRD structures was achieved by selecting the coefficients responsible for the carrier accumulation process. The SID structure model contained the inductance of the SID case. Before its turn-on, the SID structure was represented by the OrCAD diode structure model (DSID), which had the same operating area as the SID structure and the same dependence of its own capacitance on the applied voltage. The cathode of the DSID structure corresponds to the anode of the SID structure. Ideal switch K1 is connected in series with the DSID. Two additional circuits are connected in parallel to the DSID-K1 circuit, which characterize the SID structure after switching on. The first additional circuit consists of resistor R2 and ideal switch K2, and the second one consists of resistor R3 and ideal switch K3. In the initial state and during the application of the triggering voltage pulse, switch K1 is closed, and switches K2 and K3 are open. When the voltage across the DSID reaches the known turn-on voltage of the SID structure, switch K2 is turned on and switch K1 is turned off. As a result, the SID structure is modeled by resistor R2. The resistance value of R2 changes exponentially and is determined so that during the known turn-on time of the SID structure, the voltage across DSID decreases to a value equal to the voltage across the SID structure in the first moments after its turn-on. Switch K3 is turned on with a delay relative to the moment of switching on of K2 equal to the SID structure turn-on time. When K3 is turned on, K2 is turned off. As a result, the SID structure is modeled by resistor R3, the resistance value of which gradually decreases. Simulation of the DSRD block turn-off processes and the turn-on processes of the SID structures was performed using OrCAD software using previously developed simple DSRD and SID models, which provided sufficient agreement with the experimental results. The DSRD block model contained the block’s inductance and resistor R1, simulating energy losses. The block utilized OrCAD diode models, the operating area of which was equal to the operating area of the DSRD structures. The coefficients in the formulae describing the capacitance-voltage characteristics of these diodes were selected to ensure agreement with the experimentally determined capacitance-voltage characteristics of the DSRD structures. The equality of the turn-off times of the OrCAD diode models to the turn-off times of the DSRD structures was achieved by selecting the coefficients responsible for the carrier accumulation process. The SID structure model contained the inductance of the SID case. Before its turn-on, the SID structure was represented by the OrCAD diode structure model (DSID), which had the same operating area as the SID structure and the same dependence of its own capacitance on the applied voltage. The cathode of the DSID structure corresponds to the anode of the SID structure. Ideal switch K1 is connected in series with the DSID. Two additional circuits are connected in parallel to the DSID-K1 circuit, which characterize the SID structure after switching on. The first additional circuit consists of resistor R2 and ideal switch K2, and the second one consists of resistor R3 and ideal switch K3. In the initial state and during the application of the triggering voltage pulse, switch K1 is closed, and switches K2 and K3 are open. When the voltage across the DSID reaches the known turn-on voltage of the SID structure, switch K2 is turned on and switch K1 is turned off. As a result, the SID structure is modeled by resistor R2. The resistance value of R2 changes exponentially and is determined so that during the known turn-on time of the SID structure, the voltage across DSID decreases to a value equal to the voltage across the SID structure in the first moments after its turn-on. Switch K3 is turned on with a delay relative to the moment of switching on of K2 equal to the SID structure turn-on time. When K3 is turned on, K2 is turned off. As a result, the SID structure is modeled by resistor R3, the resistance value of which gradually decreases. Quantification and statistical analysis There are no quantification or statistical analyses to include in this study. Contributor Information Sergey Korotkov, Email: korotkov@mail.ioffe.ru. Alexander Zhmodikov, Email: alexander.zhmodikov@mail.ioffe.ru. References Associated Data Data Availability Statement •All data reported in this paper will be shared by the lead contact upon request. All data reported in this paper will be shared by the lead contact upon request. •No new code was generated in this study. No new code was generated in this study. •Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request. Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.